A system for processing scrap automobile residues based on residue separation

Abstract

The application provides a kind of based on residue separation's scrap car residue processing system, relating to scrap car residue processing technical field, including overall pretreatment module, targeted pretreatment module, intelligent identification multi-path sorting module and adaptive process optimization module.The application introduces targeted pretreatment module, effectively solves the problem of material sticking caused by surface adhesion, creates necessary conditions for subsequent accurate sorting, significantly improves the overall sorting efficiency and purity;Through adaptive process optimization module, real-time sensing of incoming material composition fluctuation and dynamic matching of process parameters are realized, so that the system can continuously operate in the optimal state, improving the stability and economy of resource recovery.

Description

Technical Field

[0001] This invention belongs to the field of end-of-life vehicle residue treatment technology, specifically relating to an end-of-life vehicle residue processing system based on residue separation. Background Technology

[0002] With the increasing number of scrapped vehicles each year, the efficient recycling and utilization of the residue from scrapped vehicles has become an important issue. The residue remaining after dismantling, crushing, and preliminary magnetic separation of scrapped vehicles is extremely complex in composition, consisting of a mixture of plastics, rubber, fibers, residual metals, and silt. Currently, the industry typically uses a series of sorting devices based on a single physical principle (such as magnetic separation, air separation, eddy current separation, and near-infrared separation) connected in series to process this mixed residue.

[0003] However, existing processing technologies have significant drawbacks: First, the crushed residue contains a large number of material clumps caused by surface adhesives such as uncured adhesives, paint, and oil. These clumps are made up of fragments of different materials. In subsequent sorting processes based on single material identification, such as near-infrared sorting, they may be misjudged or missed, which seriously reduces the sorting purity and recovery rate. Secondly, due to the different sources, models, and years of scrapped vehicles, the proportion of residual components fluctuates greatly. Sorting systems with fixed process parameters are difficult to adapt to this fluctuation and cannot always maintain the optimal sorting state, resulting in unstable resource recycling efficiency.

[0004] Therefore, there is an urgent need for a system that can intelligently identify and remove material adhesion, and can adaptively adjust the sorting strategy to achieve refined and high-recovery-rate sorting of complex and variable scrapped vehicle residue. Summary of the Invention

[0005] This invention provides a waste automotive residue processing system based on residue separation to solve at least one of the aforementioned technical problems.

[0006] To address the aforementioned technical problems, this invention discloses a waste automotive residue processing system based on residue separation, comprising: The overall pretreatment module is used to homogenize and perform primary impurity removal on the incoming scrapped car debris to form a uniform primary mixture. The targeted preprocessing module is used to acquire and intelligently analyze the surface images of the primary mixture, identify the agglomerates formed by surface adhesive substances in the primary mixture, and generate and execute corresponding targeted preprocessing strategies based on the type and strength of the adhesive substances to process the primary mixture into a secondary mixture without agglomerates. The intelligent identification multi-path sorting module is used to separate secondary mixtures into single-material streams based on their material, density, electrical properties, and surface characteristics. The adaptive process optimization module is used to predict and dynamically adjust the sorting parameters of the intelligent identification multi-path sorting module based on the composition characteristics of the secondary mixture.

[0007] Preferably, the overall preprocessing module includes: The screening and soil removal submodule is used to screen out foreign objects and dust with particle sizes exceeding the preset particle size range in the crushed residue of scrapped cars based on particle size differences. The wind-powered pre-sorting submodule is used to separate crushed residues whose output mass is lower than the preset mass from the screening and soil removal submodule based on density differences. The crushing homogenization submodule is used to crush the crushing residue output from the wind pre-sorting submodule to a set particle size range, forming a uniform primary mixture.

[0008] Preferably, the targeted preprocessing module includes: The surface image acquisition unit is used to perform online image acquisition of the primary mixture on the conveyor belt connected to the output end of the overall preprocessing module, and to obtain surface image data of the primary mixture. The adhesion feature analysis unit is used to identify and locate adhesion clumps in the primary mixture based on surface image data, analyze the surface material type and adhesion strength level of the adhesion clumps, and generate targeted preprocessing instructions. The multi-strategy preprocessing execution unit is used to receive targeted preprocessing instructions and drive the corresponding preprocessing device to perform the unbinding operation on the adhering clumps.

[0009] Preferably, the multi-strategy preprocessing execution unit includes: The pneumatic intervention subunit is used to process adhesive clumps with an adhesion strength score lower than the first preset threshold and surface material type of oil stains. The pneumatic intervention subunit sprays compressed air flow onto the target adhesive clumps through a high-pressure nozzle with adjustable direction and pressure. The solvent atomizing spray subunit is used to treat adhesive clumps whose surface material is either paint-bonded or adhesive-bonded. The solvent atomizing spray subunit selects a matching solvent from the built-in solvent library according to the category label and sprays it onto the bonding interface of the target adhesive clump through the atomizing nozzle. The robot-assisted deionization unit is used to process adhesive clumps with an adhesion strength score higher than a second preset threshold or a physical size exceeding a preset volume. The robot-assisted deionization unit uses a vision-guided robotic arm to grasp the target adhesive clumps and transfer them to an independent low-speed shearing and crushing unit.

[0010] Preferably, the intelligent identification multi-channel sorting module includes: The metal deep recovery submodule is used to recover various metals from secondary mixtures in stages based on magnetic and eddy current effects, forming a metal enrichment stream; The near-infrared spectroscopy identification and sorting submodule is used to identify and sort different types of plastics from the secondary mixture remaining after the metal deep recovery submodule based on near-infrared spectral features, forming a classified plastic fragment stream; The triboelectric and electrostatic co-sorting submodule is used to further separate dark-colored or painted plastics and rubbers from the secondary mixture remaining after sorting by the near-infrared spectroscopy identification sorting submodule based on the difference in the triboelectric properties of the materials, forming a rubber particle stream and a mixed plastic stream. The density sorting submodule is used to separate heavy inorganic matter in the secondary mixture remaining after the triboelectric electrostatic collaborative sorting submodule based on density differences, forming a heavy inorganic material stream.

