A method and system for controlling the graded crushing of construction waste

Abstract

This invention belongs to the field of control, specifically a method and system for graded crushing control of construction waste. It includes achieving refined crushing through multimodal sensing and collaborative control, acquiring multi-view texture images of the primary crushing chamber inlet; constructing a gray-level-gradient co-occurrence matrix and analyzing singular value entropy to represent the hardness dispersion of waste components, thereby adjusting the tilt angle of the moving jaw plate; extracting acoustic emission signals from each chamber, reconstructing the energy spectrum using wavelet packet transform, and calculating the KL divergence with the empty energy spectrum as a real-time load index; if the fractal dimension of the final stage aggregate projection profile exceeds the shaping threshold, extracting the secondary chamber load index to construct an attenuation function, generating a frequency correction factor to adjust the eccentric shaft speed, and combining the hardness and load weighted results to collaboratively control the inter-stage screening throughput, thus restoring the fractal dimension. This invention solves the problem of maintaining a balance between crushing throughput and energy consumption in the processing system, enabling intelligent graded crushing of construction waste.

Description

Technical Field

[0001] This invention belongs to the field of control, and in particular relates to a method and system for controlling the graded crushing of construction waste. Background Technology

[0002] Construction waste is typically characterized by its complex composition and wide range of sources, including a mixture of materials such as concrete, bricks, ceramics, glass, and slag. These materials exhibit differences and variations in compressive strength, hardness, and crushing characteristics. Construction waste processing relies on the coordinated operation of multi-stage crushing and screening stations. This involves using coarse, medium, and fine crushing equipment, as well as inter-stage screening devices, to process large pieces of construction waste into recycled aggregates that meet engineering standards. For example, patent CN216826392U discloses a multi-stage crushing device for construction waste processing, demonstrating a scheme for continuous crushing and screening of construction waste using a multi-stage mechanism. Traditional control methods are adequate for processing mineral rocks with a single composition, but when dealing with construction waste with highly variable composition and uneven hardness distribution, they often suffer from poor adaptability, low energy efficiency, and severe equipment wear, making it difficult to achieve precise crushing operations. Furthermore, the primary crushing stage cannot perceive the texture and hardness characteristics of the feed material in real time, nor can it adjust key geometric parameters such as the jaw plate inclination angle according to the material's hardness dispersion. This leads to problems such as insufficient crushing force causing jamming or over-crushing and generating dust when processing high-hardness concrete or low-hardness bricks. Simultaneously, monitoring the load state inside the crushing chamber using only electrical signals is insufficient to accurately detect the instantaneous energy release characteristics during the crushing process, making it impossible to accurately distinguish between no-load and overload states. The particle shape of the recycled aggregate discharged from the final crushing stage is a crucial indicator determining its application value. Due to the lack of a multi-source information-coordinated control mechanism in each of the aforementioned crushing stages, the entire processing system struggles to maintain optimal crushing throughput and energy consumption balance while ensuring high-quality aggregate particle shape. Therefore, an intelligent, staged crushing method is urgently needed. Summary of the Invention

[0003] To address the problem that existing technologies cannot achieve multi-source information collaborative control, making it difficult to maintain optimal crushing throughput and energy consumption balance while ensuring high-quality aggregate particle shape.

[0004] In the first aspect, the present invention proposes a method for controlling the graded crushing of construction waste, comprising the following steps: Multi-view texture images of construction waste at the inlet of the primary crushing chamber were acquired, and acoustic emission signals inside each crushing chamber were acquired simultaneously. A gray-level gradient co-occurrence matrix is ​​constructed based on multi-view texture images. Singular value entropy is analyzed to represent the hardness dispersion of waste components. The total energy of acoustic emission signals is extracted as an absolute hardness characterization value in combination with acoustic emission signals. Based on this, the bite angle of the primary crushing chamber is adjusted through a hydraulic system. Wavelet packet transform is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum. The KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum is calculated and the total energy growth coefficient is fused, which is used as a real-time load state index of each cavity. The projected profile of the discharged aggregate in the final crushing chamber is monitored in real time, and the boundary fractal dimension is calculated. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the deviation direction of the fractal dimension to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the rotational speed control of the eccentric shaft of the secondary crushing chamber, and the interstage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

[0005] In another aspect, the present invention also proposes a graded crushing control system for construction waste, comprising the following modules: The acquisition module is used to acquire multi-view texture images of construction waste at the inlet of the primary crushing chamber, and simultaneously acquire acoustic emission signals inside each stage of the crushing chamber. The adjustment module is used to construct a gray-level-gradient co-occurrence matrix based on multi-view texture images, analyze singular value entropy to represent the hardness dispersion of waste components, and extract the total energy of acoustic emission signals as an absolute hardness characterization value by combining acoustic emission signals. Based on this, the primary crushing chamber bite angle is adjusted through the hydraulic system. The calculation module is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum using wavelet packet transform, calculate the KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum and fuse the total energy growth coefficient, and use it as a real-time load status indicator for each level of cavity. The generation module is used to monitor the projected contour of the aggregate discharged from the final crushing chamber in real time and calculate the boundary fractal dimension. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the deviation direction of the fractal dimension to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the rotational speed control of the eccentric shaft of the secondary crushing chamber, and the inter-stage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

