Scrap aluminum classification and rapid identification method

Abstract

The invention discloses a scrap aluminum classification and rapid identification method, and relates to the field of scrap aluminum classification and rapid identification, the method comprises the following steps: S1, grading pretreatment: to-be-sorted scrap aluminum is conveyed into a three-layer vibrating screen machine through a conveying belt for grading screening, and different sizes of scrap aluminum are sorted out; and S2, impurity separation is conducted, specifically, the classified scrap aluminum is conveyed into a drum-type magnetic separator, magnetic impurities are removed, then the scrap aluminum subjected to magnetic separation is conveyed to an airflow separator, light non-metal impurities are separated, then the scrap aluminum is conveyed to an objective table through a conveying belt, full-process automatic treatment is conducted, manual dependence is reduced, the production efficiency is improved, and the production cost is reduced. A single production line can reduce the workload of sorting workers; and compared with manual sorting, due to the fact that classification is accurate, the rejection rate of the regenerated aluminum alloy can be effectively reduced, and economic benefits are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of waste mixed aluminum classification and rapid identification, and particularly relates to a waste mixed aluminum classification and rapid identification method. BACKGROUND

[0002] Efficient classification of waste mixed aluminum is the key bottleneck restricting its recycling. Different brands and sources of waste mixed aluminum have significant composition differences. If mixed and smelted, the performance of recycled aluminum alloy will be unstable, and even a large amount of waste products will be produced.

[0003] Current waste mixed aluminum classification mainly relies on manual sorting, which has inherent defects such as low efficiency, high cost, and large error. The manual sorting speed is usually only 20-30 pieces per minute, and the labor intensity is high, and long-term work can easily lead to an error rate of 10-15%.

[0004] Therefore, it is necessary to propose a waste mixed aluminum classification and rapid identification method to solve the above problems. SUMMARY

[0005] The present application relates to the field of waste mixed aluminum classification and rapid identification, and particularly relates to a waste mixed aluminum classification and rapid identification method.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a waste mixed aluminum classification and rapid identification method, comprising the following steps: S1, grading pretreatment: the waste mixed aluminum to be sorted is conveyed to a three-layer vibrating screen machine through a conveyor belt for grading screening to separate waste mixed aluminum of different sizes; S2, separation of impurities: the graded waste mixed aluminum is conveyed to a drum-type magnetic separator to remove magnetic impurities, and then the waste mixed aluminum after magnetic separation is conveyed to an air flow separator to separate light non-metallic impurities, and then a conveying belt conveys the waste mixed aluminum to a loading platform; S3, data acquisition and analysis: a multispectral imaging system is used to acquire surface images of the waste mixed aluminum on the loading platform, and reflectance feature data is extracted, and then a portable X-ray fluorescence spectrometer is used to detect the elements of the waste mixed aluminum after image acquisition to determine the main element content data of the waste mixed aluminum; S4, inputting the data obtained in the data acquisition and analysis step into a trained random forest classification model, wherein the random forest classification model contains 200-300 decision trees, the maximum depth is 15-20 layers, the Gini coefficient is used as the splitting criterion, the training data set covers 12 types of waste mixed aluminum samples, and the model outputs the classification result of the waste mixed aluminum through 12 classification nodes; S5, the classification result verification and model optimization: the output classification result, randomly extract 5% of the sample for artificial review, when the misclassification rate of a kind of waste aluminum is > 2%, the misclassified sample is added to the training data set, and the random forest classification model is retrained.

[0007] Preferably, the reflectivity feature data includes extracting the reflectivity features of 450 nm, 550 nm, 650 nm, and 850 nm wave bands, and the main element content data of the waste aluminum includes the contents of five key elements Al, Si, Cu, Mg, and Fe; Preferably, the 12 types of waste aluminum samples include 1050 pure aluminum, 2024 aluminum alloy, 3003 aluminum alloy, 4043 aluminum alloy, 5052 aluminum alloy, 6061 aluminum alloy, 7075 aluminum alloy, aluminum foil waste, aluminum profile waste, aluminum casting waste, aluminum wire waste, and composite aluminum waste, each type of sample has a quantity ≥5000 and covers different surface states of oxidation, painting, and rust.

