Construction waste recovery device based on image sensor

Abstract

The invention relates to the technical field of construction waste recovery, in particular to a construction waste recovery device based on an image sensor, comprising: a pretreatment assembly comprising a crushing bin and a preliminary screening hopper; the winnowing assembly comprises an air nozzle arranged in the primary screening hopper; the secondary screening assembly comprises a screening tray, a flattening plate and screening holes; the image acquisition assembly comprises an image sensor, construction waste fragments are subjected to secondary screening and filtering on a sieve tray, residues are subjected to secondary crushing and recycling, the crushing strength of a crushing roller is controlled by collecting residue images, the filtering precision can be adjusted by moving a flattening plate, and after a part of construction waste is blocked by sieve holes, the residues are subjected to secondary crushing and recycling. The guide block collides with the inner wall of the sieve tray and vibrates, so that blocked construction waste is vibrated and slides down, the recognition system can recognize features of residual objects and classify the residual objects, follow-up secondary crushing and recycling are facilitated, the blocking degree is judged according to the light transmittance, and overspeed vibration rotation is controlled to achieve blocking clearing.

Description

Technical Field

[0001] The invention relates to the technical field of construction waste recycling, in particular to a construction waste recycling device based on an image sensor. Background Art

[0002] Construction waste refers to construction waste generated in engineering projects due to human or natural reasons, including waste soil, abandoned soil, silt and abandoned materials. With the rapid development of the construction industry, the amount of construction waste generated is increasing day by day, which needs to be treated and utilized in time to eliminate the adverse effects on society, environment and resources. It is generally classified by the particle size of construction waste. The existing construction waste classification and recycling equipment is relatively rough, resulting in poor recycling effect.

[0003] The existing Chinese patent application with publication number CN115178573A discloses a construction waste recycling device, including a motor, a hydraulic cylinder and a base plate, a box body with a top opening is fixedly mounted on the base plate, a filter plate for separating crushed materials and dust is mounted in the box body, a crushing part is mounted on the part of the box body located above the filter plate and is provided with a crushing outlet, a dust collecting part is mounted on the part of the box body located below the filter plate and is provided with a dust outlet, the crushing part can effectively crush the garbage, and the dust collecting part can effectively reduce the flying of dust when the garbage is crushed, thereby reducing the pollution of the environment by dust when the garbage is crushed, the present invention also realizes the function of compressing and removing the garbage through the mutual linkage and cooperation of the hydraulic cylinder, the compression plate, the top material part and the pushing material part, and it does not need manual labor or additional power equipment to remove the compressed garbage, thereby saving resources.

[0004] However, due to the high dust content when sorting the above-mentioned construction waste fragments, manual sorting is difficult and hazardous. The existing machine sorting accuracy cannot meet the classification requirements, the process of adjusting the filter size is complicated, and many filter residues need to be crushed for the second time. The crushing force cannot be adjusted based on the feedback of the residue results after crushing and filtering. Multiple screening is required, and the process is complicated. In addition, construction waste contains metal objects, which are difficult to distinguish from objects such as concrete and can easily clog the filter screen holes, making them difficult to clean. Therefore, accurate identification and special recycling are required. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In view of the above problems in the prior art, the present invention is proposed.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a construction waste recycling device based on an image sensor, comprising a pre-processing assembly, including a crushing bin and a primary screening bucket, the primary screening bucket performs reciprocating oscillation, the construction waste is initially crushed after entering the crushing bin, and is initially screened in the primary screening bucket; The air selection assembly includes an air nozzle disposed in the primary screening bucket, and the air nozzle blows air when the primary screening bucket oscillates; The secondary screening assembly includes a sieve plate rotatably arranged at the lower end of the primary screening bucket, wherein the rotation modes of the sieve plate include concentric rotation and eccentric rotation, and an expansion plate is movably arranged on the sieve plate, and sieve holes are opened on the sieve plate. When the sieve plate and the expansion plate rotate relative to each other, the maximum size of the sieve holes that can pass through the construction waste is adjusted, and the sieve plate adjusts the degree of eccentric rotation according to the degree of blockage of the sieve holes; The image acquisition assembly comprises an image sensor arranged above the sieve plate. The image sensor is arranged at one end of the construction waste outlet on the sieve plate and is used for real-time acquisition of images of the construction waste on the sieve plate.