[0006] This invention integrates multi-view visual perception and acoustic emission signal analysis technologies. It uses the hardness dispersion characterized by singular value entropy in conjunction with the total acoustic emission energy to confirm absolute hardness, thereby dynamically reducing the primary cavity engagement angle and completely eliminating slippage and stalling caused by complex and extremely hard materials. Simultaneously, it innovatively introduces the energy spectrum KL divergence, which incorporates the total energy growth coefficient, as a load indicator, effectively overcoming the misjudgment of no-load caused by the proportional increase of energy in each frequency band under high load. Furthermore, this invention constructs a closed-loop feedback mechanism based on the fractal dimension of the discharged aggregate, using a bidirectional piecewise adjustment function to smoothly intervene in the secondary cavity speed to eliminate the particle shape deterioration problem caused by unidirectional adjustment. It also combines hardness and multi-level load indicators to collaboratively control the inter-stage feed throughput, replacing traditional overload shutdown with dynamic risk avoidance using second-level flow limiting. This reduces the content of needle-like and flaky particles in the finished product while achieving a significant increase in total throughput and a substantial decrease in average power consumption per ton under continuous and stable operation. Attached Figure Description

[0007] Figure 1 A flowchart of the first embodiment; Figure 2 A schematic diagram comparing the normalized energy spectra of acoustic emissions in different frequency bands; Figure 3 This is a schematic diagram of the primary crushing chamber bite angle based on hardness dispersion. Detailed Implementation

[0008] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0009] In the first embodiment, the present invention proposes a method for controlling the graded crushing of construction waste, such as... Figure 1 This includes the following steps: S1, acquire multi-view texture images of construction waste at the entrance of the primary crushing chamber, and simultaneously acquire acoustic emission signals inside each crushing chamber. Specifically, three CCD cameras are arranged at a 120° angle above and to the side of the feed hopper in the primary crushing chamber, and high-frequency strobe LED light sources are configured for supplementary lighting. The three cameras are simultaneously exposed by a trigger to acquire RGB images of the surface of the construction waste, which are then converted into single-channel grayscale images by an image processor. At the same time, wideband piezoelectric acoustic emission sensors are installed on the back of the fixed jaw plates of the primary, secondary and final crushing chambers and on the side wall of the frame using a coupling agent. These sensors are connected to a multi-channel high-speed data acquisition card through a preamplifier. A unified master clock signal source is set to align the image acquisition trigger time with the acoustic emission signal recording start time, ensuring that the image frame data and the acoustic emission waveform stream data have the same time reference.

[0010] S2, based on multi-view texture images, constructs a gray-level-gradient co-occurrence matrix, analyzes singular value entropy to represent the hardness dispersion of waste components, and extracts absolute hardness characterization values ​​by combining acoustic emission signals, thereby adjusting the primary crushing chamber bite angle through a hydraulic system.

[0011] The Sobel operator is used to calculate the gradient magnitudes of the acquired grayscale images in the horizontal and vertical directions to obtain gradient images. The pixel value ranges of the grayscale images and gradient images are normalized, and the joint probability distribution of the grayscale values ​​and gradient values ​​corresponding to pixels in the image within a preset neighborhood is statistically analyzed to generate a grayscale gradient co-occurrence matrix reflecting texture roughness and edge information. Singular value decomposition is performed on this matrix to obtain a sequence of singular values ​​in the diagonal matrix. The entropy value of this singular value sequence is calculated using the Shannon information entropy formula, and the calculated singular value entropy is used as an indicator of hardness dispersion. When the singular value entropy is greater than a preset benchmark value, and the total energy of the acoustic emission signal from the primary crushing chamber extracted synchronously is higher than a preset hardness confirmation threshold, the PLC controller outputs a control signal to drive the hydraulic pump station to inject hydraulic oil into the hydraulic cylinder behind the moving jaw of the primary crushing chamber. This controls the hydraulic cylinder to extend and push the lower part of the moving jaw plate forward towards the fixed jaw plate, thereby reducing the bite angle and increasing the extrusion force.

[0012] Hardness dispersion, also known as singular value entropy, indicates whether a material's composition is homogeneous. High hardness dispersion suggests a complex composition, while high total acoustic emission energy confirms the presence of extremely hard components within the complex material. High dispersion alone, such as soft soil mixed with ordinary waste bricks, won't cause slippage. However, given the confirmed presence of high-hardness components, reducing the bite angle to increase the crushing stroke and extrusion pressure can mitigate uneven stress caused by the mixture of soft and hard materials. This ensures that the high-hardness components, such as high-grade concrete blocks and waste granite, are forcibly crushed, preventing slippage between the jaw plates or jamming due to sudden changes in localized stress, thus maintaining the continuity and stability of the crushing process. Furthermore, even when high-hardness materials are mixed with low-hardness materials, increasing the crushing stroke and extrusion pressure will not damage the equipment. Conversely, when low-hardness materials are mixed with high-hardness materials, this invention can actually improve the crushing effect by increasing the crushing stroke and extrusion pressure.