[0008] Preferably, the air flow generating device of the air flow sorting machine adopts a centrifugal fan, the fan air pressure is 1500-2000 Pa, and a guide plate is arranged in the sorting cavity, the angle of the guide plate is adjusted within a range of 0-10° according to the density of the waste aluminum, and the separation efficiency of light non-metallic impurities is further improved.

[0009] Preferably, the first layer of the three-layer vibrating screen has a screen hole diameter of 30 mm, is used for separating large pieces of waste aluminum with a weight ≥500 g, the second layer has a screen hole diameter of 20 mm, is used for separating medium pieces of waste aluminum with a weight of 100-500 g, and the third layer has a screen hole diameter of 10 mm, is used for separating small pieces of waste aluminum with a weight ≤100 g, the vibration frequency of the vibrating screen is set to 30-40 Hz, the amplitude is set to 5-8 mm, and the processing capacity is controlled to be 200-300 kg / h. Preferably, the multi-spectral imaging system includes a multi-spectral camera with a resolution of 5 million pixels and a spectral range of 400-1000 nm, and an LED ring light source with a color temperature of 5500-6500 K, the camera acquisition distance is set to 30-50 cm, and the exposure time is set to 10-30 ms; after acquisition, the image is subjected to gray scale correction, noise reduction processing, and edge extraction.

[0010] Preferably, the object table of the multi-spectral imaging system adopts a conveyor belt type structure, the conveyor belt speed matches the processing capacity of the vibrating screen in the three-layer vibrating screen machine, realizes continuous image acquisition of the waste aluminum, and the surface of the conveyor belt adopts black matte material to reduce the interference of environmental light on image acquisition.

[0011] Preferably, the training process of the random forest classification model adopts 5-fold cross-validation method, the number of decision trees, the maximum depth and the split criterion parameters are adjusted, and the artificial review process adopts the standard sample comparison method, that is, for the extracted sample, the category is preliminarily judged by artificial observation of appearance and measurement of density, and then compared with the multispectral features and element features of the standard sample.

[0012] Preferably, the distance between the detection probe of the portable X-ray fluorescence spectrometer and the waste aluminum detection surface is fixed at 5-8mm, and a protective shield is arranged around the probe, and the lead equivalent of the shield is greater than or equal to 0.5mmPb.

[0013] Preferably, the sorting process is automatically controlled by a PLC control system, the PLC system is connected with three-layer vibration screen machines, magnetic separators, air flow sorting machines, multispectral cameras, X-ray fluorescence spectrometers and conveying belt equipment respectively, the running parameters of each equipment are collected in real time, and the running state is displayed on the touch screen.

[0014] Technical effects and advantages of the present application: 1. The present application combines multispectral surface features and X-ray fluorescence element features, and combines an optimized random forest model, so that the classification accuracy is greater than or equal to 98%, and the accuracy is improved compared with traditional single methods; for waste aluminum with complex states such as surface oxidation and painting, the misclassification rate is controlled to be less than 2%, and the problem that the traditional method is greatly affected by the surface state is solved.

[0015] 2. The three-stage physical pretreatment removes ferromagnetic impurities and non-metallic impurities, and improves the purity of waste aluminum; the surface cleaning process reduces the influence of dust and unevenness on detection, reduces the element analysis error, and provides a reliable data basis for accurate classification.

[0016] 3. The whole system processing speed is faster than traditional manual, which can meet the flow line operation demand of large-scale secondary aluminum enterprises, and the single day processing capacity is large. The automation degree is high, and the cost is reduced: the full-process automatic processing reduces the dependence on manual, and the work load of the sorting workers in a single production line can be reduced; the classification cost is reduced, and compared with manual sorting, the waste rate of secondary aluminum alloy can be effectively reduced due to accurate classification, and the economic benefit is significantly improved.