[0011] Preferably, the adaptive process optimization module includes: The historical operating condition database submodule is used to store the multidimensional component feature vectors of historical secondary mixtures, the historical sorting operation parameter combinations used by the system, and the historical sorting purity results of each output material stream. The clustering formula generation submodule is used to perform cluster analysis on the multidimensional component feature vectors of historical secondary mixtures to generate a process formula library. The process formula library contains multiple typical material component patterns and their corresponding optimal sorting operation parameter combinations and expected sorting target values. The online matching feedforward control submodule is used to obtain the real-time multidimensional component feature vector of the current secondary mixture, match the real-time multidimensional component feature vector with the typical material component patterns in the process formula library, and send the optimal sorting operation parameter combination corresponding to the matched typical material component pattern to the submodules corresponding to each intelligent identification multi-channel sorting module.

[0012] Preferably, the clustering recipe generation submodule includes: The feature extraction and standardization unit is used to read the multi-dimensional component feature vectors of historical secondary mixtures from the historical working condition database submodule, preprocess and extract features from each multi-dimensional component feature vector, and generate a standardized set of historical multi-dimensional feature vectors. The clustering analysis execution unit is used to perform clustering analysis on the standardized set of historical multidimensional feature vectors, determine the optimal number of classifications based on the silhouette coefficient of each historical multidimensional feature vector, and divide the entire set of historical multidimensional feature vectors into multiple clusters based on the optimal number of classifications; The formulation library construction unit is used to calculate the standard multidimensional feature vector of each cluster center as the typical material composition pattern corresponding to each cluster, and select one or more data points with the best historical sorting purity results from each cluster. The mean of the corresponding historical sorting operation parameter combination is set as the optimal sorting operation parameter combination for the corresponding typical material composition pattern, and the mean of the corresponding historical sorting purity results is set as the expected sorting target value for the corresponding typical material composition pattern, thereby completing the construction of the process formulation library.

[0013] Preferably, the online matching feedforward control submodule includes: The feature generation unit is used to obtain the real-time spectrum of the current secondary mixture by means of the multispectral scanning unit set at the main entrance of the intelligent identification multi-path sorting module, and to preprocess and extract features from the real-time spectrum to generate a real-time multidimensional feature vector. The pattern matching calculation unit is used to calculate the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formula library. The matching decision unit selects the most similar typical material composition pattern as the matching result and determines whether its similarity exceeds a preset similarity threshold. If it exceeds the preset similarity threshold, the optimal combination of sorting operation parameters corresponding to the most similar typical material composition pattern is sent to the control system of each sorting submodule; otherwise, the backup default combination of sorting operation parameters is used and sent to the control system of each sorting submodule, and an early warning is issued.

[0014] Preferably, the pattern matching calculation unit calculates the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material component pattern in the process formulation library: Real-time multidimensional feature vectors are The first in the process formula library The standard multidimensional feature vector of a typical material composition pattern is: The similarity between the two The calculation formula is: in, The total dimension of the multidimensional feature vector is denoted as , and its value is . ; For the preset first Weights of dimensional features; Real-time multidimensional feature vector In the The numerical value of the dimension; For the first Standard multidimensional feature vectors In the The numerical value of the dimension; For real-time feature vectors and the first The similarity between the standard feature vectors of each pattern.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) By introducing a targeted pretreatment module, the present invention effectively solves the problem of material adhesion caused by surface adhering substances, creating the necessary conditions for subsequent accurate sorting and significantly improving the overall sorting efficiency and purity; (2) The present invention realizes real-time perception of fluctuations in the composition of incoming materials and dynamic matching of process parameters through an adaptive process optimization module, enabling the system to continuously operate in the optimal state and improving the stability and economy of resource recovery. (3) This invention organically combines overall pretreatment, targeted dissociation, multi-path intelligent sorting and adaptive optimization to form a complete and intelligent solution for the deep resource utilization of scrapped automobile residues; (4) The modules of this invention work together to achieve high purity and high recovery rate separation of high-value components such as metals, plastics and rubber in the crushed residue of scrapped cars. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the waste automotive residue processing system of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] Furthermore, in this invention, the use of terms such as "first" and "second" is for descriptive purposes only and does not specifically refer to any order or sequence, nor is it intended to limit the invention. They are merely used to distinguish components or operations described using the same technical terms and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions and features of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If a combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0019] The present invention provides the following embodiments. Example 1 This invention provides a waste automotive residue processing system based on residue separation, such as... Figure 1 As shown, it includes: The overall pretreatment module is used to homogenize and perform primary impurity removal on the incoming scrapped car debris to form a uniform primary mixture. The targeted preprocessing module is used to acquire and intelligently analyze the surface images of the primary mixture, identify the agglomerates formed by surface adhesive substances in the primary mixture, and generate and execute corresponding targeted preprocessing strategies based on the type and strength of the adhesive substances to process the primary mixture into a secondary mixture without agglomerates. The intelligent identification multi-path sorting module is used to separate secondary mixtures into single-material streams based on their material, density, electrical properties, and surface characteristics. The adaptive process optimization module is used to predict and dynamically adjust the sorting parameters of the intelligent identification multi-path sorting module based on the composition characteristics of the secondary mixture.

[0020] In this embodiment, the scrapped vehicle residue includes a mixture of plastics, rubber, fibrous fabrics, residual metal shavings, glass, and silt; after homogenization and primary impurity removal, the particle size of the material is mainly distributed in the range of 20-100 mm.

[0021] In this embodiment, the material flow of a single material includes at least a plastic fragment flow, a rubber particle flow, a metal enrichment flow, and a fiber enrichment flow.

[0022] In this embodiment, the adaptive process optimization module is used to predict and dynamically adjust the sorting parameters of the intelligent identification multi-path sorting module based on the composition characteristics of the secondary mixture. The adjusted sorting parameters include the operating parameters of the metal depth recovery sub-module, the near-infrared spectral identification sorting sub-module, the triboelectric electrostatic collaborative sorting sub-module, and the density sorting module.

