Multi-sensor fused intelligent steel scrap sorting system

Abstract

The invention discloses a multi-sensor fused intelligent waste steel sorting system, and relates to the field of industrial automation. A central processing module; and a sorting execution module. According to the method, the component proportion and the three-dimensional space distribution characteristics of the whole scrap steel are synchronously collected through a multi-sensor fusion framework, accurate quantification of the economic value of the scrap steel is achieved in combination with a dynamic value scoring formula, and a three-way sorting mechanism is driven based on a scoring threshold value; the problems of cost evaluation distortion and sorting decision disjunction caused by local sampling in the prior art are effectively solved, the reprocessing energy consumption is reduced while the high-value metal recovery rate is increased, the reliability and traceability of the sorting process are guaranteed through a redetection buffer area and a block chain evidence storage mechanism, and finally the maximum utilization of waste steel resources is achieved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation, specifically to a multi-sensor fusion-based intelligent scrap steel sorting system. Background Technology

[0002] As an important recyclable metallurgical raw material, the efficient sorting of scrap steel is directly related to the production cost and resource utilization rate of steel enterprises. Traditional scrap steel sorting mainly relies on manual visual inspection or single sensor detection, which makes it difficult to accurately assess the internal composition distribution and impurity effects of the whole piece of scrap steel, resulting in low recovery rate of high-value metals and high energy consumption for reprocessing.

[0003] Existing sorting technologies can only obtain fragmented data (such as the metal content of a single cross section or the surface contamination area) through local sampling or surface detection. They cannot simultaneously quantify the ratio of high-value metal areas to low-value impurity areas and their three-dimensional spatial distribution characteristics in a whole piece of scrap steel. This leads to serious distortion in the reprocessing cost assessment. For example, when high-value metals are in a dispersed and mixed state, the actual separation energy consumption is much higher than that of a concentrated block structure, but traditional systems cannot identify such differences.

[0004] Current mainstream solutions rely on component detection sensors (such as X-ray component analyzers) or visual sorting equipment. While these can achieve basic material classification, they suffer from fundamental flaws: First, the detection data is not correlated with spatial distribution parameters, making it impossible to calculate the multiplicative impact of impurity dispersion on processing costs. Second, the decision-making mechanism is based solely on component thresholds (e.g., iron content >80% is classified as high value), without establishing a value scoring formula that integrates component proportions, distribution patterns, and market costs, leading to a disconnect between sorting results and economic benefits. Especially for large, irregular scrap steel, existing technologies, lacking a comprehensive value quantification model, often incorrectly sort scrap steel containing concentrated high-value metals but with surface impurities into low-value channels, or misjudge highly dispersed metal scrap steel as high-value, resulting in resource waste rates as high as 15%-30%. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a multi-sensor fusion-based intelligent scrap steel sorting system to solve the technical problems in the prior art, which are that the proportion of high-value metals and impurities in whole scrap steel and the three-dimensional spatial distribution characteristics cannot be simultaneously quantified, resulting in distorted reprocessing cost assessment, lack of a value scoring mechanism that integrates component proportions and distribution patterns, and thus causing a disconnect between sorting decisions and economic benefits.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-sensor fusion intelligent scrap steel sorting system, comprising:

[0007] The sensor array module consists of an X-ray composition analysis sensor (which uses dual-energy transmission technology to output the content of Fe, Cu, and Al elements in real time), a high-resolution vision sensor (equipped with a polarization filter lens to identify the area covered by oil stains / coatings / oxide layers), and a laser contour scanning sensor (which measures spatial coordinates based on the time-of-flight method, with a reconstruction accuracy of ≤0.5mm). It simultaneously collects data on the material composition, surface morphology, and three-dimensional geometric distribution of scrap steel.

[0008] Central processing module: Constructs a 3D material distribution map of scrap steel, identifies the volume ratio and spatial distribution characteristics of high-value metal regions and low-value impurity regions; calculates the comprehensive value score V using the following formula. s :

[0009]

[0010] Where: V s Assess the overall value score; α i Q represents the purity coefficient of the i-th type of metal; i P represents the weight percentage of the i-th type of metal; i Represents the real-time market unit price of the i-th type of metal; β is the spatial distribution optimization coefficient; C d C is the cost factor for impurity separation. p This is the energy consumption cost factor for reprocessing.

