A multi-level intelligent sorting system for construction waste and its adaptive control method

Abstract

This invention relates to the field of intelligent sorting control technology, and discloses a multi-stage intelligent sorting system for construction waste and its adaptive control method. The method involves: identifying the longitudinal position, equivalent particle size, aspect ratio, and average grayscale value of each material based on material image recognition; calculating the predicted trigger times for each material to reach the vibrating screen trigger point, air-powered sorting trigger point, eddy current iron removal trigger point, and air gun fine sorting trigger point based on the equivalent particle size, aspect ratio, and longitudinal position; and executing material screening according to the corresponding predicted trigger times when each material reaches these trigger points. This invention achieves piece-by-piece adaptive control throughout the entire process from coarse screening to fine screening, avoiding over- or under-triggering caused by uniform parameters for different materials, and ensuring the stability of the trigger timing during long-term operation of the sorting system.

Description

Technical Field

[0001] This invention relates to the field of intelligent sorting and control technology, and in particular to a multi-level intelligent sorting system for construction waste and its adaptive control method. Background Technology

[0002] Construction waste has an extremely complex composition, containing a mixture of materials such as concrete blocks, brick fragments, wood, plastics, and metals. Furthermore, it exhibits a wide range of particle sizes and irregular shapes. This characteristic means that a single sorting method is insufficient to meet practical processing needs. In engineering practice, a multi-stage sorting architecture is employed, sequentially connecting vibrating screens, air separation, eddy current iron removal, and air gun fine sorting. Within this architecture, the triggering timing of each actuator directly determines whether the material can be sorted within its effective working area; therefore, the accuracy of the triggering sequence is the core factor affecting the sorting effect.

[0003] However, existing equipment has significant flaws in its trigger control strategy. The trigger delays of actuators at each stage are generally pre-fixed based on the rated belt speed, neglecting the real-time deviations in belt speed caused by load fluctuations during actual operation, and failing to consider the differences in actual movement speed caused by the different frictional characteristics of materials of different particle sizes and shapes with the belt. These two types of errors accumulate continuously over long belt transmission distances, ultimately resulting in a non-negligible deviation between the actual time of material arrival at each sorting trigger point and the preset trigger time. Summary of the Invention

[0004] The main objective of this invention is to provide a multi-level intelligent sorting system for construction waste and its adaptive control method. This invention uses the material morphology characteristics obtained from image acquisition as a unified input for both trigger timing prediction and actuator parameter adjustment, solving the technical problems of fixed trigger timing and fixed execution parameters in the prior art. This enables piece-by-piece adaptive control of the entire process from coarse screening to fine screening, avoiding over-triggering or under-triggering caused by unified parameters for different materials, and ensuring the stability of the trigger timing during long-term operation of the sorting system.

[0005] To achieve the above objectives, the present invention provides an adaptive control method for multi-level intelligent sorting of construction waste, comprising the following steps: Based on material image recognition, the longitudinal position, equivalent particle size, aspect ratio, and mean gray value of each material are identified. Based on the equivalent particle size, the length-to-short axis ratio and the longitudinal position, the predicted trigger time for each material to reach the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point and the air gun fine separation trigger point is calculated. When each material reaches the trigger point of the vibrating screen, the trigger point of the air separation, the trigger point of the eddy current iron removal, and the trigger point of the air gun fine separation, the material screening is performed according to the corresponding predicted trigger time.

[0006] Optionally, in a first implementation of the first aspect of the present invention, identifying the longitudinal position, equivalent particle size, aspect ratio, and average grayscale value of each material based on the material image includes: Acquire a material image and binarize the material image and the background reference image to obtain a material foreground mask; Connectivity analysis is performed on the foreground mask of the material to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio and grayscale mean are calculated.

[0007] Optionally, in a second implementation of the first aspect of the present invention, connected component analysis is performed on the foreground mask of the material to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio, and mean gray value are calculated, including: Connectivity analysis is performed on the material foreground mask to extract the centroid longitudinal pixel coordinates of each connected region, and the centroid longitudinal pixel coordinates are converted into longitudinal positions in the belt coordinate system based on the longitudinal calibration coefficient. The equivalent particle size is calculated based on the projected area of ​​each connected domain. At the same time, the contour of each connected domain is elliptical-fitted to obtain the length of the major axis and the length of the minor axis. The ratio of the major axis to the minor axis is then calculated based on the length of the major axis and the length of the minor axis. The arithmetic mean of the gray values ​​of all pixels in each connected region is calculated to obtain the gray mean of each material.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the predicted triggering time for each material to reach the vibrating screen triggering point, the air separation triggering point, the eddy current iron removal triggering point, and the air gun fine separation triggering point is calculated based on the equivalent particle size, the aspect ratio, and the longitudinal position, including: Based on the equivalent particle size and the reference particle size, and the ratio of the major axis to the minor axis and the reference major axis, the instantaneous speed of the belt is corrected to obtain the morphological speed correction coefficient for each material. Based on the first distance calibration value of each trigger point and the longitudinal position, the remaining distance of each material to reach each trigger point is calculated. The remaining distance is divided by the product of the instantaneous speed of the belt and the morphological speed correction coefficient to obtain the predicted time delay of each material to reach each trigger point. The trigger points at each level are the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point. The predicted delay is added to the image acquisition completion time to obtain the predicted trigger time of each material reaching the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, the instantaneous speed of the belt is corrected based on the equivalent particle size and the reference particle size, and the ratio of the major axis to the minor axis and the reference major axis, to obtain the morphological velocity correction coefficient for each material, including: The equivalent particle size of each material is compared with the reference particle size to obtain the particle size ratio. The difference between the major and minor axis ratios and the preset reference major axis ratios is obtained to obtain the target difference. The particle size ratio is multiplied by the target difference and then multiplied by the morphology influence weighting coefficient to obtain the morphology correction amount of each material. Subtract the morphology correction amount from 1 to obtain the morphology velocity correction coefficient for each material.