[0023] The working principle and beneficial effects of the above technical solution are as follows: During operation, the crushed residue of scrapped vehicles first enters the overall pretreatment module, where it is homogenized into a primary mixture with a concentrated particle size range after screening, air separation, and crushing. The primary mixture then enters the targeted pretreatment module, where online image acquisition and intelligent analysis accurately identify agglomerates caused by paint, oil stains, adhesives, etc., and specifically use methods such as airflow purging, solvent spraying, or mechanical dissociation to break them up, resulting in a well-dispersed secondary mixture. The secondary mixture is then sent to the intelligent identification multi-channel sorting module, where it undergoes multiple processes such as magnetic separation, eddy current, near-infrared spectroscopy, electrostatic triboelectric separation, and density separation, gradually separating it into material streams of single materials such as metals, different types of plastics, and rubber. At the same time, the adaptive process optimization module analyzes the composition of the secondary mixture online, matches the optimal historical process formula in real time, and feeds forward to adjust the operating parameters of each sorting sub-module, so that the entire sorting process always adapts to changes in material characteristics. This invention is the first to integrate an intelligent targeted pretreatment process for material adhesion in the treatment of scrapped vehicle residue, fundamentally solving the problems of material misidentification and low sorting efficiency caused by adhesion. Simultaneously, the system possesses self-sensing, self-decision-making, and self-optimization capabilities, dynamically adapting to complex feed changes and overcoming the bottleneck of efficiency decline in traditional fixed-process sorting systems when facing composition fluctuations.

[0024] Example 2 Based on Example 1, the overall preprocessing module includes: The screening and soil removal submodule is used to screen out foreign objects and dust with particle sizes exceeding the preset particle size range in the crushed residue of scrapped cars based on particle size differences. The wind-powered pre-sorting submodule is used to separate crushed residues whose output mass is lower than the preset mass from the screening and soil removal submodule based on density differences. The crushing homogenization submodule is used to crush the crushing residue output from the wind pre-sorting submodule to a set particle size range, forming a uniform primary mixture.

[0025] In this embodiment, the screening and soil removal submodule includes a coarse screening unit and a fine screening unit. The coarse screening unit is used to screen out crushed residues with a particle size larger than the maximum value of the preset particle size range, and the fine screening unit is used to screen out crushed residues with a particle size smaller than the minimum value of the preset particle size range. The coarse screening unit can be, for example, a drum screen or a bar screen; the fine screening unit can be, for example, a vibrating screen or a tension screen.

[0026] Assuming the preset particle size range is 10-100 mm; The coarse screening unit is used to separate materials larger than 100 mm. The fine screening unit is used to remove dust and dirt smaller than 10 mm.

[0027] In this embodiment, the wind-powered pre-sorting submodule includes a suction separation unit for separating and collecting foam, film and fiber flocs based on the principle of negative pressure adsorption; the suction separation unit can be a combination of a negative pressure suction machine and a cyclone separator.

[0028] In this embodiment, the crushing homogenization submodule includes a low-speed shear crushing unit for crushing materials to a homogeneous size of 20-50 mm using shear force as the primary method; the low-speed shear crushing unit can be, for example, a twin-shaft shear crusher.

[0029] The working principle and beneficial effects of the above technical solution are as follows: The incoming material first passes through a screening and soil removal sub-module. The coarse screening unit removes oversized foreign objects, such as larger pieces that are not completely crushed, while the fine screening unit removes fine impurities such as mud and dust, thus completing primary impurity removal. The screened material then enters a pneumatic pre-sorting sub-module. Under negative pressure airflow, extremely lightweight materials such as foam, film, and fiber flocs are sucked away and separated, achieving density-based pre-enrichment. Finally, the remaining main material enters a crushing and homogenizing sub-module, where a low-speed shear crusher breaks it down to a homogeneous particle size of 20-50 mm, forming the primary mixture necessary for subsequent processing. The overall pretreatment module designed in this invention adopts a three-stage progressive process of screening, air separation, and crushing, which is highly efficient. Screening effectively removes interfering impurities, air separation separates light interfering substances in advance, reducing the burden on the subsequent main separation stage, and finally, homogeneous crushing based on shear force obtains materials of suitable particle size while minimizing internal damage or thermal aging of materials caused by impact crushing.

[0030] Example 3 Based on Example 1, the targeted preprocessing module includes: The surface image acquisition unit is used to perform online image acquisition of the primary mixture on the conveyor belt connected to the output end of the overall preprocessing module, and to obtain surface image data of the primary mixture. The adhesion feature analysis unit is used to identify and locate adhesion clumps in the primary mixture based on surface image data, analyze the surface material type and adhesion strength level of the adhesion clumps, and generate targeted preprocessing instructions. The multi-strategy preprocessing execution unit is used to receive targeted preprocessing instructions and drive the corresponding preprocessing device to perform the unbinding operation on the adhering clumps.

[0031] In this embodiment, surface adhesive substances refer to substances that adhere to the surface of fragments of scrapped vehicle debris and can cause different fragments to stick together. These mainly include incompletely cured adhesives, paints, lubricating greases, and plasticizer precipitates.

[0032] In this embodiment, the surface image acquisition unit includes a high-resolution industrial camera and a light source system with a specific lighting scheme. The high-resolution industrial camera acquires high-definition color or near-infrared images of the surface of the primary mixture at a fixed frequency.

[0033] In this embodiment, the adhesion feature analysis unit has a built-in trained deep learning recognition model. This deep learning recognition model is used to process surface image data, and the processing output includes the bounding box position of the adhesion clumps, the surface material category label, and the adhesion strength score. The training sample library of the deep learning recognition model contains a large number of labeled primary mixture surface images, and the category labels include "paint adhesion", "oil stain adhesion", "adhesive adhesion" and "other adhesion".

[0034] In this embodiment, the multi-strategy preprocessing execution unit includes a variety of physical or chemical preprocessing devices, which are selectively activated based on the category labels and adhesion strength scores output by the adhesion feature analysis unit.