[0013] In another embodiment, when the singular entropy is not greater than the benchmark value, it indicates that the material composition is uniform, such as the batch of material being pure bricks, pure concrete blocks, or pure stones. Since the singular entropy only reflects uniformity and cannot directly represent absolute hardness, the system switches to a customized operation mode, adjusting the crusher parameters according to the user's pre-set empirical values ​​based on the type of incoming material. For example, when the operator confirms that the batch of materials consists entirely of low-hardness waste bricks, the first set of control data is invoked for control. For instance, the hydraulic cylinder is extended to push the lower part of the moving jaw plate forward, increasing the bite angle to 22° to 25°. Simultaneously, the main motor is adjusted to set the eccentric shaft speed to 280 r / min to maximize throughput and screening efficiency. Conversely, if the batch of materials consists entirely of high-hardness bluestone or high-grade waste precast parts, the second set of control data is invoked for control. For instance, the hydraulic cylinder is retracted to reduce the bite angle to 16° to 18°. Simultaneously, the eccentric shaft speed is reduced to 220 r / min, and the overflow protection pressure threshold of the hydraulic system is increased to ensure material crushing without damaging the equipment housing in a low-speed, high-stroke mode. In an optional embodiment, the construction of a gray-level-gradient co-occurrence matrix based on multi-view texture images and the analysis of singular value entropy to represent the hardness dispersion of waste components includes: Singular value decomposition is performed on the constructed gray-level-gradient co-occurrence matrix, and the diagonal elements of the singular value diagonal matrix are extracted to form feature vectors; The elements in the feature vector are normalized so that the sum of all elements is 1, thus obtaining the probability distribution vector; Calculate the product of the natural logarithm of each element in the probability distribution vector and the element itself, sum all the products and take the opposite number to obtain the singular value entropy; A mapping model between singular value entropy and material hardness dispersion is established, and the calculated singular value entropy is mapped to the hardness dispersion of waste components.

[0014] The acquired multi-view texture images are converted into 256-level grayscale images, and a grayscale-gradient co-occurrence matrix with a size of 256×256 is constructed. Performing singular value decomposition on the matrix yields a diagonal matrix. Extract the singular value sequence from the diagonal of the matrix. Normalization is performed, and the probability distribution is calculated. Ensure the sum is 1. Calculate the singular value entropy using Shannon's information entropy formula. .

[0015] The entropy value E represents the complexity and disorder of the image texture. Generally, the more complex the texture of the concrete, brick, and steel reinforcement mixture in construction waste, the higher the entropy value, and the greater the hardness distribution dispersion. Increased hardness dispersion usually indicates the presence of differentiated components in the material. A preset hardness mapping range is used, for example, linearly or automatically mapping the calculated entropy value range [0.5, 4.5] to the hardness dispersion index. For example, when E=3.8, the mapping yields... This indicates that the material composition is extremely complex and has large differences in hardness, requiring appropriate crushing strategies; conversely, when E=1.2, it is mapped to... This indicates that the material has a single composition and uniform hardness.

[0016] In an optional embodiment, adjusting the primary crushing chamber engagement angle via a hydraulic system accordingly includes: Calculate the difference between the real-time hardness dispersion and the preset reference value, and use the difference to calculate the required angle compensation amount through a PID controller; If the hardness dispersion is higher than the reference value, and the absolute total energy of the synchronously acquired acoustic emission signal exceeds the preset hardness verification threshold, it is determined that there is a high-hardness foreign object, and a control command to reduce the bite angle is generated.

[0017] A fully hydraulic jaw crusher is used, with the toggle plate replaced or driven by a hydraulic cylinder. A hardness dispersion benchmark value is set. Real-time calculation of deviation Input the deviation into the PID controller and set the proportional coefficient. =1.5, integral coefficient =0.05, differential coefficient =0.2, calculate the hydraulic cylinder stroke correction amount. ,pass And the preset geometric relationships of the crusher, such as the position parameters of the moving jaw fulcrum and the hydraulic cylinder hinge point, are converted into corresponding angle compensation amounts. This allows for adjustment of the bite angle.