[0010] Optionally, in the fifth implementation of the first aspect of the present invention, when each material reaches the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point, material screening is performed according to the corresponding predicted trigger time, including: When the material reaches the vibrating screen trigger point, the corresponding first output frequency is written to the vibrating screen inverter according to the equivalent particle size and the corresponding predicted trigger time, and the vibrating screen is driven to complete the screening and separation of the material at the first output frequency. When the material reaches the wind-driven sorting trigger point, the corresponding second output frequency is written to the wind turbine inverter based on the wind-receiving area and the corresponding predicted trigger time, and the wind turbine is driven to complete the aerodynamic deflection separation of the lightweight material at the wind speed corresponding to the second output frequency. When the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material with the electromagnetic coil pulse drive current. When the material reaches the air gun's fine sorting trigger point, the air gun is driven to accurately spray and deflect the target material based on the equivalent particle size and the predicted trigger time.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, when the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material with the electromagnetic coil pulse drive current, including: When the material reaches the eddy current iron removal trigger point, the material is identified as a metal based on the average gray value. When the material is determined to be a metal, the electromagnetic coil pulse drive current is calculated based on the equivalent particle size. The electromagnetic coil is driven by the electromagnetic coil pulse drive current to generate eddy current repulsion force on the metal material, thereby completing the deflection and removal of the metal material.

[0012] Optionally, in the seventh implementation of the first aspect of the present invention, when the material reaches the air gun fine sorting trigger point, the air gun is driven to perform precise spray deflection on the target material according to the equivalent particle size and the predicted trigger time, including: When the material reaches the air gun fine sorting trigger point, the corresponding predicted trigger time is subtracted from the preset air gun valve mechanical response delay to obtain the air gun trigger command issuance time; The air gun spraying duration, material mass, and spraying pressure are determined based on the equivalent particle size. When the air gun trigger command is issued, an opening command is written to the air gun solenoid valve, and a spray pressure command is written to the pressure reducing valve. The spray pressure is maintained for the duration of the air gun spray to achieve precise spray deflection of the target material. After the spray ends, the air gun solenoid valve is closed.

[0013] Optionally, in the eighth implementation of the first aspect of the present invention, after performing material screening according to the corresponding predicted trigger time, the method further includes: The actual arrival time of each trigger point is subtracted from the predicted trigger time to obtain the time residual of each trigger point. The mean of the time residual of each trigger point is then calculated using a sliding window to obtain the mean of the residual of each trigger point. When the absolute value of the mean residual exceeds the preset residual threshold, the product of the mean residual and the instantaneous speed of the belt is used as the distance correction amount for each trigger point. The distance correction amount is then superimposed on the first distance calibration value of each trigger point to obtain the second distance calibration value.

[0014] This invention also provides a multi-level intelligent sorting system for construction waste, comprising: The recognition module is used to identify the longitudinal position, equivalent particle size, aspect ratio, and mean gray value of each material based on the material image. The calculation module is used to calculate the predicted trigger time of each material reaching the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point based on the equivalent particle size, the length-to-short axis ratio, and the longitudinal position. The material screening module is used to perform material screening according to the corresponding predicted trigger time when each material reaches the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

[0015] In summary, this invention uses the material morphology features obtained from image acquisition as a unified input for both trigger timing prediction and actuator parameter adjustment, thus solving the technical problems of trigger timing fixation and execution parameter fixation in the prior art. Regarding trigger timing, by introducing a morphological velocity correction coefficient based on equivalent particle size and length-to-short axis ratio, the instantaneous belt speed is corrected material-by-material. This ensures that the predicted trigger times at each sorting stage simultaneously reflect the real-time fluctuations in belt speed and the influence of material morphology on the movement speed, resulting in a substantial improvement in trigger timing accuracy compared to traditional fixed-delay methods. In terms of actuator parameters, the output frequency of the vibrating screen inverter, the output frequency of the fan inverter, the pulse drive current of the electromagnetic coil, and the duration and pressure of the air gun injection are all calculated independently based on the image characteristics of each material. This achieves piece-by-piece adaptive control throughout the entire process from coarse screening to fine screening, avoiding over-triggering or under-triggering caused by uniform parameters for different materials. The air gun fine sorting stage also performs pre-compensation for the mechanical response delay of the solenoid valve, ensuring precise synchronization between the injection action and the material arrival time. Furthermore, the trigger distance online correction mechanism based on the mean residual value of the sliding window enables the equipment to continuously adapt to long-term operating conditions such as belt wear and temperature rise deformation, ensuring the stability of the trigger timing during long-term operation of the sorting system. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the adaptive control method for multi-level intelligent sorting of construction waste in one embodiment of the present invention; Figure 2 This is a structural block diagram of a multi-level intelligent sorting system for construction waste in an embodiment of the present invention.

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Reference Figure 1 This embodiment provides an adaptive control method for multi-level intelligent sorting of construction waste, including the following steps: S1, based on material image recognition, the longitudinal position, equivalent particle size, aspect ratio and grayscale mean of each material; S2, based on equivalent particle size, length-to-short axis ratio and longitudinal position, calculates the predicted trigger time of each material reaching the vibrating screen trigger point, wind separation trigger point, eddy current iron removal trigger point and air gun fine separation trigger point. S3: When each material reaches the trigger point of vibrating screen, wind separation trigger point, eddy current iron removal trigger point and air gun fine separation trigger point, material screening is performed according to the corresponding predicted trigger time.

[0020] In one example, the longitudinal position, equivalent particle size, aspect ratio, and mean grayscale value of each material are identified based on the material image, including: Acquire the material image and binarize the material image and the background reference image to obtain the material foreground mask; Connectivity analysis is performed on the foreground mask of the materials to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio and gray value are calculated.

