Water garbage crushing treatment method and system

Abstract

The invention discloses an overwater garbage crushing treatment method and system, and the existing overwater garbage treatment device generally comprises a garbage collection device and a garbage treatment device, and the garbage treatment device is composed of a garbage tearing device and a garbage fragment collection box; the garbage collecting device is composed of a propeller water pump and a filter screen frame. The propeller water pump is installed at the water inlet, after the device is immersed in water, the water pump is started, the spiral blades start to rotate, and in the rotating process, the blades push water to flow to form a water flow. According to the scheme, the to-be-crushed garbage is recognized, the machine learning model is adopted to recognize information such as the type and the size of the to-be-crushed garbage, different cutters are selected and matched according to the recognition result, and the mode that all garbage uses the same cutter and the cutter is abraded is avoided; and the water garbage crushing treatment efficiency of an existing water garbage treatment device is further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of garbage treatment, and particularly relates to a method and system for crushing and treating waterborne garbage. Background Art

[0002] With the continuous development of marine resources, the utilization of marine resources has become an important part of social development. However, with the rapid rise of the economy in coastal areas and the increasing frequency of marine resource development activities, it has brought greater environmental pollution pressure to the offshore waters. Among them, due to the accelerated promotion of industrialization and urbanization, a large amount of industrial wastewater, domestic sewage, and various waste materials are directly or indirectly discharged into the offshore waters, resulting in water quality deterioration and damage to the ecosystem. In addition, domestic garbage such as plastic bags and plastic bottles generated in people's daily lives will enter the offshore waters through ways such as wind, rainfall, and surface runoff.

[0003] These garbage and waste materials entering the offshore waters have caused a large amount of waterborne garbage stock at present, affecting the balance of the marine ecosystem. Therefore, it is very necessary to crush and treat waterborne garbage. At present, the common method for crushing and treating waterborne garbage is to collect the garbage and then send it into a garbage shredder and use knives for crushing. However, due to the different types and volumes of garbage, if the knives used for crushing are not properly matched, it will lead to incomplete crushing of the garbage and wear of the knives, which will affect the efficiency of subsequent garbage landfill or incineration treatment. Therefore, there is an urgent need for a method and system for crushing and treating waterborne garbage to solve the defects of the existing technology. Summary of the Invention

[0004] The present invention aims to provide a method and system for crushing and treating waterborne garbage to solve the above technical problems. By identifying the types and volumes of waterborne garbage and matching them with knives, the efficiency of crushing and treating waterborne garbage is improved.

[0005] To solve the above technical problems, an embodiment of the present invention provides a method for crushing and treating waterborne garbage, including: Obtaining initial images of a plurality of waterborne garbage, performing image enhancement on the initial images according to a preset image enhancement method to obtain first images, and determining the type information of the waterborne garbage according to the first images and a preset first machine learning model; Extracting an edge contour feature set of the initial images according to a preset edge detection algorithm, and inputting the edge contour feature set and the type information into a preset second machine learning model to determine the volume information of the waterborne garbage; Performing tagging processing on the waterborne garbage according to the type information and volume information of the waterborne garbage, and screening out a plurality of first waterborne garbage from the waterborne garbage; Obtain the specification data of several tools, and construct a node graph between the tools and the first waterborne garbage according to the specification data of the first waterborne garbage and the tools; According to the preset tool crushing link generation algorithm, perform node linking on the node graph, construct a tool crushing link scheme, and perform crushing treatment on the waterborne garbage based on the tool crushing link scheme.

[0006] It can be understood that, compared with the prior art, the present invention enhances the initial image through an image enhancement method, which can improve the image quality and feature clarity, enabling the first machine learning model to more quickly and accurately identify the type information of waterborne garbage; through the combination of an edge detection algorithm, type information and a second machine learning model, the volume information of waterborne garbage can be efficiently determined, improving the subsequent tool adaptation efficiency by efficiently and accurately obtaining type information and volume information; by performing tagging processing on waterborne garbage, the complexity and computational amount of subsequent tool adaptation can be reduced, thereby improving the efficiency of waterborne garbage crushing treatment; by constructing a node graph based on the characteristics of the first waterborne garbage and the specification data of the tools, and performing node linking through a preset tool crushing link generation algorithm, an optimal tool crushing link scheme can be quickly and accurately generated, enabling the tools to adapt to waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by inappropriate tools used for crushing, and thus improving the efficiency of waterborne garbage crushing treatment.

[0007] As a preferred solution, the obtaining of the initial images of several waterborne garbage, the image enhancement of the initial images according to a preset image enhancement method to obtain a first image, and the determination of the type information of the waterborne garbage according to the first image and a preset first machine learning model specifically include: Obtain a sliding window and a sliding amplitude value; According to the sliding amplitude value, slide the sliding window on the initial image, obtain the median value of the pixels in the sliding window after each slide, and modify the pixel value corresponding to the center of the sliding window to the median value until the sliding window completes the sliding of the initial image to obtain a second image; Obtain the initial color channel values of the second image, and perform normalization processing on the color channel values to obtain first color channel values; Obtain the maximum color channel value in the first color channel values, and determine the brightness value and saturation value of the second image according to the maximum color channel value; Judge the corresponding channel color of the maximum color channel value, and determine the hue of the second image based on the channel color; According to the brightness value, saturation value and hue, perform color space conversion on the second image to obtain a first image; Train the preset first machine learning model according to the preset machine learning model training method to obtain a garbage type prediction model, and input the first image into the garbage type prediction model to determine the type information of the waterborne garbage.

[0008] In this preferred solution, by replacing the pixel values of the sliding window, the noise and outliers in the initial image can be filtered out, improving the feature expressiveness and image stability of the second image. Then, through color space conversion based on lightness value, saturation value, and hue, the features of the first image can be better highlighted, thereby improving the prediction accuracy of the subsequent garbage type prediction model for the type information of waterborne garbage, ensuring the accuracy of the subsequent adaptation of the tool to waterborne garbage, and avoiding incomplete garbage crushing and tool wear caused by the mismatch of the used tool, thus improving the efficiency of waterborne garbage crushing treatment.

[0009] As a preferred solution, the step of training the preset first machine learning model according to the preset machine learning model training method to obtain a garbage type prediction model, and inputting the first image into the garbage type prediction model to determine the type information of the waterborne garbage specifically includes: Query the preset waterborne garbage type database to obtain historical waterborne garbage type information and historical waterborne garbage image information; Perform one-hot encoding on the historical waterborne garbage type information to generate historical waterborne garbage type coding values; Match the historical waterborne garbage type coding values with the historical waterborne garbage image information to generate a historical waterborne garbage training set; The preset first machine learning model is a support vector machine model. Input the historical waterborne garbage training set into the support vector machine model, and adjust the regularization parameter, kernel function width, and polynomial order of the support vector machine model to update the support vector machine model, and obtain the accuracy, precision, and recall rate of the updated support vector machine model; Determine that the accuracy, precision, and recall rate of the updated support vector machine model meet the preset requirements, complete the training of the support vector machine model to obtain a garbage type prediction model; Input the first image into the garbage type prediction model to obtain the coding value output by the garbage type prediction model, and determine the type information of the waterborne garbage according to the coding value.

[0010] In this preferred solution, by performing one-hot encoding on the historical waterborne garbage type information to generate the historical waterborne garbage type encoding value, the type information of the waterborne garbage can be converted into a numerical type, thereby improving the accuracy of the subsequent garbage type prediction model. By training the support vector machine model, the support vector machine model can be optimized to enable it to quickly and accurately predict the type of waterborne garbage, ensuring the accuracy of the subsequent adaptation of the tool to the waterborne garbage, and avoiding problems such as incomplete garbage crushing and tool wear caused by the mismatch of the crushing tool, thus improving the efficiency of waterborne garbage crushing treatment.

[0011] As a preferred solution, extracting the edge contour feature set of the initial image according to the preset edge detection algorithm, and inputting the edge contour feature set and the type information into a preset second machine learning model to determine the volume information of the waterborne garbage specifically includes: Converting the color space of the initial image into a grayscale space, and filtering the initial image after color space conversion through a preset Gaussian filter to obtain a third image; Obtaining the horizontal direction gradient calculation matrix, the vertical direction gradient calculation matrix, and the brightness value of each pixel point in the third image; According to the horizontal direction gradient calculation matrix and the vertical direction gradient calculation matrix, combining the preset edge detection algorithm to determine the first pixel point set of the third image, and determining the edge contour feature set of the initial image based on the first pixel point set; Inputting the edge contour feature set and the type information into a preset second machine learning model to determine the volume information of the waterborne garbage.