[0018] When high hardness dispersion is detected and the presence of high-hardness components is confirmed, it is determined that there are materials that are difficult to break or prone to slippage. Based on the user-defined control data, the PID outputs a reverse control quantity to drive the hydraulic servo valve to extend the cylinder. The extension of the cylinder causes the lower end of the moving jaw to move closer to the fixed jaw. According to geometric relationships, the moving jaw plate tends to be upright, and the bite angle decreases from the default 21° to 19° or even 18°. Figure 3 As shown. This action not only increases the compressive force on the material but also reduces the engagement angle to meet the friction angle requirements of high-hardness materials, preventing the material from jumping upwards. Optionally, when... When the hydraulic cylinder retracts, the discharge port and engagement angle are increased to 23°, and the material is processed using a large discharge port and a small stroke to maximize throughput; or, when the hardness dispersion is not greater than the reference value, the user-defined control data is used.

[0019] S3. Wavelet packet transform is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum. The KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum is calculated and the total energy growth coefficient is fused, which is used as the real-time load status index of each cavity. The Daubechies wavelet basis function was used to perform three-level wavelet packet decomposition on the digitized acoustic emission analog signals acquired from each channel, dividing the signal frequency band into eight independent sub-bands. The sum of squares of the wavelet packet decomposition coefficients in each sub-band was calculated as the energy value of that band. Figure 2 As shown, the total energy in real time is obtained by summing the energy of the eight sub-bands. The energy values ​​of all sub-bands are then normalized by dividing the total energy to obtain the real-time crushing energy spectrum vector reflecting the probability of the current frequency energy distribution. Beforehand, under the same signal processing procedure, the standard no-load energy spectrum vector and the standard no-load total energy are obtained and stored in the database. The logarithmic weighted sum of the frequency band probability ratios of the real-time energy spectrum vector and the standard no-load energy spectrum vector is calculated item by item using the relative entropy calculation formula. The calculated KL divergence value is multiplied by the growth coefficient of the real-time total energy relative to the standard no-load total energy. This fused calculation result is defined as a real-time load state index representing the current crushing load level of the crushing chamber.

[0020] In an optional embodiment, the step of extracting acoustic emission signal features and reconstructing the fragmented energy spectrum using wavelet packet transform includes: The Db4 wavelet basis function was selected to perform three-level wavelet packet decomposition on the acoustic emission signal to obtain the sub-signal reconstruction coefficients of eight independent frequency bands. The sum of squares of the reconstruction coefficients of each sub-signal is calculated as the energy of each frequency band, and the total energy is calculated and normalized to obtain the normalized energy vector. The elements in the normalized energy vector are sorted from low to high frequency to form a fragmented energy spectrum representing the operating state of the cavity.

[0021] Considering the harsh environment inside the crushing chamber, the sensor is not placed directly inside the chamber. Instead, a waveguide rod passes through the casing and contacts the back of the wear-resistant liner, or it is magnetically fixed to the outer shell of the moving jaw bearing housing to collect high-frequency stress waves transmitted through the structure. The sampling rate is set to 200kHz to 500kHz, and a 20kHz high-pass analog filter is set to filter out the crusher's own vibration and low-frequency motor noise.

[0022] Daubechies4 was selected as the wavelet basis function, and the signal was decomposed into 3-level wavelet packet decomposition, dividing the frequency domain into 8 equal sub-bands. The wavelet coefficients of these 8 nodes were extracted, and the sum of the squares of the coefficients at each node was calculated as the absolute energy of that frequency band. The total energy was then calculated. And obtain the normalized feature vector. The energy spectrum separates the fracture sound emission from the mechanical operating background sound.

[0023] In an optional embodiment, the calculation of the KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum, and the fusion of the total energy growth coefficient as a real-time load state index for each stage of the cavity, includes: Read the pre-stored standard open-load energy spectrum Q and standard open-load total energy. The energy spectrum is the normalized energy vector of the crusher when it is idling without feeding. Obtain the currently calculated real-time fragmentation energy spectrum P and real-time total energy. ; Using formula Calculate the KL divergence, where i represents the frequency band number; Calculate the energy growth coefficient ; Calculate the fusion load divergence ; The calculated fusion load divergence is processed by moving average filtering, and the smoothed value is output as a real-time load status indicator.

[0024] During the preheating phase after equipment startup, acoustic emission data were continuously collected for 5 minutes. The average normalized energy vector was calculated as the standard no-load energy spectrum Q, and the average no-load total energy was recorded. During the operation, the energy spectrum P and total energy within the current time window are calculated in real time. The information entropy difference between P and Q can be expressed using the KL divergence formula: .

[0025] If the equipment is unloaded, P≈Q and The fusion divergence approaches 0; as the crushing load increases, the proportion of high-frequency energy generated by ore crushing increases, the spectral structure shifts sharply relative to the unloaded spectrum, and the absolute total energy amplitude increases. Significantly increased, at this time The value increases rapidly. An energy growth coefficient is introduced. This avoids the misjudgment of "no-load" load caused by the KL divergence remaining unchanged when the acoustic emission energy of each frequency band increases proportionally and rapidly. To eliminate the interference of instantaneous impulse noise, a moving average filter with a length of N=50 is used. The value is smoothed, and the output value L is used as a real-time load indicator. Compared with the traditional current detection method, this indicator is more sensitive to the instantaneous crushing state and can detect the oversaturation crushing trend earlier.