[0008] As a preferred solution of the image sensor-based construction waste recycling device of the present invention, wherein: two crushing rollers are arranged in the crushing bin, the crushing rollers rotate in opposite directions, a first elastic member is arranged between the crushing bin and the primary screening bucket, and a vibration motor is also arranged on the primary screening bucket; It also includes an identification system, which is used to receive images of construction waste and identify and classify them according to size parameters, determine the degree of blockage by comparing the light transmittance of the sieve holes, and adjust the distance between the two crushing rollers according to the average size specification parameters of the construction waste.

[0009] As a preferred solution of the construction waste recycling device based on image sensor described in the present invention, a telescopic column is provided on the sieve plate, the other end of the telescopic column is connected to the image sensor, and a concentration sensor is also provided at the lower end of the telescopic column. The telescopic column drives the image sensor to extend and retract according to the dust concentration.

[0010] As a preferred solution of the image sensor-based construction waste recycling device described in the present invention, the primary screening bucket is provided with an interlayer, and a primary filter groove is provided in a linear array on the interlayer. The remaining construction waste after a complete rotation on the screen plate is input into the crushing bin for secondary crushing.

[0011] As a preferred solution of the image sensor-based construction waste recycling device of the present invention, a dust cover is rotatably provided on the sieve plate, the spreading plate includes scrapers distributed in an array and a connecting plate connecting the scrapers in series, the connecting plate slides through the inside of the dust cover and is provided with a roller; An eccentric disk is arranged at the center of the sieve disk, a step layer is arranged on the eccentric disk, and the roller slides in accordance with the step layer.

[0012] As a preferred solution of the image sensor-based construction waste recycling device described in the present invention, the lower end face of the eccentric disk is provided with a slider, the outer wall of the slider is slidably engaged with a slide groove, the slider is also penetrated by a slide rod, and the outer wall of the slide groove is fixed with a rotating shaft.

[0013] As a preferred solution of the construction waste recycling device based on image sensor described in the present invention, a counterweight ball is provided at one end of the slide rod passing through the slide groove, a second elastic member is provided between the counterweight ball and the slide block, and a hub is provided between the rotating shaft and the outer wall of the screen plate.

[0014] As a preferred solution of the image sensor-based construction waste recycling device described in the present invention, a buffer column is slidably provided at the end of the expansion plate, a guide block is provided at the end of the buffer column, and a third elastic member is provided between the guide block and the buffer column.

[0015] As a preferred solution of the image sensor-based construction waste recycling device of the present invention, the recognition system further includes a classification strategy, which is used to recognize and classify each feature block in the image, and at the same time recognize the light transmittance of the sieve hole and output the blockage degree value of the sieve hole; The identification system also includes an execution unit. The identification system can determine the degree of blockage of the sieve holes based on the light transmittance detection of the sieve holes and output the corresponding value to the execution unit. The execution unit controls the eccentric disk to vibrate at a high frequency.

[0016] As a preferred solution of the construction waste recycling device based on image sensor of the present invention, the classification strategy includes: S1. Image preprocessing: convert the color image captured by the image sensor into a grayscale image to reduce the complexity of image processing and the interference caused by color information. Then, perform noise reduction on the grayscale image, extract the edge information of the image, detect the feature block contours of each object and record them separately.

[0017] S2. Extract features and establish a database. At the same time, extract the geometric shape features and texture features of the feature blocks, record them in the object data with corresponding labels, store the extracted geometric shape features and texture features in a structured database, and attach a label of "metal" or "non-metal" to each sample.

[0018] S3. Perform machine learning training; S3.1. Divide the dataset into a training set, a validation set, and a test set: a training set for model training, a validation set for adjusting hyperparameters, and a test set for final model evaluation. S3.2. Data standardization and normalization: standardization of geometric shape features and texture features; S3.3. Construct a convolutional neural network model; S3.4. Train the model. After setting the training parameters, input the training data into the model for training, and use the validation set to evaluate the model during the training process.

