Method and system for variable diameter feed control of a waste shredder based on machine vision

Abstract

This invention provides a machine vision-based variable-diameter feeding control method and system for a waste crusher, relating to the field of waste treatment technology. The specific steps include: continuously acquiring images of the waste before it enters the crusher; extracting the waste's outline and obtaining its dimensions; determining an initial target opening size based on the dimensions, generating a target opening adjustment curve after continuous constraint and amplitude correction, and storing it as a displacement-opening correspondence table; detecting the feeding displacement in real time during the feeding process and obtaining the target opening size from the table; controlling the variable-diameter mechanism to adjust the feed inlet, simultaneously acquiring a position feedback signal to determine the actual opening size, comparing the actual opening with the target opening, and correcting the opening adjustment of the variable-diameter mechanism in a closed loop based on the comparison result. This invention achieves adaptive dynamic adjustment of the feed inlet opening size, effectively avoiding blockage and overload, and improving the crusher's operational stability and processing efficiency.

Description

Technical Field

[0001] This invention relates to the field of waste treatment technology, specifically to a machine vision-based method and system for controlling variable diameter feeding in a waste shredder. Background Technology

[0002] In the field of waste crushing and processing, traditional crushers generally adopt a fixed-size feed inlet design. This design cannot adapt to waste with irregular shapes and significant size differences. When processing waste whose dimensions vary drastically along its length, the fixed opening may be too small, causing blockages and shutdowns, or too large, leading to drastic fluctuations in the crusher rotor load, a surge in energy consumption, and even equipment damage. This inherent contradiction is particularly prominent in the processing of waste straw (such as branches, pineapple stems and leaves, sugarcane tops, etc.), and has become a key bottleneck restricting the processing efficiency and operational stability of crushers.

[0003] In existing technologies, some crushers are equipped with simple manual or timed adjustment of the feed inlet, but this method relies on human experience and judgment, resulting in a serious response lag and an inability to achieve dynamic matching with the real-time changes in the shape of the waste. Other technologies attempt to use photoelectric switches or ultrasonic sensors for single-point size detection, but can only obtain local cross-sectional information of the waste and cannot establish a planning of the opening size along the entire length of the feed direction. This leads to the variable diameter control exhibiting a "blind man touching an elephant" style of local optimization, making it difficult to fundamentally solve the problem of coordinating overall throughput and load stability.

[0004] In recent years, machine vision technology has been widely used in industrial inspection, but its application in the feeding control of waste shredders still faces many challenges. On the one hand, the complex working conditions of waste during transportation, such as speed fluctuations, posture deflection, and uneven surface reflection, make it difficult to guarantee the accuracy and stability of image acquisition and contour extraction. On the other hand, there is a lack of systematic engineering methods for the conversion from discrete image data to continuous mechanical motion. The connection between image processing, size quantization, curve planning, and real-time closed-loop control is loose, and a complete "perception-decision-execution" technical closed loop has not yet been formed. As a result, vision-guided variable diameter feeding control remains at the proof-of-concept stage and cannot meet the stringent requirements of industrial sites for reliability, real-time performance, and automation. Summary of the Invention

[0005] The purpose of this invention is to provide a machine vision-based variable diameter feeding control method and system for waste crushers to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A machine vision-based method for controlling variable-diameter feeding in a waste shredder, comprising the following steps: Step 1: Before the waste enters the crusher, continuously acquire images of the waste to be fed to obtain continuous image data of the waste along the feeding direction; Step 2: Extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction to form a sequence of dimensions; Step 3: Based on the aforementioned external dimension sequence, determine the target opening size of the variable diameter mechanism under the corresponding feed displacement, and generate the target opening adjustment curve of the variable diameter mechanism; Step 4: During the process of waste entering the crusher, the feeding displacement of the waste is detected in real time, and the feeding displacement is mapped to the target opening adjustment curve to determine the corresponding target opening size; Step 5: Control the diameter changing mechanism to adjust the feed inlet opening according to the target opening size, and obtain the position feedback signal of the diameter changing mechanism; Step 6: Determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

[0007] Furthermore, the specific method for continuously acquiring images of the waste to be fed is as follows: an industrial camera set above the feeding conveyor channel continuously captures images of the waste to be fed. Based on the sampling frequency of the industrial camera and the conveying speed of the feeding conveyor channel, the continuously acquired multiple frames of images are processed in time synchronization, and a correspondence between each frame of image and the position in the feeding direction is established, so that each frame of image corresponds to a different feeding length position of the waste in the feeding direction.

[0008] Furthermore, establishing the correspondence between each frame image and the position in the feeding direction specifically includes: calculating the displacement increment of the waste along the feeding direction between the acquisition times of two adjacent frames and the conveying speed of the feeding conveying channel, and stitching the positions of each frame image in the feeding direction according to the displacement increment to form a continuous image sequence corresponding to the actual feeding path of the waste.

[0009] Furthermore, the specific method for extracting the outline of waste from continuous image data is as follows: each frame of the continuous image sequence is sequentially processed by grayscale conversion, noise filtering, waste region segmentation and edge detection to obtain the waste boundary in each frame of the image, and based on the positional correspondence of each frame of the image in the feeding direction, the waste boundary in each frame of the image is stitched together to obtain the continuous outline of the waste along the feeding direction. The noise filtering process includes smoothing filtering of each frame of the image; the waste region segmentation process includes separating the waste region from the image background; the edge detection process includes extracting the contour boundary of the waste region to reduce the impact of background interference on the waste shape contour extraction result.

[0010] Furthermore, the step of obtaining the external dimensions of the waste at each feeding length position along the feeding direction, perpendicular to the feeding direction, specifically includes: establishing multiple cross-sectional lines perpendicular to the feeding direction on the continuous external contour at preset length intervals, calculating the intersection distance between each cross-sectional line and the external contour of the waste, and using the intersection distance as the external dimension at the corresponding feeding length position, arranging each external dimension according to the positional order of the feeding direction to form the external dimension sequence; After forming the external dimension sequence, the method further includes: smoothing the external dimension sequence and removing outliers. Specifically, when the difference between the external dimension at a certain feed length position and the external dimensions at at least two adjacent feed length positions exceeds a preset size threshold, the external dimension is determined to be an outlier. The outlier is then corrected using the interpolation result of the external dimensions at adjacent feed length positions to obtain a corrected external dimension sequence.

[0011] Furthermore, the specific method for determining the target opening size of the variable diameter mechanism at the corresponding feeding displacement based on the external dimension sequence is as follows: add the preset feeding safety margin to the external dimension at each feeding length position after correction to obtain the initial target opening size corresponding to each feeding displacement. The specific method for generating the target opening adjustment curve of the variable diameter mechanism is as follows: the initial target opening sizes arranged sequentially along the feeding direction are subjected to continuous constraint processing. When the difference between the initial target opening sizes corresponding to two adjacent feeding displacements is greater than a preset change threshold, the initial target opening size corresponding to the next feeding displacement is limited and corrected according to the preset change threshold so that the generated target opening adjustment curve meets the adjustment continuity requirements of the variable diameter mechanism. After generating the target opening adjustment curve, the method further includes: sequentially associating the target opening sizes corresponding to each feeding displacement in ascending order of feeding displacement to form a displacement-opening correspondence table. When detecting the feeding displacement of waste in real time, the target opening size corresponding to the current feeding displacement is queried first based on the displacement-opening correspondence table.

[0012] Furthermore, the real-time detection of the feed displacement of the waste specifically includes: acquiring the rotation pulse signal of the feed drive roller shaft through a rotary encoder set on the feed drive roller shaft; calculating the rotation angular displacement of the feed drive roller shaft based on the obtained rotation pulse signal; converting the rotation angular displacement into the real-time feed displacement of the waste by combining the roller diameter parameter of the feed drive roller shaft; and mapping the obtained real-time feed displacement to the corresponding displacement position in the target opening adjustment curve to determine the target opening size corresponding to the current feed displacement.

[0013] Furthermore, the specific method for determining the actual opening size of the variable diameter mechanism based on the position feedback signal is as follows: obtain the current position feedback value of the actuator of the variable diameter mechanism, and convert the current position feedback value into the current actual opening size according to the pre-established conversion relationship between the mechanism position and the feed inlet opening size.

[0014] Furthermore, the step of correcting the opening adjustment of the variable diameter mechanism based on the comparison result specifically includes: comparing the current actual opening size with the target opening size corresponding to the current feed displacement to obtain the opening deviation; when the obtained opening deviation is greater than a preset positive deviation threshold, controlling the variable diameter mechanism to perform a correction action of reducing the opening; when the opening deviation is less than a preset negative deviation threshold, controlling the variable diameter mechanism to perform a correction action of increasing the opening. After each correction action is performed, the position feedback signal is reacquired, and the actual opening size of the variable diameter mechanism is re-determined. Based on the deviation between the re-determined actual opening size and the current target opening size, the control command for the next correction action is updated to form a closed-loop opening adjustment process. Repeatedly execute position feedback acquisition, actual opening size determination, deviation comparison and correction actions until the deviation between the actual opening size and the target opening size falls within the preset allowable deviation range, or the real-time feeding displacement of the waste reaches the end position corresponding to the target opening adjustment curve.

[0015] The present invention also provides a machine vision-based variable diameter feeding control system for a waste crusher. This machine vision-based variable diameter feeding control system is used to execute the aforementioned machine vision-based variable diameter feeding control method for a waste crusher, including: Image acquisition module: used to continuously acquire images of the waste to be fed into the crusher before the waste enters the crusher, and obtain continuous image data of the waste along the feeding direction; Dimension extraction module: used to extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction, forming a dimension sequence; Adjustment curve generation module: used to determine the target opening size of the variable diameter mechanism under the corresponding feed displacement according to the external dimension sequence, and generate the target opening adjustment curve of the variable diameter mechanism; Displacement mapping matching module: used to detect the feeding displacement of waste in real time during the process of waste entering the crusher, and map the feeding displacement to the target opening adjustment curve to determine the corresponding target opening size; Variable diameter adjustment feedback module: used to control the variable diameter mechanism to adjust the feed inlet opening according to the target opening size, and to obtain the position feedback signal of the variable diameter mechanism; Deviation closed-loop correction module: used to determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention uses an industrial camera to continuously acquire images of waste before feeding and establishes a precise correspondence between frame images and feeding displacement. This enables complete perception of the waste's overall dimensions along the feeding direction, overcoming the limitations of existing single-point detection technologies that lack information. It provides a target opening planning data foundation covering the entire length of the waste for the variable diameter mechanism, fundamentally solving the jamming and overload problems caused by traditional fixed openings or local detection.

[0017] This invention adopts a hierarchical control architecture of "offline planning - online table lookup - closed-loop correction". The computationally intensive image processing and curve optimization are completed offline. In the online execution stage, only the displacement feedback of the rotary encoder and the displacement-opening correspondence table are used for fast table lookup mapping, which significantly reduces the computational load of the real-time controller. At the same time, a closed-loop feedback correction mechanism based on the deviation threshold is introduced. By iteratively comparing the difference between the actual opening and the target opening and dynamically updating the control command, the tracking error caused by factors such as mechanical transmission backlash, motor response delay and load disturbance is effectively suppressed, ensuring the high accuracy and stability of the variable diameter control.

[0018] This invention designs a continuous amplitude limiting processing algorithm for the kinematic constraints of the variable diameter mechanism. It applies physical boundary restrictions to the opening change rate between adjacent positions and uses cubic spline interpolation to generate a smooth target opening adjustment curve, avoiding mechanical shocks and tracking failures caused by command jumps. At the same time, by setting a corrected trigger dead zone and iterative convergence criteria, it reduces the frequent reciprocating motion of the mechanism while ensuring control accuracy, thus extending the service life of the equipment. It is particularly suitable for the crushing of waste straw (such as branches, pineapple stems and leaves, sugarcane tops, etc.), and achieves a synergistic improvement in the crusher's operational stability, processing efficiency, and mechanical reliability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 The external dimension sequence processing curve provided in the embodiments of the present invention; Figure 3 This is a diagram illustrating the process of generating the target opening adjustment curve according to an embodiment of the present invention. Figure 4 This is a structural block diagram of the overall system of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0021] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0022] Example: Please see Figures 1 to 3 The present invention provides a technical solution: A machine vision-based method for controlling variable-diameter feeding in a waste shredder, comprising the following steps: Step 1: Before the waste enters the crusher, continuously acquire images of the waste to be fed to obtain continuous image data of the waste along the feeding direction.

[0023] In waste crushing operations, traditional fixed-size feed inlets are ill-suited for processing waste with irregular shapes and varying sizes. If the feed inlet opening is too small, larger waste items cannot pass through smoothly, easily causing blockages, equipment downtime, and reduced production efficiency. If the feed inlet opening is too large, the crusher rotor load fluctuates wildly when processing small waste items, increasing energy consumption and potentially damaging the equipment due to momentary overload. Therefore, achieving adaptive adjustment of the feed inlet opening size is crucial for improving the crusher's operational stability and processing efficiency.

[0024] To achieve this goal, it is first necessary to pre-sensor the waste material before it enters the crusher to obtain its complete external morphological information. In this embodiment, an image acquisition device is installed above the feeding conveyor channel. As the waste material moves at a constant speed towards the crusher along the conveyor channel, it is continuously photographed, acquiring a sequence of image data covering the entire length of the waste material along its feeding direction. The core purpose of this process is to obtain image information that can completely describe the waste's external outline in advance without contacting or interfering with the waste's transport. This provides a reliable data foundation for subsequently extracting quantitative dimensions and dynamically planning the real-time opening size of the crusher's feed inlet.