[0026] S4. Real-time monitoring of the projected profile of the discharged aggregate in the final crushing chamber and calculation of the boundary fractal dimension. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the dimension deviation direction (too high or too low) to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the eccentric shaft speed control of the secondary crushing chamber, and the inter-stage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

[0027] Specifically, a high-resolution linear array camera is installed directly above the discharge conveyor belt of the final crushing chamber to continuously acquire images of the discharged aggregate flow. The images are binarized using the OTSU method and the edge contours of the aggregates are extracted using the Canny operator. The box-counting dimension algorithm is used to calculate the slope of the number of contour-covered grids as a function of grid size in a double logarithmic coordinate system as the boundary fractal dimension. The threshold range that meets the building aggregate shaping standard is set to 1.1 to 1.2. Once the real-time calculated fractal dimension value is not within the specified range, the current load status index L of the secondary crushing chamber is automatically read. Depending on whether the fractal dimension is greater than 1.2 or less than 1.1, L is substituted into the preset piecewise adjustment function to calculate the corresponding frequency correction factor, which is less than or greater than 1. This factor is sent to the inverter of the secondary crushing chamber motor via the Modbus communication protocol to dynamically raise and lower the output frequency, thereby bidirectionally adjusting the eccentric shaft speed. At the same time, the singular entropy value obtained in the previous steps is linearly weighted and summed with the real-time load status index of each stage chamber according to the preset weight coefficient. Based on the summation result, a 0 to 20mA analog current signal is output to control the inverter frequency of the interstage screening feeder. By changing the frequency, the material throughput into the final stage chamber is adjusted. The above adjustment action is repeated cyclically with a period of 500ms until the monitored fractal dimension stably falls within the preset shaping threshold range.

[0028] In an optional embodiment, the real-time monitoring of the projected profile of the aggregate discharged from the final crushing chamber and the calculation of the boundary fractal dimension include: Binarize and edge-detect the aggregate projection image to extract single-pixel closed contours; Using the box-counting dimension method, with a side length of The grid covers the outline, and the number of grids containing the outline is counted. ; Change the grid side length Get multiple sets Data points; The data points are linearly fitted using the least squares method, and the slope of the fitted line is extracted as the boundary fractal dimension. This dimension is used to represent the roughness and angularity of the aggregate shape.

[0029] Aggregate images were acquired using a camera, binarized using the Otsu algorithm, and the Canny operator was applied to extract aggregate edges. A series of progressively increasing grid sizes were selected using the box-counting dimension method. Count the number of grids required to cover the aggregate outline. .

[0030] Construct a coordinate system and draw right The scatter plot is generated, and the line y = D × x + b is fitted using the least squares method. The slope D of this line is the fractal dimension. Ideally, shaped aggregates have a lower D value, while needle-like, flaky, or multi-faceted aggregates have a higher D value. The shaping threshold range is set to [1.1, 1.2]. If the calculated D value exceeds this range, the feedback adjustment mechanism is triggered.

[0031] In an optional embodiment, the step of extracting the real-time load state index of the secondary crushing chamber to construct a piecewise adjustment function to generate a frequency correction factor, and applying the frequency correction factor to the eccentric shaft speed control of the secondary crushing chamber, includes: When the fractal dimension is greater than the upper limit of the interval (1.2), the decay function is invoked: At this point, the frequency correction factor takes values ​​in the range of (0,1); When the fractal dimension is less than the lower limit of the interval (1.1), call the gain function: ,in The acceleration factor is the acceleration coefficient, and the frequency correction factor ranges from 1 to 2. ; Where L is the real-time load status index of the secondary crushing chamber. The preset overload threshold, This is the sensitivity coefficient; Calculate the corresponding function as the frequency correction factor; Multiply the rated eccentric shaft speed of the secondary crushing chamber by the frequency correction factor to obtain the corrected target speed.

[0032] To avoid control oscillations caused by relying solely on threshold switching, a bidirectional smoothing adjustment logic targeting the direction of deviation is employed. An overload threshold is set. Sensitivity coefficient acceleration coefficient When the real-time load index L is much smaller than Furthermore, when the dimension is normal, the correction factor f(L)≈1, and the crusher maintains its rated speed.

