A recycling and crushing system and method for renewable resources

Abstract

This invention relates to the field of renewable resource recycling and processing technology, and in particular to a renewable resource recycling crushing system and method, comprising a multimodal material characteristic sensing module, a pre-cutting dynamic programming module, a three-level closed-loop control module, a harmonic monitoring module, and a digital twin optimization module. Material characteristics are acquired through multimodal sensing technology, and the cutter head motion trajectory is generated by combining model predictive control. Harmonic monitoring, graded response, and digital twin optimization are used to achieve self-evolution of system parameters. This invention solves or at least mitigates the problem of entanglement and jamming of long strip-shaped materials in the crushing chamber, providing a renewable resource recycling crushing system and method. Through multimodal sensing, dynamic programming, harmonic monitoring, and digital twin optimization, this invention achieves accurate identification and efficient processing of complex materials, reduces the risk of entanglement, and improves equipment operational stability.

Description

Technical Field

[0001] This invention belongs to the field of renewable resource recycling and processing technology, and particularly relates to a renewable resource recycling crushing system and crushing method. Background Technology

[0002] Current recycling crushing systems often encounter material entanglement and accumulation within the crushing chamber when crushing long, strip-shaped materials, affecting the normal operation and processing efficiency of the equipment. While optimizing the system's structural parameters and control strategies enhances its adaptability to complex materials, existing technologies for identifying and dynamically responding to long, strip-shaped materials have limitations, potentially leading to cutter head obstruction or increased energy consumption. Plastic bottles are highly malleable, with an elongation at break ≥300%. During crushing, they produce ribbon-like fragments that are less prone to brittle fracture. Structural areas such as bottle labels and bottle neck threads can create localized toughness zones, causing the cutter to exhibit a tendency to "draw fibers." When the cutter linear speed is ≤35m / s, insufficient shearing force can result in uncut strips, increasing the cutter gap and causing the plastic to be stretched rather than cut, producing fibrous fragments.

[0003] Therefore, by using multimodal sensing and dynamic programming technology to enhance the perception of material properties and rationally plan the movement trajectory of the cutterhead, the probability of material entanglement can be reduced. Conventional crushing systems have weak harmonic monitoring and graded response capabilities, which can easily lead to delayed fault warnings caused by current harmonic distortion when processing high-density or high-toughness materials. Existing systems lack digital twin optimization methods for cutterhead movement parameters and lack the ability for system parameters to self-evolve, making it difficult to achieve long-term stable operation.

[0004] This invention patent, CN103657806B, discloses a double-blade plastic crusher, including a frame. The frame has a crushing chamber and a feed hopper. Two-stage crushing equipment is installed in the crushing chamber, with shearing crushing occurring above the crushing crushing. The shearing crushing rate is slower than the crushing crushing rate. A drive shaft and a driven shaft, linked by a reduction mechanism, are installed in the crushing chamber. The drive shaft is connected to a motor, and a high-speed blade is installed on the drive shaft. A low-speed blade is installed on the driven shaft. Correspondingly, high-speed fixed blades and low-speed fixed blades are installed in the crushing chamber to cooperate with the high-speed and low-speed blades. The beneficial effects of this invention are: 1. Because the low-speed blade has a low rotational speed and high torque, larger materials are easily sheared and then fall into the crushing range of the high-speed blade below, thus achieving advantages such as no material impact and rebound, stable operation, low noise, and safety and speed. This solution can alleviate material entanglement to some extent, but its adaptability to complex working conditions is still insufficient. Improvements in harmonic monitoring, graded response, and digital twin model optimization will greatly enhance the reliability and intelligence of the crushing system. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies, solve or at least alleviate the problem of entanglement and jamming of long strip-shaped materials in the crushing chamber, and provide a recycling crushing system and method for recyclable resources. This invention utilizes multimodal sensing, dynamic programming, harmonic monitoring, and digital twin optimization technologies to achieve accurate identification and efficient processing of complex materials, reduce the risk of entanglement, and improve equipment operational stability.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a recycling and crushing system for renewable resources, comprising:

[0007] The multimodal material feature sensing module includes an interleaved linear array of photoelectric sensors and a millimeter-wave radar, used to acquire material aspect ratio, surface roughness and dielectric constant parameters.