[0012] In this preferred solution, by converting the color space of the initial image into a grayscale space and filtering, the noise in the initial image can be removed, reducing the complexity of the image data, thereby improving the accuracy of edge detection; then, through the brightness value of the pixel points and the gradient calculation matrix, the change of the pixel values can be accurately and quickly captured, thus accurately reflecting the edge and texture information of the third image, ensuring the accuracy of the edge contour feature set, and further improving the prediction accuracy of the volume information of the waterborne garbage based on the edge contour feature set, the type information, and the preset second machine learning model, ensuring the accuracy of the subsequent adaptation of the tool to the waterborne garbage, and avoiding problems such as incomplete garbage crushing and tool wear caused by the mismatch of the crushing tool, thus improving the efficiency of waterborne garbage crushing treatment.

[0013] As a preferred solution, according to the horizontal direction gradient calculation matrix and the vertical direction gradient calculation matrix, combining the preset edge detection algorithm to determine the first pixel point set of the third image, and determining the edge contour feature set of the initial image based on the first pixel point set specifically includes: Calculate the horizontal gradient of each pixel point according to the horizontal direction gradient calculation matrix and the brightness value of each pixel point; calculate the vertical gradient of each pixel point according to the vertical direction gradient calculation matrix and the brightness value of each pixel point; determine the gradient amplitude and gradient direction of each pixel point according to the horizontal gradient and vertical gradient of each pixel point; Obtain the gradient discrete direction, match the gradient direction of each pixel point with the gradient discrete direction, and obtain the discrete gradient direction of each pixel point; Compare the gradient amplitude of each pixel point with that of adjacent pixel points in the discrete gradient direction. If the gradient amplitude of the pixel point is less than that of the adjacent pixel point, modify the gradient amplitude of the pixel point to a preset gradient amplitude; Obtain the gradient amplitude threshold, screen out the pixel points whose gradient amplitude is greater than or equal to the gradient amplitude threshold, and obtain the first pixel point set; Construct the edge image of the waterborne garbage according to the first pixel point set, extract the edge points of the edge image according to the preset contour detection algorithm, and determine the edge contour feature set of the initial image.

[0014] This preferred solution can improve the accuracy of edge detection through the precise calculation of gradient amplitude and direction. After that, by comparing the magnitudes of gradient amplitudes in the discrete gradient direction, non-edge pixel points can be suppressed, thereby improving the robustness of the edge image. Furthermore, it improves the accuracy of extracting the edge contour feature set from the edge image subsequently, improves the accuracy of predicting the volume information of waterborne garbage subsequently, ensures the accuracy of the subsequent adaptation of the tool to the waterborne garbage, and avoids incomplete garbage crushing and tool wear caused by the mismatch of the crushing tool. Thus, the efficiency of waterborne garbage crushing treatment is improved.

[0015] As a preferred solution, the waterborne garbage is labeled according to the type information and volume information of the waterborne garbage, and several first waterborne garbage are screened from the waterborne garbage, which specifically includes: Obtain the type coding value of the waterborne garbage according to the type information of the waterborne garbage; Generate the pairing vector of the waterborne garbage according to the type coding value and volume information of the waterborne garbage; Perform clustering processing on the pairing vectors of the waterborne garbage according to the preset clustering algorithm to obtain several clusters and the clustering center of each cluster; Assign the same label to all the waterborne garbage within each cluster; Screen the waterborne garbage according to the pairing vector corresponding to the clustering center of each cluster to obtain several first waterborne garbage.

[0016] In this preferred solution, a pairing vector is generated by integrating the type coding value and volume information of the floating garbage, and then clustering processing is performed, so that floating garbage with relatively similar volumes and types can be classified into one category, and representative first floating garbage is extracted, thereby reducing the calculation amount and complexity in subsequent adaptation of floating garbage and tools, and further improving the efficiency of floating garbage crushing and processing.

[0017] As a preferred solution, obtaining the specification data of a plurality of tools, and constructing a node graph between the tools and the first floating garbage according to the specification data of the first floating garbage and the tools specifically includes: Obtaining the specification data of a plurality of tools, where the specification data includes: the material and size of the tools; generating a tool specification vector of the tools according to the material and size of the tools; Generating a floating garbage node according to the volume information and type coding value of the first floating garbage, and constructing a tool node according to the tool specification vector; Querying a preset tool adaptation database according to the material of the tool node and the type coding value of the floating garbage node, and obtaining a first adaptation value between each tool node and each floating garbage node; Querying a preset tool adaptation database according to the size of the tool node and the volume information of the floating garbage node, and obtaining a second adaptation value between each tool node and each floating garbage node; Constructing a node graph between the tools and the first floating garbage according to the first adaptation value and the second adaptation value.

[0018] In this preferred solution, by querying a preset tool adaptation database, the adaptation relationship between the first floating garbage and the tools can be characterized more accurately, so as to ensure the accuracy of the subsequent tool crushing link scheme, and avoid incomplete garbage crushing and tool wear caused by inappropriate tools used for crushing, thereby improving the efficiency of floating garbage crushing and processing.

[0019] As a preferred solution, performing node linking in the node graph according to a preset tool crushing link generation algorithm to construct a tool crushing link scheme specifically includes: Querying a preset tool adaptation database to obtain the optimal type coding value and the optimal volume adaptation value of each tool node; According to the first adaptation value and the second adaptation value, linking a floating garbage node to each tool node to generate a plurality of initial tool crushing links; where the floating garbage nodes in the initial tool crushing links are defined as first floating garbage nodes, and the floating garbage nodes in the node graph that are not linked to the initial tool crushing links are defined as second floating garbage nodes, and the number of tool nodes in each initial tool crushing link is one; Construct a node adaptation function based on the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value; Calculate the node adaptation degree between each current second waterborne garbage node and each initial tool crushing link according to the node adaptation function; Link the second waterborne garbage nodes to the initial tool crushing links based on the node adaptation degree, update the initial tool crushing links, the first waterborne garbage nodes, and the second waterborne garbage nodes, and then update the node adaptation degree between each second waterborne garbage node and each initial tool crushing link until the number of the second waterborne garbage nodes is zero, complete the update of the initial tool crushing links, and obtain several optimal tool crushing links; Construct a tool crushing link scheme according to the optimal tool crushing links;

[0020] In this preferred solution, by constructing tool crushing links for the tool nodes, the first waterborne garbage nodes, and the second waterborne garbage nodes in the node graph, and calculating the node adaptation degree, the adaptation degree between the waterborne garbage and the tool can be comprehensively considered, thereby improving the adaptability between the tool and the waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by the mismatch of the tools used for crushing, and thus improving the efficiency of waterborne garbage crushing treatment.

[0021] As a preferred solution, the construction of the node adaptation function according to the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value specifically includes: Construct a volume guiding function of the initial tool crushing link according to the tool nodes and the waterborne garbage nodes in the tool crushing link, in combination with the optimal volume adaptation value, where the volume guiding function is: ; Wherein, is the volume guiding value of the rd initial tool crushing link, is the optimal volume adaptation value of the tool node in the th initial tool crushing link, is the number of the first waterborne garbage nodes in the th initial tool crushing link; is the th first waterborne garbage node in the th initial tool crushing link; is the second adaptation value between the th first waterborne garbage node and the tool node in the th initial tool crushing link; Construct the type guiding function of the initial tool crushing link according to the tool nodes and water garbage nodes in the tool crushing link, in combination with the optimal type coding value, where the type guiding function is: ; where is the type guiding value of the -th initial tool crushing link, is the optimal type coding value of the tool nodes in the -th initial tool crushing link, represents the natural logarithm with the mathematical constant e as the base, is the number of the first water garbage nodes in the -th initial tool crushing link, is the -th first water garbage node in the -th initial tool crushing link, is the first adaptation value between the -th first water garbage node and the tool nodes in the -th initial tool crushing link; represents rounding down ; Construct the node adaptation function of the initial tool crushing link according to the volume guiding function and the type guiding function; where the node adaptation function is: ; In the formula, is the node adaptation degree between the -th second water garbage node and the -th initial tool crushing link in the node graph; is the type coding value of the -th second water garbage node; is the type guiding value of the -th initial tool crushing link, is the number of the first water garbage nodes in the -th initial tool crushing link, is the -th first water garbage node in the -th initial tool crushing link, is the type coding value of the -th first water garbage node in the -th initial tool crushing link, is the volume of the -th second water garbage node, is the volume guiding value of the -th initial tool crushing link, is the optimal volume adaptation value of the tool node in the th initial tool crushing link, is the th in the volume of the

[0022] In this preferred solution, a node adaptation function is constructed through the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value, fully exploring the adaptation relationship between the waterborne garbage and the tool in terms of type and volume, enabling the node adaptation function to more accurately represent the adaptation relationship between the waterborne garbage and the tool, thereby improving the adaptability of the tool to the waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by inappropriate tools used for crushing, and thus improving the efficiency of waterborne garbage crushing and treatment.