[0011] The formula for calculating the spatial distribution optimization coefficient β is:

[0012]

[0013] Where: k is the correction constant for high-value metal types; σ is the spatial distribution variance of high-value metal regions; and μ is the average clustering radius of high-value metal regions.

[0014] Impurity separation cost factor C d The calculation formulas include:

[0015] C o =S o ·U o

[0016] C f =M i ·U f

[0017] Where: C o Cost of oil / coating treatment; S o For the contaminated area; U o Unit processing cost; C f Cost of crushing non-metallic impurities; M i For impurity mass; U fThis refers to the unit energy consumption cost of crushing.

[0018] Sorting execution module: Performs sorting based on scoring thresholds.

[0019] V s When the score is >85, sort it to the high-value processing channel;

[0020] V s When the score is less than 60 points, sort it into the low-value recycling channel;

[0021] 60≤V s If the score is ≤85, the sample is sorted to the re-inspection buffer (equipped with a rotary secondary scanning platform for multi-angle re-inspection);

[0022] The threshold is dynamically optimized through a machine learning model, and the decision data is stored on the blockchain in real time by blockchain units.

[0023] In summary, the present invention has the following main beneficial effects:

[0024] This invention uses a multi-sensor fusion architecture to simultaneously collect the composition ratio and three-dimensional spatial distribution characteristics of whole scrap steel. Combined with a dynamic value scoring formula, it achieves accurate quantification of the economic value of scrap steel. Based on the scoring threshold, it drives a three-way sorting mechanism, effectively solving the problems of cost assessment distortion and sorting decision disconnect caused by local sampling in traditional technologies. While improving the recovery rate of high-value metals, it reduces reprocessing energy consumption. Furthermore, the re-detection buffer and blockchain evidence storage mechanism ensure the reliability and traceability of the sorting process, ultimately maximizing the utilization of scrap steel resources. Attached Figure Description

[0025] Figure 1 This is a system architecture diagram of the present invention;

[0026] Figure 2 This is a schematic diagram of the three-dimensional material distribution map and region identification of the present invention;

[0027] Figure 3 This is a schematic diagram of the blockchain evidence storage process of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0029] The embodiments of the present invention will now be described.

[0030] Example 1

[0031] A multi-sensor fusion-based intelligent scrap steel sorting system includes:

[0032] Sensor array module: Synchronous acquisition of data from multiple sensors.

[0033] X-ray composition analysis sensor: It adopts dual-energy X-ray transmission technology (such as 120kV / 140kV dual source) to obtain the characteristic energy spectrum decay curves of Fe, Cu and Al elements after penetrating scrap steel. The content ratio of each metal is output in real time through elemental spectrum decomposition algorithm (e.g. Fe: 75%, Cu: 8%, Al: 5%).

[0034] High-resolution vision sensor: Equipped with a polarizing filter lens (e.g., linear polarization ≥ 95%), it acquires images of scrap steel surfaces under an LED ring light source. Image segmentation algorithms identify the coverage area of ​​oil / coating / oxide layers (e.g., oil area accounting for 12%).

[0035] Laser contour scanning sensor: Based on the time-of-flight (TOF) method, it emits a 905nm laser pulse, calculates the surface spatial coordinates by receiving the time difference of the reflected signal, and reconstructs a three-dimensional point cloud model of scrap steel (accuracy error ≤0.5mm).

[0036] Data fusion mechanism: The three sensors are synchronized via timestamps (error <1ms), and the central processing module maps the composition data to three-dimensional point cloud coordinates to generate a three-dimensional material distribution map. The spatial locations of high-value metals (such as Cu blocks) and impurities (such as plastic embeddings) in the map are marked with different color blocks.