[0025] Furthermore, the specific method for continuously acquiring images of the waste to be fed is as follows: an industrial camera set above the feeding conveyor channel continuously captures images of the waste to be fed. Based on the sampling frequency of the industrial camera and the conveying speed of the feeding conveyor channel, the continuously acquired multiple frames of images are processed in time synchronization, and a correspondence between each frame of image and the position in the feeding direction is established, so that each frame of image corresponds to a different feeding length position of the waste in the feeding direction.

[0026] In actual waste crushing operations, the feeding and conveying channel typically employs a belt conveyor or roller conveyor mechanism. During this process, the waste moves towards the crusher at a constant or near-constant speed. To obtain complete shape information of the waste along the feeding direction, multiple frames need to be continuously captured at certain time intervals as the waste passes through the collection area. These discrete time-series images are then converted into spatial location sequence images, establishing a precise "time-space" mapping relationship. The accuracy of this mapping relationship directly determines the precision of subsequent outline splicing and the responsiveness of diameter control.

[0027] In this embodiment, the industrial camera is installed directly above the feed conveyor channel, with its optical axis perpendicular to the bearing plane of the conveyor channel. This ensures that the shooting angle is a direct downward view, minimizing dimensional measurement errors caused by perspective distortion. The selection of an industrial camera requires comprehensive consideration of factors such as field of view, resolution, frame rate, and environmental adaptability. The field of view should cover the effective width of the conveyor channel and the maximum possible size of the waste in the height direction; the resolution must meet the requirement of resolving minimal shape changes, typically requiring that the actual physical size corresponding to a single pixel does not exceed one-third to one-fifth of the target measurement accuracy; the frame rate is closely related to the conveying speed and the required sampling density.

[0028] This embodiment follows the basic principle of the Nyquist sampling theorem in setting the sampling frequency, namely, the sampling frequency should not be less than twice the spatial frequency of the highest change in the shape of the waste, in order to avoid loss of shape information or aliasing distortion due to insufficient sampling. In actual engineering, considering the abrupt changes in the shape of waste (such as sharp edges, bifurcation, fracture surfaces, etc.), a higher oversampling factor is usually used. Let the conveying speed of the feeding conveyor channel be... The unit is The sampling frequency of the industrial camera is The unit is Then the displacement step size of the waste along the feeding direction between two adjacent image frames. The calculation formula is: The displacement step size This refers to the equivalent sampling interval of the image acquisition in the spatial direction. To ensure the continuity and smoothness of subsequent contour stitching, It should be much smaller than the minimum characteristic length of the waste's external dimensions in the feeding direction. This embodiment will... The value is limited to between 10mm and 50mm. The lower limit of this range corresponds to low-speed, fine-grained acquisition scenarios, while the upper limit corresponds to high-speed, coarse-grained acquisition scenarios. The specific value depends on the transmission speed. And the required contour accuracy is achieved by adjusting accomplish.

[0029] The core of time synchronization processing lies in eliminating the asynchronous deviation between the internal clock of the industrial camera and the control clock of the conveyor system, ensuring that the acquisition time of each frame can be accurately calibrated. This embodiment employs a hardware-triggered synchronization method, where the motion controller of the conveyor system outputs periodic trigger pulse signals to the industrial camera. Upon receiving the rising (or falling) edge of the trigger pulse, the camera immediately performs exposure and acquisition operations. The frequency of the trigger pulse is the sampling frequency. Its cycle Hardware triggering allows the image acquisition time to be locked onto the same time base as the movement state of the conveying system, eliminating the timing jitter inherent in software polling methods.

[0030] Let the acquisition time of the k-th frame be . Where N is the total number of frames, and due to the use of periodic hardware triggering, the time interval between adjacent frames is constant: .

[0031] When establishing the correspondence between each frame of image and the position along the feeding direction, it is first necessary to determine a spatial reference origin. In this embodiment, the moment when the front end of the waste (the foremost point along the feeding direction) first enters the center line of the industrial camera's field of view is defined as... The spatial position corresponding to this moment is the zero point of the feed length, denoted as . For the k-th frame image, its corresponding feed length position It is determined by the product of the time delay of this frame relative to the first frame and the transmission speed, and the mathematical expression is: in, This represents the length and position of the waste corresponding to the k-th frame image along the feeding direction, i.e., the distance measured from the front end of the waste along the feeding direction. This formula establishes the relationship between the frame index k and the feeding length position. The one-to-one correspondence between them ensures that each frame of the image uniquely corresponds to a specific cross-sectional location on the waste.

[0032] In actual operation, the conveying speed Slight fluctuations may occur due to factors such as load changes and motor speed adjustments. To compensate for the position mapping error caused by speed fluctuations, this embodiment introduces a speed feedback correction mechanism. Let the conveying speed measured in real time by the rotary encoder be... Then, the position mapping of the k-th frame image is corrected using an integral form formula: in, Indicates at time The instantaneous conveying speed is measured in real time by the rotary encoder. Indicates an infinitesimal time interval Inside, the infinitesimal displacement of the waste moving along the feeding direction; Indicates from the initial time The total displacement of the waste along the feeding direction during the time up to the acquisition time of the k-th frame, which is the feeding length position corresponding to the k-th frame image. ; This represents the instantaneous velocity value fed back by the encoder at the i-th sampling moment. This discrete integration method incorporates the cumulative effect of velocity fluctuations into the position calculation, significantly improving the position mapping accuracy of long-sized waste objects throughout the entire acquisition process.

[0033] Through the aforementioned time synchronization processing and position mapping, the continuously acquired multi-frame discrete images are converted into a spatially sampled sequence uniformly distributed along the feeding direction. Each frame image is positioned according to its corresponding feeding length position. Sort the images to form an ordered set. ,in Corresponding position This ordered image set serves as the data foundation for subsequent contour extraction and dimensional measurement. Its spatial uniformity and positional accuracy directly determine the performance ceiling of the entire variable diameter feeding control system.

[0034] Furthermore, establishing the correspondence between each frame image and the position in the feeding direction specifically includes: calculating the displacement increment of the waste along the feeding direction between the acquisition times of two adjacent frames and the conveying speed of the feeding conveying channel, and stitching the positions of each frame image in the feeding direction according to the displacement increment to form a continuous image sequence corresponding to the actual feeding path of the waste.

[0035] Based on the time synchronization processing of continuously acquired multi-frame images, this embodiment further transforms the discrete multi-frame images into a spatially continuous image sequence. The core of this process lies in accurately calculating the displacement increment between adjacent frames and achieving seamless stitching of the images based on this displacement increment. This step is a crucial bridge connecting "temporal domain sampling" and "spatial domain reconstruction," directly determining the completeness and accuracy of subsequent contour extraction.

[0036] Let the displacement increment between the k-th frame and the (k+1)-th frame be... Under ideal uniform motion conditions, constant value If the transmission speed fluctuates, the displacement increment between each frame needs to be calculated separately based on the instantaneous speed fed back by the encoder in real time. The specific calculation formula is as follows: in, This represents the instantaneous velocity measured by the encoder at the acquisition time of the k-th frame. This segmented calculation method isolates the impact of velocity fluctuations frame by frame, preventing errors from accumulating and propagating throughout the sequence.

[0037] Based on displacement increment In this embodiment, the absolute position coordinates of each frame image in the feeding direction are constructed. Let the starting position of the first frame image be... The center position of the k-th frame image is... for: These absolute position coordinates establish a unique spatial location for each frame of the image within the entire length of the waste, ensuring that the dimensions extracted from any subsequent frame can correspond to the accurate feed length position.

[0038] During the positional stitching process, this embodiment employs a "virtual canvas" mechanism to construct a continuous image sequence. A sufficiently long blank canvas is set, its length covering the estimated maximum value of the total length of the waste, and its pixel resolution is consistent with that of a single frame image. For the k-th frame image, its starting ordinate on the canvas is calculated. ,Right now ; Where p is the pixel equivalent, in mm / pixel. This represents the rounding operation. It writes the k-th frame image pixel by pixel into the canvas, starting from the x-coordinate... to The region is defined by H, where H represents the number of pixels in a single frame. By performing this operation on all frames sequentially, a vertically extending sequence of images corresponding to the actual waste feeding path can be formed on the canvas. .

[0039] because Typically, the values ​​are not integers; rounding can introduce sub-pixel-level splicing misalignments. To eliminate image jaggedness or ghosting caused by this misalignment, this embodiment uses bilinear interpolation for pixel resampling. For integer coordinate positions on the canvas... Its corresponding theoretical continuity position is This location may fall within the transition region between frame k and frame (k+1). By locating the adjacent frames and corresponding local coordinates of this location in the original frame sequence, the pixel value at this location is calculated using bilinear interpolation. The specific calculation formula is as follows: in, The interpolation weights are determined by the distance between the theoretical position and the center of pixels in adjacent frames, satisfying the following conditions: This interpolation operation ensures that the stitched consecutive image sequences maintain visual smoothness in the inter-frame transition areas, avoiding abrupt edge changes caused by hard stitching.

[0040] Furthermore, because the field of view of industrial cameras may be larger than the effective width of the conveyor channel in the width direction (perpendicular to the feed direction), continuous image sequences... There is redundant background area in the width direction. In this embodiment, after stitching is completed, [the following is done]: Cropping is performed in the width direction, retaining only the image area within the effective width of the conveyor channel, reducing the computational load and background interference for subsequent contour extraction. Let the effective width of the conveyor channel be... The camera's field of view is Then the redundant cutting amount on both sides for: And the effective image width after cropping for: .

[0041] After the above displacement increment calculation, absolute position positioning, virtual canvas stitching, bilinear interpolation resampling, and width cropping, the final continuous image sequence is formed. There is a strict spatial linear relationship in the feeding direction: the actual feed length position corresponding to any ordinate y in the sequence. for: .

[0042] This continuous image sequence fully preserves the shape, texture, and structural information of the waste along the feeding direction, and each pixel position can be accurately mapped to physical space coordinates, providing a directly operable spatial data foundation for subsequent steps to extract waste boundaries, calculate cross-sectional dimensions, and generate variable diameter control curves.

[0043] Step 2: Extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction to form a sequence of dimensions.

[0044] After completing the continuous image acquisition of the waste and establishing the correspondence between each frame of image and the position of the feeding direction through the aforementioned steps, the core task of this step is to use these image data to completely extract the overall external shape of the waste along the feeding direction and quantify it into a series of external dimensions that can be used for subsequent calculations, thereby serving as the direct basis for determining the opening size of the variable diameter mechanism.

[0045] From image data to the final dimensional sequence, the processing flow in this embodiment follows a progressive logic of "image preprocessing – contour extraction – dimensional quantization – data postprocessing". First, the original image needs to be converted into a form easier for computer analysis, and irrelevant information introduced by uneven lighting, sensor noise, or transmission interference needs to be suppressed. This accurately separates the waste area from the background and precisely locates its boundaries. The continuous boundary of the waste obtained in this step is the key bridge connecting image data and physical dimensions. After obtaining the continuous boundary contour, the width of the waste perpendicular to the feeding direction can be measured at set intervals along the feeding direction. This "width" is actually the local dimensional dimension of the waste at that feeding length position, which directly determines the minimum opening diameter of the crusher's feed inlet to accommodate the waste's feeding requirements. By traversing the entire feeding length and arranging the width values ​​at each position sequentially, a dimensional sequence corresponding one-to-one with the feeding displacement is formed. This sequence is essentially a two-dimensional projection simplification of the complex three-dimensional shape of the waste to be fed. It enables the system to "predict" changes in waste size throughout the feeding process, thus allowing for advance planning of the entire movement of the diameter-changing mechanism. This provides a direct and high-precision data source for generating a smooth and reasonable dynamic target opening adjustment curve, fundamentally different from traditional fixed-opening or passively sensing feeding methods.

[0046] Furthermore, the specific method for extracting the outline of waste from continuous image data is as follows: each frame of the continuous image sequence is sequentially processed by grayscale conversion, noise filtering, waste region segmentation and edge detection to obtain the waste boundary in each frame of the image, and based on the positional correspondence of each frame of the image in the feeding direction, the waste boundary in each frame of the image is stitched together to obtain the continuous outline of the waste along the feeding direction. The noise filtering process includes smoothing filtering of each frame of the image; the waste region segmentation process includes separating the waste region from the image background; the edge detection process includes extracting the contour boundary of the waste region to reduce the impact of background interference on the waste shape contour extraction result.

[0047] In obtaining continuous image sequences Next, it needs to be converted into binary boundary data that can be used for geometric measurement. This conversion process follows a progressive processing logic of "dimensionality reduction-noise reduction-separation-extraction", gradually stripping away irrelevant information in the image and finally condensing it into the outline boundary of the waste.

[0048] Continuous image sequence Typically, images are RGB color images, containing pixel information in three channels: red, green, and blue. To reduce the computational complexity of subsequent processing and highlight brightness contrast features, grayscale conversion is performed first. Let the red, green, and blue components of a pixel in the original color image be... Then the grayscale value of that pixel The weighted average method is used for calculation, and the specific formula is as follows: The weighting coefficients are set based on the differences in human eye sensitivity to different wavelengths of light, with green having the highest weight, followed by red, and blue having the lowest. After grayscale processing, the image is compressed from a three-dimensional color space to a one-dimensional brightness space, with pixel values ​​typically ranging from [0, 255].

[0049] Subsequently, noise filtering was performed on each frame of the image. Noise is inevitably introduced during the acquisition of grayscale images, mainly including Gaussian noise (caused by sensor thermal motion, manifested as random fluctuations in pixel values) and impulse noise (caused by dust, reflective points, etc., manifested as isolated extreme pixels). The presence of noise interferes with the subsequent edge positioning accuracy, leading to jagged or broken boundaries.

[0050] This embodiment uses Gaussian smoothing filtering for noise suppression. The two-dimensional weight distribution of the Gaussian filter kernel is as follows: in, The local coordinates are with the center of the filter kernel as the origin. The standard deviation of the Gaussian distribution controls the smoothing strength of the filter. The larger the value, the stronger the smoothing effect, but the greater the degree of edge blurring; The smaller the value, the better the edge is preserved, but the noise suppression effect is weakened.