[0021] In this example, an industrial area scan camera is fixedly installed above the feed conveyor belt, and the camera's trigger acquisition cycle is preset to 5ms, so that the camera continuously acquires grayscale image frames P covering the entire width of the conveyor belt at a fixed frequency. n Simultaneously, a frame of the conveyor belt without material interference was acquired while the equipment was running unloaded, serving as the background reference image P. bg And the background reference map P bg The image is stored in the image buffer module of the control system. During each acquisition cycle, the currently acquired grayscale image frame P is... n Compared with the background reference map P bg Perform pixel-by-pixel difference calculations, according to D n =|P n P bg |Calculate the difference image D n By using differential operations, static interference components caused by belt texture and fixed structure are eliminated, while retaining the brightness variation information of the material area relative to the background, and the differential image D is processed. n Apply grayscale threshold T 灰 =30 performs binarization processing. When the grayscale value of a pixel is greater than or equal to 30, it is identified as a foreground pixel and assigned a value of 1. When the grayscale value is less than 30, it is identified as a background pixel and assigned a value of 0, thus obtaining the material foreground mask M. n This separates the material area from the background area in the image space. The control system performs a connected component labeling algorithm on the material foreground mask, clustering all pixels with a value of 1 according to the 8-neighborhood connectivity rule. Each independent connected region is identified as an independent material individual i, and its geometric feature parameters are calculated for each connected region. During the geometric feature calculation, the number of pixels in the connected region is counted, and the projected area A is obtained by combining the pixel ratio obtained from calibration with the actual size ratio. i (Unit: mm) 2 Then calculate the centroid pixel coordinates (u) of the connected component. i ,v i ), where v iThe vertical pixel coordinates are determined along the belt movement direction, and then the vertical calibration coefficient K obtained during the installation and commissioning phase is used. v (Unit: px / mm) Convert the vertical pixel coordinates to the actual vertical position x in the belt coordinate system. i , satisfying x i =v i / K v This allows for the acquisition of the longitudinal position of each material; in terms of equivalent particle size calculation, it is based on the projected area A of the connected domain. i Using D eq,i =2×√(A i Calculate the equivalent circle diameter D of the material using the formula / π). eq,i Equivalent particle size is used to characterize the size scale of irregular materials in the sense of planar projection; in terms of morphological feature extraction, the major axis length L is obtained by fitting the contour of the connected domain with the minimum bounding rectangle or ellipse. i With minor axis length W i And calculate the major-minor axis ratio AR i =L i / W i This is used to characterize the degree of stretching and irregularity of material morphology; simultaneously, in the grayscale feature calculation stage, the difference image D is used. n Based on this, the original grayscale values ​​or differential grayscale values ​​within the corresponding regions of the connected component are statistically analyzed, and the average grayscale value μ of all pixels within the connected component is calculated. i The average gray level is used as a characterization index of the reflectivity of the material surface.

[0022] In one example, connected component analysis is performed on the foreground mask of the materials to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio, and mean gray value are calculated, including: Connectivity analysis is performed on the material foreground mask to extract the centroid vertical pixel coordinates of each connected region, and the centroid vertical pixel coordinates are converted into the vertical position in the belt coordinate system based on the vertical calibration coefficient. The equivalent particle size is calculated based on the projected area of ​​each connected domain. At the same time, the contour of each connected domain is elliptical fitted to obtain the length of the major axis and the length of the minor axis. The ratio of the major axis to the minor axis is then calculated based on the length of the major axis and the length of the minor axis. The arithmetic mean of the gray values ​​of all pixels in each connected region is calculated to obtain the gray mean of each material.

[0023] In this example, the material foreground mask M nConnectivity analysis is performed, clustering all pixels with a value of 1 in the mask according to the 8-neighborhood connectivity rule. This ensures that each group of adjacent and connected foreground pixels forms an independent connected component, and each independent connected component corresponds to a material element i. After connecting component labeling, the coordinates of all pixels within each connected component are calculated, and the centroid pixel coordinates (u) of the connected component are calculated. i ,v i ), where u i v represents the horizontal pixel coordinate. i Let v be the vertical pixel coordinate along the direction of belt movement, and v be the vertical pixel coordinate of the centroid. i The vertical calibration coefficient K is obtained by taking the arithmetic mean of the vertical coordinates of all pixels within the connected component. v (Unit: px / mm), the vertical pixel coordinate of the centroid v i Convert to actual longitudinal position x in belt coordinate system i The transformation relation satisfies x i =v i / K v This yields the longitudinal position of the material in millimeters. For equivalent particle size calculation, the number of pixels within each connected region is counted, and the projected area A of that region is calculated by combining the pixel count with the calibrated ratio of the actual size to the total pixel count. i (Unit: mm) 2 After performing connected component analysis on the material foreground mask, the number of pixels N contained in each connected component is counted. px,i And convert the number of pixels into the actual projected area A using a pixel area calibration coefficient. i The pixel area calibration coefficient is obtained from camera calibration and is used to establish the conversion relationship between image pixel coordinates and actual spatial dimensions. Let the pixel area conversion coefficient be Ka, then the projected area satisfies A. i = N px,i / K a K a This is a pixel area conversion factor, in pixels (px). 2 / mm 2 N px,i Let be the number of pixels in the i-th connected component. This conversion relationship allows us to convert the pixel area in the image to the actual area, expressed in mm. 2 Material projection area A i According to D eq,i =2×√(A i Calculate the equivalent particle size D using / π) eq,iThis method transforms irregularly shaped materials into equivalent circle diameters for uniform scale description under the concept of area equivalence. Simultaneously, to obtain material morphological features, the outer contour of each connected region is extracted, and an ellipse fitting operation is performed on the contour point set to obtain the major axis length L of the fitted ellipse. i With minor axis length W i Therefore, based on AR i =L i / W i Calculate the major-minor axis ratio AR i This is used to characterize the stretching and irregularity of the material's shape. In the grayscale feature extraction stage, the grayscale values ​​of all pixels corresponding to the original grayscale image or difference image within each connected component are statistically analyzed, and the arithmetic mean of all pixel grayscale values ​​is calculated to obtain the grayscale mean μ. i The mean grayscale value is used to reflect the differences in the reflective properties of the material surface.

[0024] Before calculating the prediction delay based on the morphological velocity correction coefficient, belt instantaneous speed, and longitudinal position, the process includes: acquiring the number of pulses output by the incremental encoder within the sampling period at the end shaft of the belt drive roller; dividing the product of the pulse number and the circumference of the drive roller by the product of the encoder resolution and the sampling period to obtain the belt instantaneous linear velocity at each sampling moment; smoothing the belt instantaneous linear velocity at multiple consecutive sampling moments using an arithmetic mean using a sliding window to obtain the smoothed belt instantaneous velocity; wherein, when the difference between the belt instantaneous linear velocities at adjacent sampling moments exceeds a preset fluctuation threshold, the smoothed belt instantaneous velocity replaces the belt instantaneous linear velocity at the current sampling moment in the subsequent prediction delay calculation; when the difference does not exceed the preset fluctuation threshold, the belt instantaneous linear velocity at the current sampling moment is directly used in the calculation; synchronizing and aligning the smoothed belt instantaneous velocity with the timestamp corresponding to the image acquisition completion moment to obtain the belt instantaneous velocity strictly corresponding to the image acquisition moment of each material, which is used for the calculation of the morphological velocity correction coefficient and prediction delay for each material.