[0037] Central Processing Module: Performs the following calculations:

[0038] 1. Identify region features:

[0039] High-value metal zone: continuous volume with metal purity ≥ 90% (e.g., Cu block volume accounting for 15%);

[0040] Low-value impurity areas: areas where non-metallic or contaminated areas account for more than 30% (e.g., Fe blocks encased in oil with a volume of 10%).

[0041] 2. Calculate the comprehensive value score V s :

[0042]

[0043] For example:

[0044] α i (Purity coefficient): Cu is 0.95 (containing a small amount of oxides), Fe is 0.85;

[0045] Q i (Weight percentage): Calculated by density integration of three-dimensional spectra, Cu = 15%, Fe = 60%;

[0046] P i (Real-time unit price): Obtained from the metals trading API, Cu = 68,000 yuan / ton, Fe = 3,200 yuan / ton;

[0047] β (space optimization coefficient):

[0048] Let the spatial variance of the Cu region be σ = 0.8, the average cluster radius be μ = 1.2 cm, the correction constant be k = 0.1, and β = 1.08.

[0049] C d (Impurity separation cost):

[0050] C o =S o ·U o =0.12m 2 *50 yuan / m 2 =6 yuan

[0051] C f =M i ·U f =20kg × 8 yuan / kg = 160 yuan

[0052] C p (Reprocessing energy consumption cost): Calculated at 85 yuan based on the electricity consumption model.

[0053] Calculation result: V s =78 points.

[0054] Sorting execution module: Performs sorting based on scoring thresholds.

[0055] V s When the score is >85, sort it to the high-value processing channel;

[0056] V s When the score is less than 60 points, sort it into the low-value recycling channel;

[0057] 60≤V s If the score is ≤85, the sample is sorted to the re-inspection buffer (equipped with a rotary secondary scanning platform for multi-angle re-inspection);

[0058] For example, V s =78 points (between 60 and 85 points), sorted to the re-inspection buffer.

[0059] Perform the following actions:

[0060] 1. The conveyor belt branching controller directs the scrap steel to the buffer zone;

[0061] 2. The rotary secondary scanning platform drives the scrap steel to rotate 120°, 240°, and 360°;

[0062] 3. The sensor array re-collects three sets of data, and the central processing module takes the median score (if it is still within the threshold range, it is sorted into the low-value channel according to a conservative strategy).

[0063] Example 2:

[0064] Dynamic threshold adjustment

[0065] Historical data input: 10,000 sorting records from the past (including metal prices and actual reprocessing costs);

[0066] Machine learning models (such as XGBoost): aim to maximize economic benefits and output threshold optimization suggestions;

[0067] When the price of Cu rises by 10%, the first threshold is raised from 85 points to 88 points;

[0068] When the energy consumption cost of crushing increases, the second threshold is lowered from 60 points to 55 points.

[0069] Blockchain evidence storage process

[0070] The central processing module generates a sorting decision data package (including a 3D map, scoring parameters, and timestamps);

[0071] Uploaded to the consortium blockchain via blockchain unit (integrated with HyperledgerFabric SDK);

[0072] Generate a unique hash value (such as 0x9a3b...c7d2) and store it in an SQL database for traceability queries.

[0073] Example 3

[0074] Special scrap steel processing

[0075] Object: A scrapped engine block with a thick layer of oil on its surface (containing a concentrated Cu coil inside).

[0076] Initial scan results:

[0077] The visual sensor identified an oil stain coverage rate of 95%, C o Extremely high cost;

[0078] X-ray analysis showed that the internal Cu content was 18%, Q i High value but C d offset;

[0079] Rating V s =58 points (below 60 points).

[0080] Re-detection buffer processing:

[0081] After rotation, laser scanning revealed an area on the side free of oil stains.

[0082] Recalculate the contaminated area S o =40%; V s =65 points

[0083] Final decision: Sorting to the high-value channel (manual re-inspection confirmed a 22% increase in Cu recovery rate).

