Search

CN-122024141-A - PET plastic bottle recycling system and method

CN122024141ACN 122024141 ACN122024141 ACN 122024141ACN-122024141-A

Abstract

The invention discloses a PET plastic bottle recycling system and method, and relates to the technical field of intelligent sorting. The method comprises the steps of monitoring the flowing state of bottles on a conveyor belt in real time through a sensor to generate a preliminary delay risk assessment sequence, screening high risk bottle data based on the sequence to determine a potential processing delay bottle list, extracting original images of the bottles in the list, calculating a preliminary material complexity score, marking high priority bottle images based on the score to form an adjusted processing queue, reordering the processing queue to obtain an optimized bottle processing sequence, verifying identification accuracy of first images in the sequence in an edge computing environment, and obtaining an enhanced bottle classification basis. According to the invention, by fusing the multi-source sensor data and the image characteristics, the dynamic optimization of the bottle processing priority and the accurate decision of the sorting path are realized, and the sorting efficiency and the accuracy of PET plastic bottle recycling under the complex working condition are effectively improved.

Inventors

  • WANG QIYOU
  • ZHANG TING
  • YANG HE
  • Zhu Qirun

Assignees

  • 成都环投循环科技有限公司

Dates

Publication Date
20260512
Application Date
20260203

Claims (9)