[0019] S4. Model evaluation, based on the accuracy of the model evaluated on the test set.

[0020] S5. Model deployment and application: Load the trained model in actual applications for prediction, and use the model to predict the real-time collected image data in the production environment. By inputting the image into the model, the prediction result is obtained.

[0021] The beneficial effects of the present invention are as follows: by sorting and crushing construction waste, the construction waste fragments are subjected to secondary screening and filtration on the sieve plate after being stirred and disturbed by the spreading plate, and the residues are secondary crushed and recycled. The crushing force of the crushing roller is controlled by collecting the residue image, and the filtration accuracy can also be adjusted by moving the spreading plate. After the sieve holes are partially blocked by construction waste, the guide block collides with the inner wall of the sieve plate and vibrates, thereby vibrating the blocked construction waste off. The recognition system can identify the characteristics of the residual objects and classify them, which is convenient for subsequent secondary crushing and recycling. The degree of blockage is determined according to the light transmittance and the overspeed vibration rotation is controlled to achieve blockage clearing. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 It is an overall schematic diagram of the construction waste recycling device based on image sensor in the present invention.

[0023] Figure 2 This is an enlarged view of the internal structure details of the pretreatment assembly in the present invention.

[0024] Figure 3 It is a schematic diagram of the connection structure between the primary screening bucket and the sieve plate in the present invention.

[0025] Figure 4 It is a schematic diagram of the back structure of the sieve plate in the present invention.

[0026] Figure 5 It is an enlarged view of the sieve hole and the spreading plate in the present invention.

[0027] Figure 6 It is a schematic diagram of the identification system flow in the present invention.

[0028] Figure 7 It is a schematic diagram of the classification strategy flow in the present invention.

[0029] Among them, in the figure: 100, crushing bin; 101, primary screening bucket; 1001, crushing roller; 1002, first elastic member; 1003, vibration motor; 1004, telescopic column; 1005, interlayer; 1006, primary filter tank; 200, wind nozzle; 300, sieve plate; 301, flat plate; 302, sieve hole; 3001, eccentric plate; 3002, dust cover; 3003, scraper; 3004, connecting plate; 3005, roller; 3006, step layer; 3007, slider; 3008, slide groove; 3009, slide rod; 3011, weight ball; 3012, second elastic member; 3013, rotating shaft; 3014, wheel hub; 3015, buffer column; 3016, guide block; 3017, third elastic member; 400. Image sensor. DETAILED DESCRIPTION

[0030] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0032] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0033] Example 1 Reference Figure 1 to Figure 7, which is the first embodiment of the present invention, provides a construction waste recycling device based on an image sensor, including a pretreatment assembly, a wind separation assembly, a secondary screening assembly and an image acquisition assembly, which sorts and crushes the construction waste, and performs secondary screening and secondary crushing after filtering, and controls the crushing force of the crushing roller 1001 by collecting the residue image.

[0034] Specifically, it includes: a pre-treatment assembly, including a crushing bin 100 and a primary screening bucket 101, the primary screening bucket 101 performs reciprocating oscillation, the construction waste is initially crushed after entering the crushing bin 100, and is initially screened in the primary screening bucket 101; The air selection assembly includes an air nozzle 200 disposed in the primary screening bucket 101, and the air nozzle 200 blows air when the primary screening bucket 101 oscillates; The secondary screening assembly includes a sieve plate 300 rotatably arranged at the lower end of the primary screening bucket 101. The rotation modes of the sieve plate 300 include concentric rotation and eccentric rotation. A spreading plate 301 is movably arranged on the sieve plate 300. The sieve holes 302 are opened on the sieve plate 300. When the sieve plate 300 and the spreading plate 301 rotate relative to each other, the maximum size of the sieve holes 302 that can pass through the construction waste is adjusted. The sieve plate 300 adjusts the degree of eccentric rotation according to the degree of blockage of the sieve holes 302. The image acquisition assembly includes an image sensor 400 disposed above the sieve plate 300 . The image sensor 400 is disposed at one end of the construction waste outlet on the sieve plate 300 and is used to acquire images of the construction waste on the sieve plate 300 in real time.