[0008] The pre-cutting dynamic planning module connects to the feature perception module and integrates a model prediction controller, outputting the tool head motion trajectory planning instruction;

[0009] The three-level closed-loop control module includes an execution layer closed-loop unit, an early warning layer closed-loop unit, and a system layer closed-loop unit. Data interaction between the units is achieved through an industrial bus.

[0010] The harmonic monitoring module includes a high-precision current sensor installed on the main crusher motor to collect the 3rd, 5th, and 7th harmonic components in real time.

[0011] The digital twin optimization module includes a virtual calibration engine and a physical system image model, and outputs parameter optimization instructions to each actuator.

[0012] Preferably, the pre-cutting dynamic programming module includes:

[0013] Adjustable radial bearing cutter head assembly with an axial swing stroke of ±15mm;

[0014] The blade angle adjustment mechanism enables dynamic adjustment of the entry angle from 50 to 60 degrees.

[0015] A model predictive control algorithm is used, with the optimization objectives of minimizing energy consumption and reducing entanglement risk. A method for recycling and crushing recyclable resources includes the following steps:

[0016] S1. Obtain material feature vectors through multimodal sensing technology;

[0017] S2. Generate pre-cut dynamic programming parameters based on model predictive control;

[0018] S3. Implement three-level closed-loop control to achieve coordinated optimization of execution, early warning, and system;

[0019] S4. Monitor current harmonic components and trigger a graded response mechanism;

[0020] S5. Use a digital twin model to achieve self-evolution of system parameters.

[0021] To further realize the present invention, the following technical solutions may be preferred:

[0022] Preferably, step S1 includes the following steps:

[0023] S101. Calculate the aspect ratio of the material projection using the light curtain occlusion rate;

[0024] S102. Obtain surface roughness parameters by Doppler frequency shift of millimeter-wave radar;

[0025] S103. Construct a three-dimensional feature vector of [aspect ratio, roughness, dielectric constant].

[0026] Preferably, in step S2:

[0027] When establishing the objective function for optimizing cutting parameters, a winding risk weight coefficient of 0.78±0.05 is assigned; a rolling time-domain optimization algorithm is used to generate the cutter head motion trajectory.

[0028] The blade cut angle setting is updated every 200ms.

[0029] Preferably, step S3 includes the following steps:

[0030] S301, First-level execution closed loop: Real-time adjustment loop based on feedback of the cutting cross-section shape;

[0031] S302, Second-level early warning closed loop: Hierarchical response loop based on harmonic characteristics;

[0032] S303, Third-level system closed loop: Parameter self-optimization loop based on digital twin.

[0033] Preferably, step S301 includes the following steps:

[0034] S3011. Acquire actual cut surface images using a high-speed industrial camera;

[0035] S3012. Calculate the deviation between the cross-sectional flatness and the preset target.

[0036] S3013. When the deviation value > 15%, adjust the sensitivity coefficient of the photoelectric sensor in the reverse direction by ±0.5-3.2%; Step S302 includes the following steps:

[0037] S3021. Continuously monitor the temporal changes of the 3rd, 5th, and 7th harmonic distortion rates;

[0038] S3022. Construct a regression prediction model for harmonic components and entanglement probability;

[0039] S3023. When the predicted probability > 40%, trigger a three-stage action: axial oscillation, rotational speed coupling, and emergency reversal. Step S303 includes the following steps:

[0040] S3031. Synchronize the physical system's operating data to the virtual mirror model;

[0041] S3032. Iteratively optimize the control strategy through reinforcement learning algorithm;

[0042] S3033, Automatically migrate optimized parameters to physical actuators.

[0043] Preferably, step S4 includes the following steps:

[0044] S401. Construct a regression model of harmonic distortion rate-entanglement probability;

[0045] S402. Activate Level 3 response when the predicted probability > 40%.

[0046] Level 1 triggers the axial oscillation of the cutter head and reduces the feed speed by 15%;

[0047] Level 2 simultaneously increases the cutter head speed by 20% and reduces the conveyor belt speed by 15%;

[0048] Level 3 executes an emergency reverse rotation of the main crusher and activates the audible and visual alarms.