[0025] In one example, the predicted trigger times for each material reaching the vibrating screen trigger point, air separation trigger point, eddy current iron removal trigger point, and air gun fine separation trigger point are calculated based on equivalent particle size, aspect ratio, and longitudinal position, including: Based on the equivalent particle size and reference particle size, and the ratio of the major axis to the minor axis and the reference major axis, the instantaneous speed of the belt is corrected to obtain the morphological speed correction coefficient for each material. Based on the first distance calibration value and longitudinal position of each trigger point, the remaining distance of each material to reach each trigger point is calculated. The remaining distance is divided by the product of the belt instantaneous speed and the morphological speed correction coefficient to obtain the predicted time delay of each material to reach each trigger point. Among them, the trigger points at each level are the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point. The predicted delay is added to the image acquisition completion time to obtain the predicted trigger time of each material reaching the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

[0026] In this example, the instantaneous belt speed is morphologically corrected to eliminate the speed deviation caused by differences in rolling, sliding, and friction between materials of different particle sizes and shapes during actual movement. Instantaneous belt speed v 皮带 Measured by an incremental rotary encoder mounted on the end shaft of the drive roller, during the sampling period T s =N number of pulses counted within 1ms 脉冲 And based on v 皮带 =N 脉冲 ×(2πr / P enc ) / T s The calculation yielded that the encoder resolution P enc =1000 pulses / revolution, drive roller radius r=150mm, thus obtaining the instantaneous linear velocity in mm / ms, and introducing a shape velocity correction coefficient κ. i Its calculation formula is κ i =1 λ×(D eq,i / D ref )×(AR i AR ref ), where λ=0.04 is the morphological influence weighting coefficient, D ref =200mm is the reference particle size, AR ref =3.0 is the reference major axis ratio. This expression is used to make the material equivalent particle size D... eq,i Increased and larger aspect ratio AR i When the deviation is 3, the correction factor κ i The corresponding reduction reflects the fact that the actual movement speed of large-sized or irregularly shaped materials is lower than the belt linear speed. This completes the shape correction coefficient κ. i After calculation, first distance calibration values ​​L1, L2, L3, and L4 are pre-set for the trigger points of vibrating screen, air separation, eddy current iron removal, and air gun fine separation, respectively. These first distance calibration values ​​are the longitudinal physical distances of each trigger point relative to the image acquisition reference position, and are used as the spatial reference for time-driven calculations during operation. For any material i, the remaining distance ΔL to each trigger point is calculated. n,i =L n x i Where n takes values ​​of 1, 2, 3, and 4, corresponding to the four trigger points respectively, the actual spatial difference between the current position of the material and each trigger point is obtained. The remaining distance is divided by the instantaneous belt speed v. 皮带With the shape velocity correction coefficient κ i The product of these two values ​​yields the predicted delay T for material i to reach the nth trigger point. 预测,n,i =ΔL n,i / (v 皮带 ×κ i This formula maps spatial distance to time delay, while also taking into account the speed correction effects caused by real-time changes in belt speed and differences in material shape. The predicted time delay T... 预测,n,i By overlaying the image acquisition completion time with the predicted trigger time of material i at each trigger point, the predicted trigger time is obtained. The image acquisition completion time is the time reference for the current material being captured by the image system and completing feature extraction. The consistency and traceability of time calculation are ensured by a unified system clock. Thus, the predicted trigger times for the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point are all obtained.

[0027] In one example, the instantaneous belt speed is corrected based on the equivalent particle size and reference particle size, and the aspect ratio and reference aspect ratio, to obtain the morphological velocity correction coefficient for each material, including: The equivalent particle size of each material is compared with the reference particle size to obtain the particle size ratio. The difference between the length-to-short axis ratio and the preset reference length-to-short axis ratio is obtained to obtain the target difference. The particle size ratio is multiplied by the target difference and then multiplied by the morphology influence weighting coefficient to obtain the morphology correction amount of each material. Subtract the shape correction amount from 1 to obtain the shape velocity correction coefficient for each material.

[0028] In this example, the equivalent particle size D eq,i With the preset reference particle size D ref Perform ratio calculations, where a preset reference particle size D is used. ref Taking 200 mm, the particle size ratio D is obtained. eq,i / D ref This is used to characterize the degree of scaling up or down of the current material size relative to the standard particle size; simultaneously, it measures the material's aspect ratio (AR). i Compared with the preset reference major axis ratio AR ref Perform interpolation calculations, where a preset reference major axis ratio AR is used. ref Take 3.0, and calculate AR i AR ref The target difference is obtained to reflect the degree to which the material shape deviates from the reference shape. When AR i A length-to-length ratio greater than the reference ratio indicates that the material shape is more elongated or irregular. i A length-to-width ratio close to or less than the reference ratio indicates that the material shape tends to be regular. To comprehensively consider the combined effects of size and shape on velocity deviation, the particle size ratio is multiplied by the target difference, i.e., (D...eq,i / D ref )×(AR i AR ref The overall morphological deviation is obtained, and then multiplied by the morphological influence weighting coefficient λ. The morphological influence weighting coefficient λ is preset to 0.04 to characterize the weighting ratio of particle size and morphology on velocity, thus obtaining the morphological correction amount λ×(D) for each material. eq,i / D ref )×(AR i AR ref Subtracting the shape correction amount from 1 yields the shape velocity correction coefficient κ for each material. i , i.e. κ i =1 λ×(D eq,i / D ref )×(AR i AR ref This expression allows for an increase in morphology correction as particle size and irregularity increase, thereby affecting the morphology velocity correction coefficient κ. i Reducing the equivalent velocity in subsequent arrival time calculations reflects the physical characteristic that the actual conveying speed of large-diameter or high aspect ratio materials is lower than the instantaneous belt speed due to increased rolling friction and frequent posture changes. Conversely, when the material particle size is close to the reference particle size and the shape is close to the reference aspect ratio, the shape correction amount approaches zero, and the shape velocity correction coefficient κ... i A value close to 1 indicates that the actual speed of the material movement is basically consistent with the linear speed of the belt.