[0035] Two crushing rollers 1001 are arranged in the crushing bin 100 , and the crushing rollers 1001 rotate in opposite directions. A first elastic member 1002 is arranged between the crushing bin 100 and the primary screening bucket 101 , and a vibration motor 1003 is also arranged on the primary screening bucket 101 .

[0036] Furthermore, it also includes an identification system, which is used to receive images of construction waste and identify and classify them according to size parameters, and determine the degree of blockage by comparing the light transmittance of the sieve holes 302, and adjust the distance between the two crushing rollers 1001 according to the average size specification parameters of the construction waste. The more the residual construction waste and the larger the average size, the two crushing rollers 1001 are controlled to be closer to each other to increase the crushing force.

[0037] Preferably, the primary screening bucket 101 is tilted at 30° to facilitate the rolling of construction waste during vibration and improve screening efficiency. Four air nozzles 200 are arranged in a linear array and parallel to the primary screening bucket 101. An opening is provided on the other side of the primary screening bucket 101. Lighter plastic waste and wood waste fly out of the opening under the action of wind and are filtered and dust-removed for collection and reuse. The primary screening bucket 101 is closed to prevent dust from escaping.

[0038] More preferably, a telescopic column 1004 is provided on the sieve plate 300, the other end of the telescopic column 1004 is connected to the image sensor 400, and a concentration sensor is also provided at the lower end of the telescopic column 1004. The telescopic column 1004 drives the image sensor 400 to extend and retract according to the dust concentration.

[0039] Furthermore, an interlayer 1005 is provided on the primary screening bucket 101, and primary filter slots 1006 are provided in a linear array on the interlayer 1005. The construction waste remaining after a complete circle on the sieve plate 300 is input into the crushing bin 100 for secondary crushing.

[0040] Among them, the higher the dust concentration, the higher the height of the telescopic column 1004. At the same time, the focal length parameters of the image sensor 400 are adjusted to ensure that the image collection area does not change. The interlayer 1005 is set parallel to the primary screening bucket 101. The construction waste filtered below the interlayer 1005 is input into the sieve plate 300 for secondary screening, and the construction waste remaining above the interlayer 1005 enters the crushing bin 100 for secondary crushing.

[0041] In summary, when in use, the construction waste first enters the crushing bin 100 for pre-processing and crushing, and then falls into the primary screening bucket 101 for primary screening. At the same time, the lighter plastic and wood fragments are blown away by the airflow in the air nozzle 200. While oscillating, the construction waste fragments smaller than the primary filter slot 1006 fall under the interlayer 1005 and enter the sieve plate 300 for secondary screening. In this process, the overlap degree between the spreading plate 301 and the sieve hole 302 is adjusted to the preset size of the construction waste fragments. When the sieve plate 300 rotates, it is adjusted to the eccentric rotation mode to perform secondary screening on the construction waste fragments. After a complete rotation on the sieve plate 300, the remaining construction waste is input into the crushing bin 100 for secondary crushing, and the image information on the sieve plate 300 is collected by the image sensor 400.

[0042] Example 2 Reference Figure 1 to Figure 7 , which is the second embodiment of the present invention. This embodiment is based on the previous embodiment, but the difference is that the spreading plate 301 does not rotate but the sieve plate 300 rotates. After the sieve holes 302 are partially blocked by the construction waste, the guide block 3016 collides with the inner wall of the sieve plate 300 and vibrates, thereby vibrating the blocked construction waste to slide off.

[0043] Specifically, a dust cover 3002 is rotatably provided on the sieve plate 300 , and the spreading plate 301 includes scrapers 3003 distributed in an array and a connecting plate 3004 connecting the scrapers 3003 in series. The connecting plate 3004 slides through the inside of the dust cover 3002 and is provided with a roller 3005 .