[0049] Preferably, the method for determining the grading threshold of the three-level response in step S402 includes the following steps: S402a, establishing a Weibull probability distribution model based on historical fault data;

[0050] S402b: Determine the threshold confidence interval through kernel density estimation;

[0051] S402c, dynamically scales the threshold band range based on material density.

[0052] Preferably, step S5 includes the following steps:

[0053] S501, the digital twin model injects 23 sets of boundary test cases per work cycle;

[0054] S502, Update the control strategy network parameters using a deep learning algorithm;

[0055] S503, automatically migrates the optimized PID parameters to the physical controller.

[0056] The beneficial effects of this invention are:

[0057] This invention obtains the feature vector of the material based on multi-mode perception and obtains dynamic programming parameters through model predictive control, thereby realizing the acquisition of the characteristics of different long strip materials and rationally planning the movement trajectory of the cutter head to reduce the risk of material entanglement. In addition, the harmonic monitoring module is designed with a hierarchical response method, which can respond quickly to different working conditions, especially for high toughness or high density materials, and solves the problem of poor current stability caused by current harmonic distortion, which leads to delayed fault warning.

[0058] Furthermore, by integrating information between the execution layer, early warning layer, and system layer in the three-level closed-loop control module, multi-level collaborative optimization is achieved, improving the flatness of the cutting section. Sensitivity compensation and weight adjustment are used to improve the accuracy of material feature acquisition. Through the linkage between the virtual calibration engine and the physical system mirror model in the digital twin optimization module, the self-evolution of system parameters is achieved, ensuring long-term stable operation.

[0059] Furthermore, by adjusting the grading threshold and combining the Weibull probability distribution model with kernel density estimation, the system achieves adaptive response to changes in material density, enhancing its robustness and further improving its intelligence level. Attached Figure Description

[0060] Figure 1 This is a system block diagram of the crushing system of the present invention.

[0061] Figure 2 This is a flowchart of the crushing method of the present invention.

[0062] Figure 3 This is a flowchart of step S3 of the present invention.

[0063] Figure 4 This is a flowchart of step S4 of the present invention. Detailed Implementation

[0064] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Example 1

[0067] Currently, when crushing long strip-shaped materials, the recycling crushing system often encounters problems such as the materials becoming entangled and stuck in the crushing chamber, affecting the normal operation and processing efficiency of the crushing equipment. Conventional optimization measures require optimizing the structural parameters and control strategies of the crushing system to improve its adaptability to complex materials such as long strips. However, the existing technology has limited ability to identify and dynamically respond to long strip-shaped materials, which to some extent leads to the cutter head being easily obstructed and energy consumption being high during crushing.

[0068] This technical field provides a recycling and crushing system for renewable resources, including a multimodal material feature sensing module, a pre-cutting dynamic planning module, a three-level closed-loop control module, a harmonic monitoring module, and a digital twin optimization module. These modules are connected via an industrial bus and work collaboratively according to the following logic to achieve efficient crushing of complex materials: The multimodal material feature sensing module includes an alternating array of linear photoelectric sensors and a millimeter-wave radar. The linear photoelectric sensors are horizontally mounted above the feed inlet to sense the material's projected area and obstruction rate. The millimeter-wave radar is mounted on the side and alternating with the linear photoelectric sensors to sense the material's surface roughness and dielectric constant. The linear photoelectric sensors and the millimeter-wave radar are connected to the feature parameter extraction module via a signal acquisition unit.