[0033] When the fractal dimension exceeds the upper limit, it indicates that the aggregate has too many needle-like and flaky edges, and the load is close to or exceeds the threshold. The exponential term increases rapidly, causing the decay function f(L) to decrease. For example, if f(2.5)≈0.27 is calculated, the target rotation speed is adjusted to 81 rpm. Reducing the rotation speed increases the material filling rate within the chamber, enhancing the lamination crushing effect and improving the mutual grinding and shaping effect between materials. This improves the output particle shape and reduces the fractal dimension while reducing the load. Conversely, when the fractal dimension is detected to be below 1.1, it indicates that the aggregate surface is excessively rounded and the crushing rate is too high. A correction factor of 1.15 is generated using the gain function, increasing the rotation speed above the rated value. Increasing the rotation speed increases the frequency of single impacts and reduces the material filling rate, making the crushing mode biased towards impact and cleavage. This reduces the residence and grinding time of the material within the chamber, thereby increasing the retention rate of the aggregate's edges and corners, bringing the fractal dimension back to the normal range, and completely solving the problem of error deterioration caused by unidirectional adjustment.

[0034] In an optional embodiment, the weighted result of combining hardness dispersion with the real-time load status index of each stage cavity to coordinately adjust the interstage material screening throughput includes: Constructing a flux regulation model: ; in, For the target screening throughput, To maximize the planned throughput, This represents the normalized hardness dispersion. This is a normalized comprehensive real-time load status index for each stage of the crushing chamber. and These are the weighting coefficients; The real-time load status indicators of each crushing chamber are obtained by weighting and summing them according to the influence weight of each crushing chamber in the material conveying path, and then normalizing them to the [0,1] interval. according to Adjust the vibration frequency of the interstage screening feeder. When the hardness or the load of each stage cavity increases, reduce the throughput to balance the load of the crushing chamber.

[0035] Set maximum flux Weight Focusing on the influence of raw material hardness, Focusing on the impact of load on each cavity level, and taking , , To highlight the role of the secondary fracture chamber in the shaping effect , , Weights are assigned to the influence of each cavity level.

[0036] Assuming the current raw material composition is extremely complex and contains materials with extremely high hardness, meaning both the singular entropy and the absolute energy of acoustic emission are at a high level, the corresponding hardness dispersion mapping value... , , , ,in, , , Calculate the primary, secondary, and final crushing loads respectively. The load is the weighted summation of the loads from each crushing stage. To enable calculations with hardness dispersion and prevent the influence of different dimensions, [the following is omitted]. After normalization Then the weighted sum Target flux The target throughput is the weight of material processed per hour. Based on this target value, the frequency of the inverter for the interstage vibrating feeder will be reduced from 50Hz to approximately 8Hz. The throughput limit allows the secondary crushing system sufficient time for digestion and shaping, preventing stalling accidents caused by the accumulation of high-hardness, high-load materials. The high throughput will automatically resume once the target value decreases.

[0037] A total of 1000 tons of mixed demolition waste was selected as raw material and divided into two groups. The raw material components included C40 reinforced concrete blocks, clay bricks, and ceramic fragments with significantly different strengths. The first group used a traditional fixed-parameter control mode, setting the jaw crusher's bite angle to a constant 21°, the secondary crusher's speed to a fixed 300 r / min, and the interstage screening feed frequency to a fixed 45 Hz. The second group utilized the control system described in this invention, integrating hardness dispersion sensing based on singular value entropy and total acoustic emission energy, load monitoring based on the energy ratio fused from acoustic emission KL divergence, and bidirectional speed shaping adjustment based on the aggregate fractal dimension.

[0038] When processing densely fed sections of high-hardness concrete, the second vision system calculated a singular entropy of 3.9 in the gray-level gradient co-occurrence matrix. Simultaneously, the absolute energy of the high-frequency acoustic emission channel surged, exceeding the preset hardness confirmation threshold. The system determined that the hardness dispersion was extremely high and that the material contained extremely hard material. The PID controller then drove the hydraulic cylinder to retract, automatically reducing the engagement angle to 18.5°. At the same time, the vision system detected that the fractal dimension of the final-stage discharged aggregate exceeded the standard by 1.25. Furthermore, when the acoustic emission monitoring system detected an overload threshold exceeding 2.0 for the fusion dispersion smoothing value, it generated a frequency correction factor based on a preset piecewise nonlinear adjustment function, temporarily adjusting the secondary crusher speed to 120 r / min and reducing the feed throughput to 55% of the maximum planned throughput. Subsequently, when localized over-grinding was detected, causing the fractal dimension to drop to 1.08, the system automatically invoked a gain function to slightly increase the speed to suppress over-grinding. In contrast, under the same high-hardness conditions, the first group experienced repeated slippage and jumping of the material due to an excessively large bite angle, resulting in the current peak exceeding 120% of the rated current three times and causing two machine stalls. Discharge testing showed that the average boundary fractal dimension of the aggregate produced by the first group was 1.27, while the second group, by optimizing the rotation speed to enhance the lamination effect and dynamic impact cleavage feedback, controlled the average fractal dimension to 1.12.