[0023] Correspondingly, an embodiment of the present invention provides a waterborne garbage crushing and treatment system, including: a type information acquisition module, a volume information acquisition module, a tagging processing module, a node graph construction module, and a tool crushing link acquisition module; Among them, the type information acquisition module is used to acquire initial images of several waterborne garbage, perform image enhancement on the initial images according to a preset image enhancement method to obtain first images, and determine the type information of the waterborne garbage according to the first images and a preset first machine learning model; The volume information acquisition module is used to extract the edge contour feature set of the initial image according to a preset edge detection algorithm, input the edge contour feature set and type information into a preset second machine learning model, and determine the volume information of the waterborne garbage; The tagging processing module is used to perform tagging processing on the waterborne garbage according to the type information and volume information of the waterborne garbage, and screen several first waterborne garbage from the waterborne garbage; The node graph construction module is used to acquire the specification data of several tools, and construct a node graph between the tools and the first waterborne garbage according to the first waterborne garbage and the specification data of the tools; The tool crushing link acquisition module is used to perform node linking on the node graph according to a preset tool crushing link generation algorithm, construct a tool crushing link solution, and perform crushing treatment on the waterborne garbage based on the tool crushing link solution.

[0024] It can be understood that, compared with the prior art, the present system enhances the initial image through an image enhancement method, which can improve the image quality and feature clarity, enabling the first machine learning model to more quickly and accurately identify the type information of the waterborne garbage; through the combination of an edge detection algorithm, the type information and the second machine learning model, the volume information of the waterborne garbage can be efficiently determined, so as to improve the subsequent efficiency of tool adaptation in an efficient and accurate manner for obtaining the type information and the volume information; by performing tagging processing on the waterborne garbage, the complexity and computational amount of subsequent tool adaptation can be reduced, thereby improving the efficiency of waterborne garbage crushing processing; by constructing a node graph based on the characteristics of the first waterborne garbage and the specification data of the tool, and performing node linking through a preset tool crushing link generation algorithm, an optimal tool crushing link scheme can be quickly and accurately generated, so that the tool can be adapted to the waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by inappropriate tools used for crushing, thereby improving the efficiency of waterborne garbage crushing processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 : is a flowchart of the steps of a method for crushing and processing waterborne garbage provided by an embodiment of the present invention; Figure 2 : is a schematic structural diagram of a system for crushing and processing waterborne garbage provided by an embodiment of the present invention; Among them, 201: type information acquisition module; 202: volume information acquisition module; 203: tagging processing module; 204: node graph construction module; 205: tool crushing link acquisition module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Embodiment 1 Please refer to Figure 1 , which is a flowchart of the steps of a method for crushing and processing waterborne garbage provided by an embodiment of the present invention, including steps S101 to S105.

[0028] Step S101: Obtain initial images of a plurality of waterborne garbage, enhance the initial images according to a preset image enhancement method to obtain first images, and determine the type information of the waterborne garbage according to the first images and a preset first machine learning model.

[0029] In this embodiment, obtaining the initial images of several pieces of floating garbage, enhancing the initial images according to a preset image enhancement method to obtain first images, and determining the type information of the floating garbage according to the first images and a preset first machine learning model specifically includes: Obtaining a sliding window and a sliding amplitude value; Sliding the sliding window on the initial image according to the sliding amplitude value, obtaining the median value of the pixels in the sliding window after each sliding, and modifying the pixel value corresponding to the center of the sliding window to the median value of the pixels until the sliding of the sliding window on the initial image is completed to obtain a second image; Obtaining the initial color channel values of the second image and performing normalization processing on the color channel values to obtain first color channel values; Obtaining the maximum color channel value in the first color channel values and determining the brightness value and saturation value of the second image according to the maximum color channel value; Judging the corresponding channel color of the maximum color channel value and determining the hue of the second image based on the channel color; Performing color space conversion on the second image according to the brightness value, saturation value and hue to obtain a first image; Training the preset first machine learning model according to a preset machine learning model training method to obtain a garbage type prediction model, and inputting the first image into the garbage type prediction model to determine the type information of the floating garbage.

[0030] In an alternative embodiment, the sizes of the sliding window and the sliding amplitude value can be determined according to the size of the initial image. In this embodiment, the size of the sliding window is set to , the size of the sliding amplitude value is set to 2, the sliding window is slid on the initial image, the number of grids for each sliding is the sliding amplitude value, all the pixel values in the sliding window are obtained after each sliding, and they are sorted in ascending order, and the median value of the sorted pixel values is used as the median value of the pixels, and the median value of the pixels is used to replace the pixel value corresponding to the center of the sliding window; then sliding is performed in sequence until the sliding window traverses the initial image and the sliding is completed to obtain a second image.

[0031] In an alternative embodiment, the color space of the second image is the RGB color space, which has three primary colors: red, green, and blue. Therefore, each pixel point in the second image has three color channels: red, green, and blue, and each color channel has its corresponding color channel value. In this embodiment, the color channel values corresponding to the color channels of each pixel point are traversed and normalized to obtain the first color channel value; the color channel value with the largest value is selected from the first color channel values as the maximum color channel value, and this maximum color channel value is used as the lightness value and the saturation value. Then, the corresponding channel color of the maximum color channel value (i.e., one of red, green, and blue) is used as the hue, so as to realize the color space conversion of the second image based on the lightness value, the saturation value, and the hue, that is, to convert the color space of the second image from the RGB color space to the HSV color space, and obtain the first image.

[0032] In this embodiment, by replacing the pixel values of the sliding window, the noise and outliers in the initial image can be filtered out, and the feature expressiveness and image stability of the second image can be improved. Then, by performing color space conversion based on the lightness value, the saturation value, and the hue, the features of the first image can be better highlighted, thereby improving the prediction accuracy of the subsequent garbage type prediction model for the information of the types of waterborne garbage, ensuring the accuracy of the subsequent adaptation of the tool to the waterborne garbage, and avoiding incomplete crushing of the garbage and tool wear caused by the mismatch of the crushing tool, thus improving the efficiency of waterborne garbage crushing treatment.

[0033] In this embodiment, training the preset first machine learning model according to the preset machine learning model training method to obtain a garbage type prediction model, and inputting the first image into the garbage type prediction model to determine the type information of the waterborne garbage specifically includes: Query the preset database of waterborne garbage types to obtain historical waterborne garbage type information and historical waterborne garbage image information; Perform one-hot encoding on the historical waterborne garbage type information to generate historical waterborne garbage type coding values; Match the historical waterborne garbage type coding values with the historical waterborne garbage image information to generate a historical waterborne garbage training set; The preset first machine learning model is a support vector machine model. Input the historical waterborne garbage training set into the support vector machine model, and adjust the regularization parameter, kernel function width, and polynomial order of the support vector machine model to update the support vector machine model, and obtain the accuracy rate, precision rate, and recall rate of the updated support vector machine model; Determine that the accuracy rate, precision rate, and recall rate of the updated support vector machine model meet the preset requirements, complete the training of the support vector machine model, and obtain a garbage type prediction model; Input the first image into the garbage type prediction model to obtain the encoded value output by the garbage type prediction model, and determine the type information of the floating garbage according to the encoded value.