[0069] like Figure 2The diagram shows the dynamic planning module of the pre-cutting module. Its main components include an adjustable radial bearing cutterhead assembly mounted on the output shaft of the main crusher, a blade angle adjustment mechanism mounted on the outer edge of the cutterhead, and a model predictive controller. The adjustable radial bearing cutterhead assembly can swing axially by ±15mm under the action of a hydraulic drive. The blade angle adjustment mechanism adjusts the cutting angle of each blade, driven by a stepper motor mounted on the outer edge of the cutterhead, and dynamically adjusted within a range of 50-60 degrees. The model predictive controller uses the feature vector sensed by the multimodal material feature sensing module as input, with minimum energy consumption and reduced entanglement risk as optimization objectives. It uses a rolling time-domain optimization algorithm to generate cutterhead motion trajectory planning instructions, and updates the cutting angle setpoint every 200ms before sending them to the blade angle adjustment mechanism for execution via a servo control system. The three-level closed-loop control module includes an execution-level closed-loop unit, an early warning-level closed-loop unit, and a system-level closed-loop unit, which exchange information via an industrial bus. The execution layer closed-loop unit monitors the flatness of the cut section image against the expected shape. If the flatness fluctuation exceeds 15%, the photoelectric sensor sensitivity compensation module is activated. The early warning layer closed-loop unit determines whether the current operating condition exceeds its handling capacity based on the monitoring results of the harmonic monitoring module. If so, it requests a tiered response. The system layer closed-loop unit comprehensively analyzes historical operating data and the current status to determine the relative importance of each module's work under the current conditions and dynamically reallocates weight ratios accordingly to ensure more accurate subsequent processing.

[0070] The key component of the harmonic monitoring module is a high-precision current sensor, installed in the power supply circuit of the main crusher, used to acquire its 3rd, 5th, and 7th harmonic components. For example... Figure 3 As shown, the graded response execution unit triggers different response measures based on the harmonic distortion rate. Level 1: Activate axial oscillation + ±10mm reciprocating motion + reduce feed speed (15%). Level 2: Activate speed coupler + conveyor belt speed reduction + cutter head speed increase (15%→20%) to clear blockage. Level 3: Activate emergency reverse rotation + main crusher reverse rotation to clear obstacles + activate audible and visual alarms to remind personnel to check.

[0071] The digital twin optimization module includes a virtual calibration engine and a physical system mirror model, which are interconnected via a high-speed communication network. The virtual calibration engine injects 23 boundary test cases per cycle and uses deep learning algorithms to update the control strategy network parameters. The physical system mirror model simulates the operation of physical equipment to verify whether the optimized PID parameters meet performance requirements. Once verified, the new parameters are migrated to the physical controller via the high-speed communication network, enabling the system parameters to self-evolve.

[0072] Example 2

[0073] To improve the performance of the system in Embodiment 1, this embodiment provides a method for recycling and crushing renewable resources, which includes the following steps:

[0074] S1. Obtain material feature vectors through multimodal sensing technology;

[0075] S2. Generate pre-cut dynamic programming parameters based on model predictive control;

[0076] S3. Implement three-level closed-loop control to achieve coordinated optimization of execution, early warning, and system;

[0077] S4. Monitor current harmonic components and trigger a graded response mechanism;

[0078] S5. Use a digital twin model to achieve self-evolution of system parameters.

[0079] Step S1 includes the following steps:

[0080] S101. A cross detection light curtain with a spacing of 15cm is formed by a linear array of photoelectric sensors. The aspect ratio of the material projection is calculated based on the time difference of the light curtain occlusion. The measurement accuracy reaches ±0.7mm.

[0081] The S102 millimeter-wave radar acquires the reflected signals from the material surface at a scanning frequency of 80Hz, and analyzes the surface roughness parameters through Doppler frequency shift, with a resolution of 0.3μm.

[0082] S103. Synchronously collect the dielectric constant characteristic value of the material and establish a three-dimensional feature matrix including aspect ratio, roughness, and dielectric constant. The weight distribution ratio of each parameter is 5:3:2.

[0083] The dynamic programming in step S2 includes:

[0084] a) A rolling time-domain optimization algorithm is used to update the control parameters every 200ms;

[0085] b) In the objective function setting, the weight coefficient for the entanglement risk reduction rate is set to 0.78±0.05, and the weight for energy consumption optimization is 0.22±0.05;

[0086] c) The dynamic adjustment range of the blade entry angle is limited to 50-60 degrees, and the adjustment step size shall not exceed 0.5 degrees / time;

[0087] d) The ratio of the cutter head speed to the conveyor belt speed is maintained in the range of 1.25-1.38, and the speed difference is automatically compensated according to the material density.

[0088] Step S3 includes the following steps:

[0089] S301, First-level execution closed loop: Real-time adjustment loop based on cutting section shape feedback, which includes;

[0090] S3011. Acquire actual cut surface images using a high-speed industrial camera;

[0091] S3012. Calculate the deviation between the cross-sectional flatness and the preset target.