[0035] The working principle and beneficial effects of the above technical solution are as follows: When the primary mixture passes through the conveyor belt, the surface image acquisition unit continuously captures images of its surface to obtain high-definition image data; the deep learning model in the adhesion feature analysis unit analyzes these images in real time, accurately identifies the adhesive clumps in the images, and determines whether the surface is paint, oil, or adhesive, while assessing the tightness of the adhesion; subsequently, the analysis results are converted into control commands and sent to the multi-strategy preprocessing execution unit. The multi-strategy preprocessing execution unit intelligently selects and activates the corresponding preprocessing device according to the commands. For example, high-pressure air blowing is used for oil adhesion, specific solvent spraying is activated for paint adhesion, and robotic arms are dispatched to grab and transfer the highly adhesive clumps to an independent workstation for processing. This invention creatively applies machine vision and deep learning technologies to the online diagnosis of residue adhesion, enabling the identification of adhesion problems. Unlike traditional, blind, and uniform pretreatment methods, this invention can distinguish the cause and intensity of adhesion and implement differentiated processing strategies. This intelligent targeted pretreatment mode effectively removes adhesion and ensures the accuracy of subsequent sorting while minimizing unnecessary processing or damage to non-adhesive materials or the material matrix, significantly improving the accuracy and economy of the pretreatment process.

[0036] Example 4 Based on Example 3, the multi-strategy preprocessing execution unit includes: The pneumatic intervention subunit is used to process adhesive clumps with an adhesion strength score lower than the first preset threshold and surface material type of oil stains. The pneumatic intervention subunit sprays compressed air flow onto the target adhesive clumps through a high-pressure nozzle with adjustable direction and pressure. The solvent atomizing spray subunit is used to treat adhesive clumps whose surface material is either paint-bonded or adhesive-bonded. The solvent atomizing spray subunit selects a matching solvent from the built-in solvent library according to the category label and sprays it onto the bonding interface of the target adhesive clump through the atomizing nozzle. The robot-assisted deionization unit is used to process adhesive clumps with an adhesion strength score higher than a second preset threshold or a physical size exceeding a preset volume. The robot-assisted deionization unit uses a vision-guided robotic arm to grasp the target adhesive clumps and transfer them to an independent low-speed shearing and crushing unit.

[0037] In this embodiment, the first preset threshold and the second preset threshold are values ​​set by the system based on historical processing data, and the second preset threshold is greater than the first preset threshold.

[0038] In this embodiment, the independent low-speed shearing and crushing unit has a separate discharge port. The material after being separated by it is transported back to the conveyor belt upstream of the surface image acquisition unit for a new round of identification and sorting cycle.

[0039] The working principle and beneficial effects of the above technical solution are as follows: The multi-strategy pretreatment execution unit performs specific operations according to the type of instruction received: For slight oil stains, the precise directional high-pressure airflow of the pneumatic intervention subunit can blow them apart; for paint or adhesives, the solvent atomization spray subunit will precisely spray out a small amount of atomized solvent, causing a dissolution or swelling reaction at the adhesion interface, weakening the bonding force; for those large or abnormally firmly bonded clumps, the robot-assisted deionization unit will grab them and send them to an independent low-speed shearing and crushing unit for forced and gentle dissociation, and the dissociated material is returned to the main process for reprocessing; The multi-strategy execution mechanism designed in this invention embodies a deep match between processing methods and problem characteristics. The pneumatic approach is energy-saving and environmentally friendly, suitable for simple adhesions; micro-solvent spraying enables precise and low-consumption application of chemical energy; and robot assistance solves extremely difficult cases. This not only significantly improves the success rate and efficiency of adhesion removal but also minimizes the negative impact of the dissociation process on the material's properties, such as plastic molecular chains, ensuring the quality of recycled materials.

[0040] Example 5 Based on Example 1, the intelligent identification multi-path sorting module includes: The metal deep recovery submodule is used to recover various metals from secondary mixtures in stages based on magnetic and eddy current effects, forming a metal enrichment stream; The near-infrared spectroscopy identification and sorting submodule is used to identify and sort different types of plastics from the secondary mixture remaining after the metal deep recovery submodule based on near-infrared spectral features, forming a classified plastic fragment stream; The triboelectric and electrostatic co-sorting submodule is used to further separate dark-colored or painted plastics and rubbers from the secondary mixture remaining after sorting by the near-infrared spectroscopy identification sorting submodule based on the difference in the triboelectric properties of the materials, forming a rubber particle stream and a mixed plastic stream. The density sorting submodule is used to separate heavy inorganic matter in the secondary mixture remaining after the triboelectric electrostatic collaborative sorting submodule based on density differences, forming a heavy inorganic material stream.

[0041] In this embodiment, the metal deep recovery submodule includes a high-intensity magnetic separation unit, an eddy current separation unit, and a metal sensing ejection unit arranged in series; wherein: The strong magnetic separation unit is used to adsorb and separate ferromagnetic metals from the secondary mixture and incorporate them into the metal enrichment stream; Eddy current separation unit is used to separate the remaining secondary mixture after the strong magnetic separation unit. Based on the eddy current effect, non-ferrous metals are ejected and separated and merged into the metal enrichment stream. The metal sensing ejection unit is used to detect the remaining secondary mixture after the eddy current separation unit. It uses sensors to identify and high-pressure air knife ejection to remove the residual metal pieces and then feeds them into the metal enrichment stream.

[0042] In this embodiment, the strong magnetic separation unit is a permanent magnet drum or an electromagnetic pulley; the eddy current separation unit is a high-speed rotating permanent magnet drum separator; and the metal sensing ejection unit includes a combination of an inductive sensor and a high-pressure air valve. The near-infrared spectral identification and sorting submodule includes a high-speed online spectral scanning unit and an array-type high-pressure pneumatic jet actuator; wherein: The high-speed online spectral scanning unit is used to scan the secondary mixture after the metal deep recycling submodule in real time, obtain its near-infrared spectral data, and compare and analyze it with the built-in material spectral database to identify different types of plastics. The array-type high-pressure pneumatic jetting actuator triggers the corresponding high-pressure nozzle at a specific position on the material flight trajectory based on the identification results of the high-speed online spectral scanning unit. This blows the identified target plastic fragments out of the main material flow, forming a classified plastic fragment flow. The material that is not blown out continues to enter the subsequent sorting stage along the main material flow. The high-speed online spectral scanning unit is a linear array scanning near-infrared spectrometer; the array-type high-pressure pneumatic jet actuator is a multi-nozzle array controlled by solenoid valves.