[0051] This embodiment will The value is limited to between 0.8 and 2.0 pixels. The lower limit of this range is suitable for scenes with low noise levels and rich edge details; the upper limit is suitable for harsh conditions with uneven lighting and severe dust interference. The specific value is adaptively adjusted according to the signal-to-noise ratio (SNR) of the image, and the specific formula is as follows: in, This is the baseline smoothing coefficient (usually taken as 1.0). For reference signal-to-noise ratio threshold, This is an estimate of the actual signal-to-noise ratio (SNR) for the current image. It automatically increases when the actual SNR is low. To enhance the smoothing effect; conversely, to reduce it. To preserve more edge details.

[0052] The Gaussian filter kernel is convolved with the grayscale image to obtain the smoothed image. The calculation formula is: in, The half-window size of the filter kernel ensures coverage of over 99% of the energy in the Gaussian distribution.

[0053] The smoothed image still contains the transport channel background, the waste material itself, and possible shadow areas. To separate the waste area from the background, this embodiment employs an adaptive thresholding method.

[0054] Let a pixel in the image The neighborhood window is Window size is (usually taken) (matching the filter kernel size); then the local mean within that window. and local standard deviation The calculation formula is: Adaptive threshold It is dynamically determined based on local statistical characteristics, and its calculation formula is as follows: Here, c is the offset coefficient, which controls the strictness of segmentation. When c is positive, the threshold is higher than the local mean, tending to classify darker areas as background; when c is negative, the threshold is lower than the local mean, tending to retain more dark areas. In this embodiment, the value range of the offset coefficient c is limited to between -0.5 and 1.5, calibrated according to the typical brightness characteristics of waste.

[0055] Next, the pixel values ​​are compared with an adaptive threshold to generate a segmented binary image. Its mathematical expression is: in, This indicates that the pixel belongs to the waste foreground area. This indicates that the area belongs to the background region. This adaptive strategy can effectively address issues such as uneven lighting and differences in surface reflectivity of waste, avoiding oversegmentation or undersegmentation in local areas caused by a globally fixed threshold.

[0056] After obtaining the binarized waste region, its contour boundary needs to be further extracted. This embodiment uses the Canny edge detection operator, which achieves sub-pixel-level edge localization through multi-stage processing.

[0057] First, calculate the magnitude and direction of the image gradient. Then, use the Sobel operator to calculate the partial derivatives in the horizontal (x-direction) and vertical (y-direction) directions respectively: The gradient components are then: gradient magnitude The calculation formula is: gradient direction The calculation formula is: In the gradient direction, only the pixels with the largest local gradient magnitude are retained, and the weak response at non-edge locations is suppressed, so that the edge lines are refined to a single pixel width.

[0058] Next, this embodiment uses dual-threshold hysteresis processing to determine the final edge. A high threshold is set. and low threshold (generally The gradient magnitude is higher than... The pixels are identified as strong edges; the gradient magnitude is lower than Pixels that are strong edges are identified as non-edges; pixels that fall between these two types are identified as weak edges and preserved if they are connected to strong edge pixels, otherwise they are suppressed.

[0059] After the above processing, a single frame image is obtained. Waste boundary pixel set This set is stored in the form of a coordinate chain, which records the closed contour of the waste in the frame image.

[0060] Finally, based on the correspondence between each frame image and the position in the feeding direction, the boundary set of each frame image is determined. According to its corresponding feed length position Perform spatial sorting and stitching. Let the pixel coordinates of the k-th frame boundary be... Where i is the boundary point index, Let be the vertical coordinate relative to the image frame. Mapping this coordinate to the global coordinate system of the continuous image sequence, the mathematical expression is: in, Let be the starting y-coordinate of the k-th frame in the global canvas; global coordinates This is the unique spatial location of the boundary point within the entire outline of the waste.

[0061] All frame boundary points are sorted according to the global ordinate Y. Duplicate points in the overlapping area of ​​adjacent frame boundaries are deduplicated (points with higher gradient magnitudes are retained), ultimately forming a continuous, unbroken waste outline curve extending along the feeding direction. Its mathematical expression is: Where M is the total number of contour points. Strictly monotonically increasing (along the feed direction). This indicates the boundary coordinates of the waste at this location in the width direction. This is the continuous outline. The complete description of the geometry of the waste along the feeding direction provides a precise geometric basis for subsequent dimensional extraction.

[0062] Furthermore, the step of obtaining the external dimensions of the waste at each feeding length position along the feeding direction, perpendicular to the feeding direction, specifically includes: establishing multiple cross-sectional lines perpendicular to the feeding direction on the continuous external contour at preset length intervals, calculating the intersection distance between each cross-sectional line and the external contour of the waste, and using the intersection distance as the external dimension at the corresponding feeding length position, arranging each external dimension according to the positional order of the feeding direction to form the external dimension sequence; After forming the external dimension sequence, the method further includes: smoothing the external dimension sequence and removing outliers. Specifically, when the difference between the external dimension at a certain feed length position and the external dimensions at at least two adjacent feed length positions exceeds a preset size threshold, the external dimension is determined to be an outlier. The outlier is then corrected using the interpolation result of the external dimensions at adjacent feed length positions to obtain a corrected external dimension sequence.

[0063] The continuous outline of the waste along the feeding direction is obtained through the aforementioned steps. Subsequently, the profile, in the form of a discrete set of coordinate points, completely records the spatial distribution of the waste boundary in the global coordinate system. However, this set of coordinate points itself cannot be directly used for diameter control decisions; it needs to be further transformed into a scalar sequence, or external dimension sequence, characterized by an ordered arrangement along the feed direction and representing the width of the waste at each location. The core of this transformation process lies in: sampling cross-sections on the profile curve at fixed spatial intervals, reducing the two-dimensional profile information to a one-dimensional dimension sequence, thereby establishing a direct mapping relationship between "feed length position and external dimension," providing a discrete but complete data foundation for the subsequent generation of the target opening adjustment curve.

[0064] In this embodiment, a preset length interval is used. The setting needs to comprehensively consider both measurement accuracy requirements and computational efficiency. Too small an interval will lead to data redundancy and increase the computational burden of subsequent processing; too large an interval may miss local abrupt changes in the external dimensions, resulting in a lag in the response of the diameter-changing mechanism. This embodiment will... The value is limited to between 50mm and 200mm. The lower limit of this range applies to waste materials with complex shapes and drastic dimensional variations, while the upper limit applies to waste materials with regular shapes and gradual dimensional changes. Specific values ​​are based on the contour curve. The total length and the required sampling density are determined.

[0065] Assume a continuous outer contour The total coverage area in the feeding direction is According to the preset length interval If Q equally spaced cross-sectional positions are established along the feeding direction, then the global ordinate of the q-th cross-section is... for: in, , indicating the total number of cross-sectional lines.

[0066] Each cross-sectional line is a straight line perpendicular to the feed direction (vertical axis), and its mathematical expression is: Calculate the relationship between this cross-sectional line and the outer contour of the waste. The intersection points between them, that is, finding all points in the set of contour points that satisfy the condition. The point, among which This is the tolerance threshold used to compensate for the discreteness of the contour points; in this embodiment, it is set to... ,Right now One pixel equivalent.

[0067] Since the waste profile is a closed curve, each cross-sectional line typically intersects the profile at two points, corresponding to the left and right boundaries of the waste at that location. Let the x-coordinates of the intersection points of the q-th cross-sectional line and the profile be... and ,and Then the external dimensions at the feed length position The distance between the two intersection points is: This external dimension The physical meaning is: at the feed length position At this point, the actual width of the waste perpendicular to the feeding direction is given. Since the contour point set is discretely sampled, the x-coordinate of the intersection point may not be directly obtainable; therefore, linear interpolation is used for calculation. Let two adjacent points on the contour... and satisfy The x-coordinate of the intersection point is then calculated using the following interpolation formula: Arrange the corresponding external dimensions of each cross-section line into a sequence according to the positional order of the feeding direction, to obtain the external dimension sequence W, i.e. This sequence is related to the feed length position sequence. One-to-one correspondence, fully describing the dimensional distribution of waste along the feeding direction.

[0068] After generating the external dimension sequence, it is necessary to perform smoothing and outlier removal. Due to factors such as image acquisition noise, contour extraction errors, and cross-section interpolation calculations, the external dimension sequence W may contain random fluctuations and outliers. Random fluctuations can cause the sequence to exhibit high-frequency jitter. If directly used to generate the target opening adjustment curve, this will lead to frequent fine-tuning and mechanical vibration of the diameter-changing mechanism, accelerating equipment wear. Outliers, on the other hand, may cause severe deviations in local opening dimensions, leading to material jamming or over-opening problems. Therefore, post-processing of the original external dimension sequence is essential to improve its smoothness and reliability.

[0069] This embodiment first uses a moving average filter to smooth the sequence. Let the moving window length be... The smoothed outer dimensions The calculation formula is: For sequence boundary points (q close to 0 or Q-1), this embodiment adopts a boundary mirroring filling strategy to ensure that there are always points within the window. Each valid data point is used in the calculation. Moving average filtering effectively suppresses high-frequency random noise while preserving low-frequency trend changes in dimensions. Window length. The choice requires a trade-off between smoothing effects and feature preservation. The larger the value, the stronger the smoothing effect, but it may blur real local size abrupt changes; The smaller the size, the better the feature retention, but the noise suppression effect is weakened. This embodiment, based on the typical complexity of the waste's shape, will... Setting it to 5 means averaging over five adjacent locations centered on the current location. This setting effectively filters out single-point noise fluctuations while maintaining sensitivity to general size changes.

[0070] After smoothing, outlier removal is required. The logic for outlier determination is based on the principle of local consistency: in the normal shape of waste materials, the dimensions of adjacent locations are usually continuous, without drastic isolated jumps. Therefore, when the dimension of a certain location differs significantly from multiple adjacent locations within its neighborhood, that dimension is determined to be an outlier.

[0071] The specific determination rule is as follows: for the smoothed outer dimension at the q-th position... Calculate the difference in external dimensions between the object and its two adjacent forward and backward positions. The mathematical expression is as follows: Set preset size threshold This threshold is determined based on the statistical variation range of the waste's external dimensions and the measurement system error. In this embodiment, The value ranges from 10mm to 30mm, with the specific value determined based on the width of the conveyor channel and the typical size of the waste; if both conditions are met... All greater than Then The value is identified as an outlier. This criterion requires that the difference between the current position's dimensions and its two adjacent positions (a total of four adjacent positions) exceeds a threshold, ensuring the rigor and reliability of outlier detection and avoiding misjudgments caused by single-point noise or normal local abrupt changes.

[0072] For the location q identified as an outlier, correction is required using the interpolation results of the external dimensions at adjacent feed length locations. This embodiment employs cubic spline interpolation, utilizing the values ​​before and after the outlier location. A spline curve is constructed from 10 normal data points (in this embodiment, 10 normal data points are used). ), calculate the interpolation result at that location. As the corrected external dimensions, cubic spline interpolation ensures that the corrected data maintains the continuity of the second derivative with the neighboring data at outlier locations, allowing the corrected results to naturally and smoothly integrate into the overall sequence trend, avoiding the angle discontinuity problem that may be introduced by linear interpolation.

[0073] After smoothing and outlier removal, the corrected external dimension sequence is obtained. Its expression is: in, (Normal position) or (Outlier Correction Location). The corrected dimensional sequence retains the true trend of waste shape changes while eliminating noise interference and anomalous jumps, providing high-quality and reliable data input for subsequent determination of target opening size and generation of adjustment curves.

[0074] Table 1: Comparison of External Dimension Sequence Processing To verify the effectiveness of the smoothing and outlier removal processes described above, this embodiment uses a 1200mm long piece of waste straw (taking sugarcane tops as an example) as a sample, and then applies it at preset length intervals. Cross-sectional sampling was performed, and this interval is suitable for waste types with relatively gentle shape changes. The results of the shape and size sequence processing are shown in Table 1.

[0075] As can be seen from Table 1, at the feed length position Original external dimensions The smoothed dimensions at each of its two adjacent positions. The differences between the dimensions (i.e., 45.0mm, 89.2mm, 95.6mm, and 135.2mm) are 107.8mm, 63.6mm, 42.2mm, and 17.6mm, respectively, with the first three values ​​far exceeding the preset size threshold. Therefore, this point was determined to be an outlier. The outlier originated from the reflective interference of soil adhering to the straw branching node at this location, causing excessive expansion of the foreground region during image segmentation. After cubic spline interpolation correction, the external dimensions at this location were restored to 112.3 mm, consistent with the trend of adjacent locations, effectively avoiding drastic jumps in the target opening size caused by single-point measurement errors. The external dimensions at other locations were smoothed using moving averages, significantly suppressing high-frequency random noise. The overall sequence exhibited a true external dimension change trend of "thinner at the beginning—thicker in the middle—locally thinner—abnormal jump—return to normal—thicker at the end."

[0076] Please see Figure 2 The figure shows a curve representation of the smoothing and outlier removal process described above, using a 3000mm long piece of discarded straw (taking sugarcane tops as an example) as a sample, with preset length intervals. A total of 16 cross-sectional locations were collected for external dimensional data. The black dashed lines in the figure represent the original external dimensional sequence. ,like Figure 2 As shown, at the feed length position A noticeable abnormal peak exists (original value approximately 155mm), caused by excessive expansion of the foreground region in the image segmentation due to reflective interference from soil adhering to the straw forks. The blue solid line represents the area after moving average filtering (window length). Smoothed sequence after ) High-frequency random noise was effectively suppressed, and the overall trend of the sequence became clearer. The red dot marks the location of the outlier; after cubic spline interpolation correction, the dimensions at this location were restored to 112 mm (the corrected sequence is marked with a green triangle). The corrected dimensional sequence maintains a continuous trend with adjacent positions, eliminating drastic changes in external dimensions caused by single-point measurement errors. It fully preserves the true external shape characteristics of the waste material, from a thin initial segment to a gradually thickening middle section, followed by localized thinning, and then a rise at the tail end, providing a highly reliable data foundation for determining the subsequent target opening size.