[0029] Before obtaining the material's wind-receiving area by multiplying the projected area of ​​the connected domain corresponding to the equivalent particle size with the sine of the main shaft deflection angle, the process includes: extracting the major axis direction vector of the fitted ellipse from the contour ellipse fitting results of each connected domain; performing a dot product operation between the major axis direction vector and the unit vector of the belt's longitudinal movement direction; taking the inverse cosine of the dot product result to obtain the main shaft deflection angle between the main shaft of each material and the belt's movement direction; when the main shaft deflection angle is greater than 90°, replacing the main shaft deflection angle with 180° minus the main shaft deflection angle, limiting the range of the main shaft deflection angle to between 0° and 90° to obtain the corrected main shaft deflection angle; where the closer the main shaft deflection angle is to 90°, the closer the material's major axis direction is to being perpendicular to the belt's movement direction, and the larger the material's wind-receiving projected area; taking the sine of the corrected main shaft deflection angle, multiplying the sine value with the projected area of ​​the corresponding connected domain to obtain the wind-receiving area of ​​each material, which is used for the subsequent calculation of the target output frequency of the fan inverter.

[0030] In one example, when each material reaches the trigger point of the vibrating screen, the trigger point of the air separation, the trigger point of the eddy current iron removal, and the trigger point of the air gun fine separation, material screening is performed according to the corresponding predicted trigger time, including: When the material reaches the vibrating screen trigger point, the corresponding first output frequency is written to the vibrating screen inverter according to the equivalent particle size and the corresponding predicted trigger time, and the vibrating screen is driven to complete the screening and separation of the material at the first output frequency. When the material reaches the wind separation trigger point, the corresponding second output frequency is written to the wind turbine inverter based on the wind-receiving area and the corresponding predicted trigger time, and the wind turbine is driven to complete the aerodynamic deflection separation of the lightweight material at the wind speed corresponding to the second output frequency. When the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material with the electromagnetic coil pulse drive current. When the material reaches the air gun's fine sorting trigger point, the air gun is driven to accurately spray and deflect the target material based on the equivalent particle size and the predicted trigger time.

[0031] In this example, the control system uses a unified system clock as a reference for time-driven scheduling. When the system clock reaches the predicted trigger time corresponding to the vibrating screen trigger point, the equivalent particle size D of the material is read. eq,i The first output frequency f of the vibrating screen is determined based on the equivalent particle size range. vib , where D eq,i When the value is greater than 200mm, take f. vib =14Hz, putting the vibrating screen in a low-frequency, large-amplitude mode to enhance the tumbling stroke of large materials, when 50mm≤D eq,i When ≤200mm, according to the linear interpolation relationship f vib =14+10×(200 D eq,i ) / 150 calculates the intermediate frequency, when D eq,i When the diameter is less than 50mm, take f. vib =24Hz, putting the vibrating screen in a high-frequency, low-amplitude mode to accelerate the separation of fine particles. The frequency is written before the clock reaches the predicted trigger time corresponding to the vibrating screen trigger point, ensuring that the vibrating screen motor is already running stably at the corresponding first output frequency when the material reaches the effective action area of ​​the screen surface, thereby completing the particle size classification and screening of the material; when the system clock reaches the predicted trigger time corresponding to the wind separation trigger point, the equivalent particle size D of the material is also read. eq,i The second output frequency f is generated by combining the windward area calculation results. wind The wind turbine reference frequency f base =35Hz, adjust gain γ=8Hz, reference windward area S wind,ref =10000mm 2 and according to fwind =f base +γ×(S wind,ref S wind,i ) / S wind,ref To ensure the stability of the fan operation, the control system calculates the target frequency and limits the calculated fan output frequency to ensure that the output frequency meets the target frequency requirement. min ≤ f wind ≤ f max , where f min with f max These are the minimum and maximum permissible operating frequencies of the wind turbine inverter, respectively. Let f be the value. min = 20Hz, f max = 50Hz. When the calculated f wind Less than f min f min When the calculation result is greater than f max f max This ensures that the wind turbine operates within a safe and stable range. The wind-receiving area is calculated from the angle between the projected area of ​​the connected region and the direction of the principal axis, and the wind-receiving area satisfies S... wind,i = A i × sin(θ i ), where A i Let θ be the projected area of ​​the connected domain. i This is the angle between the material main shaft and the belt's direction of movement. Then, the second output frequency f is written to the fan inverter. wind This makes the fan output match f wind The corresponding wind speed creates a matching airflow intensity the instant the material arrives at the wind-driven sorting area. This allows the lightweight material to gain sufficient lift under the appropriate wind speed to complete aerodynamic deflection and separation, while the heavy aggregate continues to be conveyed along its original trajectory. When the system clock reaches the predicted trigger time corresponding to the eddy current iron removal trigger point, if the grayscale mean μ... i Greater than the metal recognition threshold T 金属 If the particle size is 180, it is determined to be a metallic material, and the equivalent particle size D is used as the basis for classification. eq,i Calculate the pulse drive current I of the electromagnetic coil coil The reference current I base =5A, reference metal grain size D 参考金属 =100mm, the calculation relationship is I coil =I base ×(D eq,i / D 参考金属)^0.5, and then send the corresponding peak current pulse to the electromagnetic coil drive module, so that the electromagnetic coil generates a pulse magnetic field of matching strength at the predicted trigger time corresponding to the eddy current removal trigger point. Through eddy current induction and Lorentz force, the metal material is deflected and removed; when the system clock reaches the predicted trigger time corresponding to the air gun fine sorting trigger point, according to the equivalent particle size D eq,i Calculate the injection duration τ gun Where α = 0.05 ms / mm and β = 8 ms, satisfying τ gun =α×D eq,i +β, combined with the valve's inherent mechanical response delay Δt valve =3.5ms advance compensation is performed to advance the control command issuance time by 3.5ms. At this time, the opening command is written to the air gun solenoid valve controller and continues for τ gun The time is calculated, and the injection pressure P is also calculated based on the estimated material mass. gun , where P base =0.5MPa, m ref =0.5kg, satisfying P gun =P base ×(m i / m ref The parameter λ^0.4 is used to ensure the air gun accurately deflects the material upon arrival at the specified pressure and duration, enabling the four-stage sorting equipment to drive the vibrating screen, fan, electromagnetic coil, and air gun to perform corresponding actions at their respective predicted trigger times. The parameters λ in the morphology-velocity correction coefficient, α and β in the air gun spray duration formula, the electromagnetic coil drive current index of 0.5, and the spray pressure index of 0.4 are all obtained through regression fitting of experimental data. By collecting experimental data on the movement speed of materials of different particle sizes and shapes on the conveyor belt, the air gun spray deflection effect, and the eddy current iron removal repulsion effect, a minimum error objective function is constructed. The optimal values ​​of the above parameters are then determined through regression fitting to minimize the material triggering timing error and sorting deviation.