[0043] In this embodiment, the triboelectric electrostatic collaborative sorting submodule includes a vibration triboelectric charging unit and a high-voltage electrostatic roller sorting unit; The vibration-triboelectric charging unit is used to cause the remaining secondary mixture from the near-infrared spectroscopy identification and sorting submodule to be charged with different polarities or intensities of material fragments on its polytetrafluoroethylene-lined bed surface through vibration, collision and friction. The high-voltage electrostatic roller sorting unit receives materials that have been tactilely charged. The materials move along different trajectories due to the difference in charge in the high-voltage electrostatic field, causing rubber particles to deflect to one side and fall into the rubber particle stream collection bin, while the remaining plastic and other materials deflect to the other side to form a mixed plastic stream. The vibration-friction charging unit is a vibratory feeder lined with polytetrafluoroethylene; the high-voltage electrostatic roller sorting unit includes a high-voltage electrode and a grounded rotating roller.

[0044] In this embodiment, the density sorting submodule includes a hydrocyclone unit; The hydrocyclone unit is used to separate the remaining secondary mixture (mainly mixed plastic stream and residual inorganic matter) from the triboelectric electrostatic co-sorting submodule based on density differences; In the high-speed rotating water flow inside the hydrocyclone, heavy inorganic materials with a density greater than the set value (such as glass and sand) are thrown against the wall of the hydrocyclone and sink, and are discharged from the bottom outlet, forming a heavy inorganic material stream. The hydrocyclone unit is a hydrocyclone, and its separation density threshold is controlled by adjusting the inlet pressure and the underflow outlet size.

[0045] The working principle and beneficial effects of the above technical solution are as follows: After pretreatment, the secondary mixture first enters the metal deep recovery submodule, where ferromagnetic metals are adsorbed by strong magnetic separation, non-ferrous metals are ejected by eddy current separation, and residual metal pieces are removed by the metal sensing ejection unit, thus achieving deep metal recovery. The material after metal removal enters the near-infrared spectral identification and sorting submodule, where it is spectrally scanned on a high-speed conveyor belt. The system identifies different plastics such as PP and ABS in real time and uses high-pressure air nozzles to accurately blow them into the corresponding collection bins. The remaining material enters the triboelectric electrostatic collaborative sorting submodule, where the material becomes charged through vibration and friction. After passing through a high-voltage electrostatic field, the rubber and plastic separate due to the difference in charge. Finally, the remaining light mixture enters the hydrocyclone of the density sorting submodule, where the heavy inorganic matter is separated out under centrifugal force. This invention integrates multiple physical sorting technologies and optimizes their sequential arrangement according to the logical order of metal priority, plastic fine separation, rubber extraction, and heavy material rejection, forming a highly efficient sorting chain. Deep metal recovery employs a three-stage separation process to maximize metal recovery rate; near-infrared sorting achieves refined separation of plastic types; triboelectric sorting effectively solves the spectral identification problem of dark-colored materials; and final density sorting ensures product purity. This multi-technology collaborative and streamlined sorting design fully leverages the advantages of each technology, creating a complementary effect, ultimately achieving efficient and high-purity separation and enrichment of multiple valuable components in complex mixtures.

[0046] Example 6 Based on Example 1, the adaptive process optimization module includes: The historical operating condition database submodule is used to store the multidimensional component feature vectors of historical secondary mixtures, the historical sorting operation parameter combinations used by the system, and the historical sorting purity results of each output material stream. The clustering formula generation submodule is used to perform cluster analysis on the multidimensional component feature vectors of historical secondary mixtures to generate a process formula library. The process formula library contains multiple typical material component patterns and their corresponding optimal sorting operation parameter combinations and expected sorting target values. The online matching feedforward control submodule is used to obtain the real-time multidimensional component feature vector of the current secondary mixture, match the real-time multidimensional component feature vector with the typical material component patterns in the process formula library, and send the optimal sorting operation parameter combination corresponding to the matched typical material component pattern to the submodules corresponding to each intelligent identification multi-channel sorting module.

[0047] In this embodiment, the multidimensional component feature vector specifically includes the numerical values ​​of the following dimensions: : The intensity or concentration prediction of the characteristic spectral absorption peak of polypropylene (PP), expressed as a percentage; : The intensity or concentration prediction of the characteristic spectral absorption peak of acrylonitrile-butadiene-styrene copolymer (ABS), with the concentration prediction expressed as a percentage; : The intensity or concentration prediction value corresponding to the characteristic spectral absorption peak of polyvinyl chloride (PVC), with the concentration prediction value expressed as a percentage; : The secondary mixture of materials in a specific wavelength band of visible light, such as Average reflectance and near-infrared band The ratio of average reflectance is used as an indicator of the likelihood of the presence of a metal. Material in the shortwave infrared band The average radiation intensity serves as an auxiliary indicator of metal content; The average moisture content of the material is predicted by near-infrared spectroscopy analysis, and the average moisture content of the material is expressed as a percentage.

[0048] In this embodiment, the combination of sorting operation parameters includes: the magnetic separation intensity and eddy current separator speed of the metal deep recovery submodule, the airflow velocity of the near-infrared spectroscopy identification sorting submodule, the voltage value and electrode spacing of the triboelectric electrostatic synergistic sorting submodule, and the water flow velocity of the density sorting submodule.

[0049] In this embodiment, each output material flow refers to the metal enrichment flow, classified plastic fragment flow, rubber particle flow, heavy inorganic flow, and mixed plastic flow formed after sorting by the intelligent identification multi-channel sorting module.

[0050] In this embodiment, the historical sorting purity result refers to the actual quantitative evaluation index obtained by sampling and laboratory analysis of the final products of each discharge port after the system has processed a batch of historical secondary mixture materials, namely the metal enrichment stream, the classified plastic fragment stream, the rubber particle stream, the heavy inorganic stream and the mixed plastic stream. Specifically, this includes: the percentage of ferromagnetic metal recovery in the metal enrichment stream, and the percentage of non-ferrous metal recovery; the percentage of target plastic types such as polypropylene and acrylonitrile-butadiene-styrene copolymer in the classified plastic debris stream, and the percentage of non-target plastics such as impurities; the percentage of rubber such as EPDM rubber in the rubber particle stream; the percentage of the total mass of target inorganic materials such as glass and sand in the heavy inorganic stream; and the sorting purity of the mixed plastic stream, which is usually treated as low-value or unclassified residues, can be expressed by its yield, i.e., its proportion of the total feed mass or the proportion of its main components.