[0044] An eccentric disk 3001 is provided at the center of the sieve disk 300 , a step layer 3006 is provided on the eccentric disk 3001 , and the roller 3005 slides against the step layer 3006 .

[0045] Among them, in this embodiment, the scraper 3003 is an arc-shaped plate and gradually increases from the center of the circle to the outside. The scraper 3003 itself is made of magnetic and wear-resistant metal material. The broken ferrous metal in the construction waste is collected and adsorbed, and the image is captured and identified through the image sensor 400, and then specially sorted and recycled.

[0046] Preferably, the sieve hole 302 itself is consistent in shape with the scraper 3003, and the two can be completely overlapped to close the sieve hole 302. In this embodiment, there are 5 spreader plates 301, and the corresponding sieve holes 302 are arranged in a circular array of 7 sets. When the spreader plate 301 and the sieve holes 302 rotate relative to each other, the construction waste on the sieve plate 300 is driven to move and smaller construction waste is pushed into the sieve holes 302.

[0047] Exemplarily, as a preferred solution, in other embodiments, the scraper 3003 can correspond to the number of sieve holes 302, and the sieve plate 300 and the spreader plate 301 are fixed by removable screws. In the initial state, the scraper 3003 and the sieve holes 302 completely overlap, resulting in the sealing of the sieve plate 300. By controlling the rotation of the roller 3005, the connecting plate 3004 and the scraper 3003 slide radially outward along the sieve plate 300, and a gap is generated between the scraper 3003 and the sieve holes 302. The greater the rotation of the roller 3005, the larger the gap, and the larger the size of the construction waste that can fall, thereby realizing the adjustment of the size of the secondary screening of construction waste. In this embodiment, two layers of sieve plates 300 and spreader plates 301 can be set, the second layer adopts the preferred solution, and the first layer adopts the solution of this embodiment. The combination of the two makes the filtering effect better.

[0048] More preferably, the end surface of the dust cover 3002 is connected to the telescopic column 1004 and remains stationary. As the roller 3005 rotates along with the rotating shaft and the sieve plate 300, the roller 3005 circulates up and down on the step layer 3006, thereby sliding back and forth along the dust cover 3002.

[0049] Furthermore, a slider 3007 is provided on the lower end surface of the eccentric disk 3001 , and a slide groove 3008 is slidably engaged on the outer wall of the slider 3007 . A slide rod 3009 is also passed through the slider 3007 , and a rotating shaft 3013 is fixedly provided on the outer wall of the slide groove 3008 .

[0050] Among them, a weight ball 3011 is provided at one end of the slide rod 3009 passing through the slide groove 3008, a second elastic member 3012 is provided between the weight ball 3011 and the slide block 3007, and a hub 3014 is provided between the rotating shaft 3013 and the outer wall of the sieve plate 300.

[0051] Furthermore, the slider 3007 is fixed at the center position of the eccentric disk 3001, and the second elastic member 3012 is a spring. In the initial state, the slide rod 3009, the slider 3007 and the counterweight ball 3011 are pulled to slide toward the center of the circle, and the slider 3007 deviates from the center position of the slide groove 3008 and remains stationary on one side of the inner wall of the slide groove 3008. At this time, when the rotating shaft 3013 rotates, it drives the slider 3007 to rotate eccentrically through the slide groove 3008.

[0052] A buffer column 3015 is slidably disposed at the end of the spreading plate 301 , a guide block 3016 is disposed at the end of the buffer column 3015 , and a third elastic member 3017 is disposed between the guide block 3016 and the buffer column 3015 .

[0053] Among them, the guide block 3016 is in the shape of an inclined arc, one end of which is in contact with the inner wall of the sieve plate 300, and the other end extends along the arc toward the center of the sieve plate 300, so that the moving construction waste is guided toward the center of the circle to prevent accumulation and caking on the surrounding walls of the circle and cause blockage.