[0039] The first group had an average hourly output of 112 tons, with a finished aggregate containing 21.5% needle-like and flaky particles, and an average power consumption of 3.3 kWh per ton. The second group increased its average hourly output to 148 tons, reduced the finished aggregate containing needle-like and flaky particles to 7.8%, and decreased its average power consumption per ton to 2.5 kWh, while achieving zero downtime throughout the entire process. Although the second group's system would reduce the feed rate when encountering extremely hard materials, for example, to 32 tons / hour, this was only a dynamic avoidance action on a second or minute level. Compared to the first group's downtime of several hours due to engine stalling, the second group's continuous and stable operation accumulated a higher total throughput over a macroscopic time axis. The 32.1% increase in output of the second group was due to the establishment of a mapping model between singular value entropy and material hardness dispersion, the introduction of absolute hardness auxiliary determination based on the total acoustic emission energy, and the maintenance of continuous operation of the crushing chamber by reducing the bite angle in real time. The improvement in finished product quality and energy efficiency is attributed to the synergistic feedback mechanism of acoustic emission fusion energy spectrum and fractal dimension. By accurately adjusting the rotation speed in both directions through a piecewise adjustment function, the equipment is prevented from operating in the inefficient zone, ensuring the grinding and shaping effect under all working conditions.

[0040] In a second embodiment, the present invention also provides a construction waste grading and crushing control system, comprising the following modules: The acquisition module is used to acquire multi-view texture images of construction waste at the inlet of the primary crushing chamber, and simultaneously acquire acoustic emission signals inside each stage of the crushing chamber. The adjustment module is used to construct a gray-level-gradient co-occurrence matrix based on multi-view texture images, analyze singular value entropy to represent the hardness dispersion of waste components, and extract the total energy of acoustic emission signals as an absolute hardness characterization value by combining acoustic emission signals. Based on this, the primary crushing chamber bite angle is adjusted through the hydraulic system. The calculation module is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum using wavelet packet transform, calculate the KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum and fuse the total energy growth coefficient, and use it as a real-time load status indicator for each level of cavity. The generation module is used to monitor the projected contour of the aggregate discharged from the final crushing chamber in real time and calculate the boundary fractal dimension. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the deviation direction of the fractal dimension to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the rotational speed control of the eccentric shaft of the secondary crushing chamber, and the inter-stage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

[0041] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0042] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for controlling the graded crushing of construction waste, characterized in that, Includes the following steps: Multi-view texture images of construction waste at the inlet of the primary crushing chamber were acquired, and acoustic emission signals inside each crushing chamber were acquired simultaneously. A gray-level gradient co-occurrence matrix is ​​constructed based on multi-view texture images. Singular value entropy is analyzed to represent the hardness dispersion of waste components. The total energy of acoustic emission signals is extracted as an absolute hardness characterization value in combination with acoustic emission signals. Based on this, the bite angle of the primary crushing chamber is adjusted through a hydraulic system. Wavelet packet transform is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum. The KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum is calculated and the total energy growth coefficient is fused, which is used as a real-time load state index of each cavity. The projected profile of the discharged aggregate in the final crushing chamber is monitored in real time, and the boundary fractal dimension is calculated. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the deviation direction of the fractal dimension to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the rotational speed control of the eccentric shaft of the secondary crushing chamber, and the interstage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

2. The method according to claim 1, characterized in that, The construction of a gray-level-gradient co-occurrence matrix based on multi-view texture images, and the analysis of singular value entropy to represent the hardness dispersion of waste components, includes: Singular value decomposition is performed on the constructed gray-level-gradient co-occurrence matrix, and the diagonal elements of the singular value diagonal matrix are extracted to form feature vectors; The elements in the feature vector are normalized so that the sum of all elements is 1, thus obtaining the probability distribution vector; Calculate the product of the natural logarithm of each element in the probability distribution vector and the element itself, sum all the products and take the opposite number to obtain the singular value entropy; A mapping model between singular value entropy and material hardness dispersion is established, and the calculated singular value entropy is mapped to the hardness dispersion of waste components.

3. The method according to claim 1, characterized in that, The adjustment of the primary crushing chamber engagement angle via the hydraulic system includes: Calculate the difference between the real-time hardness dispersion and the preset reference value, and use the difference to calculate the required angle compensation amount through a PID controller; If the hardness dispersion is higher than the reference value, and the absolute total energy of the synchronously acquired acoustic emission signal exceeds the preset hardness verification threshold, then a control command to reduce the bite angle is generated.

4. The method according to claim 1, characterized in that, The step of extracting acoustic emission signal features and reconstructing the fragmented energy spectrum using wavelet packet transform includes: The Db4 wavelet basis function was selected to perform three-level wavelet packet decomposition on the acoustic emission signal to obtain the sub-signal reconstruction coefficients of eight independent frequency bands. The sum of squares of the reconstruction coefficients of each sub-signal is calculated as the energy of each frequency band. The energy of each frequency band is accumulated to obtain the real-time total energy. The real-time total energy is then used to normalize the energy of each frequency band to obtain a normalized energy vector. The elements in the normalized energy vector are sorted from low to high frequency to form a fragmented energy spectrum representing the operating state of the cavity.