[0017] 4. The model is iterative and has strong adaptability. Through the sample review mechanism and dynamic training set update, the model can continuously learn new sample features. When new waste aluminum is introduced, only the sample needs to be supplemented to realize accurate identification, solving the problem of poor generalization ability of traditional models; the equipment can adjust parameters according to the source of waste aluminum, and adapt to the characteristics of waste aluminum in different fields such as automobile, building and electronics. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 It is a waste aluminum classification and rapid identification method process schematic diagram. Detailed Implementation

[0019] This invention provides, for example Figure 1 The method for classifying and rapidly identifying scrap aluminum shown includes the following steps: S1. Grading and pretreatment: The waste aluminum to be sorted is transported to the three-layer vibrating screen through a conveyor belt equipped with a wind-powered dust removal mechanism for grading and screening, and is used to remove dust from the waste aluminum to be sorted. Specifically, a three-layer vibrating screen is used, with screen aperture diameters of 10mm, 20mm, and 30mm, to classify scrap aluminum by weight. The vibration frequency is set to 30-40Hz, the amplitude to 5-8mm, and the processing capacity to 200-300kg / h. The first layer, with 30mm apertures, separates large pieces of scrap aluminum weighing ≥500g; the second layer, with 20mm apertures, separates medium-sized pieces weighing 100-500g; and the third layer, with 10mm apertures, separates small pieces weighing ≤100g. This classification ensures that scrap aluminum of different sizes enters the corresponding detection channels, avoiding interference from size differences in subsequent testing.

[0020] S2. Impurity Separation: The graded waste aluminum is conveyed to a drum magnetic separator to remove magnetic impurities. The magnetically separated waste aluminum is then conveyed to an air classifier to separate light non-metallic impurities. Subsequently, a conveyor belt equipped with a wind-driven dust removal mechanism transports the waste aluminum to a loading platform. The drum magnetic separator operates at a speed of 30-40 r / min, with the magnetic field strength increasing from 1000 Gs to 1200 Gs along the drum axis. This ensures a removal rate of ≥99% for iron impurities (including micro-iron powder) of varying magnetic strengths, minimizing the impact of iron impurities on subsequent spectral analysis. The airflow angle of the air classifier is 30-45° to the horizontal, with a wind speed controlled at 5-8 m / s. This allows for precise separation of light non-metallic impurities such as plastics and rubber with a density ≤1 g / cm³, achieving a separation efficiency ≥95%. The purity of the pre-treated waste aluminum is increased to over 98%, laying the foundation for subsequent testing.

[0021] S3. Data Acquisition and Analysis: Surface images of scrap aluminum on the stage are acquired using a multispectral camera, and reflectance characteristic data are extracted. Subsequently, a portable X-ray fluorescence spectrometer is used to perform elemental analysis on the scrap aluminum after image acquisition to determine the content data of the main elements in the scrap aluminum. For different surface conditions of scrap aluminum (oxidized, painted, smooth), a multispectral camera with a spectral range of 400-1000nm and a resolution of 5 megapixels was used to acquire images of the scrap aluminum surface. The camera acquisition distance was 30-50cm, the exposure time was 10-30ms, and an LED ring light source with a color temperature of 5500-6500K was used to ensure image stability.

[0022] Four characteristic bands of reflectance were extracted: 450 nm (reflecting surface coating), 550 nm (metallic color), 650 nm (oxidation layer characteristics), and 850 nm (near-infrared reflectance). Specifically, the reflectance at 450 nm was R1, the reflectance at 550 nm was R2, the difference between the reflectance at 650 nm and the reflectance at 850 nm was R3-R4, and the ratio of R2 / R1 was calculated (to distinguish between painted and unpainted surfaces) and the difference R3-R4 was calculated (to judge the degree of oxidation).

[0023] The collected images were pre-processed, including gray scale correction using a standard white board, noise reduction using a 3x3 pixel window median filter, and edge extraction using a Canny operator with a threshold of 100-200. Finally, a 128-dimensional feature vector was formed to quantitatively describe the surface characteristics.