[0034] In an alternative embodiment, the one-hot encoding of the historical floating garbage type information to generate the historical floating garbage type encoded value specifically includes: according to the historical floating garbage type information, assign a unique encoded value to it. In this embodiment, glass bottles are encoded as the number 1, discarded fishing nets are encoded as the number 2, and so on. Each different type of floating garbage is assigned a corresponding code. Therefore, after the garbage type prediction model outputs the corresponding encoded value, the type information of the floating garbage can be determined according to the previously calibrated relationship between the code and the type.

[0035] The specific encoding can be adaptively modified by the experimenter according to specific classification needs, and will not be elaborated here.

[0036] It should be noted that the Support Vector Machines (SVM) model is a binary classification model and has a wide range of applications in the field of machine learning. In this embodiment, by adjusting the regularization parameter, kernel function width, and polynomial order of the support vector machine model currently, the support vector machine model can be made more adaptable to the prediction of the types of floating garbage.

[0037] In this embodiment, by performing one-hot encoding on the historical floating garbage type information to generate the historical floating garbage type encoded value, the type information of the floating garbage can be converted into a numerical type, thereby improving the accuracy of the subsequent garbage type prediction model. By training the support vector machine model, the support vector machine model can be optimized to enable it to achieve rapid and accurate prediction of the types of floating garbage, ensuring the accuracy of the subsequent adaptation of the tool to the floating garbage, and avoiding problems such as incomplete garbage crushing and tool wear caused by the mismatch of the crushing tool, thereby improving the efficiency of floating garbage crushing treatment.

[0038] Step S102: Extract the edge contour feature set of the initial image according to a preset edge detection algorithm, and input the edge contour feature set and the type information into a preset second machine learning model to determine the volume information of the floating garbage.

[0039] In this embodiment, the extraction of the edge contour feature set of the initial image according to a preset edge detection algorithm, inputting the edge contour feature set and the type information into a preset second machine learning model to determine the volume information of the floating garbage specifically includes: Convert the color space of the initial image into a grayscale space, and filter the initial image after color space conversion through a preset Gaussian filter to obtain a third image; Obtain the horizontal gradient calculation matrix, the vertical gradient calculation matrix, and the brightness value of each pixel point in the third image; According to the horizontal gradient calculation matrix and the vertical gradient calculation matrix, in combination with a preset edge detection algorithm, determine the first pixel point set of the third image, and based on the first pixel point set, determine the edge contour feature set of the initial image; Input the edge contour feature set and the category information into a preset second machine learning model to determine the volume information of the floating garbage.

[0040] In an alternative embodiment, the preset second machine learning model may be a CNN machine learning model. The CNN machine learning model (Convolutional Neural Network), that is, the convolutional neural network, is a machine learning model widely used in tasks such as image recognition, object detection, and natural language processing. CNN mainly consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Since there are already relatively mature technologies for adaptively training the CNN machine learning model, the training process of the CNN machine learning model will not be described in detail in this embodiment.

[0041] In this embodiment, by converting the color space of the initial image into a grayscale space and performing filtering, the noise in the initial image can be removed, and the complexity of the image data can be reduced, thereby improving the accuracy of edge detection; then, through the brightness value of the pixel points and the gradient calculation matrix, the change of the pixel value can be accurately and quickly captured, so as to accurately reflect the edge and texture information of the third image, ensure the accuracy of the edge contour feature set, and further improve the prediction accuracy of the volume information of the floating garbage based on the edge contour feature set, the category information, and the preset second machine learning model, ensure the accuracy of the subsequent adaptation of the cutting tool to the floating garbage, and avoid incomplete crushing of the garbage and tool wear caused by the mismatch of the cutting tool used for crushing, thereby improving the efficiency of floating garbage crushing treatment.

[0042] In this embodiment, the step of determining the first pixel point set of the third image according to the horizontal gradient calculation matrix and the vertical gradient calculation matrix, in combination with a preset edge detection algorithm, and determining the edge contour feature set of the initial image based on the first pixel point set specifically includes: Calculate the horizontal gradient of each pixel point according to the horizontal gradient calculation matrix and the brightness value of each pixel point; calculate the vertical gradient of each pixel point according to the vertical gradient calculation matrix and the brightness value of each pixel point; determine the gradient amplitude and gradient direction of each pixel point according to the horizontal gradient and vertical gradient of each pixel point; Obtain the discrete gradient directions, match the gradient direction of each pixel point with the discrete gradient directions, and obtain the discrete gradient direction of each pixel point; Compare the gradient magnitudes of each pixel point with its adjacent pixel points in the discrete gradient direction. If the gradient magnitude of a pixel point is less than that of its adjacent pixel point, modify the gradient magnitude of the pixel point to a preset gradient magnitude; Obtain a gradient magnitude threshold, and filter out the pixel points whose gradient magnitudes are greater than or equal to the gradient magnitude threshold to obtain a first set of pixel points; Construct an edge image of the waterborne garbage based on the first set of pixel points, and extract the edge points of the edge image according to a preset contour detection algorithm to determine the edge contour feature set of the initial image.

[0043] In an alternative embodiment, the horizontal direction gradient calculation matrix is defined as , and the vertical direction gradient calculation matrix is defined as . Then, calculate the convolution of the brightness value of each pixel point with the horizontal direction gradient calculation matrix to obtain the horizontal gradient of each pixel point ; calculate the convolution of the brightness value of each pixel point with the vertical direction gradient calculation matrix to obtain the vertical gradient of each pixel point . Then, calculate the horizontal gradient and the vertical gradient according to the following formula to obtain the gradient magnitude and the gradient direction of each pixel point, where , , where is the gradient magnitude, is the gradient direction, and is the partial derivative operation; Then, obtain the discrete gradient directions, which include: 45°, 90°, 135°, and 180°. Then, calculate the difference between the gradient direction and the discrete gradient directions, and select the discrete gradient direction with the smallest difference as the discrete gradient direction of the pixel point; Then, compare the gradient magnitudes of each pixel point with its adjacent pixel points in the discrete gradient direction. If the gradient magnitude of a pixel point is less than that of its adjacent pixel point, modify the gradient magnitude of the pixel point to 1 (i.e., the preset gradient magnitude); Then, obtain a gradient magnitude threshold, which is set to 5 in this embodiment, and filter out the pixel points whose gradient magnitudes are greater than or equal to the gradient magnitude threshold to obtain a first set of pixel points; Then, construct an edge image of the waterborne garbage based on the first set of pixel points, and then extract the edge points of the edge image through a preset contour detection algorithm to determine the edge contour feature set of the initial image, where the preset contour detection algorithms include: Sobel operator detection algorithm, Prewitt operator detection algorithm, and Canny operator detection algorithm.

[0044] In this embodiment, by calculating the accurate gradient magnitude and direction, the accuracy of edge detection can be improved. Then, by comparing the gradient magnitudes in the discrete gradient directions, non-edge pixel points can be suppressed, thereby improving the robustness of the edge image. Furthermore, the accuracy of extracting the edge contour feature set from the edge image in the subsequent process is improved, the accuracy of predicting the volume information of the floating garbage is improved, the accuracy of the subsequent adaptation of the tool to the floating garbage is ensured, and the incomplete crushing of the garbage and tool wear caused by the mismatch of the tool used for crushing are avoided, thereby improving the efficiency of the floating garbage crushing process.

[0045] Step S103: According to the type information and volume information of the floating garbage, perform labeling processing on the floating garbage, and screen out several first floating garbage from the floating garbage.

[0046] In this embodiment, the step of performing labeling processing on the floating garbage according to the type information and volume information of the floating garbage and screening out several first floating garbage from the floating garbage specifically includes: Obtain the type coding value of the floating garbage according to the type information of the floating garbage; Generate a pairing vector of the floating garbage according to the type coding value and volume information of the floating garbage; Perform clustering processing on the pairing vectors of the floating garbage according to a preset clustering algorithm to obtain several clusters and the clustering center of each cluster; Assign the same label to all the floating garbage within each cluster; Screen the floating garbage according to the pairing vector corresponding to the clustering center of each cluster to obtain several first floating garbage.

[0047] In an optional embodiment, the form of the pairing vector can be expressed as (1, 370), which means that the floating garbage is a glass bottle with a volume of ; then, the pairing vectors are clustered by the K-Means clustering algorithm, so that the pairing vectors within each cluster have relatively similar data values, that is, the type coding values and volumes are relatively similar. The same label (such as using letters A, B, C, etc. for labeling) is assigned to all the floating garbage within each cluster, and the first floating garbage is obtained according to the pairing vector corresponding to the clustering center.