[0051] In this embodiment, the expected sorting target value refers to the standard value of sorting purity of each output material stream that is pre-set for each typical material composition mode in the process formula library and is expected to be achieved by the optimal combination of sorting operation parameters. It is a target set based on historical best data, rather than actual measurement results.

[0052] In this embodiment, the typical material composition pattern is a material category with similar composition characteristics summarized from historical data by the clustering formula generation submodule. Each pattern is defined by a standard multidimensional feature vector.

[0053] In this embodiment, both the multidimensional component feature vector and the real-time multidimensional component feature vector refer to the set of quantitative data that characterizes the content of various chemical components in the material, obtained through spectral analysis; specifically, they are obtained by a dedicated multispectral scanning unit set at the main entrance of the intelligent identification multi-path sorting module, i.e., before the metal deep recovery submodule.

[0054] The working principle and beneficial effects of the above technical solution are as follows: The multispectral scanning unit continuously analyzes the components of the incoming material to form a real-time multidimensional feature vector. The online matching feedforward control submodule performs rapid similarity matching between this vector and various typical material patterns pre-stored in the process formula library. Once the most similar pattern is matched, the system immediately calls the optimal set of sorting equipment parameters corresponding to this pattern, such as magnetic separation intensity, airflow velocity, and electrostatic voltage, and sends these parameters to the control system of each sorting submodule in advance. At the same time, the component data, parameters used, and final product purity results are stored in the database as a new set of historical operating conditions for optimization and enrichment of the formula library. This invention overcomes the limitations of traditional residue sorting systems, which suffer from fixed parameters and poor adaptability, achieving a leap from fixed processes to flexible, formula-based processes. This component-aware, feedforward dynamic optimization ensures that the system always operates close to its historical optimal state when facing complex and ever-changing feeds, thereby stably outputting high-purity products and greatly improving the efficiency, effectiveness, and stability of the entire resource recovery process.

[0055] Example 7 Based on Example 6, the clustering recipe generation submodule includes: The feature extraction and standardization unit is used to read the multi-dimensional component feature vectors of historical secondary mixtures from the historical working condition database submodule, preprocess and extract features from each multi-dimensional component feature vector, and generate a standardized set of historical multi-dimensional feature vectors. The clustering analysis execution unit is used to perform clustering analysis on the standardized set of historical multidimensional feature vectors, determine the optimal number of classifications based on the silhouette coefficient of each historical multidimensional feature vector, and divide the entire set of historical multidimensional feature vectors into multiple clusters based on the optimal number of classifications; The formulation library construction unit is used to calculate the standard multidimensional feature vector of each cluster center as the typical material composition pattern corresponding to each cluster, and select one or more data points with the best historical sorting purity results from each cluster. The mean of the corresponding historical sorting operation parameter combination is set as the optimal sorting operation parameter combination for the corresponding typical material composition pattern, and the mean of the corresponding historical sorting purity results is set as the expected sorting target value for the corresponding typical material composition pattern, thereby completing the construction of the process formulation library.

[0056] In this embodiment, the clustering analysis algorithm adopts Algorithm, Optimal Number of Classes By calculating different presets value( from arrive The average silhouette coefficient of all historical multidimensional feature vectors under ) Determine, select The largest Value as the optimal number of categories The silhouette coefficient measures the compactness of samples within the same cluster and the separation between samples in different clusters.

[0057] In this embodiment, each cluster center is the arithmetic mean of all historical multidimensional feature vectors within that cluster across all dimensions; specifically, for a cluster containing... indivual Clusters of 1D eigenvectors Its cluster center vector The The method for calculating the dimension is as follows: in, Cluster The Middle The first historical multidimensional feature vector Dimensional value.

[0058] In this embodiment, the optimal sorting purity result among one or more data points with the best historical sorting purity results selected from each cluster is defined as: the highest comprehensive score calculated according to a preset evaluation function for the historical sorting purity results of each output material stream corresponding to that historical operating condition data point; the evaluation function can be a weighted sum of various key purity indicators, for example: Select the data with the highest overall score, or the average of the top-ranked data.

[0059] In this embodiment, typical material composition patterns summarized by cluster analysis include: Mode A (high PP, medium metal content, low moisture content): Its standard multidimensional feature vector is characterized as follows: High value (high polypropylene content). The value is moderate (medium metal content). Low value (low moisture content) , , The value is within a specific range; The corresponding optimal combination of sorting operation parameters is: ; The corresponding expected sorting target value is: .

[0060] Mode B (high ABS and glass, low metal): Its standard multidimensional feature vector is characterized as follows: High value (high ABS content). High value (high content of inorganic substances such as glass). Low value (low metal content). , , The value is within a specific range; The corresponding optimal combination of sorting operation parameters is: ; Its corresponding expected sorting target value is: .

[0061] The working principle and beneficial effects of the above technical solution are as follows: First, the material composition characteristics are standardized. Then, using clustering algorithms such as K-Means, hidden patterns in historical data are automatically discovered, classifying thousands of materials with different compositions into a few to a dozen representative typical material composition patterns, such as "high PP material" and "high ABS material." For each classified pattern, the system identifies the few records with the best sorting results when processing this type of material in history, takes the average of the process parameters used in these records as the optimal combination of sorting parameters for that pattern, and uses the achieved sorting effect as the expected sorting target value. Finally, all patterns, their corresponding parameter combinations, and target values ​​together constitute an online queryable process formula library. This invention uses unsupervised machine learning methods to enable the system to automatically extract core knowledge from massive amounts of operational data, avoiding the subjectivity and limitations of relying on human experience and making the division of typical patterns and the determination of optimal parameters more scientific, objective, and data-driven.