[0054] Preferably, the third elastic member 3017 is a spring, which always pushes the guide block 3016 to slide radially in all directions until it contacts the inner wall of the sieve plate 300. When the eccentricity of the eccentric plate 3001 increases, the movement amplitude of the spreader plate 301 increases, and the guide block 3016 begins to collide with the inner wall of the sieve plate 300 to generate vibration.

[0055] Among them, when the rotating shaft 3013 drives the sieve plate 300 to work normally, the centrifugal force of the counterweight ball 3011 at the current speed drives the slide bar 3009 to slide outward, and friction deceleration occurs with the spreader plate 301, and the counterweight ball 3011 rebounds and resets, so that the slider 3007 disengages from the center position of the slide groove 3008 and starts to rotate eccentrically, causing vibration, which vibrates the blocked construction waste and slides it down, and at the same time the accumulated construction waste is flattened onto the sieve plate 300.

[0056] In summary, during normal operation, the rotation of the rotating shaft 3013 drives the eccentric disk 3001 to rotate concentrically therewith, and each spreading plate 301 moves back and forth, disturbing the construction waste through the scraper 3003 and returning the qualified construction waste to the sieve hole 302; when part of the construction waste is blocked in the sieve hole 302, the counterweight ball 3011 rebounds from the equilibrium position, so that the slider 3007 and the slide groove 3008 rotate eccentrically, and the guide block 3016 collides with the inner wall of the sieve plate 300 and vibrates, thereby vibrating the blocked construction waste to slide down, and at the same time, the accumulated construction waste is vibrated and rolled down and flattened onto the sieve plate 300, which is convenient for the image sensor 400 to collect information.

[0057] Example 3 Reference Figure 1-Figure 7, which is the third embodiment of the present invention. This embodiment is based on the previous embodiment, but the difference is that the recognition system can classify and recognize objects according to image features.

[0058] Specifically, the recognition system also includes a classification strategy, which is used to identify each feature block in the image and classify each feature block in the image, and at the same time identify the light transmittance of the sieve hole 302 and output the blockage degree value of the sieve hole 302.

[0059] Specifically, it includes: S1. Image preprocessing, converting the color image captured by the image sensor 400 into a grayscale image to reduce the complexity of image processing and reduce the interference caused by color information, then performing noise reduction on the grayscale image, extracting the image edge information, detecting the feature block contours of each object and recording them separately.

[0060] In this embodiment, only the objects on the spreading plate 301 are identified and classified. In other embodiments, the objects on the entire sieve plate 300 may also be classified.

[0061] Preferably, the extreme values ​​of the image are processed using a median filtering method, the outline of the object can be effectively detected using a Canny algorithm, each object is labeled and recorded separately, and the corresponding labeled objects are subsequently processed according to the classification results, the metal materials adsorbed on the flattening block are recycled, and the remaining materials are input into the crushing bin 100 for secondary crushing.

[0062] S2. Perform feature extraction and establish a database. At the same time, extract the geometric shape features and texture features of the feature blocks, record them in the object data with corresponding labels, store the extracted geometric shape features and texture features in a structured database, and attach a label of "metal" or "non-metal" to each sample.

[0063] Among them, the geometric shape features include the aspect ratio, roundness and rectangularity of the object, which comprehensively judge whether the shape of the object is a regular geometric shape. The texture features include contrast, homogeneity, entropy value and other indicators collected using the grayscale co-occurrence matrix. Whether the surface of the object is metal is judged based on whether the contrast is higher than the set judgment threshold and whether the entropy value is lower than the set value.

[0064] S3. Perform machine learning training, S3.1. Divide the dataset into training set, validation set and test set: a training set for model training, a validation set for adjusting hyperparameters and a test set for final model evaluation.

[0065] In this embodiment, the division ratio is 70% training set, 15% validation set, and 15% test set, and the proportion of metal and non-metal categories in each subset is made as consistent as possible to avoid deviations caused by data imbalance.

[0066] S3.2. Data standardization and normalization: standardize the geometric shape features and texture features so that the mean of each feature is 0 and the variance is 1.