[0024] A portable X-ray fluorescence spectrometer was used to determine the elemental content of scrap aluminum. The spectrometer tube voltage was set to 30-50 kV, the tube current was 50-100 μA, the detection spot diameter was 3-5 mm, and the content analysis of Al, Si, Cu, Mg, and Fe could be completed within 2-3 seconds with a detection error of ≤±0.2%. Among them, the Al content was ≥80%, and the Si, Cu, Mg, and Fe contents were each 0.1%-10%.

[0025] Before detection, high-pressure air with a pressure of 0.4-0.6 MPa was used to blow the surface of scrap aluminum to remove dust and impurities with a particle size of ≤50 μm. For scrap aluminum with uneven surfaces, mechanical adjustment was used to ensure that the detection surface had a flatness of ≤0.5 mm, ensuring detection accuracy. The elemental detection data formed a 5-dimensional feature vector, reflecting the intrinsic composition characteristics of scrap aluminum.

[0026] S4, inputting the data obtained in the data acquisition and analysis step into the trained random forest classification model; The random forest classification model contains 200-300 decision trees with a maximum depth of 15-20 layers, and uses the Gini coefficient as the splitting criterion. The model input is a fusion vector of multi-spectral features (128-dimensional) and elemental features (5-dimensional), and the classification result is output through 12 classification nodes (corresponding to 12 types of scrap aluminum).

[0027] The training data set covers 12 types of scrap aluminum samples, each with a sample size of ≥5000, including scrap aluminum with different surface states (oxidation, painting, and rust). Five-fold cross-validation was used to optimize the model parameters, ensuring a classification accuracy of ≥98%.

[0028] A classification result verification link is arranged, 5% of the samples are manually reviewed, when the single-class misclassification rate is greater than 2%, the misclassified samples are added to the training set to retrain the model until the classification accuracy requirement is met. At the same time, the feature database of each type of waste aluminum is supplemented with more than 1000 samples per month to realize dynamic optimization of the model to adapt to the identification needs of new types of waste aluminum.

[0029] 12 types of waste aluminum specifically include: 1050 pure aluminum, 2024 aluminum alloy, 3003 aluminum alloy, 4043 aluminum alloy, 5052 aluminum alloy, 6061 aluminum alloy, 7075 aluminum alloy, aluminum foil waste, aluminum profile waste, aluminum casting waste, aluminum wire waste, and composite aluminum waste, and each type of feature database is updated in real time (more than 1000 samples are supplemented per month).

[0030] S5, classification result verification and model optimization: 5% of the samples are randomly selected for manual review, and when the misclassification rate of a certain type of waste aluminum is greater than 2%, the misclassified samples of this type are added to the training data set, and the random forest classification model is retrained.

[0031] The present application fuses multi-spectral surface features and X-ray fluorescence element features, combines an optimized random forest model, and has a classification accuracy of greater than or equal to 98%, which is 13-28 percentage points higher than that of traditional single methods (accuracy of 70%-85%); for waste aluminum in complex states such as surface oxidation and painting, the misclassification rate is controlled to be within 2%, solving the problem that traditional methods are greatly affected by the surface state.

[0032] The whole system processing speed reaches 150-200 pieces per minute, which is 5-10 times that of manual sorting (20-30 pieces per minute); the X-ray fluorescence detection time is shortened to 2-3 seconds per piece, which is 1-3 times higher than that of traditional equipment (5-10 seconds per piece), and can meet the flow line operation needs of large-scale secondary aluminum enterprises, and the single-day processing capacity can reach 200-300 tons.

[0033] The three-stage physical pretreatment makes the removal rate of ferromagnetic impurities greater than or equal to 99%, and the removal rate of non-metallic impurities greater than or equal to 95%, and the purity of waste aluminum is increased to more than 98%; the surface cleaning process reduces the influence of dust and unevenness on detection, so that the element analysis error is less than or equal to ±0.2%, providing a reliable data basis for accurate classification.