[0062] Example 8 Based on Example 6, the online matching feedforward control submodule includes: The feature generation unit is used to obtain the real-time spectrum of the current secondary mixture by means of the multispectral scanning unit set at the main entrance of the intelligent identification multi-path sorting module, and to preprocess and extract features from the real-time spectrum to generate a real-time multidimensional feature vector. The pattern matching calculation unit is used to calculate the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formula library. The matching decision unit selects the most similar typical material composition pattern as the matching result and determines whether its similarity exceeds a preset similarity threshold. If it exceeds the preset similarity threshold, the optimal combination of sorting operation parameters corresponding to the most similar typical material composition pattern is sent to the control system of each sorting submodule; otherwise, the backup default combination of sorting operation parameters is used and sent to the control system of each sorting submodule, and an early warning is issued.

[0063] In this embodiment, the preset similarity threshold is: .

[0064] In this embodiment, the backup default sorting operation parameter combination is a set of validated, conservative process parameters applicable to most material conditions. Its design principle is to prioritize system operation stability and equipment safety while ensuring basic sorting effects such as metal recovery rate >95% and core plastic purity >85%. This combination is usually set by engineers based on long-term experience and is activated when the system is initialized or when a suitable formula cannot be matched.

[0065] In this embodiment, the control system sent to each sorting submodule refers to adjusting the operating parameters of the submodule before the secondary mixture arrives at the physical actuators of the corresponding sorting submodule, such as magnetic separators, blow valves, and electrostatic electrodes, via the conveyor belt from the multispectral scanning point. This achieves prediction-based feedforward control to match the optimal processing conditions when the material arrives.

[0066] The working principle and beneficial effects of the above technical solution are as follows: When the secondary mixture flows through the multispectral scanning unit at the main inlet, the feature generation unit completes component analysis and generates a real-time feature vector within milliseconds. The pattern matching calculation unit immediately performs a rapid similarity calculation between this vector and all standard pattern vectors in the process formula library. The matching decision unit makes a judgment based on the calculation results: if a matching pattern with a similarity greater than 0.85 is found, the current material is determined to belong to the known category, and the optimal parameter combination corresponding to that category is immediately sent to each downstream sorting device; if the matching degree is insufficient, it indicates that the incoming material is relatively special, and the system automatically switches to the robust default parameter group for operation, while issuing a warning, indicating that manual attention or updating of the formula library may be necessary. This invention, through high similarity threshold matching and preset default strategies, ensures that the system can call precisely optimized solutions to achieve the best sorting effect in most cases, while automatically downgrading to a safe and reliable backup solution when encountering rare materials, thus ensuring the robustness and continuity of the entire process. This predictive control method customizes process parameters for materials before they even reach the sorting equipment. Compared to traditional detection, feedback, and adjustment lag control modes, it responds faster and adjusts more promptly, effectively avoiding batch quality fluctuations caused by parameter mismatches.

[0067] Example 9 Based on Example 8, the pattern matching calculation unit calculates the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formula library: Real-time multidimensional feature vectors are The first in the process formula library The standard multidimensional feature vector of a typical material composition pattern is: The similarity between the two The calculation formula is: in, The total dimension of the multidimensional feature vector is denoted as , and its value is . ; For the preset first Weights of dimensional features; Real-time multidimensional feature vector In the The numerical value of the dimension; For the first Standard multidimensional feature vectors In the The numerical value of the dimension; For real-time feature vectors and the first The similarity between the standard feature vectors of each pattern.

[0068] In this embodiment, satisfy The weights are set according to the degree of influence of each component on the sorting effect, for example, as follows: (PP) (ABS) (PVC) set as , (Metal indicator) (Inorganic material indicator) set to , (Moisture content) is set as .

[0069] In this embodiment, The range is The larger the value, the more similar the two numbers are.

[0070] The working principle and beneficial effects of the above technical solution are as follows: The calculation of the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formula library not only calculates the cosine value of the angle between the two vectors in the spatial direction to measure their directional consistency, but also introduces a weight factor. The weight factor reflects the influence of different component features on the final sorting effect. For example, the weight of the main target plastics such as PP and ABS is set higher, while the weight of moisture content is relatively lower. Through weighted calculation, the matching result focuses more on the key components that have a decisive influence on the selection of sorting process parameters, so that the most similar pattern matched is also the most relevant and effective in terms of process guidance. This invention employs weighted cosine similarity as the matching algorithm, which is both scientific and practical. Cosine similarity excels at measuring the similarity of component structures, overcoming the drawback of simple numerical Euclidean distance, which may be affected by absolute values. Introducing weighting coefficients quantifies the experience of process experts and integrates it into the automated decision-making process, ensuring that the system can accurately identify the historical pattern most closely related to the current material's processing characteristics from complex component data, providing the most reliable theoretical basis for subsequent parameter retrieval.

[0071] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A system for processing and treating scrapped vehicle residue based on residue separation, characterized in that: include: The overall pretreatment module is used to homogenize and perform primary impurity removal on the incoming scrapped car debris to form a uniform primary mixture. The targeted preprocessing module is used to acquire and intelligently analyze the surface images of the primary mixture, identify the agglomerates formed by surface adhesive substances in the primary mixture, and generate and execute corresponding targeted preprocessing strategies based on the type and strength of the adhesive substances to process the primary mixture into a secondary mixture without agglomerates. The intelligent identification multi-path sorting module is used to separate secondary mixtures into single-material streams based on their material, density, electrical properties, and surface characteristics. An adaptive process optimization module is used to predict and dynamically adjust the sorting parameters of the intelligent identification multi-path sorting module based on the composition characteristics of the secondary mixture. The adaptive process optimization module includes: The historical operating condition database submodule is used to store the multidimensional component feature vectors of historical secondary mixtures, the historical sorting operation parameter combinations used by the system, and the historical sorting purity results of each output material stream. The clustering formula generation submodule is used to perform cluster analysis on the multidimensional component feature vectors of historical secondary mixtures to generate a process formula library. The process formula library contains multiple typical material component patterns and their corresponding optimal sorting operation parameter combinations and expected sorting target values. The online matching feedforward control submodule is used to obtain the real-time multidimensional component feature vector of the current secondary mixture, match the real-time multidimensional component feature vector with the typical material component patterns in the process formula library, and send the optimal sorting operation parameter combination corresponding to the matched typical material component pattern to the submodules corresponding to each intelligent identification multi-channel sorting module.