  1. 1. The PET plastic bottle recycling method is characterized by comprising the following steps of: Step 1, acquiring position data and a motion track of a bottle body on a conveyor belt in real time through a sensor, calculating the flow speed deviation of the bottle body and the interval time fluctuation of the bottle body, and judging whether the bottle body is in an abnormal flow state or not by combining the load distribution of the conveyor belt and the position offset of the bottle body to obtain a preliminary delay risk assessment sequence; Step 2, analyzing abnormal bottle movement track and bottle arrival frequency change according to a preliminary delay risk assessment sequence, screening out high-risk bottle data by combining sensor data noise interference and conveyor belt operation stability, and determining a potential processing delay bottle list; Step 3, extracting corresponding high-risk bottle original images from the potential processing delay bottle list, detecting the label surface contamination proportion and the bottle shape distortion degree, and calculating a preliminary material complexity score by combining the material surface texture roughness and the label edge fuzzy index; Step 4, aiming at the preliminary material complexity score, further analyzing the color distribution unevenness and material reflection gloss difference of the bottle, fusing the label character readability score and the bottle structural integrity index, and marking the bottle image as a high-priority bottle image if the comprehensive score is higher than a preset threshold value to obtain an adjusted processing queue; step 5, extracting bottle interval time fluctuation and bottle position offset characteristics of the high-priority bottle images according to the adjusted processing queue, and determining an optimized bottle processing sequence by combining the label surface contamination proportion and the bottle shape distortion degree and reordering the processing priority; Step 6, analyzing the surface texture roughness of materials and the uneven distribution characteristics of the color of the bottle body in an edge computing environment aiming at the optimized first image of the bottle body processing sequence, and if the analysis result is matched with a preset standard, confirming the identification accuracy to obtain the reinforced bottle body classification basis; And 7, according to the reinforced bottle classification basis, fusing the bottle structural integrity index and the label edge fuzzy index of the subsequent bottle image, and determining a final sorting path distribution instruction sequence by iteratively comparing the bottle flow speed deviation and the conveyor belt load distribution.
  2. 2. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 1 comprises: acquiring position information and motion trail data of a bottle body on a conveyor belt in real time through a sensor, and storing the position information and the motion trail data as an initial data set; Preprocessing an initial data set to remove noise interference and obtain cleaned position track data; According to the cleaned position track data, calculating the flow speed and interval time of the bottle body, and generating a speed distribution table and a time interval table; Adopting a preset threshold range, and marking as a speed abnormal point if the calculated flow speed exceeds the threshold range; if the interval time exceeds the preset range, marking the interval time as a time abnormal point to obtain an abnormal marking data set; Extracting speed abnormal points and time abnormal points from the abnormal mark data set, and analyzing the influence degree of load distribution on speed abnormality and time abnormality by combining the load distribution data of the conveyor belt to generate a load association analysis result; Calculating the position offset of the bottle body according to the load association analysis result, and generating an offset distribution table; If the position offset exceeds a preset offset threshold, judging that the position is in an abnormal state, and obtaining a position abnormal mark set; According to the position anomaly marking set, combining the speed anomaly point and the time anomaly point, and comprehensively evaluating the flow state of the bottle body; If the speed abnormality, the time abnormality and the position abnormality exist in the flowing state at the same time, judging that the flowing state is in a high-risk flowing state, and generating a flowing state risk level table; calculating a delay risk assessment sequence through a flow state risk level table, classifying risk levels by adopting a logistic regression model, and determining a final delay risk level sequence; And acquiring a delay risk level sequence, generating detailed analysis data of the bottle flow abnormality, and storing the detailed analysis data as a risk assessment file for subsequent flow tracking.
  3. 3. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 2 comprises: acquiring related data of a bottle motion path and an arrival frequency through a preliminary delay risk assessment sequence, and performing preliminary marking on abnormal fluctuation and variation conditions existing in the related data to obtain a bottle data set with abnormal marks; according to the bottle body data set with the abnormal mark, analyzing noise interference conditions in sensor data, and if the noise interference exceeds a preset threshold range, cleaning the data to obtain cleaned bottle body motion data; Aiming at the cleaned bottle motion data, analyzing the correlation between abnormal fluctuation and stability expression by combining with the stability expression of the operation of a conveyor belt, and marking the bottle as an operation abnormal bottle if the stability expression is lower than a preset standard to obtain an operation abnormal marking set; Extracting bottle data with high risk characteristics from the operation anomaly marking set, and comprehensively evaluating anomaly fluctuation and variation conditions by adopting a preset classification rule to obtain a high risk bottle classification result; According to the high-risk bottle classification result, the influence degree of abnormal fluctuation of the motion path and the arrival frequency on the processing delay is analyzed by combining the delay risk evaluation sequence, and a potential delay bottle list is determined; Acquiring a potential delay bottle list, and generating a priority sequence of bottle delay processing according to bottle data in the list and the running state of a conveyor belt by combining sensor data to obtain a processing priority sequence; And generating tracking identifiers of subsequent flows aiming at the bottle data related to the high risk characteristics and the processing delay through the processing priority sequence, and determining a final delay processing tracking list.