[0034] Full-flow automatic processing reduces the dependence on manual labor, and a single production line can reduce 80% of the sorting workers; the classification cost is reduced to 15-20 yuan / ton, which is reduced by 75%-85% compared with manual sorting (80-100 yuan / ton); at the same time, due to accurate classification, the waste rate of secondary aluminum alloy is reduced from 15%-20% to 3%-5%, which significantly improves the economic benefits.

[0035] Through the 5% sample review mechanism and dynamic training set update, the model can continuously learn new sample characteristics. When new types of waste aluminum are introduced, only 500-1000 samples need to be supplemented to achieve accurate identification, solving the problem of poor generalization ability of traditional models; the device can adjust parameters according to the source of waste aluminum, and adapt to the characteristics of waste aluminum in different fields such as automobiles, construction, and electronics. Embodiment

[0036] This embodiment uses the above technical solution to identify 1050 pure aluminum (Al≥99.5%, Fe≤0.4%).

[0037] Physical pretreatment: vibration screen frequency 30Hz, amplitude 5mm, 30mm screen hole separates large blocks (≥500g) of 1050 pure aluminum; magnetic separator speed 30r / min, magnetic field strength 1000Gs; air flow separation wind speed 5m / s, angle 30°.

[0038] Multispectral imaging: collection distance 50cm, exposure time 30ms, light source color temperature 5500K, extract R1=450nm reflectance 0.35, R2=550nm reflectance 0.42, R2 / R1=1.2, R3-R4=0.05.

[0039] X-ray fluorescence analysis: tube voltage 30kV, tube current 50μA, detection time 3 seconds, detection result Al=99.6%, Fe=0.32%, Si=0.05%.

[0040] Machine learning classification: input multispectral features and element features into a random forest model containing 200 decision trees and 15 layers of depth, output classification result as 1050 pure aluminum, compared with standard samples, accurate identification. Processing speed 150 pieces / minute, single-class misclassification rate 0.8%.

[0041] Embodiment 2: Identify 2024 aluminum alloy Identify medium-sized (100-500g) 2024 aluminum alloy (Cu=3.8%-4.9%, Mg=1.2%-1.8%).

[0042] Physical pretreatment: vibration screen frequency 35Hz, amplitude 7mm, 20mm screen hole; magnetic separator speed 35r / min, magnetic field strength 1100Gs; air flow separation wind speed 7m / s, angle 40°.

[0043] Multispectral imaging: collection distance 40cm, exposure time 20ms, light source color temperature 6000K, extract R1=0.29, R2=0.34, R2 / R1=1.17, R3-R4=0.09.

[0044] X-ray fluorescence analysis: tube voltage 40 kV, tube current 75 μΑ, detection time 2.5 seconds, detection results: Cu = 4.3%, Mg = 1.5%, Al = 93.8%.

[0045] Machine learning classification: input contains a random forest model with 250 decision trees, depth 17 layers, classification is correct. Speed 170 pieces / min, misclassification rate 0.9%.

[0046] Example 3: Identification of 3003 aluminum alloy Identify small pieces (≤100g) of 3003 aluminum alloy (Mn = 1.0% ~ 1.5%, Fe ≤ 0.7%).

[0047] Physical pretreatment: vibration screen frequency 40 Hz, amplitude 8 mm, 10 mm screen hole; magnetic separator speed 40 r / min, magnetic field strength 1200 Gs; air flow sorting wind speed 8 m / s, angle 45°.

[0048] Multispectral imaging: acquisition distance 30 cm, exposure time 10 ms, light source color temperature 6500 K, extraction R1 = 0.31, R2 = 0.37, R2 / R1 = 1.19, R3-R4 = 0.05.

[0049] X-ray fluorescence analysis: tube voltage 50 kV, tube current 100 μΑ, detection time 2 seconds, detection results: Mn = 1.5%, Fe = 0.7%, Al = 97.6%.