[0077] Step 3: Based on the aforementioned external dimension sequence, determine the target opening size of the variable diameter mechanism under the corresponding feed displacement, and generate the target opening adjustment curve of the variable diameter mechanism.

[0078] The core task of this step is to transform the discrete data sequence representing the width of the waste into a target opening size command sequence that the variable diameter mechanism can directly execute, and further generate a continuous and smooth target opening adjustment curve. This transformation process is the key bridge connecting "waste shape perception" and "feed inlet mechanical execution". Its essence is to plan an opening size trajectory that changes continuously with the feed displacement for the variable diameter mechanism while ensuring the smooth passage of the waste.

[0079] From the perspective of practical engineering needs, determining the target opening size must simultaneously consider two mutually restrictive objectives: first, ensuring passage capacity, meaning the opening size must be large enough to allow waste to pass through the feed inlet unimpeded, avoiding blockages and shutdowns due to insufficient size; second, controlling economic efficiency, meaning the opening size should not be excessively expanded to prevent drastic fluctuations in the crusher rotor load, increased energy consumption, and problems such as small-sized waste tumbling and entanglement within the feed inlet due to an overly large opening. Therefore, the target opening size is not a direct replication of the external dimensions, but rather requires introducing an appropriate safety margin based on the external dimensions and implementing continuous constraints to ensure that the generated adjustment curve satisfies both passage capacity requirements and the kinematic constraints of the mechanical system.

[0080] Furthermore, as a mechanical actuator, the variable diameter mechanism's drive motor, transmission mechanism, and movable baffle all have limited response speed and acceleration, making it impossible to instantly complete large-scale opening size changes. If the opening size change between adjacent positions is too drastic, it will lead to excessive tracking errors, severe mechanical shock, and even trigger the overload protection of the drive system. Therefore, when generating the target opening adjustment curve, a continuity constraint must be imposed on the rate of change of the opening size to ensure the smoothness of the curve and mechanical feasibility.

[0081] Furthermore, the specific method for determining the target opening size of the variable diameter mechanism at the corresponding feeding displacement based on the external dimension sequence is as follows: add the preset feeding safety margin to the external dimension at each feeding length position after correction to obtain the initial target opening size corresponding to each feeding displacement.

[0082] After obtaining the corrected external dimension sequence Next, this needs to be converted into a target opening size command that the variable diameter mechanism can directly respond to. However, the external dimension sequence only reflects the theoretical width value of the waste in the image measurement coordinate system. Directly using this value as the target opening size poses significant risks in practical engineering applications. Waste is inevitably affected by various uncertainties during transport: First, mechanical vibration of the conveyor channel may cause lateral displacement of the waste in the width direction, making its actual occupied space larger than the static measurement value; second, some waste (such as flexible woven bags and elastic rubber parts) may undergo elastic deformation or local expansion under its own weight or conveyor belt tension, causing its actual shape to be larger than the measurement value at the moment of image acquisition at the instant it passes through the inlet; third, the image measurement system itself has inherent system errors, including lens distortion, pixel quantization error, and sub-pixel level deviation of the contour extraction algorithm in boundary determination. The superposition of these three types of uncertainties necessitates a certain safety clearance between the theoretical external dimension and the minimum opening required for the waste to actually pass through.

[0083] To quantify this safety clearance requirement, this embodiment introduces a preset feed safety margin. This margin is essentially an engineering conservatism factor. Its function is to reserve an additional buffer space beyond the theoretical dimensions to accommodate potential dimensional expansions caused by the aforementioned uncertainties, thereby reducing the risk of jamming to an acceptable engineering level. Feeding safety margin. The value of must follow the principle of balancing security and economy. If If the value is too small, the safety buffer is insufficient. If the actual size of the waste exceeds the theoretical value, it will directly cause the feed inlet to become blocked, leading to the crusher stopping, the conveyor channel becoming blocked, and even triggering the motor overload protection, causing production interruption and equipment damage. If the value is too large, the opening size will expand excessively. When the crusher rotor processes small-sized waste, the load will fluctuate violently, and the instantaneous power demand peak will increase. This will not only increase energy consumption, but may also cause impact damage to the rotor bearings due to sudden load changes. At the same time, an excessively large opening will cause the waste to lose effective restraint in the feed inlet, resulting in unstable feeding postures such as rolling, deflection or entanglement, which will reduce the crushing efficiency.

[0084] Based on the above considerations, this embodiment will The value range is limited to 10mm to 50mm. The lower end of this range, 10mm, is suitable for waste types that are regularly shaped, hard, and not easily deformed, such as metal profiles, hard plastic blocks, and glass fragments. These wastes have stable geometry, minimal positional offset during transport, high image measurement accuracy, and a relatively small safety margin. The higher end of this range, 50mm, is suitable for waste types that are irregularly shaped, soft, or easily expandable, such as woven bags, foam materials, wet organic matter, and elastic rubber parts. These wastes are prone to deformation, expansion, or tilting during transport, and the image measurement boundaries are blurred, resulting in greater uncertainty and requiring a larger safety margin to ensure passability. In practical applications, The specific values ​​are calibrated offline based on the physical characteristics of the waste and the guiding accuracy of the conveying channel. The calibration method is as follows: under standard test conditions, multiple sets of waste samples with known external dimensions are collected, the opening size is gradually reduced until blockage occurs, the difference between the critical passing size and the theoretical external dimension is recorded, the distribution pattern is statistically analyzed, and the 95% confidence level quantile value is taken as the safety margin calibration value for this type of waste.

[0085] This embodiment, based on the theoretical width of the waste material, adds a fixed safety buffer amount to obtain the minimum theoretical opening value that the variable diameter mechanism should open at that position. Let the q-th feed length position be... The external dimensions at the location are Then the initial target opening size corresponding to that position for: ;in, This represents the initial target opening size at the q-th feed length position, in mm; This indicates the corrected external dimensions at the same location, in mm. This indicates the preset feed safety margin, in mm.

[0086] Subsequently, according to the positional order of the feeding direction, the initial target opening sizes corresponding to each feeding length position are arranged into a sequence to obtain the initial target opening size sequence. ,Right now This sequence is related to the feed length position sequence. This provides a one-to-one correspondence, fully describing the initial target opening value that the diameter-changing mechanism should achieve at each position along the feeding direction. The word "initial" here indicates that this sequence has not yet undergone mechanical continuity constraint processing, and the change in opening size between adjacent positions may exceed the physical execution capability of the diameter-changing mechanism. Therefore, this sequence serves only as the input reference for subsequent continuity constraint processing, rather than the final execution instruction.

[0087] Furthermore, the specific method for generating the target opening adjustment curve of the variable diameter mechanism is as follows: the initial target opening sizes arranged sequentially along the feeding direction are subjected to continuity constraint processing. When the difference between the initial target opening sizes corresponding to two adjacent feeding displacements is greater than a preset change threshold, the initial target opening size corresponding to the next feeding displacement is limited and corrected according to the preset change threshold so that the generated target opening adjustment curve meets the adjustment continuity requirements of the variable diameter mechanism.

[0088] The initial target opening size sequence obtained through the aforementioned steps Although the waste passage requirements are met independently at each discrete location, the change in opening size between adjacent locations may exhibit drastic jumps. The root cause of these jumps lies in the potential abrupt changes in the waste's shape (such as sharp edges, bifurcation, or fracture surfaces), leading to a revised sequence of dimensions. Significant differences arise between adjacent sampling locations; in addition, a fixed safety margin is added. After superposition, this difference is directly transmitted to the initial target opening size sequence. However, as a mechanical actuating component, the variable diameter mechanism's drive motor, transmission mechanism (such as lead screw, connecting rod, or rack and pinion), and movable baffle all have limited inertia, response speed, and acceleration, making it impossible to instantly complete large-scale opening size switching. If the initial target opening size sequence is directly used as the control command, the mechanism will generate significant dynamic lag and tracking error during actual tracking, causing the actual opening size to deviate from the target value, triggering mechanical shock, vibration, and noise, accelerating the wear of transmission components, and in severe cases, even triggering the overcurrent protection of the drive motor or the encoder's step loss alarm, causing the control system to shut down.

[0089] To address the aforementioned problem, this embodiment introduces a continuity constraint processing mechanism. The core idea is to impose physical boundary restrictions on the rate of change of the initial target opening size sequence, ensuring that the generated target opening adjustment curve satisfies the kinematic constraints of the variable diameter mechanism. The key parameter of this mechanism is a preset change threshold. It characterizes the maximum allowable change in opening size of the variable diameter mechanism between two adjacent sampling positions.

[0090] The preset change threshold The physical nature of the variable diameter mechanism is determined by its mechanical kinematic parameters and the sampling time interval. Let the rated speed of the drive motor of the variable diameter mechanism be... The unit is r / min; let the reduction ratio be... This represents the ratio of the motor shaft speed to the output shaft speed; let the lead of the leadscrew be... The unit is mm / r, representing the linear displacement converted by the lead screw per revolution; the actual time interval between adjacent sampling positions. Based on conveying speed and external dimension sampling interval Determined, that is Its physical meaning is the speed at which waste travels on the conveyor channel. Distance traveled during exercise The required time is also the available time window for the variable diameter mechanism to adjust the opening between two adjacent control command points.

[0091] Under the constraints of the above parameters, the variable diameter mechanism in The maximum linear displacement that can be achieved within a given time period is the upper limit of its theoretical motion capability. The formula for calculating this upper limit is: in, This indicates the rated speed of the output shaft of the variable diameter mechanism, in r / min. Divide this rated speed by 60 to convert the unit to r / s, then divide it by the lead screw. Multiply, the result is This is expressed as the linear motion speed driven by the output shaft, with units of mm / s; finally, this linear motion speed is compared with the available time window. Multiply, we get This refers to the maximum displacement change that can be achieved within the specified time window, expressed in mm.

[0092] However, the theoretical upper limit of motion capability is the limit under ideal conditions of no load, no friction, and no transmission backlash. In actual engineering, mechanical transmission chains have non-ideal factors such as gear meshing clearance, lead screw nut preload clearance, and coupling elastic deformation. Motors require a certain response time to accelerate from rest to rated speed, and the position and speed loops of the control system also have steady-state errors and dynamic lags. To ensure reliable tracking of the target opening adjustment curve after continuity constraint processing in actual execution, this embodiment also introduces a safety factor. The theoretical upper limit is reduced by the following mathematical expression: Safety factor The value of reflects the tolerance of the control system to mechanical uncertainties. This embodiment will... The value range is set between 0.6 and 0.8. When A value of 0.6 indicates a conservative control strategy, reserving a 40% margin in motion capability to cope with sudden load changes and mechanical hysteresis. This is suitable for heavy-duty variable-diameter mechanisms with large inertia and significant transmission backlash. A value of 0.8 indicates an aggressive control strategy with only a 20% margin, suitable for lightweight variable-diameter mechanisms with high-dynamic servo drives and precision ball screw transmissions. The specific value is determined based on the actual mechanical parameters of the variable-diameter mechanism and the tracking performance tests during the commissioning phase.

[0093] In addition, this embodiment will also The value range is limited to 50mm to 200mm. The lower limit of 50mm is suitable for variable diameter mechanisms driven by stepper motors or ordinary asynchronous motors, with long transmission chains and large mechanical inertia. Such mechanisms have limited dynamic response capabilities and require a small change threshold to ensure smooth tracking. The upper limit of 200mm is suitable for variable diameter mechanisms using servo motor direct drive, high rigidity transmission, and low inertia design. Such mechanisms have high dynamic response capabilities and allow for a larger change threshold to improve the following speed to sudden changes in the shape of waste.

[0094] This embodiment employs a forward traversal limiting algorithm for handling continuous constraints. This algorithm starts from the initial feeding position and scans point by point along the feeding direction, forcibly limiting and correcting opening changes that exceed the mechanism's motion capability. Let... This represents the target opening size at position q after continuity constraint processing. During algorithm initialization, let... That is, the initial target opening size does not need to be corrected from the starting position of the feed.

[0095] For the q-th position, calculate the initial opening change between it and the next position, expressed mathematically as: in, The sign of the opening indicates the direction of the change: a positive value indicates that the opening is increasing, and a negative value indicates that the opening is decreasing; its absolute value... This indicates the magnitude of the change in the opening.

[0096] according to and The comparison results are handled in three ways in this embodiment: like This indicates that the change in opening between adjacent positions is within the allowable range of the variable diameter mechanism, and the mechanism can operate within the available time window. The tracking of this change is completed internally without the need for amplitude limiting correction, therefore... .

[0097] like This indicates that the required opening increase rate at the next position exceeds the mechanism's capacity; that is, the waste's shape suddenly widens at this location, and the mechanism cannot open it sufficiently in a short time. In this case, a positive limiting correction is needed for the next position, restricting its target opening size to the current target value plus the maximum allowable increment. Therefore, let... The physical meaning of this correction is: the mechanism opens the opening at its maximum permissible speed, but due to the limited time window, it can only reach a certain position upon reaching the next stage. rather than ideal This means that at this location, the actual opening may be slightly smaller than the sum of the waste's dimensions and the safety margin, posing a certain risk of blockage. However, due to the safety margin... As long as the pre-set limit adjustment opening size is still larger than the actual dimensions of the waste, the passage can still be guaranteed. If the limit adjustment results in the opening size being smaller than the dimensions of the waste, it indicates that the system cannot safely handle the waste under the current conveying speed and mechanism parameter configuration, and an early warning should be triggered to reduce the conveying speed or suspend feeding.