[0067] S3.3. Construct a convolutional neural network model. The convolution layer can be used to extract local edge and texture space features. In each layer, multiple convolution kernels are used to extract different features. The activation function is selected, and the size of the feature map is reduced by maximum pooling to extract the high-level features of the image. The convolution and pooling layers are performed and the feature map is flattened and passed to the fully connected layer. The fully connected layer is used to map the high-dimensional features of the image to the classification results.

[0068] S3.4. Train the model. After setting the training parameters, input the training data into the model for training. Use the validation set to evaluate the model during the training process, adjust the parameters, and monitor the loss value or accuracy on the validation set. If there is no improvement after several consecutive rounds, stop training early.

[0069] The training parameter batch size is 64 to ensure the stability and speed of training. The learning rate is 0.0005 to prevent non-convergence caused by excessive learning rate. The number of training rounds is set to 20.

[0070] S4. Model evaluation and optimization: Evaluate the accuracy and loss function of the model on the test set to ensure that the model can generalize to unseen data.

[0071] Furthermore, cross-validation is used to further optimize the model and avoid overfitting.

[0072] S5. Model deployment and application: Save the trained model, load the trained model in actual applications for prediction, and use the model to predict the real-time collected image data in the production environment. By inputting the image into the CNN model, the prediction result "metal" or "non-metal" can be obtained.

[0073] Example 4 Reference Figure 1-Figure 7 , which is the fourth embodiment of the present invention, is based on the previous embodiment, but differs in that high-frequency vibration can also be performed by an execution unit.

[0074] Specifically, the recognition system also includes an execution unit. The recognition system can determine the degree of blockage of the sieve holes based on the light transmittance detection of the sieve holes, and output the corresponding value to the execution unit. The execution unit controls the eccentric disk 3001 to vibrate at a high frequency.

[0075] Among them, the rotation speed of the rotating shaft 3013 is controlled according to the degree of blockage of the sieve holes, so that the rotating shaft 3013 runs at an overspeed while the counterweight ball drives the slider 3007 to slide in the opposite direction, leaving the central position of the slide groove 3008, and rotating eccentrically at a high speed, thereby performing high-frequency oscillation to eliminate blockage.

[0076] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A construction waste recycling device based on an image sensor, characterized in that: The pre-treatment assembly comprises a crushing bin (100) and a primary screening bucket (101), wherein the primary screening bucket (101) performs reciprocating oscillation, and construction waste is initially crushed after entering the crushing bin (100), and is initially screened in the primary screening bucket (101); The air selection assembly comprises an air nozzle (200) arranged in the primary screening bucket (101), wherein the air nozzle (200) blows air when the primary screening bucket (101) oscillates; The secondary screening assembly comprises a sieve plate (300) rotatably arranged at the lower end of the primary screening bucket (101), wherein the rotation modes of the sieve plate (300) include concentric rotation and eccentric rotation, wherein an expansion plate (301) is movably arranged on the sieve plate (300), and sieve holes (302) are opened on the sieve plate (300), and when the sieve plate (300) and the expansion plate (301) rotate relative to each other, the maximum size of the construction waste that can be passed through the sieve holes (302) is adjusted, and the sieve plate (300) adjusts the degree of eccentric rotation according to the degree of blockage of the sieve holes (302); The image acquisition assembly comprises an image sensor (400) arranged above the sieve plate (300); the image sensor (400) is arranged at one end of the construction waste outlet on the sieve plate (300) and is used to acquire images of the construction waste on the sieve plate (300) in real time.

2. The construction waste recycling device based on image sensor according to claim 1, characterized in that: Two crushing rollers (1001) are arranged in the crushing bin (100), and the crushing rollers (1001) rotate in opposite directions. A first elastic member (1002) is arranged between the crushing bin (100) and the primary screening bucket (101), and a vibration motor (1003) is also arranged on the primary screening bucket (101); It also includes an identification system, which is used to receive images of construction waste and identify and classify them according to size parameters, and to determine the degree of blockage by comparing the light transmittance of the sieve holes (302), and to adjust the distance between the two crushing rollers (1001) according to the average size specification parameters of the construction waste.