[0050] Machine learning classification: input contains a random forest model with 300 decision trees, depth 20 layers, classification is correct. Speed 200 pieces / min, misclassification rate 0.7%.

[0051] Example 4: Identification of 5052 aluminum alloy Identify 5052 aluminum alloy (Mg = 2.2% ~ 2.8%, Cr = 0.15% ~ 0.35%).

[0052] Physical pretreatment: vibration screen frequency 32 Hz, amplitude 6 mm, 20 mm screen hole; magnetic separator speed 32 r / min, magnetic field strength 1050 Gs; air flow sorting wind speed 6 m / s, angle 35°.

[0053] Multispectral imaging: acquisition distance 45 cm, exposure time 25 ms, light source color temperature 5800 K, extraction R1 = 0.33, R2 = 0.39, R2 / R1 = 1.18, R3-R4 = 0.07.

[0054] X-ray fluorescence analysis: tube voltage 35 kV, tube current 60 μΑ, detection time 2.8 seconds, detection results: Mg = 2.2%, Cr = 0.15%, Al = 97.4%.

[0055] Machine learning classification: Random forest model with 220 decision trees, depth 16 layers, input contains, classification correct. Speed 160 pieces / min, misclassification rate 0.9%.

[0056] Example 5: Recognizing 6061 aluminum alloy Recognizing 6061 aluminum alloy (Si = 0.4% ~ 0.8%, Mg = 0.8% ~ 1.2%).

[0057] Physical pretreatment: vibration screen frequency 35 Hz, amplitude 7 mm, 20 mm screen hole; magnetic separator speed 35 r / min, magnetic field strength 1100 Gs; air flow sorting wind speed 7 m / s, angle 40°.

[0058] Multispectral imaging: acquisition distance 40 cm, exposure time 20 ms, light source color temperature 6000 K, extraction R1=0.34, R2=0.40, R2 / R1=1.18, R3-R4=0.09.

[0059] X-ray fluorescence analysis: tube voltage 40 kV, tube current 75 μΑ, detection time 2.5 seconds, detection result Si = 0.6%, Mg = 1.0%, Al = 98.1%.

[0060] Machine learning classification: Random forest model with 250 decision trees, depth 17 layers, input contains, classification correct. Speed 170 pieces / min, misclassification rate 0.6%.

[0061] Example 6: Recognizing 7075 aluminum alloy Recognizing 7075 aluminum alloy (Zn = 5.1% ~ 6.1%, Mg = 2.1% ~ 2.9%).

[0062] Physical pretreatment: vibration screen frequency 40 Hz, amplitude 8 mm, 30 mm screen hole; magnetic separator speed 40 r / min, magnetic field strength 1200 Gs; air flow sorting wind speed 8 m / s, angle 45°.

[0063] Multispectral imaging: acquisition distance 30 cm, exposure time 10 ms, light source color temperature 6500 K, extraction R1=0.28, R2=0.33, R2 / R1=1.18, R3-R4=0.10.

[0064] X-ray fluorescence analysis: tube voltage 50 kV, tube current 100 μΑ, detection time 2 seconds, detection result Zn = 6.1%, Mg = 2.9%, Al = 90.5%.

[0065] Machine learning classification: Random forest model with 300 decision trees, depth 20 layers, input contains, classification correct. Speed 200 pieces / min, misclassification rate 1.0%.

[0066] Example 7: Identifying aluminum foil scrap Identifying aluminum foil scrap (thickness < 0.2 mm, Al > 99%).

[0067] Physical pretreatment: vibrating screen frequency 30 Hz, amplitude 5 mm, 10 mm screen aperture; magnetic separator speed 30 r / min, magnetic field strength 1000 Gs; air flow sorting wind speed 5 m / s, angle 30°.

[0068] Multispectral imaging: acquisition distance 50 cm, exposure time 30 ms, light source color temperature 5500 K, extraction R1=0.40, R2=0.45, R2 / R1=1.13, R3-R4=0.04.