[0098] like This indicates that the required opening reduction rate at the next position exceeds the mechanism's capability; that is, the waste's shape suddenly narrows at this position, and the mechanism cannot reduce the size sufficiently in a short time. Therefore, a negative limiting correction is applied to the next position, limiting its target opening size to the current target value minus the maximum allowable reduction. The physical meaning of this correction is: the mechanism contracts the opening at its maximum permissible speed, but due to inertia, it can only contract to a certain extent when reaching the next position. At this point, the opening size is larger than the ideal value. While this will not cause blockage, it may lead to fluctuations in the crusher's load and increased energy consumption. However, the core priority of continuity constraints is to ensure the physical feasibility of the mechanism's movement; the secondary objective is the economic optimization of the opening size.

[0099] The target aperture size sequence is generated through the above forward traversal limiting process. The following continuity constraints must be met: This constraint ensures that the rate of change of the target opening size between adjacent sampling positions remains within the physical execution capability of the diameter-changing mechanism, fundamentally avoiding mechanical shocks and tracking failures caused by command jumps. Simultaneously, the unidirectional scanning characteristic of the forward traversal algorithm guarantees the causality of the constraint processing; that is, the correction of the next position depends only on the current position and previous historical information, conforming to the physical causality law of real-time control systems and facilitating online implementation.

[0100] To further improve the smoothness of the target opening adjustment curve, this embodiment, after amplitude limiting processing, performs sequence... Perform cubic spline interpolation to generate a continuous opening adjustment curve. The reason why cubic spline interpolation is used instead of nonlinear interpolation in this embodiment is that linear interpolation only guarantees the continuity of the function value at the nodes, and its first derivative (rate of change of opening) has a jump at the nodes, causing the variable diameter mechanism to be subjected to acceleration impact at the node positions; while cubic spline interpolation guarantees the continuity of the function value, first derivative, and second derivative at the nodes, making the rate of change of opening and the acceleration bounded, which meets the kinematic smoothness requirements of the mechanical system. This curve uses the feed length position Y as the independent variable and the target opening size as the variable variable. As the dependent variable, it provides a continuous, smooth, and mechanically executable target trajectory reference for subsequent real-time displacement mapping and closed-loop control.

[0101] Please see Figure 3 This figure shows a phased comparison of the generation process of the target opening adjustment curve, as described above. Figure 2 The corrected external dimension sequence shown For input, feed safety margin Preset change threshold The gray dotted lines in the diagram represent the corrected external dimensions. Its numerical range is approximately 60-130 mm, directly reflecting the theoretical width distribution of the waste. The light blue dashed triangle represents the initial target opening size. It can be seen that in There is an upward jump of approximately 40mm (a sudden increase from 90mm to 128mm), which far exceeds the kinematic constraint capability of the diameter-changing mechanism. The red dotted line represents the target opening size after forward traversal limiting processing. When the initial opening change between adjacent positions exceeds At this time, the target opening at the next position is forcibly limited to the value at the previous position plus or minus 20mm, thereby compressing the jump amplitude within the trackable range of the mechanism. The black solid line is the final cubic spline interpolation curve. At the nodes, the function values, first derivative, and second derivative are ensured to be continuous, making the rate of change of the opening and the acceleration bounded, thus avoiding the discontinuity problem at the bend that may be introduced by linear interpolation. The green arrow on the left side of the figure indicates the safety margin. The vertical offset visually represents the buffer space of the target opening relative to its external dimensions. This curve serves as the benchmark for generating the displacement-opening correspondence table, providing a continuous, smooth, and mechanically executable target trajectory for subsequent real-time table lookup mapping.

[0102] Furthermore, after generating the target opening adjustment curve, the method further includes: sequentially associating the target opening sizes corresponding to each feeding displacement in ascending order of feeding displacement to form a displacement-opening correspondence table; when detecting the feeding displacement of waste in real time, the target opening size corresponding to the current feeding displacement is queried first based on the displacement-opening correspondence table.

[0103] After continuity constraint processing and cubic spline interpolation, this embodiment has obtained a continuous and smooth target opening adjustment curve. This curve fully describes the ideal trajectory of the target opening size of the variable diameter mechanism as the feed displacement changes. However, in the actual deployment of embedded real-time control systems, directly storing and calling it as a continuous function presents significant engineering obstacles. On the one hand, cubic spline interpolation involves storing and evaluating the coefficients of piecewise polynomials in real time, including multiple floating-point exponentiation, multiplication, and addition operations. On resource-constrained microcontroller platforms, this results in significant computational delays, making it difficult to meet the real-time requirements of millisecond-level control cycles. On the other hand, the boundary condition processing and node search and location logic of piecewise polynomials increase the complexity of the control software, reducing the system's reliability and maintainability. Therefore, it is necessary to transform the continuous curve into a discrete, compact, and easily retrieval-friendly data structure to adapt to the computing power and storage resources of industrial field controllers.

[0104] This embodiment uses a displacement-opening correspondence table as the engineering representation of the target opening adjustment curve. The essence of this table is to represent a continuous curve... By resampling at equal intervals along the feed displacement dimension, the infinitely continuous functional relationship is discretized into a finite number of record entries. Each record contains a feed displacement value and a corresponding target opening size value, and the two are mapped one-to-one through a table index. This table lookup mechanism transforms the complex floating-point operations in the real-time control stage into simple integer index calculations and linear interpolation, achieving an optimal balance between computational efficiency and storage overhead.

[0105] This embodiment first determines the sampling resolution of the displacement-opening correspondence table. This parameter determines the feed displacement interval between adjacent table records, directly affecting the table's storage capacity and query accuracy. The selection needs to consider a variety of factors: if If the value is too small, the number of table records increases, and the storage space usage increases, but the query accuracy improves and the interpolation error decreases; if... If the size is too large, the number of table records decreases, saving storage space, but the variation in the opening size between adjacent records may exceed the applicable range of the linear assumption, increasing interpolation error. This embodiment will... Set as the sampling interval relative to the outline size sequence Equal, that is This setting ensures that the sampling density of the relation table in the feed displacement dimension is consistent with the original sensing data, which avoids data redundancy caused by oversampling and ensures that the interpolation accuracy meets the control requirements.

[0106] Let the total coverage area of ​​the feed displacement be ,in The moment when the waste first enters the feed inlet. The value corresponding to the moment when the tail end of the waste is about to completely enter the feed inlet is determined by the total length of the waste along the feeding direction. Therefore, the total number of records M in the displacement-opening correspondence table is calculated as follows: The feed displacement of the m-th record in the displacement-opening correspondence table for: Target opening size ; in, This represents the feed displacement value corresponding to the m-th record, in mm, starting from 0 and incrementing by 1. The step size increases incrementally until the entire length of the waste is covered; This indicates that the feed displacement is adjusted by the target opening curve. The calculated target opening size is in mm. The continuity constraint has been applied, and adjacent records in the table satisfy the following conditions: That is, the rate of change of the opening size is always within the allowable range of the mechanism.

[0107] The displacement-opening correspondence table constructed in this embodiment is stored in the controller's non-volatile memory in array form. The table uses compact binary encoding, with each record occupying a fixed number of bytes. For example, if the feed displacement and target opening size are each represented by 4 bytes of single-precision floating-point numbers, the total storage capacity is 8MB. For waste materials with typical lengths ranging from 1m to 3m, the following is taken... If M is between 100 and 300, the total storage capacity is approximately 0.8KB to 2.4KB, which is far below the storage capacity limit of typical industrial controllers, and has good deployability.

[0108] During the real-time control phase, as the waste enters the shredder, the system uses a rotary encoder to detect the current feeding displacement of the waste in real time. The system prioritizes querying the target opening size corresponding to the current feed displacement based on the displacement-opening correspondence table. The query process is divided into two cases: direct hit and interpolation calculation.

[0109] When a direct hit occurs, if the current feed displacement... The value is exactly equal to the offset of a certain record, that is, there exists an integer. Make Then, directly look up the table to output the target opening size corresponding to that record, i.e. The calculation process in this case is extremely simple, requiring only one integer division to determine the index. One array address read This can be completed in hundreds of nanoseconds on a typical 32-bit microcontroller.

[0110] When difference calculation is required, if the current feed displacement An integer exists between two adjacent records. Make Then, linear interpolation is used to calculate the target opening size. The rationale for linear interpolation is based on the target opening adjustment curve. Local smoothness: Due to the continuity constraint processing, it is ensured that the opening variation between adjacent records does not exceed [the specified value]. ,and The choice of [a specific parameter] makes the change approximately linear within a local range. Therefore, linear interpolation can accurately estimate the target opening size at any intermediate position, while avoiding the computational complexity of higher-order interpolation. The specific formula for linear interpolation is: This interpolation operation means that, in order to and Based on two adjacent records, according to the current displacement The relative positional proportions within this interval are used to calculate a weighted average of the two reference opening dimensions. Wherein, This represents the distance of the current displacement relative to the lower limit record. The ratio of the two values ​​is the interpolation weight coefficient, where is the total length of the interval.

[0111] Table 2: Displacement-Opening Correspondence Table Please refer to Table 2. To verify the engineering feasibility of the displacement-opening correspondence table, this embodiment uses a local segment with index m=20 to m=50 as an example to demonstrate the data structure and continuity constraint effect of the lookup mechanism. This segment covers the section where the feed displacement ranges from 200mm to 500mm, corresponding to the complete change process of the waste shape from the transition section to the widest local position (the enlarged part of the main stem of the straw, m=25-27, opening about 156mm), then narrowing through a thinner segment (the thinner segment at the tip of the straw, m=36-39, opening about 134mm), and then rising again to the second peak (m=47-48, opening about 153mm).

[0112] As shown in Table 2, the variation in opening size between adjacent records is within the range of 0.4 mm to 2.8 mm, which is much smaller than the preset variation threshold. This satisfies the kinematic constraints of the variable diameter mechanism. The practical significance of this variation range lies in: controlling the conveying speed... Calculate the time interval between adjacent table records. The theoretical maximum displacement capacity of the variable diameter mechanism within 33.3 ms is... The actual change is only about 10% of the capability, leaving sufficient dynamic margin for mechanical transmission clearance, motor response delay and load disturbance.

[0113] During the real-time control phase, when the rotary encoder feeds back the current feed displacement... At that time, the controller performs a table lookup index calculation. It directly hits the record with m=23 in Table 2 and outputs the target opening. ;like If the value falls between m=23 and m=24, then linear interpolation can be used for calculation. The specific formula is as follows: This interpolation operation involves only two array addressings, one subtraction, one division, and one multiplication. It can be completed within 500ns on a typical 32-bit microcontroller, meeting the real-time requirements of millisecond-level control cycles.

[0114] This embodiment uses a displacement-opening correspondence table as the engineering implementation of the target opening adjustment curve, which has the following technical advantages: real-time calculation, the table lookup-interpolation mechanism simplifies the complex spline evaluation to basic arithmetic operations, meeting the real-time constraints of embedded systems; compact storage, the fixed-interval discrete sampling significantly reduces data storage requirements, facilitating deployment in resource-constrained controllers; reliable execution, the table lookup operation does not have iterative convergence or numerical stability issues, and the output results are predictable and definite, meeting the strict deterministic requirements of industrial control; and convenient maintenance, the relationship table is stored in array form, facilitating offline modification, online updates and version management, and supporting parameterized configuration for different waste types.

[0115] Through the above processing, this embodiment realizes the transformation from continuous mathematical curves to discrete engineering data structures, providing an efficient, reliable, and quickly queryable target trajectory data source for subsequent real-time displacement mapping and closed-loop variable diameter control.

[0116] Step 4: During the process of waste entering the crusher, the feeding displacement of the waste is detected in real time, and the feeding displacement is mapped to the target opening adjustment curve to determine the corresponding target opening size.

[0117] The preceding steps completed the comprehensive perception of the waste's shape and the pre-planning of the opening curve, all done offline before the waste enters the crusher. This step marks a crucial shift in the entire control process from offline planning to online execution—as the waste actually enters the crusher, the system must track its feeding position in real time and dynamically determine the target opening size that the diameter-changing mechanism should achieve at the current moment based on the pre-generated target opening adjustment curve. The practical consideration behind this phased design is that computationally intensive tasks such as image processing, contour extraction, and curve optimization can be completed offline, without real-time constraints; while the online execution phase must rapidly complete displacement detection and curve mapping under strict timing requirements to ensure that the diameter-changing mechanism's movements are synchronized with the actual movement of the waste. If there is a lag or error in displacement detection, the diameter-changing mechanism's opening adjustment will be earlier or later than the actual arrival of the waste, resulting in the waste losing constraint when the opening is too large, or getting stuck when the opening is too small, causing the entire diameter-changing control to fail. The core challenge of real-time detection of feed displacement is to find a sensing solution that can accurately measure linear displacement, adapt to harsh industrial environments (dust, vibration, electromagnetic interference), and whose output signal can be quickly read and processed by the controller.

[0118] The process of mapping the feed displacement to the target opening adjustment curve is essentially a process of mapping a one-dimensional displacement variable. Through the pre-defined functional relationship Convert to one-dimensional opening size variable The physical meaning of this mapping is that, regardless of the waste's position in the feeding process, the system can predict its dimensions in advance and pre-adjust the opening, achieving a "opening before material arrives" feedforward control effect. This feedforward control, combined with feedback correction in subsequent steps, constitutes a complete closed-loop control strategy. The specific implementation of the mapping calculation relies on the aforementioned displacement-opening correspondence table. In the real-time control phase, the controller reads the current pulse count value of the rotary encoder at fixed intervals and converts it into the actual feeding displacement. Then, the corresponding target opening size is quickly obtained by looking up the table index and using linear interpolation. This target value serves as the setting input for the position control of the variable diameter mechanism in subsequent steps, driving the actuator to move towards the target position.