  4. 4. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 3 comprises: Acquiring original image data of a high-risk bottle body from a potential processing delay bottle body list, and dividing a label area in an image and the whole outline of the bottle body to obtain a separated label image and a bottle body outline image; detecting a distribution area of label offset by adopting a preset gray level analysis method aiming at the separated label image, and calculating an offset ratio value by combining the ratio of the offset area to the total area of the label to obtain a quantification result of the label offset; Analyzing the deviation condition of shape distortion by a geometric shape comparison method aiming at the separated bottle contour image, and if the deviation value exceeds a preset threshold range, marking the bottle contour image as distortion abnormality, and determining a distortion judgment result of the bottle shape; According to the original image of the bottle body, extracting texture features of the surface of the material, analyzing the roughness of the surface of the material by combining with a preset roughness evaluation standard, and if the roughness is higher than the preset standard, marking the surface as a high-roughness material to obtain evaluation data of texture of the material; Calculating an edge fuzzy index through an edge region of the label image, analyzing edge definition by combining a preset fuzzy detection tool, and marking as fuzzy abnormality if the definition is lower than a preset threshold value to obtain a quantitative index of edge fuzzy; Comprehensively generating a complexity score by adopting a weighted calculation method according to the quantization result of label offset, the distortion judgment result of the bottle shape, the evaluation data of the texture and the quantization index of the edge blurring, and determining the texture complexity classification result of the high-risk bottle; and generating a priority list of the bottle processing sequence according to the material complexity classification result and the bottle data in the delay list, and obtaining a final processing sequencing basis.
  5. 5. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 4 comprises: Acquiring original image data of the surface of the bottle body, and aiming at the color distribution condition of the surface of the bottle body, carrying out region segmentation by adopting a preset color analysis tool, and separating out the distribution ranges of different color regions to obtain preliminary classification data of color distribution; Analyzing the color contrast difference of each region according to the preliminary classification data of the color distribution, identifying the region range of uneven color distribution by a preset contrast detection method, and determining the specific region identification of uneven color distribution; Aiming at specific area marks with uneven color distribution, combining the reflective gloss characteristics of the surface of the bottle body, extracting the intensity distribution of gloss reflection by using a preset gloss detection tool, judging the difference area of the reflective gloss, and obtaining a distribution record of the gloss difference; analyzing the visible area of the label characters by combining the original image data of the surface of the bottle body through the distribution record of the gloss difference, and evaluating the readability degree of the label characters by adopting an image text recognition tool to obtain scoring data of the character readability; According to scoring data of the text readability, detecting the structural integrity by using a geometric analysis method aiming at the overall outline of the surface of the bottle body, if the detection result shows that the structural integrity is lower than a preset threshold range, marking the structural integrity as abnormal, and determining an evaluation result of the structural integrity; aiming at the evaluation result of the structural integrity, combining the region identification with uneven color distribution, the distribution record of the gloss difference and the scoring data of the text readability, and generating a comprehensive score by a weighted calculation method; In the specific implementation process, the degree of uneven color distribution is quantified through the standard deviation of color distribution, the difference of material reflection gloss is quantified through the deviation of reflection intensity, the character readability score is obtained based on the character recognition accuracy, and the structural integrity index of the bottle body is obtained through calculation through the integral outline defect proportion of the bottle body; The method comprises the steps of carrying out standardized processing on all indexes before entering comprehensive score calculation, uniformly mapping the numerical values of the indexes to a value interval of 0-1 so as to ensure comparability among different physical quantities, calculating to obtain a comprehensive score value based on standardized index values and corresponding weight parameters, judging that the bottle is a high-priority bottle when the comprehensive score value is higher than a preset threshold range, and generating priority ordering of a processing queue according to the high-priority bottle; According to the priority ordering of the processing queues, a processing sequence list of the bottle images is generated, the queue data is updated through a preset ordering tool, and the final processing sequence basis is determined.
  6. 6. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 5 comprises: Acquiring bottle interval data of high-priority bottle images, analyzing time fluctuation conditions corresponding to the bottle intervals by adopting a preset time recording tool, and determining distribution records of the bottle interval time fluctuation by comparing time interval differences of adjacent bottles; extracting relevant data of the position deviation of the bottle body according to the distribution record of the interval time fluctuation of the bottle body, and detecting the position deviation condition of the bottle body in the conveying process by using a preset positioning analysis tool to obtain a specific range mark of the position deviation of the bottle body; The method comprises the steps of obtaining the offset proportion information of the label surface aiming at a specific range mark of the position offset of a bottle body, separating an offset area and a normal area of the label surface through an image scanning tool, and judging the distribution condition of the offset proportion of the label; If the distribution condition of the label contamination proportion exceeds a preset threshold range, analyzing the distribution of the areas of the bottle shape distortion by adopting a geometric detection method in combination with the overall contour data of the bottle image, and determining the specific mark of the shape distortion; According to the specific mark of the shape distortion, combining the record of the interval time fluctuation and the position deviation of the bottle body, and integrating the priority orders in the processing queue by using a weighted calculation method to obtain a rearranged priority ordering list; And aiming at the rearranged priority ranking list, adjusting the sequence of the bottle processing sequence through a preset sequence updating tool, and determining optimized bottle processing sequence data.