[0048] In this embodiment, by integrating the type coding value and volume information of the floating garbage to generate a pairing vector and then performing clustering processing, the floating garbage with relatively similar volumes and types can be grouped into one category, and the representative first floating garbage can be extracted, thereby reducing the calculation amount and complexity in the subsequent adaptation of the floating garbage and the tool, and further improving the efficiency of the floating garbage crushing process.

[0049] Step S104: Obtain the specification data of several tools, and construct a node graph between the tools and the first waterborne garbage according to the specification data of the first waterborne garbage and the tools.

[0050] In this embodiment, the obtaining the specification data of several tools and constructing a node graph between the tools and the first waterborne garbage according to the specification data of the first waterborne garbage and the tools specifically includes: Obtain the specification data of several tools, where the specification data includes: the material and size of the tools; generate a tool specification vector of the tools according to the material and size of the tools; Generate a waterborne garbage node according to the volume information and category coding value of the first waterborne garbage, and construct a tool node according to the tool specification vector; According to the material of the tool node and the category coding value of the waterborne garbage node, query a preset tool adaptation database to obtain a first adaptation value between each tool node and each waterborne garbage node; According to the size of the tool node and the volume information of the waterborne garbage node, query a preset tool adaptation database to obtain a second adaptation value between each tool node and each waterborne garbage node; Construct a node graph between the tools and the first waterborne garbage according to the first adaptation value and the second adaptation value.

[0051] In an alternative embodiment, the first adaptation value and the second adaptation value are used to characterize the adaptation degree of the tool and the waterborne garbage in terms of category and volume. Among them, due to the different materials and sizes of the tools, the crushing and cutting efficiency of waterborne garbage of different categories and volumes is different. Therefore, by querying the database, the adaptation degree between the two can be quickly calibrated. In this embodiment, the first adaptation value and the second adaptation value between the tool and the waterborne garbage with a higher adaptation degree are set to be higher, that is, the higher the first adaptation value and the second adaptation value, the more adaptable the tool and the waterborne garbage are. The specific first adaptation value and second adaptation value can be calibrated by experimenters according to actual needs and historical data, and will not be elaborated here.

[0052] In this embodiment, by querying the preset tool adaptation database, the adaptation relationship between the first waterborne garbage and the tool can be characterized more accurately, so as to ensure the accuracy of the subsequent tool crushing link scheme, and avoid problems such as incomplete garbage crushing and tool wear caused by the inadaptability of the tools used for crushing, thereby improving the efficiency of waterborne garbage crushing treatment.

[0053] It should be noted that some tools may have better crushing efficiency for a certain type of floating garbage. However, if floating garbage of different volumes of this type is mixed together, the crushing efficiency will be further reduced, that is, the type and volume of floating garbage will affect each other. Therefore, in the following embodiments, by performing node linking on the node graph, the overall matching degree between the tool and the floating garbage, and between the floating garbage and the floating garbage is considered, thereby improving the crushing efficiency of the tool.

[0054] Step S105: Perform node linking on the node graph according to a preset tool crushing link generation algorithm to construct a tool crushing link scheme, and perform crushing processing on the floating garbage based on the tool crushing link scheme.

[0055] In this embodiment, the performing node linking on the node graph according to a preset tool crushing link generation algorithm to construct a tool crushing link scheme specifically includes: Query the preset tool adaptation database to obtain the optimal type coding value and the optimal volume adaptation value of each tool node; According to the first adaptation value and the second adaptation value, link a floating garbage node to each tool node to generate a number of initial tool crushing links; wherein, the floating garbage nodes in the initial tool crushing links are defined as the first floating garbage nodes, and the floating garbage nodes in the node graph that are not linked to the initial tool crushing links are defined as the second floating garbage nodes, and the number of tool nodes in each initial tool crushing link is one; Construct a node adaptation function according to the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value; Calculate the node adaptation degree between each current second floating garbage node and each initial tool crushing link according to the node adaptation function; Based on the node adaptation degree, link the second floating garbage nodes to the initial tool crushing links, update the initial tool crushing links, the first floating garbage nodes, and the second floating garbage nodes, and then update the node adaptation degree between each second floating garbage node and each initial tool crushing link until the number of second floating garbage nodes is zero, complete the update of the initial tool crushing links, and obtain a number of optimal tool crushing links; Construct a tool crushing link scheme according to the optimal tool crushing links.

[0056] In an alternative embodiment, the node adaptation degree between each current second floating garbage node and each initial cutter crushing link is calculated according to the node adaptation function. Then, for each initial cutter crushing link in sequence, the second floating garbage node with the highest node adaptation degree is selected and linked to the initial cutter crushing link, updating the initial cutter crushing link, the first floating garbage node, and the second floating garbage node. After that, the node adaptation degree between each second floating garbage node and each initial cutter crushing link is calculated again, and the linking is repeated until the number of second floating garbage nodes is zero, that is, all the first floating garbage nodes in the node graph are linked to the initial cutter crushing link, obtaining several optimal cutter crushing links.

[0057] In this embodiment, by constructing cutter crushing links for the cutter nodes, the first floating garbage nodes, and the second floating garbage nodes in the node graph, and calculating the node adaptation degree, the adaptation degree between the floating garbage and the cutter can be comprehensively considered, thereby improving the adaptability between the cutter and the floating garbage, avoiding incomplete garbage crushing and cutter wear caused by the mismatch of the cutters used for crushing, and thus improving the efficiency of floating garbage crushing and treatment.

[0058] In an alternative embodiment, a cutter crushing link scheme is constructed according to the optimal cutter crushing link, which specifically includes: sequentially selecting the first floating garbage nodes from the optimal cutter crushing link, extracting the floating garbage with the same label from the floating garbage according to the label to which the first floating garbage node belongs (for the assignment of the label, please refer to step S103), and using the cutter corresponding to the cutter node in the optimal cutter crushing link for crushing treatment until all the first floating garbage nodes in the optimal cutter crushing link have been selected, completing the crushing treatment of the floating garbage in the optimal cutter crushing link. After that, the above steps are repeated until the floating garbage in all the optimal cutter crushing links has been completed in crushing treatment, completing the crushing treatment of the floating garbage.

[0059] It should be noted that in this embodiment, a volume guiding function is constructed to characterize the ideal volume of the next floating garbage of the current initial cutter crushing link, and a type guiding function is constructed to characterize the ideal type of the next floating garbage of the current initial cutter crushing link. Then, by combining the volume guiding function and the type guiding function, a node adaptation function is constructed, comprehensively considering the ideal volume and the ideal type, so as to be able to add the best floating garbage nodes to each initial cutter crushing link.

[0060] In this embodiment, the construction of the node adaptation function according to the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value specifically includes: Construct a volume - guiding function for the initial tool - crushing link based on the tool nodes and water - borne garbage nodes in the tool - crushing link, in combination with the optimal volume adaptation value, where the volume - guiding function is: ; where is the volume - guiding value of the th initial tool - crushing link, is the optimal volume adaptation value of the tool node in the th initial tool - crushing link, is the number of the first water - borne garbage nodes in the th initial tool - crushing link; is the th first water - borne garbage node in the th initial tool - crushing link; is the second adaptation value between the th first water - borne garbage node and the tool node in the th initial tool - crushing link. Construct a type - guiding function for the initial tool - crushing link based on the tool nodes and water - borne garbage nodes in the tool - crushing link, in combination with the optimal type - coding value, where the type - guiding function is: ; where is the type - guiding value of the th initial tool - crushing link, is the optimal type - coding value of the tool node in the th initial tool - crushing link, represents the natural logarithm with base \(e\), is the number of the first water - borne garbage nodes in the th initial tool - crushing link, is the th first water - borne garbage node in the th initial tool - crushing link, is the first adaptation value between the th first water - borne garbage node and the tool node in the th initial tool - crushing link; represents rounding down . Construct a node - adaptation function for the initial tool - crushing link according to the volume - guiding function and the type - guiding function; where the node - adaptation function is: ; In the formula, is the The fitness of the second waterborne garbage node with the nodes of the th initial tool crushing link; The type coding value of the th second waterborne garbage node; The type guiding value of the th initial tool crushing link; The number of the first waterborne garbage nodes in the th initial tool crushing link; The type coding value of the th first waterborne garbage node in the th initial tool crushing link; The volume of the

[0061] In this embodiment, a node adaptation function is constructed through the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value, fully exploring the adaptation relationship between waterborne garbage and tools in terms of type and volume, enabling the node adaptation function to more accurately represent the adaptation relationship between waterborne garbage and tools, thereby improving the adaptability between tools and waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by inappropriate tools for crushing, and thus improving the efficiency of waterborne garbage crushing and treatment.