[0069] X-ray fluorescence analysis: tube voltage 30 kV, tube current 50 μΑ, detection time 3 seconds, detection result Al=99.2%, Fe=0.2%.

[0070] Machine learning classification: input contains a random forest model with 200 decision trees, depth 15 layers, classification correct. Speed 150 pieces / min, misclassification rate 1.2%.

[0071] Example 8: Identifying aluminum profile scrap Identifying building aluminum profile scrap (6063 series, Si=0.2%-0.6%).

[0072] Physical pretreatment: vibrating screen frequency 35 Hz, amplitude 7 mm, 20 mm screen aperture; magnetic separator speed 35 r / min, magnetic field strength 1100 Gs; air flow sorting wind speed 7 m / s, angle 40°.

[0073] Multispectral imaging: acquisition distance 40 cm, exposure time 20 ms, light source color temperature 6000 K, extraction R1=0.36, R2=0.42, R2 / R1=1.17, R3-R4=0.08.

[0074] X-ray fluorescence analysis: tube voltage 40 kV, tube current 75 μΑ, detection time 2.5 seconds, detection result Si=0.4%, Mg=0.7%, Al=98.6%.

[0075] Machine learning classification: input contains a random forest model with 250 decision trees, depth 17 layers, classification correct. Speed 170 pieces / min, misclassification rate 0.7%.

[0076] Example 9: (Identifying aluminum casting scrap) Identifying automobile aluminum casting scrap (ADC12, Si=9.6%-12%).

[0077] Physical pretreatment: vibrating screen frequency 38 Hz, amplitude 7 mm, 30 mm mesh; magnetic separator speed 38 r / min, magnetic field strength 1150 Gs; air flow sorting wind speed 7.5 m / s, angle 42°.

[0078] Multispectral imaging: acquisition distance 35 cm, exposure time 15 ms, light source color temperature 6300 K, extraction R1=0.25, R2=0.30, R2 / R1=1.20, R3-R4=0.12.

[0079] X-ray fluorescence analysis: tube voltage 45 kV, tube current 90 μΑ, detection time 2.2 seconds, detection results Si=10.5%, Cu=2.5%, Al=86.7%.

[0080] Machine learning classification: input contains 280 decision trees, random forest model with depth 19 layers, classification correct. Speed 190 pieces / min, misclassification rate 0.9%.

[0081] Example 10: Composite aluminum waste (high limit parameters) Identify aluminum-plastic composite waste (aluminum ratio 60%-70%).

[0082] Physical pretreatment: vibrating screen frequency 40 Hz, amplitude 8 mm, 20 mm mesh; magnetic separator speed 40 r / min, magnetic field strength 1200 Gs; air flow sorting wind speed 8 m / s, angle 45°.

[0083] Multispectral imaging: acquisition distance 30 cm, exposure time 10 ms, light source color temperature 6500 K, extraction R1=0.22, R2=0.26, R2 / R1=1.18, R3-R4=0.15.

[0084] X-ray fluorescence analysis: tube voltage 50 kV, tube current 100 μΑ, detection time 2 seconds, detection results Al=65%, others are non-metallic.

[0085] Machine learning classification: input contains 300 decision trees, random forest model with depth 20 layers, classification correct. Speed 200 pieces / min, misclassification rate 1.1%. After 5% sample review, misclassification rate is reduced to 0.8% after adjusting the training set.