[0032] In one example, when the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material using the electromagnetic coil pulse drive current, including: When the material reaches the eddy current iron removal trigger point, the material is identified as a metal based on the average gray value. When the material is determined to be a metal, the electromagnetic coil pulse drive current is calculated based on the equivalent particle size. The electromagnetic coil is driven by the electromagnetic coil pulse drive current to generate eddy current repulsion force on the metal material, thereby completing the deflection and removal of the metal material.

[0033] In this example, when the system clock reaches the predicted trigger time, the material type is determined based on the grayscale statistical features output by the image recognition module, where the grayscale mean μi Derived from the difference image D n The set of gray values ​​of pixels in the corresponding connected region is used to calculate the mean gray value μ by taking the arithmetic mean of all gray values ​​in the connected region. i Simultaneously, high-gray pixels are statistically analyzed within the connected component's grayscale distribution, and pixels with grayscale values ​​greater than a preset high-gray threshold T are identified. 高 The number of pixels is denoted as N. 高 And let N be the total number of pixels in the connected component. 连 By calculating the proportion of high gray pixels η i =N 高 / N 连 The proportion of bright spots in the connected components is obtained, where the high gray threshold is preset to 180. The control system calculates the grayscale mean μ. i With high grayscale pixel ratio η i As a joint feature for metal category identification, when the gray-scale mean μ i The proportion of high gray pixels (η) is greater than the preset grayscale threshold. i If the bright spot percentage exceeds a preset threshold, the material is determined to be a metal. The preset grayscale threshold is 180, and the preset high gray pixel percentage threshold is 0.35. After determining the material to be a metal, the eddy current iron removal process begins, and the material's equivalent particle size D is read. eq,i The electromagnetic coil pulse drive current I is calculated based on the equivalent particle size. coil The calculation uses I coil =I base ×(D eq,i / D 参考金属 )^0.5, where I base The default value is 5A and D. 参考金属 The preset size is 100mm, and the exponent is 0.5 to reflect the geometrical impact of particle size variation on the eddy current induction area and repulsive force requirements. This allows for a higher pulse drive current for larger metal particles to achieve a stronger repulsive effect, while a lower pulse drive current is used for smaller metal particles to avoid excessive magnetic field interference with the trajectories of adjacent non-metallic materials. Write I to the electromagnetic coil driver. coil The peak pulse trigger command drives the electromagnetic coil to generate a pulse magnetic field at the predicted trigger time corresponding to the eddy current iron removal trigger point. This induces eddy currents inside the metal material, which interact with the external magnetic field to generate a Lorentz repulsive force. This deflects the metal material off the belt track and ejects it into the metal recycling tank, achieving precise deflection, removal, and recycling of the metal material.

[0034] In one example, when the material reaches the air gun's fine sorting trigger point, the air gun is driven to precisely deflect the target material based on the equivalent particle size and the predicted trigger time, including: When the material reaches the air gun fine sorting trigger point, the corresponding predicted trigger time is subtracted from the preset air gun valve mechanical response delay to obtain the air gun trigger command issuance time; The air gun spraying duration, material mass, and spraying pressure are determined based on the equivalent particle size. When the air gun trigger command is issued, an opening command is written to the air gun solenoid valve, and an injection pressure command is written to the pressure reducing valve. The injection pressure is used to maintain the air gun injection duration to achieve precise spray deflection of the target material. After the spraying is completed, the air gun solenoid valve is closed.

[0035] In this example, the predicted trigger time corresponding to the air gun fine sorting trigger point is used as the time reference, and the mechanical response delay Δt of the air gun valve measured during the equipment calibration phase is also used. valve Pre-compensation is performed, including a preset mechanical response delay Δt for the air gun valve. valve Taking 3.5ms, the predicted trigger time is subtracted from the mechanical response delay to obtain the air gun trigger command issuance time, ensuring that the air gun valve aligns precisely with the actual arrival time of the material in the air gun spray area after completing its mechanical opening action. This is based on the material's equivalent particle size D. eq,i Calculate the air gun spray duration τ gun The injection duration satisfies τ gun =α×D eq,i +β, α is set to 0.05 ms / mm and β is set to 8 ms, so that larger particle sizes correspond to longer injection times to accumulate sufficient airflow impulse. At the same time, the equivalent mass index m is estimated according to the density model corresponding to the material type. i The equivalent particle size is derived from the equivalent circle diameter calculated from the projected area of ​​the connected domain. Since construction waste often exhibits polyhedral, flaky, or porous structures after crushing, there is no unique correspondence between the actual volume and the two-dimensional projected area. Therefore, in the spray control calculation, the mass derived from the equivalent particle size is defined as the equivalent mass index. This index represents the material's inertial scale, not its actual mass, and is used as a relative control quantity for adjusting the air gun pressure. The concrete material mass satisfies m... i =2000×(π / 6)×(D eq,i / 1000) 3 The quality of the timber materials meets m i =600×(π / 6)×(D eq,i / 1000) 3 The quality of the plastic material meets m i =900×(π / 6)×(D eq,i / 1000) 3 The mass is estimated by converting the equivalent particle size to volume and multiplying by the density. After obtaining the material mass m_i, the injection pressure P is calculated. gun The injection pressure satisfies P gun=P base ×(m i / m ref )^0.4, P base The preset value is 0.5 MPa and m ref The preset pressure is 0.5 kg. The non-linear growth relationship of the required injection pressure for different material masses is adjusted using an exponent of 0.4. This allows heavier materials to receive higher pressure for effective deflection, while preventing over-injection of lighter materials that could cause trajectory disturbances. When the system clock reaches the point where the air gun trigger command is issued, the control system simultaneously writes an opening command to the air gun solenoid valve and the target injection pressure P to the pressure reducing valve. gun This allows the air gun to open at the predicted trigger moment corresponding to the air gun's fine sorting trigger point, and to apply pressure P. gun Continuous injection τ gun Time applies airflow impact to the target material to achieve precise spray deflection; the spray duration τ gun After completion, the control system automatically writes a closing command to the air gun solenoid valve to reset the valve, thus completing one air gun fine sorting control cycle.