[0084] Key component implementation details:

[0085]

[0086] This invention achieves the quantification of the composition and spatial distribution characteristics of scrap steel by synchronous acquisition of multiple sensors and three-dimensional map modeling. Based on the dynamic value scoring formula, it integrates the economic value of metals and the cost of impurity treatment to drive a three-channel sorting mechanism. The re-inspection buffer solves the problem of misjudgment caused by surface interference through a rotating scanning platform. The blockchain unit ensures the traceability of decision data, and the machine learning model continuously optimizes the scoring threshold. Ultimately, it achieves the core goal of accurate recycling of high-value metals and cost control of reprocessing.

[0087] Although embodiments of the present invention have been shown and described, these specific embodiments are merely explanations of the invention and are not intended to limit it. The specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. After reading this specification, those skilled in the art may make modifications, substitutions, and variations to the embodiments as needed without departing from the principles and spirit of the invention, but such modifications, substitutions, and variations are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A multi-sensor fusion-based intelligent scrap steel sorting system, characterized in that, include: The sensor array module consists of an X-ray composition analysis sensor, a high-resolution vision sensor, and a laser contour scanning sensor, which are used to collect material composition data, surface morphology data, and three-dimensional geometric distribution data of the scrap steel to be sorted in real time. The central processing module, connected to the sensor array module, is configured as follows: Based on the material composition data, surface morphology data, and three-dimensional geometric distribution data, a three-dimensional material distribution map of scrap steel is constructed. Identify the volume ratio and spatial distribution characteristics of high-value metal regions and low-value impurity regions in the scrap steel; The comprehensive value score of scrap steel is calculated based on the volume ratio and spatial distribution characteristics. The sorting execution module, connected to the central processing module, includes a high-speed pneumatic sorting arm and a conveyor belt routing controller, and is configured as follows: When the comprehensive value score is higher than the first threshold, the scrap steel will be sorted to the high-value processing channel; When the overall value score is below the second threshold, the scrap steel will be sorted into the low-value recycling channel. When the comprehensive value score is between the first threshold and the second threshold, the scrap steel is sorted into the re-inspection buffer.

2. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The X-ray composition analysis sensor uses dual-energy X-ray transmission technology to output the content ratio of Fe, Cu, and Al elements in scrap steel in real time.

3. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The high-resolution vision sensor is equipped with a polarizing filter lens to identify the area of ​​oil stains, coatings, and oxide layers on the surface of scrap steel.

4. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The laser contour scanning sensor uses the time-of-flight method to measure the spatial coordinates of the scrap steel surface, with a reconstruction accuracy error ≤0.5mm.

5. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The overall value score is calculated using the following formula: Where: V s Assess the overall value; α i Q represents the purity coefficient of the i-th type of metal; i P represents the weight percentage of the i-th type of metal; i Represents the real-time market unit price of the i-th type of metal; β is the spatial distribution optimization coefficient; C d C is the cost factor for impurity separation. p This is the energy consumption cost factor for reprocessing.

6. The multi-sensor fusion intelligent scrap steel sorting system according to claim 5, characterized in that: The formula for calculating the spatial distribution optimization coefficient β is as follows: Where: k is the correction constant for high-value metal types; σ is the spatial distribution variance of high-value metal regions; and μ is the average clustering radius of high-value metal regions.

7. The multi-sensor fusion intelligent scrap steel sorting system according to claim 5, characterized in that: The impurity separation cost factor C d Calculation formula include: C o =S o ·U o C f =M i ·U f Where: C o Cost of oil / coating treatment; S o For the contaminated area; U o Unit processing cost; C f Cost of crushing non-metallic impurities; M i For impurity mass; U f This refers to the unit energy consumption cost of crushing.

8. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The first threshold is 85 points, the second threshold is 60 points, and the scoring range is dynamically optimized based on historical processed data through a machine learning model.

9. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The re-inspection buffer is equipped with a rotating secondary scanning platform, which triggers the sensor array module to re-inspect the scrap steel from at least three angles.

10. The multi-sensor fusion intelligent scrap steel sorting system according to claim 1, characterized in that: The central processing module integrates a blockchain unit to upload and store scrap steel sorting decision data and value scores on the blockchain in real time.

Images