[0062] It can be understood that, compared with the prior art, in this embodiment, the initial image is enhanced by an image enhancement method, which can improve the image quality and feature clarity, enabling the first machine learning model to more quickly and accurately identify the type information of the waterborne garbage; through the combination of the edge detection algorithm, the type information and the second machine learning model, the volume information of the waterborne garbage can be efficiently determined, and the efficiency of subsequent tool adaptation is improved by efficiently and accurately obtaining the type information and the volume information; by performing tagging processing on the waterborne garbage, the complexity and computational amount of subsequent tool adaptation can be reduced, thereby improving the efficiency of waterborne garbage crushing processing; by constructing a node graph based on the characteristics of the first waterborne garbage and the specification data of the tool, and performing node linking through a preset tool crushing link generation algorithm, an optimal tool crushing link scheme can be quickly and accurately generated, so that the tool can be adapted to the waterborne garbage, avoiding incomplete garbage crushing and tool wear caused by inappropriate tools used for crushing, thereby improving the efficiency of waterborne garbage crushing processing.

[0063] Embodiment 2 Please refer to Figure 2 , which is a schematic structural diagram of a waterborne garbage crushing processing system provided by an embodiment of the present invention, including: a type information acquisition module 201, a volume information acquisition module 202, a tagging processing module 203, a node graph construction module 204, and a tool crushing link acquisition module 205.

[0064] Among them, the type information acquisition module 201 is used to acquire the initial images of a plurality of waterborne garbage, enhance the initial images according to a preset image enhancement method to obtain a first image, and determine the type information of the waterborne garbage according to the first image and a preset first machine learning model.

[0065] In this embodiment, the type information acquisition module 201 includes: a type information acquisition unit; The type information acquisition unit is used to acquire a sliding window and a sliding amplitude value; According to the sliding amplitude value, slide the sliding window on the initial image, acquire the pixel median value in the sliding window after each slide, and modify the pixel value corresponding to the center of the sliding window to the pixel median value until the sliding of the sliding window on the initial image is completed to obtain a second image; Acquire the initial color channel value of the second image, and perform normalization processing on the color channel value to obtain a first color channel value; Acquire the maximum color channel value in the first color channel value, and determine the brightness value and saturation value of the second image according to the maximum color channel value; Determine the corresponding channel color of the maximum color channel value, and determine the hue of the second image based on the channel color; Perform color space conversion on the second image according to the lightness value, saturation value, and hue to obtain a first image; Train the preset first machine learning model according to a preset machine learning model training method to obtain a garbage type prediction model, and input the first image into the garbage type prediction model to determine the type information of the waterborne garbage.

[0066] In this embodiment, the type information acquisition unit includes: a type information acquisition subunit; The type information acquisition subunit is used to query a preset waterborne garbage type database to obtain historical waterborne garbage type information and historical waterborne garbage image information; Perform one-hot encoding on the historical waterborne garbage type information to generate a historical waterborne garbage type coding value; Match the historical waterborne garbage type coding value with the historical waterborne garbage image information to generate a historical waterborne garbage training set; The preset first machine learning model is a support vector machine model. Input the historical waterborne garbage training set into the support vector machine model, and adjust the regularization parameter, kernel function width, and polynomial order of the support vector machine model to update the support vector machine model, and obtain the accuracy, precision, and recall rate of the updated support vector machine model; Determine that the accuracy, precision, and recall rate of the updated support vector machine model meet the preset requirements, complete the training of the support vector machine model, and obtain a garbage type prediction model; Input the first image into the garbage type prediction model to obtain the coding value output by the garbage type prediction model, and determine the type information of the waterborne garbage according to the coding value.

[0067] The volume information acquisition module 202 is used to extract the edge contour feature set of the initial image according to a preset edge detection algorithm, and input the edge contour feature set and type information into a preset second machine learning model to determine the volume information of the waterborne garbage.

[0068] In this embodiment, the volume information acquisition module 202 includes: a volume information acquisition unit; The volume information acquisition unit is used to convert the color space of the initial image into a grayscale space, and filter the initial image after color space conversion through a preset Gaussian filter to obtain a third image; Obtain a horizontal direction gradient calculation matrix, a vertical direction gradient calculation matrix, and the brightness value of each pixel point in the third image; Based on the horizontal direction gradient calculation matrix and the vertical direction gradient calculation matrix, combine a preset edge detection algorithm to determine the first pixel point set of the third image, and determine the edge contour feature set of the initial image based on the first pixel point set; Input the edge contour feature set and the category information into a preset second machine learning model to determine the volume information of the waterborne garbage.

[0069] In this embodiment, the volume information acquisition unit includes: an edge contour feature set acquisition subunit; The edge contour feature set acquisition subunit is used to calculate the horizontal gradient of each pixel point according to the horizontal direction gradient calculation matrix and the brightness value of each pixel point; calculate the vertical gradient of each pixel point according to the vertical direction gradient calculation matrix and the brightness value of each pixel point; determine the gradient amplitude and gradient direction of each pixel point according to the horizontal gradient and vertical gradient of each pixel point; Obtain the gradient discrete direction, and match the gradient direction of each pixel point with the gradient discrete direction to obtain the discrete gradient direction of each pixel point; Compare the gradient amplitude of each pixel point with that of adjacent pixel points in the discrete gradient direction. If the gradient amplitude of the pixel point is less than that of the adjacent pixel point, modify the gradient amplitude of the pixel point to a preset gradient amplitude; Obtain a gradient amplitude threshold, and screen out the pixel points whose gradient amplitude is greater than or equal to the gradient amplitude threshold to obtain a first pixel point set; Construct an edge image of the waterborne garbage according to the first pixel point set, and extract the edge points of the edge image according to a preset contour detection algorithm to determine the edge contour feature set of the initial image.

[0070] The tagging processing module 203 is used to perform tagging processing on the waterborne garbage according to the category information and volume information of the waterborne garbage, and screen out several first waterborne garbage from the waterborne garbage.

[0071] In this embodiment, the tagging processing module 203 includes: a tagging processing unit; The tagging processing unit is used to obtain the category coding value of the waterborne garbage according to the category information of the waterborne garbage; Generate a pairing vector of the waterborne garbage according to the category coding value and volume information of the waterborne garbage; Perform clustering processing on the pairing vectors of the waterborne garbage according to a preset clustering algorithm to obtain several clusters and the clustering center of each cluster; Assign the same label to all the waterborne garbage within each cluster; Screen the above - water garbage according to the pairing vectors corresponding to the clustering centers of each cluster to obtain a number of first above - water garbage.

[0072] The node graph construction module 204 is used to obtain the specification data of a number of tools, and construct a node graph between the tools and the first above - water garbage according to the first above - water garbage and the specification data of the tools.

[0073] In this embodiment, the node graph construction module 204 includes: a node graph construction unit; The node graph construction unit is used to obtain the specification data of a number of tools, where the specification data includes: the material and size of the tool; generate a tool specification vector of the tool according to the material and size of the tool; Generate an above - water garbage node according to the volume information and type coding value of the first above - water garbage, and construct a tool node according to the tool specification vector; Query a preset tool adaptation database according to the material of the tool node and the type coding value of the above - water garbage node, and obtain a first adaptation value between each tool node and each above - water garbage node; Query a preset tool adaptation database according to the size of the tool node and the volume information of the above - water garbage node, and obtain a second adaptation value between each tool node and each above - water garbage node; Construct a node graph between the tools and the first above - water garbage according to the first adaptation value and the second adaptation value.

[0074] The tool crushing link acquisition module 205 is used to perform node linking in the node graph according to a preset tool crushing link generation algorithm, construct a tool crushing link scheme, and perform crushing processing on the above - water garbage based on the tool crushing link scheme.