  7. 7. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 6 comprises: aiming at the first image in the bottle sequence, acquiring the original data of the material surface by adopting a preset image acquisition tool, focusing on the initial information of texture roughness and color distribution, and obtaining a preliminary surface data record; According to the preliminary surface data record, an image processing tool is applied to separate and extract texture rough information of the material surface in an edge computing environment, and a specific distribution range of texture rough is determined; Aiming at a specific distribution range of rough texture, acquiring uneven data of color distribution through a preset color analysis module, and judging deviation conditions of the color distribution in different areas to obtain detailed marks of the color distribution; If the detailed mark display deviation of the color distribution exceeds a preset standard, combining the distribution range of the texture roughness, and adopting a support vector machine algorithm to comprehensively compare the two to determine the overall consistency evaluation result of the material surface; Obtaining matching degree information of the material surface and a preset standard according to the overall consistency evaluation result, and if the matching degree accords with a preset threshold value, confirming identification accuracy to obtain enhanced classification basis data; Aiming at the enhanced classification basis data, adjusting subsequent processing logic of the bottle sequence through a preset classification updating tool, and determining an optimized classification priority order; And rearranging the bottle sequence by adopting an automatic sequence adjustment module according to the optimized classification priority order to obtain a final processing sequence record.
  8. 8. The method for recycling PET plastic bottles as claimed in claim 1 wherein step 7 comprises: acquiring original data of a structural integrity index and an edge fuzzy index through a subsequent acquisition tool of the bottle body image, and determining a preliminary image analysis record; according to the preliminary image analysis record, a preset comparison module is adopted to carry out standardization processing on the structural integrity index and the edge fuzzy index, and standardized bottle body state data are obtained; acquiring real-time monitoring information of the flow speed deviation of the bottle body according to normalized bottle body state data, if the flow speed deviation exceeds a preset threshold value, smoothing the deviation information by adjusting a data filtering tool, and judging a corrected speed deviation range; acquiring dynamic data of load distribution of the conveyor belt according to the corrected speed deviation range, and if the difference of the load distribution in different areas exceeds a preset standard, recalculating the distributed data through a load balancing module to determine a balanced load distribution state; For the balanced load distribution state, combining the bottle classification basis and bottle state data, and carrying out preliminary division on the sorting paths by adopting a path distribution model to obtain an initial scheme of path distribution, wherein the path distribution model can be selected as a support vector machine model or a decision tree model; generating a corresponding allocation instruction sequence according to an initial scheme of path allocation, and finally confirming a sorting path through an instruction distribution tool to determine a complete path allocation instruction sequence; and aiming at the complete path allocation instruction sequence, real-time adjustment is carried out on the bottle sorting path through an automatic execution module, so that a final sorting execution record is obtained.
  9. 9. A PET plastic bottle recycling system for performing a PET plastic bottle recycling method according to any one of claims 1 to 8, comprising: The real-time acquisition and preliminary risk assessment module is used for acquiring position data and motion tracks of the bottles on the conveyor belt in real time through the sensor, calculating the flow speed deviation of the bottles and the interval time fluctuation of the bottles, and judging whether the bottles are in an abnormal flow state or not by combining the load distribution of the conveyor belt and the position offset of the bottles to obtain a preliminary delay risk assessment sequence; The high-risk bottle screening module is used for analyzing abnormal bottle movement track and bottle arrival frequency change aiming at the preliminary delay risk assessment sequence, screening out high-risk bottle data by combining sensor data noise interference and conveyor belt operation stability, and determining a potential processing delay bottle list; the material complexity preliminary scoring module is used for extracting corresponding high-risk bottle original images from the potential processing delay bottle list, detecting the label surface contamination proportion and the bottle shape distortion degree, and calculating a preliminary material complexity score by combining the material surface texture roughness and the label edge fuzzy index; the high-priority bottle marking module is used for further analyzing the color distribution unevenness and the material reflection gloss difference of the bottle aiming at the preliminary material complexity score, fusing the label character readability score and the bottle structural integrity index, and marking the bottle image as a high-priority bottle image if the comprehensive score is higher than a preset threshold value to obtain an adjusted processing queue; The processing priority reordering module is used for extracting bottle interval time fluctuation and bottle position offset characteristics of the high-priority bottle images according to the adjusted processing queue, and determining an optimized bottle processing sequence by combining the label surface contamination proportion and the bottle shape distortion degree and reordering the processing priority; The recognition accuracy enhancing module is used for analyzing the texture roughness of the material surface and the uneven distribution characteristics of the color of the bottle body in the edge computing environment aiming at the optimized bottle body processing sequence first image, and if the analysis result is matched with a preset standard, the recognition accuracy is confirmed, so that the enhanced bottle body classification basis is obtained; And the sorting path distribution module is used for fusing the bottle structure integrity index and the label edge fuzzy index of the subsequent bottle images according to the reinforced bottle classification basis, and determining a final sorting path distribution instruction sequence by iteratively comparing the bottle flow speed deviation and the conveyor belt load distribution.