Claims

1. A method for classifying and rapidly identifying scrap aluminum, characterized in that: Includes the following steps: S1. Grading and pretreatment: The waste aluminum to be sorted is conveyed to the three-layer vibrating screen by the conveyor belt for grading and screening to separate waste aluminum of different sizes. S2. Separation of impurities: The graded waste aluminum is fed into a drum magnetic separator to remove magnetic impurities. Then, the magnetically separated waste aluminum is fed into an air classifier to separate light non-metallic impurities. Finally, the waste aluminum is conveyed to the loading platform by a conveyor belt. S3. Data Acquisition and Analysis: Surface images of scrap aluminum on the stage are acquired using a multispectral imaging system, and reflectance characteristic data are extracted. Subsequently, a portable X-ray fluorescence spectrometer is used to perform elemental analysis on the scrap aluminum after image acquisition to determine the content data of the main elements in the scrap aluminum. S4. Input the data obtained in the data collection and analysis steps into the trained random forest classification model. The random forest classification model contains 200-300 decision trees with a maximum depth of 15-20 layers. The Gini coefficient is used as the splitting criterion. The training dataset covers 12 types of waste aluminum samples. The model outputs the classification result of waste aluminum through 12 waste aluminum samples. S5. Classification result verification and model optimization: 5% of the output classification results are randomly selected for manual review. When the misclassification rate of a certain type of waste aluminum is >2%, the misclassified samples of that type are added to the training dataset and the random forest classification model is retrained.

2. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The reflectivity feature data includes the reflectivity features extracted from the 450nm, 550nm, 650nm, and 850nm wavelength bands, and the main element content data of the waste aluminum is the content of five elements: Al, Si, Cu, Mg, and Fe.

3. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The 12 categories of scrap aluminum samples include 1050 pure aluminum, 2024 aluminum alloy, 3003 aluminum alloy, 4043 aluminum alloy, 5052 aluminum alloy, 6061 aluminum alloy, 7075 aluminum alloy, aluminum foil scrap, aluminum profile scrap, aluminum casting scrap, aluminum wire scrap, and composite aluminum scrap. Each category has ≥5000 samples and covers different surface states such as oxidation, painting, and rust.

4. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The airflow generator of the airflow separator uses a centrifugal fan with a fan pressure of 1500-2000Pa. A guide plate is installed inside the separation chamber, and the angle of the guide plate is adjustable within the range of 0-10° according to the density of the waste aluminum.

5. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The three-layer vibrating screen has a first layer with a screen hole diameter of 30mm, used to separate large pieces of waste aluminum weighing ≥500g; a second layer with a screen hole diameter of 20mm, used to separate medium-sized pieces of waste aluminum weighing 100-500g; and a third layer with a screen hole diameter of 10mm, used to separate small pieces of waste aluminum weighing ≤100g. The vibration frequency of the vibrating screen is set to 30-40Hz, the amplitude is set to 5-8mm, and the processing capacity is controlled at 200-300kg / h.

6. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The multispectral imaging system includes a multispectral camera with a resolution of 5 megapixels and a spectral range of 400-1000nm, and an LED ring light source with a color temperature of 5500-6500K. The camera acquisition distance is set to 30-50cm and the exposure time is set to 10-30ms. After acquisition, the image is subjected to grayscale correction, noise reduction processing and edge extraction.

7. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The stage of the multispectral imaging system adopts a conveyor belt structure. The speed of the conveyor belt is matched with the processing capacity of the vibrating screen in the three-layer vibrating screen machine to realize continuous image acquisition of waste aluminum. The surface of the conveyor belt is made of black matte material.

8. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The training process of the random forest classification model adopts the 5-fold cross-validation method. By adjusting the number of decision trees, the maximum depth, and the splitting criterion parameters, the manual verification process adopts the standard sample comparison method. That is, for the extracted samples, the category is initially judged by manually observing the appearance and measuring the density, and then compared with the multispectral features and elemental features of the standard samples.

9. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The distance between the detection probe of the portable X-ray fluorescence spectrometer and the detection surface of the waste aluminum is fixed at 5-8 mm, and a protective shield is set around the probe, with a lead equivalent of ≥0.5 mmPb.

10. The method for classifying and rapidly identifying scrap aluminum according to claim 1, characterized in that: The sorting process is automated through a PLC control system. The PLC system is connected to the three-layer vibrating screen, magnetic separator, air classifier, multispectral camera, X-ray fluorescence spectrometer and conveyor belt equipment, respectively, to collect the operating parameters of each device in real time and display the operating status on the touch screen.

Images