[0075] In this embodiment, the tool crushing link acquisition module 205 includes: a tool crushing link acquisition unit; The tool crushing link acquisition unit is used to query a preset tool adaptation database to obtain the optimal type coding value and the optimal volume adaptation value of each tool node; Link a water - garbage node to each of the tool nodes according to the first adaptation value and the second adaptation value to generate a number of initial tool crushing links; where the water - garbage nodes in the initial tool crushing links are defined as first water - garbage nodes, and the water - garbage nodes in the node graph that are not linked to the initial tool crushing links are defined as second water - garbage nodes, and the number of tool nodes in each of the initial tool crushing links is one; Construct a node adaptation function according to the first adaptation value, the second adaptation value, the optimal type coding value, and the optimal volume adaptation value; Calculate the node adaptation degree between each current second waterborne garbage node and each initial tool crushing link according to the node adaptation function; Link the second waterborne garbage nodes to the initial tool crushing links based on the node adaptation degree, update the initial tool crushing links, the first waterborne garbage nodes and the second waterborne garbage nodes, and then update the node adaptation degree between each second waterborne garbage node and each initial tool crushing link until the number of the second waterborne garbage nodes is zero, complete the update of the initial tool crushing links, and obtain several optimal tool crushing links; Construct a tool crushing link scheme according to the optimal tool crushing links.

[0076] In this embodiment, the tool crushing link acquisition unit includes: a node adaptation function construction subunit; The node adaptation function construction subunit is used to construct a volume guiding function of the initial tool crushing link according to the tool nodes and waterborne garbage nodes in the tool crushing link, in combination with the optimal volume adaptation value, where the volume guiding function is: ; Where is the volume guiding value of the th initial tool crushing link, is the optimal volume adaptation value of the tool node in the th initial tool crushing link, is the number of the first waterborne garbage nodes in the th initial tool crushing link; is the th first waterborne garbage node in the th initial tool crushing link; is the second adaptation value between the th first waterborne garbage node and the tool node in the th initial tool crushing link; Construct a type guiding function of the initial tool crushing link according to the tool nodes and waterborne garbage nodes in the tool crushing link, in combination with the optimal type coding value, where the type guiding function is: ; Where is the type guiding value of the th initial tool crushing link, is the optimal type coding value of the tool node in the th initial tool crushing link, represents the natural logarithm with the mathematical constant e as the base, is the The number of the first waterborne waste nodes in the initial tool crushing link For the th first waterborne waste node in the initial tool crushing link For the th first waterborne waste node in the initial tool crushing link, the first adaptation value between the It means to round down According to the volume guiding function and the type guiding function, construct the node adaptation function of the initial tool crushing link; wherein, the node adaptation function is: ; In the formula, is the node adaptation degree between the th second waterborne waste node in the node graph and the th initial tool crushing link For the th type coding value of the second waterborne waste node For the th type guiding value of the initial tool crushing link For the th number of the first waterborne waste nodes in the initial tool crushing link For the th first waterborne waste node in the initial tool crushing link is the th type coding value of the first waterborne waste node in the th initial tool crushing link is the volume of the th second waterborne waste node is the th volume guiding value of the initial tool crushing link is the optimal volume adaptation value of the tool node in the th initial tool crushing link is the th volume of the first waterborne waste node in the th initial tool crushing link

[0077] ​In this embodiment, the initial image is enhanced by an image enhancement method, which can improve the image quality and feature clarity, enabling the first machine learning model to more quickly and accurately identify the type information of the floating garbage; through the combination of the edge detection algorithm, the type information and the second machine learning model, the volume information of the floating garbage can be efficiently determined, and the efficiency of subsequent tool adaptation is improved by efficiently and accurately obtaining the type information and the volume information; by performing labeling processing on the floating garbage, the complexity and computational amount of subsequent tool adaptation can be reduced, thereby improving the efficiency of floating garbage crushing processing; by constructing a node graph based on the characteristics of the first floating garbage and the specification data of the tool, and performing node linking through a preset tool crushing link generation algorithm, an optimal tool crushing link scheme can be quickly and accurately generated, so that the tool can be adapted to the floating garbage, avoiding incomplete garbage crushing and tool wear caused by the mismatch of the tools used for crushing, thereby improving the efficiency of floating garbage crushing processing.

[0078] In summary, in the embodiment of the present invention, the initial image is enhanced by an image enhancement method, which can improve the image quality and feature clarity, enabling the first machine learning model to more quickly and accurately identify the type information of the floating garbage; through the combination of the edge detection algorithm, the type information and the second machine learning model, the volume information of the floating garbage can be efficiently determined, and the efficiency of subsequent tool adaptation is improved by efficiently and accurately obtaining the type information and the volume information; by performing labeling processing on the floating garbage, the complexity and computational amount of subsequent tool adaptation can be reduced, thereby improving the efficiency of floating garbage crushing processing; by constructing a node graph based on the characteristics of the first floating garbage and the specification data of the tool, and performing node linking through a preset tool crushing link generation algorithm, an optimal tool crushing link scheme can be quickly and accurately generated, so that the tool can be adapted to the floating garbage, avoiding incomplete garbage crushing and tool wear caused by the mismatch of the tools used for crushing, thereby improving the efficiency of floating garbage crushing processing.

[0079] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for crushing and treating water garbage, characterized in that: include: Acquire a number of initial images of water garbage, perform image enhancement on the initial images according to a preset image enhancement method to obtain a first image, and determine type information of the water garbage according to the first image and a preset first machine learning model; Extracting an edge contour feature set of the initial image according to a preset edge detection algorithm, inputting the edge contour feature set and type information into a preset second machine learning model, and determining volume information of the water garbage; labeling the water garbage according to the type information and volume information of the water garbage, and screening out a number of first water garbage from the water garbage; Acquire specification data of a plurality of cutting tools, and construct a node graph between the cutting tools and the first aquatic garbage according to the specification data of the first aquatic garbage and the cutting tools; The nodes in the node graph are linked according to a preset tool crushing link generation algorithm, a tool crushing link scheme is constructed, and the water garbage is crushed based on the tool crushing link scheme.

2. A method for crushing and treating water garbage as claimed in claim 1, characterized in that: The method of obtaining a plurality of initial images of the garbage on the water, performing image enhancement on the initial images according to a preset image enhancement method to obtain a first image, and determining the type information of the garbage on the water according to the first image and a preset first machine learning model specifically includes: Get the sliding window and sliding amplitude values; According to the sliding amplitude value, the sliding window is slid on the initial image, a median value of pixels in the sliding window after each sliding is obtained, and a pixel value corresponding to the center of the sliding window is modified to the median value of pixels, until the sliding of the sliding window on the initial image is completed, so as to obtain a second image; Acquire initial color channel values ​​of the second image, and perform normalization processing on the color channel values ​​to obtain first color channel values; Acquire a maximum color channel value among the first color channel values, and determine a brightness value and a saturation value of the second image according to the maximum color channel value; Determining a channel color corresponding to the maximum color channel value, and determining a hue of the second image based on the channel color; Performing color space conversion on the second image according to the brightness value, saturation value and hue to obtain a first image; The preset first machine learning model is trained according to a preset machine learning model training method to obtain a garbage type prediction model, and the first image is input into the garbage type prediction model to determine the type information of the water garbage.

3. A method for crushing and treating water garbage as claimed in claim 2, characterized in that: The step of training the preset first machine learning model according to the preset machine learning model training method to obtain a garbage type prediction model, and inputting the first image into the garbage type prediction model to determine the type information of the water garbage specifically includes: Query a preset database of water garbage types to obtain historical water garbage type information and historical water garbage image information; Performing one-hot encoding on the historical water garbage type information to generate a historical water garbage type code value; Matching the historical water garbage type code value with the historical water garbage image information to generate a historical water garbage training set; The preset first machine learning model is a support vector machine model, the historical aquatic garbage training set is input into the support vector machine model, the regularization parameter, kernel function width and polynomial order of the support vector machine model are adjusted to update the support vector machine model, and the accuracy, precision and recall of the updated support vector machine model are obtained; Determine whether the accuracy, precision and recall of the updated support vector machine model meet the preset requirements, complete the training of the support vector machine model, and obtain a garbage type prediction model; The first image is input into the garbage type prediction model to obtain a coding value output by the garbage type prediction model, and the type information of the water garbage is determined according to the coding value.