Description

PET plastic bottle recycling system and method Technical Field The invention relates to the technical field of intelligent sorting, in particular to a PET plastic bottle recycling system and method. Background Under the background of modern environmental protection and resource recycling, efficient recycling of plastic bottles is an important link for resource recycling. PET plastic bottles are the most common beverage containers, and their recycling efficiency is directly related to environmental protection and economic benefits. At present, an image recognition technology is widely adopted on an industrial recycling line to automatically sort plastic bottles on a conveyor belt. In the prior art, a plastic bottle sorting method based on image recognition generally comprises the steps of collecting bottle images through a camera, recognizing bottle characteristics (such as labels, colors and shapes) through an image processing algorithm, and controlling a mechanical arm or an airflow nozzle to sort according to recognition results. However, these methods have the following technical limitations in practical applications: the arrival time, interval and position of the bottles on the conveyor belt are random, and the existing method is generally based on static image analysis, and does not consider real-time change of the flow state of the bottles. When the bottle body is abnormal in movement due to accumulation, sliding or vibration of a conveyor belt, the system cannot adjust the processing sequence in time, and sorting delay or omission is easily caused. In an actual recycling scene, the bottle body often has complex conditions such as label offset, shape distortion, rough surface texture, blurred edges and the like. In the existing method, a fixed threshold value or a simple classifier is mostly adopted, the complexity of materials is difficult to evaluate accurately, the identification accuracy is reduced, and the bottle body cannot be split in time. The existing system is usually processed sequentially according to the arrival sequence of the bottles, and real-time priority rearrangement is carried out on the flow state and the image characteristics of the uncombined bottles. When high-complexity bottles are mixed with standard bottles, the system cannot dynamically adjust the processing queue, so that the overall sorting efficiency is reduced, and even the conveyor belt is jammed. The existing method generally only depends on image characteristics to carry out sorting decisions, and real-time working condition data such as load distribution of a conveyor belt, flow speed deviation of a bottle body and the like are not fused, so that sorting path distribution is inaccurate, and the overall throughput rate of the system is affected. Therefore, how to dynamically optimize the bottle processing priority and sorting path according to the dynamic change of the arrival time of the bottle and the difference of the material complexity and the real-time fusion of the multi-source sensor data and the image characteristics becomes a key technical problem for improving the recovery efficiency and accuracy of the PET plastic bottle under the edge computing environment. Disclosure of Invention The invention aims to provide a PET plastic bottle recycling system and method, which solve the technical problems of fixed image recognition processing sequence, low sorting efficiency and easiness in delay and congestion caused by uncertain arrival time of bottles and difference of material complexity in a dynamic conveyor environment. In order to solve the technical problems, the invention adopts the following technical scheme: A PET plastic bottle recycling method comprises the following steps: Step 1, acquiring position data and a motion track of a bottle body on a conveyor belt in real time through a sensor, calculating the flow speed deviation of the bottle body and the interval time fluctuation of the bottle body, and judging whether the bottle body is in an abnormal flow state or not by combining the load distribution of the conveyor belt and the position offset of the bottle body to obtain a preliminary delay risk assessment sequence; Step 2, analyzing abnormal bottle movement track and bottle arrival frequency change according to a preliminary delay risk assessment sequence, screening out high-risk bottle data by combining sensor data noise interference and conveyor belt operation stability, and determining a potential processing delay bottle list; Step 3, extracting corresponding high-risk bottle original images from the potential processing delay bottle list, detecting the label surface contamination proportion and the bottle shape distortion degree, and calculating a preliminary material complexity score by combining the material surface texture roughness and the label edge fuzzy index; Step 4, aiming at the preliminary material complexity score, further analyzing the color distribution unevenness and material ref