3. The construction waste recycling device based on image sensor according to claim 2, characterized in that: A telescopic column (1004) is provided on the sieve plate (300), the other end of the telescopic column (1004) is connected to the image sensor (400), and a concentration sensor is also provided at the lower end of the telescopic column (1004). The telescopic column (1004) drives the image sensor (400) to extend and retract according to the dust concentration.

4. The construction waste recycling device based on image sensor as claimed in claim 3, characterized in that: The primary screening bucket (101) is provided with an interlayer (1005), and the interlayer (1005) is provided with primary filter slots (1006) in a linear array. The construction waste remaining after a complete rotation on the sieve plate (300) is input into the crushing bin (100) for secondary crushing.

5. The construction waste recycling device based on image sensor according to claim 4, characterized in that: The sieve plate (300) is rotatably provided with a dust cover (3002), the spreading plate (301) comprises scrapers (3003) distributed in an array and a connecting plate (3004) connecting the scrapers (3003) in series, the connecting plate (3004) slides through the inside of the dust cover (3002) and is provided with a roller (3005); An eccentric disk (3001) is provided at the center of the sieve disk (300), a step layer (3006) is provided on the eccentric disk (3001), and the roller (3005) slides in contact with the step layer (3006).

6. The construction waste recycling device based on image sensor according to claim 5, characterized in that: The lower end surface of the eccentric disk (3001) is provided with a slider (3007), the outer wall of the slider (3007) is slidably engaged with a slide groove (3008), the slider (3007) is also penetrated by a slide rod (3009), and the outer wall of the slide groove (3008) is fixedly provided with a rotating shaft (3013).

7. The construction waste recycling device based on image sensor according to claim 6, characterized in that: A weight ball (3011) is provided at one end of the slide rod (3009) passing through the slide groove (3008), a second elastic member (3012) is provided between the weight ball (3011) and the slide block (3007), and a hub (3014) is provided between the rotating shaft (3013) and the outer wall of the sieve plate (300).

8. The construction waste recycling device based on image sensor according to claim 7, characterized in that: A buffer column (3015) is slidably provided at the end of the expansion plate (301), a guide block (3016) is provided at the end of the buffer column (3015), and a third elastic member (3017) is provided between the guide block (3016) and the buffer column (3015).

9. The construction waste recycling device based on image sensor according to claim 8, characterized in that: The recognition system also includes a classification strategy, wherein the classification strategy is used to identify each feature block in the image and classify each feature block in the image, and simultaneously identify the light transmittance of the sieve hole (302) and output a blockage degree value of the sieve hole (302); The identification system also includes an execution unit. The identification system can determine the degree of blockage of the sieve holes based on the light transmittance detection of the sieve holes and output the corresponding value to the execution unit. The execution unit controls the eccentric disk (3001) to vibrate at a high frequency.

10. The construction waste recycling device based on image sensor according to claim 9, characterized in that: The classification strategy includes: S1. Image preprocessing, converting the color image collected by the image sensor (400) into a grayscale image to reduce the complexity of image processing and reduce the interference caused by color information, then performing noise reduction processing on the grayscale image, extracting image edge information, detecting the feature block contours of each object and recording them with labels respectively; S2. Extract features and establish a database. At the same time, extract the geometric shape features and texture features of the feature blocks, record them in the object data of the corresponding labels, store the extracted geometric shape features and texture features in a structured database, and attach a label of "metal" or "non-metal" to each sample; S3. Perform machine learning training; S3.

1. Divide the dataset into a training set, a validation set, and a test set: a training set for model training, a validation set for adjusting hyperparameters, and a test set for final model evaluation. S3.

2. Data standardization and normalization: standardization of geometric shape features and texture features; S3.

3. Construct a convolutional neural network model; S3.

4. Train the model. After setting the training parameters, input the training data into the model for training, and use the validation set to evaluate the model during the training process; S4. Model evaluation, based on the accuracy of the model evaluated on the test set; S5. Model deployment and application: Load the trained model in actual applications for prediction, and use the model to predict the real-time collected image data in the production environment. By inputting the image into the model, the prediction result is obtained.

Images