[0119] In actual operation, the waste conveying process experiences speed fluctuations (motor speed adjustment, load changes), and the actual feeding displacement is not strictly linear with time. Therefore, displacement detection must be based on real-time pulse feedback from the encoder for integral calculation, rather than simply multiplying time by the nominal speed, to ensure the accuracy of displacement measurement. Furthermore, once the tail end of the waste material passes through the feed inlet, the system must identify the end of feeding, reset the displacement counter, and prepare for the processing of the next piece of waste, thus achieving cyclical execution of the control process.

[0120] Furthermore, the real-time detection of the feed displacement of the waste specifically includes: acquiring the rotation pulse signal of the feed drive roller shaft through a rotary encoder set on the feed drive roller shaft; calculating the rotation angular displacement of the feed drive roller shaft based on the obtained rotation pulse signal; converting the rotation angular displacement into the real-time feed displacement of the waste by combining the roller diameter parameter of the feed drive roller shaft; and mapping the obtained real-time feed displacement to the corresponding displacement position in the target opening adjustment curve to determine the target opening size corresponding to the current feed displacement.

[0121] This embodiment uses a rotary encoder as the core displacement sensor. Since the feed drive roller is the core component directly driving the waste forward in the conveying channel, there is mechanical coupling between the roller surface and the waste or conveyor belt. Under ideal, slip-free conditions, the two have a strict motion synchronization relationship. Therefore, mounting the encoder on the drive roller allows direct measurement of the rotational motion at the drive end, thereby calculating the linear displacement of the waste. This avoids the structural complexity and reliability issues associated with mounting sensors on the conveyor belt surface or the waste itself. Furthermore, the drive roller is typically located on a fixed support structure of the conveying channel, resulting in relatively low vibration, providing a stable mechanical mounting foundation for the encoder and ensuring signal quality.

[0122] This embodiment uses an incremental photoelectric rotary encoder, whose output signal consists of two orthogonal pulse sequences (phase A and phase B), and a reference signal (phase Z) that outputs one pulse per revolution. The orthogonality between phase A and phase B pulses (90° phase difference) is used not only for counting but also to determine the direction of rotation by judging whether the phase is leading or lagging. The encoder's resolution is determined by the number of pulses it outputs per revolution. Decide, The selection requires a trade-off between displacement measurement accuracy and controller counting load. If If the pulse size is too small, the linear displacement corresponding to a single pulse will be too large, resulting in significant quantization errors in displacement measurement. This could lead to a rough response from the variable diameter mechanism at abrupt changes in opening size. Excessive pulse size or frequency may exceed the upper limit of the controller's counter or interrupt handling capabilities, especially under high-speed conveying conditions. This embodiment will... The value range is limited to 1000 to 5000 pulses / revolution, and this range applies to conveyor speeds. Under typical operating conditions between 0.1 m / s and 1.0 m / s, the corresponding linear displacement of a single pulse is on the order of 0.1 mm to 1.0 mm, which can meet the accuracy requirements of variable diameter control.

[0123] Let the diameter of the feed drive roller be... The unit is mm. The theoretical straight-line distance that the waste travels along the feed direction for each rotation of the roller is the circumference of the roller. For each pulse output by the encoder, the corresponding roller rotation angle increment is: The corresponding linear displacement increment of the waste for: ; The physical meaning of is the equivalent value of encoder resolution in the linear displacement dimension, which is the smallest quantization unit for system displacement measurement.

[0124] During real-time detection, the controller counts the pulses by capturing the rising (or falling) edge of the encoder's A-phase pulses, and simultaneously determines the rotation direction based on the phase relationship between phases A and B. Assuming the pulse count increases during forward rotation and decreases during reverse rotation, the current cumulative pulse count is... The rotational angular displacement of the feed drive roller shaft for: This angular displacement, expressed in radians, is equal to the cumulative number of pulses multiplied by the angular increment corresponding to a single pulse. The rotational angular displacement, combined with the roller diameter parameter, is then converted into the real-time feed displacement of the waste material. The conversion formula is: This formula indicates that the real-time feed displacement With cumulative pulse count value They exhibit a strictly linear proportional relationship, with the proportionality coefficient being the linear displacement increment corresponding to a single pulse. This linear relationship relies on two preconditions: first, there must be no relative slippage between the encoder and the drive roller (usually ensured by a rigid coupling or key connection); second, there must be no slippage between the drive roller and the waste (or conveyor belt). In actual engineering, there may be slight elastic slippage between the conveyor belt and the roller, especially during startup, braking, or sudden load changes. To compensate for this slippage error, this embodiment introduces a slippage correction coefficient during the system calibration phase. Its value ranges from 0.98 to 1.00, and is determined through actual measurement and calibration. The corrected formula for calculating the real-time feed displacement is: The encoder's Z-phase signal is used for absolute position calibration and cumulative error clearing. Whenever a new round of waste feeding begins, the system detects the rising edge of the Z-phase pulse and sets the pulse counter... Zeroing ensures the accuracy of the starting reference for displacement measurement. This mechanism eliminates displacement drift caused by accumulated errors in pulse counting or accidental step loss during long-term operation, guaranteeing the consistency and repeatability of displacement measurements across multiple batches of waste.

[0125] Through the complete process of signal acquisition, angular displacement calculation, linear displacement conversion and curve mapping described above, this embodiment realizes the real-time, accurate and reliable conversion from the original pulse signal of the rotary encoder to the target opening size of the diameter-changing mechanism, providing accurate displacement feedback and opening setting input for subsequent closed-loop diameter-changing control.

[0126] Step 5: Control the diameter changing mechanism to adjust the feed inlet opening according to the target opening size, and obtain the position feedback signal of the diameter changing mechanism.

[0127] After determining the target opening size corresponding to the current feed displacement, the core task of this step is to convert this digitized target value into the physical action of the variable diameter mechanism, so that the actual opening size of the feed inlet tracks the change of the target value. At the same time, position feedback signals are collected during the execution process to provide a data basis for deviation correction. This step is the key interface for "electrical control to mechanical" in the entire variable diameter feed control system, and its performance directly determines whether the system can accurately reproduce the pre-planned opening curve as mechanical motion.

[0128] The selection of a diameter-changing mechanism requires comprehensive consideration of factors such as adjustment range, response speed, load capacity, and control accuracy. This embodiment uses an electrically driven split-plate structure as the diameter-changing actuator. Its working principle is as follows: two synchronously operating electric actuators drive the left and right arc-shaped baffles respectively, moving them towards or away from each other along a horizontal guide rail perpendicular to the feeding direction. The opening size of the feed inlet is continuously adjusted by changing the distance between the two baffles. The advantages of this structure are that the split-plate design ensures that the center line of the opening always coincides with the center line of the feed channel, avoiding waste skewing caused by unilateral adjustment; the arc-shaped baffles match the outer contour of the crusher rotor, effectively guiding waste into the crushing area when the opening narrows, reducing dead corners and material accumulation. The drive motor of the electric actuator uses a DC servo motor or stepper motor, possessing closed-loop position control capability. Its controller receives target position commands from the main control system, drives the motor, and provides real-time feedback on the current position of the actuator through a built-in potentiometer or encoder. The linear stroke of the push rod determines the maximum adjustment range of the diameter-changing mechanism, which should cover the size range of all types of waste processed by the crusher. In this embodiment, the push rod stroke is set to 100mm to 300mm, corresponding to a minimum feed inlet opening size (mechanical clearance when fully closed) of approximately 10mm and a maximum size of approximately 310mm, which can meet the processing needs of most industrial wastes.

[0129] The main control system sends a target position command to the electric linear actuator controller at a fixed control cycle (e.g., 10ms), which is determined by the current target opening size. This was calculated by combining the geometric parameters of the mechanism. Assuming the zero position of the push rod corresponds to the fully closed state of the feed inlet, the push rod stroke... With opening size The geometric relationship between them is determined by the transmission ratio of the baffle mechanism. Decision, that is Transmission ratio The physical meaning of is the proportionality coefficient between the linear displacement of the push rod and the change in the opening size, and its value is determined by the geometric parameters of the baffle mechanism, such as the lever arm length and connecting rod angle. For a symmetrical split structure, Typically, it's 0.5, meaning that for every 1mm movement of the push rod, the single-sided baffle moves by 0.5mm, resulting in a total opening change of 1mm on both sides. This formula converts the target opening size into the target stroke position of the push rod, which serves as the position setting value for servo control.

[0130] After receiving the target stroke position, the electric linear actuator controller compares it with the current actual position, generates a position deviation signal, and calculates the motor drive voltage or current using a PID control algorithm to drive the motor to the target position. The parameters of the PID controller need to be tuned during the commissioning phase to balance response speed and overshoot suppression. Specific parameters of the PID controller include the proportional gain. Integral coefficient Differential coefficients Since the load characteristics (lateral pressure of waste on the baffle) of the variable diameter mechanism vary with the opening size and waste type, this embodiment adopts a control strategy based on adaptive adjustment of PID gain according to position error. The gain is increased to accelerate convergence when the error is small, and the gain is reduced to avoid oscillation when the error is large.

[0131] During the actuator's movement, the position feedback signal is acquired through a position sensor built into the electric actuator. This sensor is typically a potentiometer or a magnetostrictive displacement sensor, and its output signal is an analog voltage or digital pulse proportional to the actuator's stroke. The controller samples this feedback signal at the same frequency as the control cycle and converts it into the current actuator stroke position. This is then converted into the actual opening size using the following formula: .

[0132] The actual opening size This is the core feedback quantity used for deviation comparison in the aforementioned steps. The accuracy and real-time performance of the feedback signal directly affect the performance of the closed-loop control; therefore, the resolution of the position sensor must meet the control accuracy requirements of the opening size. In this embodiment, the control accuracy of the opening size is required to be better than 2mm, corresponding to a control accuracy of better than 1mm for the push rod stroke (taken as...). Therefore, a magnetostrictive displacement sensor with a resolution of 0.1 mm was selected, which has an accuracy reserve of 10 times that of the control requirements and can effectively suppress the impact of sensor noise and quantization error on control performance.

[0133] In addition, the diameter-changing mechanism is equipped with limit switches as safety protection devices. The upper limit switch is triggered when the opening reaches the maximum allowable value to prevent the push rod from overextending and causing damage to the mechanical structure; the lower limit switch is triggered when the opening reaches the minimum allowable value to prevent the baffle from colliding and causing the motor to stall. The limit switch signal is connected to the emergency stop input terminal of the controller, and once triggered, the motor drive power is immediately cut off to ensure the safety of equipment and personnel.

[0134] Through the complete process of issuing target position commands, executing push rod drives, and collecting position feedback, this embodiment realizes a closed-loop control front end from digital target opening size to physical mechanical motion, providing necessary execution actions and feedback data for subsequent deviation correction. It is a key link in the entire variable diameter feeding control system.

[0135] Step 6: Determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

[0136] In the preceding steps, the system has sent the target opening size control command to the variable diameter mechanism and acquired the position feedback signal during execution. However, due to factors such as mechanical transmission backlash, motor response delay, load disturbance, and sensor error, there will inevitably be a deviation between the actual opening size of the variable diameter mechanism and the target value. If this deviation is not detected and corrected in real time, it will accumulate over time, leading to a decrease in opening size tracking accuracy, and in severe cases, it may cause jamming or over-opening. Therefore, the core task of this step is to establish a closed-loop feedback mechanism, dynamically adjusting the control command by continuously comparing the difference between the actual opening and the target opening, so that the actual opening accurately converges to the target value. The technical significance of this closed-loop correction step is that it upgrades the entire variable diameter feeding control system from simple feedforward control to feedforward-feedback composite control. Feedforward control can adjust the opening in advance according to the pre-planned curve, with a rapid response and no overshoot; however, its inherent defect is a lack of adaptability to model errors and external disturbances. The feedback control in this step, by detecting and correcting the actual deviation in real time, effectively suppresses the uncertainties that feedforward control cannot cover, significantly improving the robustness and tracking accuracy of the system. The combination of these two factors ensures both speed and accuracy of control.

[0137] First, the actual opening size of the variable diameter mechanism is determined based on the position feedback signal obtained in the preceding steps. This feedback signal typically represents the current stroke position of the push rod in the form of analog voltage or digital pulses. It needs to be signal conditioned and calibrated to convert it into the actual opening size in engineering units (mm). Secondly, the actual opening size Target opening size corresponding to the current feed displacement Perform deviation comparison and calculate opening deviation. The mathematical expression is: The sign of the deviation indicates the direction of the actual opening relative to the target value: a positive value indicates that the actual opening is greater than the target value (over-opening), and a negative value indicates that the actual opening is less than the target value (under-opening). Its absolute value represents the degree of deviation and is the basis for determining whether correction is needed and the magnitude of the correction. Finally, the opening adjustment of the variable diameter mechanism is corrected based on the deviation comparison results. The design of the correction strategy must consider both response speed and stability. If the correction is too aggressive, it may lead to system oscillation or even divergence; if the correction is too conservative, convergence will be slow, and the deviation cannot be eliminated in time. Furthermore, the closed-loop correction process must also consider the system's working boundary conditions. When the real-time feed displacement of the waste reaches the end of the target opening adjustment curve, it indicates that the processing of that waste is about to be completed, and the system should gradually restore the opening to the default initial state to prepare for the feeding of the next waste. This state switching logic ensures the continuity and automated operation of the control flow.

[0138] Furthermore, the specific method for determining the actual opening size of the variable diameter mechanism based on the position feedback signal is as follows: obtain the current position feedback value of the actuator of the variable diameter mechanism, and convert the current position feedback value into the current actual opening size according to the pre-established conversion relationship between the mechanism position and the feed inlet opening size.