5. The method according to claim 1, characterized in that, The calculation of the KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum, and the fusion of the total energy growth coefficient, serve as a real-time load status indicator for each stage of the cavity, including: Read the pre-stored standard no-load energy spectrum Q and standard no-load total energy, wherein the energy spectrum is the normalized energy vector of the crusher in the no-feed idling state; Obtain the real-time fragmentation energy spectrum P and the real-time total energy obtained from the current calculation; Using formula Calculate the KL divergence, where i represents the frequency band number; The ratio of the real-time total energy to the standard no-load total energy is calculated as the total energy growth coefficient; Multiply the calculated KL divergence by the total energy growth coefficient to obtain the fusion load divergence; The fused load divergence is processed by moving average filtering, and the smoothed value is output as a real-time load status indicator.

6. The method according to claim 1, characterized in that, The real-time monitoring of the projected profile of the discharged aggregate in the final crushing chamber and the calculation of the boundary fractal dimension include: Binarize and edge-detect the aggregate projection image to extract single-pixel closed contours; Using the box-counting dimension method, with a side length of The grid covers the outline, and the number of grids containing the outline is counted. ; Change the grid side length Get multiple sets Data points; The least squares method is used to perform linear fitting on the data points, and the slope of the fitted line is extracted as the boundary fractal dimension.

7. The method according to claim 1, characterized in that, The step of extracting real-time load state indicators of the secondary crushing chamber to construct a piecewise adjustment function to generate a frequency correction factor, and applying the frequency correction factor to the speed control of the eccentric shaft of the secondary crushing chamber, includes: When the fractal dimension is greater than the upper limit of the preset shaping threshold range, the attenuation function is called to calculate the frequency correction factor, and the frequency correction factor takes a value range of (0,1). When the fractal dimension is less than the lower limit of the preset shaping threshold range, the gain function is called to calculate the frequency correction factor, and the frequency correction factor is greater than 1. Multiply the rated eccentric shaft speed of the secondary crushing chamber by the frequency correction factor to obtain the corrected target speed.

8. The method according to claim 1, characterized in that, The weighted result of combining hardness dispersion and real-time load status indicators of each stage cavity to coordinately adjust the interstage material screening throughput includes: Constructing a flux regulation model: ; in, For the target screening throughput, To maximize the planned throughput, This represents the normalized hardness dispersion. This is a normalized comprehensive real-time load status index for each stage of the crushing chamber. and These are the weighting coefficients; The real-time load status indicators of each crushing chamber are obtained by weighting and summing them according to the influence weight of each crushing chamber in the material conveying path, and then normalizing them to the [0,1] interval. according to Adjust the vibration frequency of the interstage screening feeder.

9. A construction waste grading and crushing control system, characterized in that, Includes the following modules: The acquisition module is used to acquire multi-view texture images of construction waste at the inlet of the primary crushing chamber, and simultaneously acquire acoustic emission signals inside each stage of the crushing chamber. The adjustment module is used to construct a gray-level-gradient co-occurrence matrix based on multi-view texture images, analyze singular value entropy to represent the hardness dispersion of waste components, and extract the total energy of acoustic emission signals as an absolute hardness characterization value by combining acoustic emission signals. Based on this, the primary crushing chamber bite angle is adjusted through the hydraulic system. The calculation module is used to extract acoustic emission signal features and reconstruct the fragmented energy spectrum using wavelet packet transform, calculate the KL divergence between the real-time energy spectrum and the standard unloaded energy spectrum and fuse the total energy growth coefficient, and use it as a real-time load status indicator for each level of cavity. The generation module is used to monitor the projected contour of the aggregate discharged from the final crushing chamber in real time and calculate the boundary fractal dimension. When the fractal dimension exceeds the preset shaping threshold range, the real-time load state index of the secondary crushing chamber is extracted according to the deviation direction of the fractal dimension to construct a piecewise adjustment function to generate a frequency correction factor. The frequency correction factor is applied to the rotational speed control of the eccentric shaft of the secondary crushing chamber, and the inter-stage material screening throughput is adjusted in conjunction with the weighted result of hardness dispersion and the real-time load state index of each stage chamber until the fractal dimension is reset.

10. The system according to claim 9, characterized in that, The construction of a gray-level-gradient co-occurrence matrix based on multi-view texture images, and the analysis of singular value entropy to represent the hardness dispersion of waste components, includes: Singular value decomposition is performed on the constructed gray-level-gradient co-occurrence matrix, and the diagonal elements of the singular value diagonal matrix are extracted to form feature vectors; The elements in the feature vector are normalized so that the sum of all elements is 1, thus obtaining the probability distribution vector; Calculate the product of the natural logarithm of each element in the probability distribution vector and the element itself, sum all the products and take the opposite number to obtain the singular value entropy; A mapping model between singular value entropy and material hardness dispersion is established, and the calculated singular value entropy is mapped to the hardness dispersion of waste components.

Images