4. A method for crushing and treating water garbage according to any one of claims 1 to 3, characterized in that: The step of extracting an edge contour feature set of the initial image according to a preset edge detection algorithm, inputting the edge contour feature set and the type information into a preset second machine learning model, and determining the volume information of the water garbage specifically includes: Converting the color space of the initial image into a grayscale space, and filtering the initial image after the color space conversion by a preset Gaussian filter to obtain a third image; Obtaining a horizontal gradient calculation matrix, a vertical gradient calculation matrix, and a brightness value of each pixel in the third image; Determine a first pixel point set of the third image according to the horizontal gradient calculation matrix and the vertical gradient calculation matrix in combination with a preset edge detection algorithm, and determine an edge contour feature set of the initial image based on the first pixel point set; The edge contour feature set and type information are input into a preset second machine learning model to determine the volume information of the water garbage.

5. A method for crushing and treating water garbage as claimed in claim 4, characterized in that: The step of determining a first pixel point set of the third image according to the horizontal gradient calculation matrix and the vertical gradient calculation matrix in combination with a preset edge detection algorithm, and determining an edge contour feature set of the initial image based on the first pixel point set, specifically includes: Calculate the horizontal gradient of each pixel point according to the horizontal gradient calculation matrix and the brightness value of each pixel point; calculate the vertical gradient of each pixel point according to the vertical gradient calculation matrix and the brightness value of each pixel point; determine the gradient amplitude and gradient direction of each pixel point according to the horizontal gradient and vertical gradient of each pixel point; Acquire a discrete gradient direction, match the gradient direction of each pixel with the discrete gradient direction, and obtain a discrete gradient direction of each pixel; Comparing the gradient amplitude of each pixel point with that of an adjacent pixel point in a discrete gradient direction, if the gradient amplitude of the pixel point is smaller than the gradient amplitude of the adjacent pixel point, modifying the gradient amplitude of the pixel point to a preset gradient amplitude; Obtaining a gradient amplitude threshold, screening out pixel points whose gradient amplitudes are greater than or equal to the gradient amplitude threshold, and obtaining a first pixel point set; An edge image of the water garbage is constructed according to the first pixel point set, and edge points of the edge image are extracted according to a preset contour detection algorithm to determine an edge contour feature set of the initial image.

6. A method for crushing and treating water garbage as claimed in claim 1, characterized in that: The labeling process of the water garbage according to the type information and volume information of the water garbage, and screening out a plurality of first water garbage from the water garbage, specifically includes: Obtaining a type code value of the water garbage according to the type information of the water garbage; Generate a pairing vector of the water garbage according to the type code value and volume information of the water garbage; Clustering the paired vectors of the water garbage according to a preset clustering algorithm to obtain a number of clusters and a cluster center of each cluster; Assign the same label to all the water debris in each cluster; The water garbage is screened according to the pairing vector corresponding to the cluster center of each cluster to obtain a plurality of first water garbage.

7. A method for crushing and treating water garbage as claimed in claim 1, characterized in that: The obtaining of specification data of a plurality of cutting tools and constructing a node graph between the cutting tools and the first aquatic garbage according to the specification data of the first aquatic garbage and the cutting tools specifically includes: Acquire specification data of a plurality of cutting tools, wherein the specification data includes: material and size of the cutting tools; generate a cutting tool specification vector of the cutting tools according to the material and size of the cutting tools; Generate a water garbage node according to the volume information and type code value of the first water garbage, and construct a tool node according to the tool specification vector; According to the material of the tool node and the type code value of the water garbage node, query the preset tool adaptation database to obtain the first adaptation value between each tool node and each water garbage node; According to the size of the cutter node and the volume information of the water garbage node, query the preset cutter adaptation database to obtain the second adaptation value between each cutter node and each water garbage node; A node graph between the tool and the first aquatic garbage is constructed according to the first adaptation value and the second adaptation value.

8. A method for crushing and treating water garbage as claimed in claim 7, characterized in that: The step of linking nodes in the node graph according to a preset tool crushing link generation algorithm to construct a tool crushing link solution specifically includes: Query the preset tool adaptation database to obtain the optimal type code value and optimal volume adaptation value of each tool node; According to the first adaptation value and the second adaptation value, each of the tool nodes is linked to an aquatic garbage node to generate a plurality of initial tool crushing links; wherein the aquatic garbage node in the initial tool crushing link is defined as a first aquatic garbage node, and the aquatic garbage node that is not linked to the initial tool crushing link in the node graph is defined as a second aquatic garbage node, and the number of tool nodes in each of the initial tool crushing links is one; constructing a node adaptation function according to the first adaptation value, the second adaptation value, the optimal category coding value and the optimal volume adaptation value; Calculating the node fitness between each current second water garbage node and each initial tool crushing link according to the node fitness function; Based on the node fitness, the second water garbage node is linked to the initial tool crushing link, the initial tool crushing link, the first water garbage node and the second water garbage node are updated, and then the node fitness between each second water garbage node and each initial tool crushing link is updated until the number of the second water garbage nodes is zero, and the initial tool crushing link is updated to obtain a plurality of optimal tool crushing links; A tool crushing link solution is constructed according to the optimal tool crushing link.

9. A method for crushing and treating water garbage as claimed in claim 8, characterized in that: The constructing a node adaptation function according to the first adaptation value, the second adaptation value, the optimal type code value and the optimal volume adaptation value specifically includes: According to the tool nodes and the water garbage nodes in the tool crushing link, combined with the optimal volume adaptation value, the volume guidance function of the initial tool crushing link is constructed, wherein the volume guidance function is: ； in, For the The volume-oriented value of the initial tool crushing link, For the The optimal volume adaptation value of the tool node in the initial tool crushing link, For the The number of first water garbage nodes in the initial tool crushing link; For the The first of the initial tool crushing links The first water garbage node; For the The first A second adaptation value between a first water garbage node and a cutter node; According to the tool nodes and the water garbage nodes in the tool crushing link, combined with the optimal category coding value, the category guidance function of the initial tool crushing link is constructed, wherein the category guidance function is: ； in, For the The type-oriented value of the initial tool crushing link, For the The optimal type encoding value of the tool node in the initial tool crushing link, represents the logarithm with the mathematical constant e as base, For the The number of the first water garbage nodes in the initial tool crushing link, For the The first of the initial tool crushing links The first water garbage node, For the The first A first adaptation value between a first water garbage node and a cutter node; Express Round down; According to the volume-oriented function and the type-oriented function, a node adaptation function of the initial tool crushing link is constructed; wherein the node adaptation function is: ； In the formula, The node graph is The second water garbage node and the The node fitness of the initial tool crushing link; For the The type code value of the second water garbage node; For the The type-oriented value of the initial tool crushing link, For the The number of the first water garbage nodes in the initial tool crushing link, For the The first of the initial tool crushing links The first water garbage node, For the The first of the initial tool crushing links The type code value of the first water garbage node, For the The volume of the second water garbage node, For the The volume-oriented value of the initial tool crushing link, For the The optimal volume adaptation value of the tool node in the initial tool crushing link, For the The first of the initial tool crushing links The volume of the first water garbage node.

10. A water garbage crushing and processing system, characterized in that: include: Type information acquisition module, volume information acquisition module, labeling processing module, node graph construction module and tool crushing link acquisition module; The type information acquisition module is used to acquire initial images of a number of water garbage, perform image enhancement on the initial images according to a preset image enhancement method to obtain a first image, and determine the type information of the water garbage according to the first image and a preset first machine learning model; The volume information acquisition module is used to extract the edge contour feature set of the initial image according to a preset edge detection algorithm, input the edge contour feature set and the type information into a preset second machine learning model, and determine the volume information of the water garbage; The labeling processing module is used to label the water garbage according to the type information and volume information of the water garbage, and screen out a number of first water garbage from the water garbage; The node graph construction module is used to obtain specification data of a plurality of cutting tools, and to construct a node graph between the cutting tools and the first aquatic garbage according to the specification data of the first aquatic garbage and the cutting tools; The tool crushing link acquisition module is used to link nodes in the node graph according to a preset tool crushing link generation algorithm, construct a tool crushing link plan, and crush the water garbage based on the tool crushing link plan.

Images