[0139] The position feedback signal of the variable diameter mechanism comes from the displacement sensor built into the electric push rod. Its output is usually in the form of an analog voltage signal or a digital pulse signal, which represents the current stroke position of the push rod, rather than the direct opening size. Therefore, it is necessary to establish a mathematical conversion relationship between the push rod stroke position and the feed inlet opening size, converting the raw sensor readings into opening size values ​​that can be directly understood in engineering.

[0140] The physical basis of this conversion relationship is the geometric transmission structure of the variable diameter mechanism. This embodiment employs a split-plate structure, with two synchronous electric push rods driving the left and right arc-shaped plates to move along a horizontal guide rail, respectively. Let the current stroke position of the push rod be... The unit is mm. This value is measured from the zero position when the push rod is fully retracted and increases towards the fully extended position. (Inlet opening size) Defined as the minimum distance between the inner edges of the left and right baffles. In a symmetrical transmission structure, there is a fixed proportional relationship between the push rod stroke and the displacement of a single-sided baffle; this ratio is determined by the transmission ratio of the mechanism. Decide.

[0141] Assume the output end of the push rod is hinged to a single-sided baffle via a connecting rod, and the effective lever arm length of the connecting rod is... The angle between the push rod axis and the direction of the baffle movement is Under the assumption of small displacement (i.e., the push rod stroke is much smaller than the connecting rod length), the axial displacement of the push rod and the lateral displacement of the baffle are approximately linearly related. Therefore, the transmission ratio can be expressed as: .

[0142] Wherein, constant coefficient The design is based on the symmetry of the split structure, that is, the movement of the push rod. Single-sided baffle movement The total change in opening on both sides is then... In an ideal vertical arrangement ( )hour, This means that for every 1mm movement of the push rod, the opening changes by 1mm; this is the most common simplified design. However, in actual engineering projects, due to space constraints... Typically within the range of 0° to 30°, the corresponding The value is between 0.5 and 0.577. In this embodiment, the value is precisely determined through 3D modeling and motion simulation during the mechanism design phase. The values ​​are determined and calibrated using standard gauge blocks after assembly to ensure that the consistency error between the theoretical and actual values ​​is less than 1%.

[0143] Based on the above transmission ratio, the push rod stroke position Size of the feed inlet The conversion relationship between them is as follows: ;in, When the push rod is fully retracted ( The mechanical residual opening is determined by the minimum clearance of the baffle structure, typically 5mm to 10mm, to prevent collision interference when the baffle closes. The inverse operation of this formula is the conversion from opening size to push rod stroke, used for generating the target position command in the aforementioned steps.

[0144] During real-time operation, the controller reads the output signal of the displacement sensor at a fixed sampling period. If the sensor output is an analog voltage signal... First, analog-to-digital conversion and linear calibration are performed. The mathematical expression is: ;in, This is the sensor sensitivity coefficient, expressed in mm / V. The zero-position offset is measured in mm and is determined by the sensor's factory calibration data or on-site calibration. If the sensor output is a digital pulse signal (such as an absolute encoder with an SSI or RS-485 interface), the current travel position is obtained by directly parsing the protocol data frame.

[0145] Obtain the putter travel position Then, substitute the values ​​into the aforementioned conversion relationship to calculate the current actual opening size. The specific formula is as follows: This refers to the actual opening size used for deviation comparison in the aforementioned steps. To improve conversion accuracy, this embodiment uses floating-point arithmetic in the software to implement the above formula and performs digital filtering on the sensor signal (such as first-order low-pass filtering or moving average filtering) to suppress the influence of high-frequency noise on the opening size calculation. The setting of the filtering time constant needs to balance noise suppression and response speed, typically within the range of 10ms to 50ms, to ensure that the filtering calculation is completed within the control cycle without introducing significant phase lag.

[0146] Furthermore, the conversion results need to be validated for reasonableness. If the calculated... Exceeding the physical limits of the mechanism, i.e., less than or greater than If the sensor signal shows an abnormal jump (the difference from the previous cycle exceeds a preset threshold), it is determined to be a sensor malfunction or signal interference. The system will trigger an alarm and switch to safe mode (maintain the current opening or restore to the default opening) to avoid control instability caused by corrections based on erroneous feedback.

[0147] Furthermore, the step of correcting the opening adjustment of the variable diameter mechanism based on the comparison result specifically includes: comparing the current actual opening size with the target opening size corresponding to the current feed displacement to obtain the opening deviation; when the obtained opening deviation is greater than a preset positive deviation threshold, controlling the variable diameter mechanism to perform a correction action of reducing the opening; and when the opening deviation is less than a preset negative deviation threshold, controlling the variable diameter mechanism to perform a correction action of increasing the opening.

[0148] To determine whether the opening deviation reaches a level that requires triggering a correction action, this embodiment introduces a preset positive deviation threshold. With respect to the preset negative deviation threshold Positive deviation threshold The upper tolerance limit for over-opening conditions is defined, and its value ranges from 2mm to 5mm; negative deviation threshold. The tolerance lower limit for under-opening conditions is used to define the range of -5mm to -2mm. The reason for the asymmetrical setting of positive and negative deviation thresholds is that while over-opening conditions do not directly cause blockage, they can exacerbate rotor load fluctuations and increase energy consumption. Furthermore, an excessively large opening may cause small-sized waste materials to tumble and become entangled within the feed inlet. Therefore, the positive deviation threshold... The absolute value is usually set relatively small (e.g., 2mm to 3mm) to strictly control the situation where the opening is too large; while the under-opening state is directly related to the passage of waste. Once the actual opening is smaller than the shape of the waste, it may cause jamming and shutdown. Therefore, the negative deviation threshold is important. The absolute value is usually set relatively large (e.g., 3mm to 5mm) to reduce unnecessary frequent corrections and avoid mechanism oscillations while ensuring safe passage. In this embodiment, the positive deviation threshold... Set to 3mm, negative deviation threshold With a setting of -4mm, this set of parameters achieves an engineering balance between the stringency of over-opening control and the safety margin of under-opening.

[0149] The triggering logic for the correction action in this embodiment is as follows: when If the actual opening is determined to be too large, the diameter-changing mechanism is controlled to perform a corrective action to reduce the opening. This corrective action is achieved by sending a retraction command to the electric push rod controller, causing the left and right baffles to move towards each other, thus reducing the opening size. The calculation of the correction amount follows the proportional control principle, and the correction proportional coefficient is set as follows: The opening adjustment amount for a single correction for: Wherein, the correction ratio is The value range is from 0.5 to 1.0, and in this embodiment, it is taken as... The negative sign indicates the direction of reducing the opening. This refers to the over-opening amount exceeding the positive deviation threshold. This formula ensures that the correction amount is proportional to the degree of over-opening; the more severe the over-opening, the greater the contraction amplitude of a single correction. The limit will not exceed the overshoot itself, thus avoiding overshoot oscillation.

[0150] when If the actual opening is deemed insufficient, the diameter-changing mechanism is controlled to perform a corrective action to increase the opening. This corrective action is achieved by sending an extension command to the electric push rod controller, causing the left and right baffles to move in opposite directions, thus increasing the opening size. The opening adjustment amount for a single correction is... for: At this time, due to and If all are negative, then and The result after multiplication A positive value indicates an increase in the opening. This formula also ensures that the correction amount is proportional to the degree of under-opening.

[0151] when When the opening deviation is within the preset allowable deviation range, no corrective action is triggered, and the diameter-changing mechanism maintains its current state. This dead zone setting avoids frequent reciprocating motion of the mechanism caused by minor noise fluctuations near the target value, reducing mechanical wear and energy consumption.

[0152] The execution of the correction action must satisfy the kinematic constraints of the diameter-changing mechanism, that is, the opening adjustment amount of a single correction. The change must not exceed the preset threshold determined in the aforementioned continuity constraint processing. If the calculated result Then limit its amplitude to ,Right now: .

[0153] in, For symbolic functions, This is the actual adjustment amount after limiting. This limiting process ensures that the correction action is within the physical execution capability of the mechanism, avoiding mechanical shock caused by excessively forceful correction commands.

[0154] The conversion relationship between the correction command and the push rod travel position is as follows: .in, This is the correction amount for the push rod's stroke. This is the transmission ratio of the mechanism. The controller subtracts (if over-opening) or adds (if under-opening) the current target stroke position of the push rod. A new target stroke command is generated and sent to the electric actuator controller, which drives the motor to perform the corresponding contraction or extension action.

[0155] Furthermore, after each correction action is performed, the position feedback signal is reacquired, and the actual opening size of the variable diameter mechanism is re-determined. Based on the deviation between the re-determined actual opening size and the current target opening size, the control command for the next correction action is updated to form a closed-loop opening adjustment process.

[0156] The execution of a single correction action cannot guarantee that the actual opening size will immediately and accurately converge to the target value. This is because the motor of the electric actuator has an inherent dynamic response process from receiving the command to completing the mechanical movement, including electrical delay, mechanical inertia, and elastic deformation of the transmission chain. Therefore, in this embodiment, after each correction action is issued, the next round of correction calculation is not performed immediately, but rather a complete control cycle is waited for. After the mechanical movement of the mechanism has fully responded, the position feedback signal is collected again, the correction effect is evaluated, and the control command is iteratively updated accordingly, forming a closed-loop adjustment mechanism of "detection-comparison-correction-redetection".

[0157] Control cycle The settings need to balance system response speed and mechanical tracking capability. If If the timeframe is too short, a new instruction is issued before the organization has completed the previous correction, leading to an accumulation of instructions and chaotic execution; if... If the opening is too long, the system will lag significantly in tracking changes in the target opening, especially at locations where the waste's shape and size change abruptly, leading to large dynamic deviations. This embodiment will... The value range is limited to between 10ms and 50ms. The lower limit of 10ms is suitable for lightweight variable diameter mechanisms using high-dynamic servo motors and low-inertia transmission designs, which have a small mechanical response time constant and can withstand a high control refresh rate. The upper limit of 50ms is suitable for heavy-duty variable diameter mechanisms driven by stepper motors or ordinary asynchronous motors with long transmission chains, whose mechanical response is relatively slow and requires a longer control cycle to ensure that the motion is fully completed. This embodiment is based on the measured electromechanical time constant of the actual mechanism. The refresh rate is set to 20ms, which means the refresh rate of the control system is 50Hz. This frequency can capture the changing trend of the target opening in a timely manner, and also provides sufficient response time for the mechanical tracking of the mechanism.

[0158] At the beginning of each control cycle, the controller first reads the current output signal of the displacement sensor through an analog-to-digital converter interface or a digital communication interface to obtain the current stroke position of the push rod. The read operation is completed instantaneously at the start of the control cycle, and the signal acquisition time is much shorter than... This can be ignored. Then, based on the conversion relationships described above, [the following is done]... Converted to the current actual opening size The conversion formula is: .in, This indicates the newly determined current actual opening size, compared to the size before the correction action was performed. To differentiate them, so as to reflect the status update in the time dimension.

[0159] get Then, it needs to be compared with the target opening size corresponding to the current feed displacement. A new round of deviation comparisons is conducted. However, the target opening size... It is not fixed, but changes with the real-time feeding and displacement of the waste. The continuous increase, along the target opening adjustment curve It evolves dynamically. Therefore, in each closed-loop iteration, the controller must first update the current feed displacement based on the latest pulse count value of the rotary encoder. Then, the latest target opening size at that displacement is obtained by querying the displacement-opening correspondence table or by interpolation. This ensures that the deviation comparison is always based on the displacement-opening correspondence at the same moment.

[0160] Recalculated opening deviation for: .according to The value and the positive and negative deviation threshold Based on the comparison results, and according to the correction action triggering logic described above, the opening adjustment amount for the new round of correction is calculated. And convert it into push rod travel correction amount. The updated target stroke command is then generated and sent to the electric actuator controller. Thus, the system completes a closed-loop iteration from the z-th control cycle to the (z+1)-th control cycle. This is a closed-loop iterative counter.

[0161] The essence of this closed-loop regulation process is a discrete-time feedback control system, and the mathematical expression of its control law is: in, The threshold corresponding to the current deviation direction, when hour, ;when hour, When the deviation is within the dead zone, no correction is triggered. .

[0162] Furthermore, the process of obtaining position feedback, determining the actual opening size, comparing deviations, and updating correction actions is repeated until the deviation between the actual opening size and the target opening size falls within a preset allowable deviation range, or the real-time feeding displacement of the waste reaches the end position corresponding to the target opening adjustment curve.

[0163] The termination condition of the closed-loop adjustment process is determined by two mutually exclusive criteria. If either criterion is met, the closed-loop correction cycle of the current waste is terminated, and the system enters the preparation state or standby state for the next waste.

[0164] The first termination criterion is the accuracy convergence criterion, which states that the deviation between the actual opening size and the target opening size falls within a preset allowable deviation range. The preset allowable deviation range consists of an allowable positive deviation... With allowable negative deviation The definition, its value is usually the same as or slightly wider than the modified trigger threshold, that is... ,and In the embodiments, this is... The mathematical expression for the criterion is: When this condition is met, it indicates that the actual opening of the variable diameter mechanism has fully converged to the target value, and the tracking accuracy meets the engineering requirements, so no further correction is needed. At this time, the system maintains the current opening state and continuously monitors the changes in the feed displacement and the target opening. If the target opening is updated due to changes in the shape of the waste, causing the deviation to exceed the allowable range again, the closed-loop correction will be automatically re-triggered.