[0036] Before calculating the material mass based on the product of equivalent particle size and material density, the process includes: comparing the average gray value with a preset metal identification gray value threshold; if the average gray value is greater than the metal identification gray value threshold, it is determined that the material has been removed in the eddy current iron removal stage, and the air gun fine sorting trigger is not performed on the material; if the average gray value is not greater than the metal identification gray value threshold, the average gray value and the aspect ratio are further input into the density category discrimination rule. The density category discrimination rule is: when the average gray value is in the concrete gray value range and the aspect ratio is less than the upper limit threshold of the aspect ratio, the material density category is determined to be concrete, and the corresponding density value is assigned. When the average grayscale value is within the wood grayscale range and the aspect ratio is not less than the upper limit threshold of the aspect ratio, the material density category is determined to be wood, and a corresponding density value is assigned. When the average grayscale value is within the plastic grayscale range and the aspect ratio is less than the upper limit threshold of the aspect ratio, the material density category is determined to be plastic, and a corresponding density value is assigned. The material mass is calculated based on the product of the density value corresponding to the density category and the cube of the equivalent particle size. The material mass is then compared with a preset reference mass, and the comparison value is taken as a preset power. The power value is multiplied by a preset benchmark injection pressure to obtain the injection pressure corresponding to the material, which is used for writing the pressure command of the air gun pressure reducing valve in the subsequent process.

[0037] In one example, after performing material screening based on the corresponding predicted trigger time, the process also includes: The actual arrival time of each trigger point is subtracted from the predicted trigger time to obtain the time residual of each trigger point. The mean of the time residual of each trigger point is then calculated using a sliding window to obtain the mean of the residual of each trigger point. When the absolute value of the residual mean exceeds the preset residual threshold, the product of the residual mean and the instantaneous speed of the belt is used as the distance correction amount for each trigger point. The distance correction amount is then superimposed on the first distance calibration value of each trigger point to obtain the second distance calibration value.

[0038] In this example, laser beam sensors or equivalent arrival detection units are respectively arranged at the trigger points of vibrating screen, air separation, eddy current iron removal, and air gun fine separation. These are used to record in real time the actual time when the material passes through the corresponding trigger position, and the actual arrival time is recorded as T. 实际,n,i Where the subscript n corresponds to the trigger points from the first to the fourth level, respectively; meanwhile, in the preceding steps, the time t after image acquisition has already been determined. cap,i Instantaneous speed of the belt v 皮带 and the morphological velocity correction coefficient κ i The predicted trigger times for each level are calculated. Then, after the material arrives at the corresponding trigger area, the difference between the actual arrival time and the predicted trigger time is calculated to obtain the timing residual e of the material at the nth level trigger point. n,i , when e n,i A positive value indicates that the actual arrival time lags behind the predicted time. n,i A negative value indicates that the actual arrival time is earlier than the predicted time. A sliding window is constructed for each trigger point, with the sliding window capacity W preset to 100, meaning that the timing residual e of the most recent 100 materials is continuously recorded. n,i And perform mean calculation, according to ē n =(1 / 100)×Σe n,i The mean residual of the nth trigger point is calculated to smooth out random fluctuations, making the statistical results better reflect the trend of persistent deviation. After obtaining the mean residual ē_n of each trigger point, the mean residual is compared with a preset residual threshold, where the preset residual threshold is 2ms. When |ē n When | is less than or equal to 2ms, the current calibration distance error is considered to be within the allowable range, and the original calibration distance value remains unchanged. However, when | ē n If the time is greater than 2ms, a persistent system deviation is determined, and the current instantaneous belt speed v is read. 皮带 And calculate the distance correction ΔL n,修正 =ē n ×v 皮带 This product converts the time deviation into an equivalent spatial deviation, because the instantaneous belt speed v 皮带 The unit is mm / ms, thus giving the product of the residual mean and the belt speed a physical meaning in millimeters. The distance correction is superimposed on the first distance calibration value L at each trigger point. n That is, execute L n ←L n +ΔL n,修正This yields the updated second distance calibration value.

[0039] Reference Figure 2 This embodiment provides a multi-level intelligent sorting system for construction waste, including: The identification module 1 is used to identify the longitudinal position, equivalent particle size, aspect ratio, and mean gray value of each material based on the material image. Calculation module 2 is used to calculate the predicted trigger time of each material reaching the vibrating screen trigger point, wind separation trigger point, eddy current iron removal trigger point and air gun fine separation trigger point based on equivalent particle size, length-to-short axis ratio and longitudinal position. Material screening module 3 is used to perform material screening according to the corresponding predicted trigger time when each material reaches the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

[0040] In this embodiment, the specific implementation of each unit in the above device embodiment is described in the above method embodiment, and will not be repeated here.

[0041] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0042] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An adaptive control method for multi-level intelligent sorting of construction waste, characterized in that, include: Based on material image recognition, the longitudinal position, equivalent particle size, aspect ratio, and mean gray value of each material are identified. Based on the equivalent particle size, the length-to-short axis ratio and the longitudinal position, the predicted trigger time for each material to reach the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point and the air gun fine separation trigger point is calculated. When each material reaches the trigger point of the vibrating screen, the trigger point of the air separation, the trigger point of the eddy current iron removal, and the trigger point of the air gun fine separation, the material screening is performed according to the corresponding predicted trigger time.

2. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 1, characterized in that, Based on material image recognition, the longitudinal position, equivalent particle size, aspect ratio, and mean grayscale value of each material are identified, including: Acquire a material image and binarize the material image and the background reference image to obtain a material foreground mask; Connectivity analysis is performed on the foreground mask of the material to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio and grayscale mean are calculated.

3. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 2, characterized in that, Connectivity analysis is performed on the foreground mask of the materials to obtain the longitudinal position of each material, and the equivalent particle size, aspect ratio, and mean gray value are calculated, including: Connectivity analysis is performed on the material foreground mask to extract the centroid longitudinal pixel coordinates of each connected region, and the centroid longitudinal pixel coordinates are converted into longitudinal positions in the belt coordinate system based on the longitudinal calibration coefficient. The equivalent particle size is calculated based on the projected area of ​​each connected domain. At the same time, the contour of each connected domain is elliptical-fitted to obtain the length of the major axis and the length of the minor axis. The ratio of the major axis to the minor axis is then calculated based on the length of the major axis and the length of the minor axis. The arithmetic mean of the gray values ​​of all pixels in each connected region is calculated to obtain the gray mean of each material.

4. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 1, characterized in that, Based on the equivalent particle size, the aspect ratio, and the longitudinal position, the predicted trigger times for each material to reach the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point are calculated, including: Based on the equivalent particle size and the reference particle size, and the ratio of the major axis to the minor axis and the reference major axis, the instantaneous speed of the belt is corrected to obtain the morphological speed correction coefficient for each material. Based on the first distance calibration value of each trigger point and the longitudinal position, the remaining distance of each material to reach each trigger point is calculated. The remaining distance is divided by the product of the instantaneous speed of the belt and the morphological speed correction coefficient to obtain the predicted time delay of each material to reach each trigger point. The trigger points at each level are the vibrating screen trigger point, the air separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point. The predicted delay is added to the image acquisition completion time to obtain the predicted trigger time of each material reaching the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

5. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 4, characterized in that, Based on the equivalent particle size and reference particle size, and the aspect ratio and reference aspect ratio, the instantaneous belt speed is corrected to obtain the morphological velocity correction coefficient for each material, including: The equivalent particle size of each material is compared with the reference particle size to obtain the particle size ratio. The difference between the major and minor axis ratios and the preset reference major axis ratios is obtained to obtain the target difference. The particle size ratio is multiplied by the target difference and then multiplied by the morphology influence weighting coefficient to obtain the morphology correction amount of each material. Subtract the morphology correction amount from 1 to obtain the morphology velocity correction coefficient for each material.

6. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 1, characterized in that, When each material reaches the trigger point of the vibrating screen, the trigger point of the air separation, the trigger point of the eddy current iron removal, and the trigger point of the air gun fine separation, material screening is performed according to the corresponding predicted trigger time, including: When the material reaches the vibrating screen trigger point, the corresponding first output frequency is written to the vibrating screen inverter according to the equivalent particle size and the corresponding predicted trigger time, and the vibrating screen is driven to complete the screening and separation of the material at the first output frequency. When the material reaches the wind-driven sorting trigger point, the corresponding second output frequency is written to the wind turbine inverter based on the wind-receiving area and the corresponding predicted trigger time, and the wind turbine is driven to complete the aerodynamic deflection separation of the lightweight material at the wind speed corresponding to the second output frequency. When the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material with the electromagnetic coil pulse drive current. When the material reaches the air gun's fine sorting trigger point, the air gun is driven to accurately spray and deflect the target material based on the equivalent particle size and the predicted trigger time.

7. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 6, characterized in that, When the material reaches the eddy current removal trigger point, the electromagnetic coil pulse drive current is calculated, and the electromagnetic coil is driven to deflect and remove the metal material using the electromagnetic coil pulse drive current, including: When the material reaches the eddy current iron removal trigger point, the material is identified as a metal based on the average gray value. When the material is determined to be a metal, the electromagnetic coil pulse drive current is calculated based on the equivalent particle size. The electromagnetic coil is driven by the electromagnetic coil pulse drive current to generate eddy current repulsion force on the metal material, thereby completing the deflection and removal of the metal material.

8. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 1, characterized in that, When the material reaches the air gun's fine sorting trigger point, the air gun is driven to precisely deflect the target material based on the equivalent particle size and the predicted trigger time, including: When the material reaches the air gun fine sorting trigger point, the corresponding predicted trigger time is subtracted from the preset air gun valve mechanical response delay to obtain the air gun trigger command issuance time; The air gun spraying duration, material mass, and spraying pressure are determined based on the equivalent particle size. When the air gun trigger command is issued, an opening command is written to the air gun solenoid valve, and a spray pressure command is written to the pressure reducing valve. The spray pressure is maintained for the duration of the air gun spray to achieve precise spray deflection of the target material. After the spray ends, the air gun solenoid valve is closed.

9. The adaptive control method for multi-level intelligent sorting of construction waste according to claim 8, characterized in that, After performing material screening based on the corresponding predicted trigger time, the process also includes: The actual arrival time of each trigger point is subtracted from the predicted trigger time to obtain the time residual of each trigger point. The mean of the time residual of each trigger point is then calculated using a sliding window to obtain the mean of the residual of each trigger point. When the absolute value of the mean residual exceeds the preset residual threshold, the product of the mean residual and the instantaneous speed of the belt is used as the distance correction amount for each trigger point. The distance correction amount is then superimposed on the first distance calibration value of each trigger point to obtain the second distance calibration value.

10. A multi-level intelligent sorting system for construction waste, characterized in that, The steps for implementing the adaptive control method for multi-level intelligent sorting of construction waste according to any one of claims 1 to 9 include: The recognition module is used to identify the longitudinal position, equivalent particle size, aspect ratio, and mean gray value of each material based on the material image. The calculation module is used to calculate the predicted trigger time of each material reaching the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point based on the equivalent particle size, the length-to-short axis ratio, and the longitudinal position. The material screening module is used to perform material screening according to the corresponding predicted trigger time when each material reaches the vibrating screen trigger point, the wind separation trigger point, the eddy current iron removal trigger point, and the air gun fine separation trigger point.

Images