2. The waste automotive residue processing system based on residue separation according to claim 1, characterized in that: The overall preprocessing module includes: The screening and soil removal submodule is used to screen out foreign objects and dust with particle sizes exceeding the preset particle size range in the crushed residue of scrapped cars based on particle size differences. The wind-powered pre-sorting submodule is used to separate crushed residues whose output mass is lower than the preset mass from the screening and soil removal submodule based on density differences. The crushing homogenization submodule is used to crush the crushing residue output from the wind pre-sorting submodule to a set particle size range, forming a uniform primary mixture.

3. The waste automotive residue processing system based on residue separation according to claim 1, characterized in that: The targeted preprocessing module includes: The surface image acquisition unit is used to perform online image acquisition of the primary mixture on the conveyor belt connected to the output end of the overall preprocessing module, and to obtain surface image data of the primary mixture. The adhesion feature analysis unit is used to identify and locate adhesion clumps in primary mixtures based on surface image data, analyze the surface material type and adhesion strength score of the adhesion clumps, and generate targeted preprocessing instructions. The multi-strategy preprocessing execution unit is used to receive targeted preprocessing instructions and drive the corresponding preprocessing device to perform the unbinding operation on the adhering clumps.

4. The waste automotive residue processing system based on residue separation according to claim 3, characterized in that: The multi-strategy preprocessing execution unit includes: The pneumatic intervention subunit is used to process adhesive clumps with an adhesion strength score lower than the first preset threshold and surface material type of oil stains. The pneumatic intervention subunit sprays compressed air flow onto the target adhesive clumps through a high-pressure nozzle with adjustable direction and pressure. The solvent atomizing spray subunit is used to treat adhesive clumps whose surface material is either paint-bonded or adhesive-bonded. The solvent atomizing spray subunit selects a matching solvent from the built-in solvent library according to the category label and sprays it onto the bonding interface of the target adhesive clump through the atomizing nozzle. The robot-assisted deionization unit is used to process adhesive clumps with an adhesion strength score higher than a second preset threshold or a physical size exceeding a preset volume. The robot-assisted deionization unit uses a vision-guided robotic arm to grasp the target adhesive clumps and transfer them to an independent low-speed shearing and crushing unit.

5. The waste automotive residue processing system based on residue separation according to claim 1, characterized in that: The intelligent identification and multi-path sorting module includes: The metal deep recovery submodule is used to recover various metals from secondary mixtures in stages based on magnetic and eddy current effects, forming a metal enrichment stream; The near-infrared spectroscopy identification and sorting submodule is used to identify and sort different types of plastics from the secondary mixture remaining after the metal deep recovery submodule based on near-infrared spectral features, forming a classified plastic fragment stream; The triboelectric and electrostatic co-sorting submodule is used to further separate dark-colored or painted plastics and rubber from the secondary mixture remaining after sorting by the near-infrared spectroscopy identification sorting submodule based on the difference in the triboelectric properties of the materials, forming a rubber particle stream and a mixed plastic stream. The density sorting submodule is used to separate heavy inorganic matter in the secondary mixture remaining after the triboelectric electrostatic collaborative sorting submodule based on density differences, forming a heavy inorganic material stream.

6. The waste automotive residue processing system based on residue separation according to claim 1, characterized in that: The clustering recipe generation submodule includes: The feature extraction and standardization unit is used to read the multi-dimensional component feature vectors of historical secondary mixtures from the historical working condition database submodule, preprocess and extract features from each multi-dimensional component feature vector, and generate a standardized set of historical multi-dimensional feature vectors. The clustering analysis execution unit is used to perform clustering analysis on the standardized set of historical multidimensional feature vectors, determine the optimal number of classifications based on the silhouette coefficient of each historical multidimensional feature vector, and divide the entire set of historical multidimensional feature vectors into multiple clusters based on the optimal number of classifications; The formulation library construction unit is used to calculate the standard multidimensional feature vector of each cluster center as the typical material composition pattern corresponding to each cluster, and select one or more data points with the best historical sorting purity results from each cluster. The mean of the corresponding historical sorting operation parameter combination is set as the optimal sorting operation parameter combination for the corresponding typical material composition pattern, and the mean of the corresponding historical sorting purity results is set as the expected sorting target value for the corresponding typical material composition pattern, thereby completing the construction of the process formulation library.

7. A waste vehicle residue processing system based on residue separation according to claim 1, characterized in that: The online matching feedforward control submodule includes: The feature generation unit is used to obtain the real-time spectrum of the current secondary mixture by means of the multispectral scanning unit set at the main entrance of the intelligent identification multi-path sorting module, and to preprocess and extract features from the real-time spectrum to generate a real-time multidimensional feature vector. The pattern matching calculation unit is used to calculate the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formula library. The matching decision unit selects the most similar typical material composition pattern as the matching result and determines whether its similarity exceeds a preset similarity threshold. If it exceeds the preset similarity threshold, the optimal combination of sorting operation parameters corresponding to the most similar typical material composition pattern is sent to the control system of each sorting submodule; otherwise, the backup default combination of sorting operation parameters is used and sent to the control system of each sorting submodule, and an early warning is issued.

8. A waste vehicle residue processing system based on residue separation according to claim 7, characterized in that: The pattern matching calculation unit calculates the similarity between the real-time multidimensional feature vector and the standard multidimensional feature vector of each typical material composition pattern in the process formulation library: Real-time multidimensional feature vectors are The first in the process formula library The standard multidimensional feature vector of a typical material composition pattern is: The similarity between the two The calculation formula is: in, The total dimension of the multidimensional feature vector is denoted as , and its value is . ; For the preset first Weights of dimensional features; Real-time multidimensional feature vector In the The numerical value of the dimension; For the first Standard multidimensional feature vectors In the The numerical value of the dimension; For real-time feature vectors and the first The similarity between the standard feature vectors of each pattern.

Images