[0165] The second termination criterion is the feeding completion criterion, which is that the real-time feeding displacement of the waste reaches the position corresponding to the end of the target opening adjustment curve. (The position corresponding to the end of the target opening adjustment curve is...) The total length of the waste along the feed direction Determined, that is This value has been accurately calculated and stored during the offline image processing and curve generation phases. Real-time feed displacement. The formula for comparing with the end position is: in, The safety margin at the end is set between 10mm and 30mm. Physically, it represents a buffer distance to ensure that the system has identified the feeding completion status and begun resetting preparation before the waste completely passes through the inlet, preventing end-of-pipe blockage due to detection lag. This embodiment uses... .

[0166] When the feeding completion criterion is met, it indicates that the current waste diameter change control process has ended, and the system performs the following closing operations: First, a reset command is sent to the electric push rod controller to drive the diameter change mechanism to restore the opening to the default initial opening size. The default value is typically set to a full-width opening in the conveyor channel or a medium opening suitable for most waste types. In this embodiment... Secondly, the pulse counter of the rotary encoder... Zeroing establishes an accurate zero-point reference for the displacement measurement of the next piece of waste; finally, the closed-loop iteration counter z is reset to 0, and the image acquisition module and the external dimension extraction module enter standby mode, waiting for the next piece of waste to enter the field of view to trigger the complete diameter change control process again.

[0167] During the repeated execution of closed-loop regulation, a maximum limit on the number of iterations also needs to be set. To prevent infinite loops caused by extreme situations such as mechanical failure, sensor malfunction, or drastic changes in the target opening. This embodiment will The maximum closed-loop adjustment time is set to 500 times. If the accuracy convergence criterion is not met within 10 seconds and the feeding is not yet complete, the system determines that the closed-loop regulation is abnormal, triggers an alarm signal, and switches to safe mode, maintaining the current opening or forcibly restoring to the default opening. At the same time, a fault code is reported to the host computer, prompting maintenance personnel to check the mechanical status of the diameter changing mechanism, sensor connection, and drive system.

[0168] Table 3: Closed-loop correction iteration record table Please refer to Table 3. To verify the dynamic response characteristics and convergence performance of the above closed-loop opening adjustment process, this embodiment uses the feed displacement... The start-up phase and Taking the final stage as an example, the closed-loop correction iteration process is recorded.

[0169] At the initial moment of the iteration The actual opening of the variable diameter mechanism Opening the door to the current target There is a deviation The absolute value of this negative deviation far exceeds the preset negative deviation threshold. The system determines the system to be in an under-opening state. The single correction amount is calculated based on the proportional correction law. This value is less than No amplitude limiting is required; it can be directly converted into push rod travel correction amount. The controller sends an extension command to the electric actuator. After one control cycle... After resampling, the actual opening was updated to 120.2 mm, and the deviation decreased to -5.5 mm. Under-opening correction was still triggered, but the correction amount was reduced to 1.2 mm. By the third iteration, the deviation... Falling within the allowable deviation range Within this phase, the system ceases its corrective actions, and the mechanism enters a stable holding state. This startup process took 40ms and converged after two active corrections, demonstrating the excellent speed of the closed-loop control.

[0170] During the stable tracking phase, the system continuously monitors at a period of 20ms after convergence. Since the target opening changes smoothly along the curve with the feed displacement (the change between adjacent sampling points is approximately 0.1-0.3mm), and the variable diameter mechanism has a tracking capability of 20mm within 20ms, the actual opening can follow the target value fluctuations without deviation. The deviation remains within the dead zone, eliminating the need for additional correction actions. This phase verifies the effectiveness of the composite control strategy of "feedforward lookup table dominance and feedback dead zone suppression," significantly reducing the frequent reciprocating motion of the mechanism.

[0171] when When the system feed is completed, the real-time feed displacement... Achieve the target opening adjustment curve end position Subtract safety margin When the threshold is reached, the system determines that the current waste processing is complete, stops the closed-loop iteration, and performs a reset operation: sending a default opening command to the electric actuator. Simultaneously, the rotary encoder pulse counter Zeroing out establishes a zero-point baseline for the feeding control of the next waste item.

[0172] In summary, the closed-loop opening adjustment process of this embodiment has rapid convergence capability in the startup phase (entering the dead zone within 40ms after 2 iterations), maintains zero-correction tracking in the steady-state phase (avoiding mechanical oscillation), and has reliable boundary switching logic (safety margin triggering reset) in the ending phase, thus realizing high-precision dynamic tracking of the actual opening of the variable diameter mechanism to the target opening.

[0173] Through the above-described closed-loop opening adjustment process, this embodiment achieves high-precision dynamic tracking of the actual opening of the variable diameter mechanism to the target opening. It organically combines the forward-looking nature of feedforward control with the robustness of feedback control, ensuring that the opening size of the crusher's feed inlet always maintains optimal matching with the real-time external dimensions of the waste throughout the entire feeding process. This effectively avoids problems such as jamming, overload, and energy waste, and significantly improves the crusher's operational stability and processing efficiency.

[0174] Please see Figure 4The present invention also provides a machine vision-based variable diameter feeding control system for a waste crusher. This machine vision-based variable diameter feeding control system is used to execute the aforementioned machine vision-based variable diameter feeding control method for a waste crusher, including: Image acquisition module: used to continuously acquire images of the waste to be fed into the crusher before the waste enters the crusher, and obtain continuous image data of the waste along the feeding direction; Dimension extraction module: used to extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction, forming a dimension sequence; Adjustment curve generation module: used to determine the target opening size of the variable diameter mechanism under the corresponding feed displacement according to the external dimension sequence, and generate the target opening adjustment curve of the variable diameter mechanism; Displacement mapping matching module: used to detect the feeding displacement of waste in real time during the process of waste entering the crusher, and map the feeding displacement to the target opening adjustment curve to determine the corresponding target opening size; Variable diameter adjustment feedback module: used to control the variable diameter mechanism to adjust the feed inlet opening according to the target opening size, and to obtain the position feedback signal of the variable diameter mechanism; Deviation closed-loop correction module: used to determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

[0175] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0176] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0177] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0178] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A machine vision-based variable diameter feeding control method for a waste shredder, characterized in that, The specific steps include: Step 1: Before the waste enters the crusher, continuously acquire images of the waste to be fed to obtain continuous image data of the waste along the feeding direction; Step 2: Extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction to form a sequence of dimensions; Step 3: Based on the aforementioned external dimension sequence, determine the target opening size of the variable diameter mechanism under the corresponding feed displacement, and generate the target opening adjustment curve of the variable diameter mechanism; Step 4: During the process of waste entering the crusher, the feeding displacement of the waste is detected in real time, and the feeding displacement is mapped to the target opening adjustment curve to determine the corresponding target opening size; Step 5: Control the diameter changing mechanism to adjust the feed inlet opening according to the target opening size, and obtain the position feedback signal of the diameter changing mechanism; Step 6: Determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

2. The method for controlling variable-diameter feeding of a waste shredder based on machine vision according to claim 1, characterized in that, The specific method for continuously acquiring images of the waste to be fed is as follows: an industrial camera set above the feeding conveyor channel continuously captures images of the waste to be fed. Based on the sampling frequency of the industrial camera and the conveying speed of the feeding conveyor channel, the continuously acquired multiple frames of images are processed in time synchronization, and a correspondence between each frame of image and the position in the feeding direction is established, so that each frame of image corresponds to a different feeding length position of the waste in the feeding direction.

3. The method for controlling the variable diameter feeding of a waste shredder based on machine vision according to claim 2, characterized in that, The establishment of the correspondence between each frame image and the position in the feeding direction specifically includes: calculating the displacement increment of the waste along the feeding direction between the acquisition times of two adjacent frames and the conveying speed of the feeding conveying channel, and stitching the positions of each frame image in the feeding direction according to the displacement increment to form a continuous image sequence corresponding to the actual feeding path of the waste.

4. The method for controlling the variable diameter feeding of a waste shredder based on machine vision according to claim 3, characterized in that, The specific method for extracting the outline of waste from continuous image data is as follows: each frame of the continuous image sequence is sequentially processed by grayscale conversion, noise filtering, waste region segmentation and edge detection to obtain the waste boundary in each frame of the image. Based on the positional correspondence of each frame of the image in the feeding direction, the waste boundary in each frame of the image is spliced ​​to obtain the continuous outline of the waste along the feeding direction. The noise filtering process includes smoothing filtering of each frame of the image; the waste region segmentation process includes separating the waste region from the image background; the edge detection process includes extracting the contour boundary of the waste region to reduce the impact of background interference on the waste shape contour extraction result.

5. The method for controlling the variable diameter feeding of a waste shredder based on machine vision according to claim 4, characterized in that, The step of obtaining the external dimensions of the waste at each feeding length position along the feeding direction, perpendicular to the feeding direction, specifically includes: establishing multiple cross-sectional lines perpendicular to the feeding direction on the continuous external contour at preset length intervals, calculating the intersection distance between each cross-sectional line and the waste external contour, and using the intersection distance as the external dimension at the corresponding feeding length position, arranging each external dimension according to the positional order of the feeding direction to form the external dimension sequence; After forming the external dimension sequence, the method further includes: smoothing the external dimension sequence and removing outliers. Specifically, when the difference between the external dimension at a certain feed length position and the external dimensions at at least two adjacent feed length positions exceeds a preset size threshold, the external dimension is determined to be an outlier. The outlier is then corrected using the interpolation result of the external dimensions at adjacent feed length positions to obtain a corrected external dimension sequence.

6. The method for controlling variable-diameter feeding of a waste shredder based on machine vision according to claim 5, characterized in that, The specific method for determining the target opening size of the variable diameter mechanism at the corresponding feeding displacement based on the external dimension sequence is as follows: add the preset feeding safety margin to the external dimension at each feeding length position after correction to obtain the initial target opening size corresponding to each feeding displacement. The specific method for generating the target opening adjustment curve of the variable diameter mechanism is as follows: the initial target opening sizes arranged sequentially along the feeding direction are subjected to continuous constraint processing. When the difference between the initial target opening sizes corresponding to two adjacent feeding displacements is greater than a preset change threshold, the initial target opening size corresponding to the next feeding displacement is limited and corrected according to the preset change threshold so that the generated target opening adjustment curve meets the adjustment continuity requirements of the variable diameter mechanism. After generating the target opening adjustment curve, the method further includes: sequentially associating the target opening sizes corresponding to each feeding displacement in ascending order of feeding displacement to form a displacement-opening correspondence table. When detecting the feeding displacement of waste in real time, the target opening size corresponding to the current feeding displacement is queried first based on the displacement-opening correspondence table.

7. The method for controlling variable-diameter feeding of a waste shredder based on machine vision according to claim 6, characterized in that, The real-time detection of the feed displacement of waste specifically includes: acquiring the rotation pulse signal of the feed drive roller shaft through a rotary encoder set on the feed drive roller shaft; calculating the rotation angular displacement of the feed drive roller shaft based on the obtained rotation pulse signal; converting the rotation angular displacement into the real-time feed displacement of waste by combining the roller diameter parameter of the feed drive roller shaft; and mapping the obtained real-time feed displacement to the corresponding displacement position in the target opening adjustment curve to determine the target opening size corresponding to the current feed displacement.

8. The method for controlling variable-diameter feeding of a waste shredder based on machine vision according to claim 7, characterized in that, The specific method for determining the actual opening size of the variable diameter mechanism based on the position feedback signal is as follows: obtain the current position feedback value of the actuator of the variable diameter mechanism, and convert the current position feedback value into the current actual opening size according to the pre-established conversion relationship between the mechanism position and the feed inlet opening size.

9. A machine vision-based variable diameter feeding control method for a waste shredder according to claim 8, characterized in that, The step of correcting the opening adjustment of the variable diameter mechanism based on the comparison result specifically includes: comparing the current actual opening size with the target opening size corresponding to the current feed displacement to obtain the opening deviation; when the obtained opening deviation is greater than a preset positive deviation threshold, controlling the variable diameter mechanism to perform a correction action of reducing the opening; when the opening deviation is less than a preset negative deviation threshold, controlling the variable diameter mechanism to perform a correction action of increasing the opening. After each correction action is performed, the position feedback signal is reacquired, and the actual opening size of the variable diameter mechanism is re-determined. Based on the deviation between the re-determined actual opening size and the current target opening size, the control command for the next correction action is updated to form a closed-loop opening adjustment process. Repeatedly execute position feedback acquisition, actual opening size determination, deviation comparison and correction actions until the deviation between the actual opening size and the target opening size falls within the preset allowable deviation range, or the real-time feeding displacement of the waste reaches the end position corresponding to the target opening adjustment curve.

10. A machine vision-based variable diameter feeding control system for a waste shredder, characterized in that, The machine vision-based variable diameter feeding control system for the waste shredder is used to execute the machine vision-based variable diameter feeding control method for the waste shredder according to any one of claims 1-9, including: Image acquisition module: used to continuously acquire images of the waste to be fed into the crusher before the waste enters the crusher, and obtain continuous image data of the waste along the feeding direction; Dimension extraction module: used to extract the outline of the waste based on the continuous image data, and obtain the dimensions of the waste perpendicular to the feeding direction at each feeding length position along the feeding direction, forming a dimension sequence; Adjustment curve generation module: used to determine the target opening size of the variable diameter mechanism under the corresponding feed displacement according to the external dimension sequence, and generate the target opening adjustment curve of the variable diameter mechanism; Displacement mapping matching module: used to detect the feeding displacement of waste in real time during the process of waste entering the crusher, and map the feeding displacement to the target opening adjustment curve to determine the corresponding target opening size; Variable diameter adjustment feedback module: used to control the variable diameter mechanism to adjust the feed inlet opening according to the target opening size, and to obtain the position feedback signal of the variable diameter mechanism; Deviation closed-loop correction module: used to determine the actual opening size of the variable diameter mechanism based on the position feedback signal, compare the deviation between the actual opening size and the target opening size, and correct the opening adjustment of the variable diameter mechanism based on the comparison result.

Images