[0092] S3013. When the deviation value is >15%, adjust the sensitivity coefficient of the photoelectric sensor in the reverse direction by ±0.5-3.2%.

[0093] S302, Second-level early warning closed loop: A hierarchical response loop based on harmonic characteristics, which includes:

[0094] S3021. Continuously monitor the temporal changes of the 3rd, 5th, and 7th harmonic distortion rates;

[0095] S3022. Construct a regression prediction model for harmonic components and entanglement probability;

[0096] S3023. When the predicted probability is >40%, trigger three-level actions: axial oscillation, rotational speed coupling, and emergency reversal.

[0097] S303, Third-level system closed loop: Parameter self-optimization loop based on digital twin, which includes:

[0098] S3031. Synchronize the physical system's operating data to the virtual mirror model;

[0099] S3032. Iteratively optimize the control strategy through reinforcement learning algorithm;

[0100] S3033, Automatically migrate optimized parameters to physical actuators.

[0101] The collaborative logic of the three-level closed loop is as follows:

[0102] The closed-loop response speed is ≤200ms, and local process deviations are handled.

[0103] The early warning closed-loop response speed is ≤2s, suppressing systemic risks;

[0104] The system completes global parameter evolution every 2.5 hours in a closed loop, achieving continuous optimization.

[0105] A clear three-tier architecture is established, comprising the execution layer (process level), the early warning layer (system level), and the system layer (strategic level). Three response timing scales are set: 200ms, 2s, and 2.5h, reflecting the rigor of the control logic. Data from the execution layer is uploaded to the early warning layer for model training, while system-level optimization parameters are downloaded to the execution layer controller, creating a closed-loop data flow. Furthermore, their functions are complementary: the execution loop addresses the "how" question, the early warning loop addresses the "when" question, and the system loop addresses the "why" question.

[0106] Step S4 includes the following steps:

[0107] S401. Construct a regression model of harmonic distortion rate-entanglement probability, and monitor the harmonic components of the main crusher current in real time with a sampling rate of 10kHz.

[0108] S402. Activate Level 3 response when the predicted probability > 40%.

[0109] When the 3rd harmonic distortion rate exceeds 7.2% for 5 seconds, a Level 1 (40-60%) response is triggered:

[0110] The cutter head performs axial reciprocating oscillation of ±10mm at a frequency of 15Hz.

[0111] The feed rate is reduced by 15% and material distribution scanning is activated;

[0112] When the 5th harmonic distortion rate exceeds 9.5%, a Level 2 (60-80%) response is triggered:

[0113] The cutter head speed is increased by 20% while the conveyor belt speed is reduced by 15% simultaneously;

[0114] Activate the enhanced mode of the cavity negative pressure dust removal system;

[0115] When the 7th harmonic distortion rate exceeds 11.8%, a Level 3 (>80%) response is triggered:

[0116] The main crusher performs an emergency reverse operation, which lasts for 3-5 seconds.

[0117] Trigger an audible and visual alarm and upload a fault code to the central control system.

[0118] The method for determining the grading threshold of the three-level response in step S402 includes the following steps:

[0119] S402a. Establish a metal / plastic differentiated threshold library based on 217 hours of bench test data: lower the threshold for each grade of plastic material by 12-15%, and raise the threshold for high-density metal material by 8-10%.

[0120] S402b: Determine the threshold confidence interval through kernel density estimation, and dynamically calibrate the threshold through the online learning module: Calculate the false alarm or missed alarm rate every 24 hours. When the false alarm rate is >15% or the missed alarm rate is >5%, trigger retraining. Automatically compensate for the threshold offset by combining the material humidity parameter (the threshold decreases by 2% for every 5% increase in humidity).

[0121] A 10-second delay confirmation mechanism is also set up to prevent accidental triggering due to momentary interference. During the delay period, a high-frequency (100Hz) harmonic re-check is initiated.

[0122] S402c, dynamically scales the threshold band range based on material density.

[0123] The critical point was determined by failure mode analysis: 40% probability corresponds to 1.25 times the benchmark value of harmonic distortion rate, 60% probability corresponds to the stress level of 70% of the material yield strength, and 80% probability corresponds to close to 90% of the system resonant frequency.

[0124] Step S5 includes the following steps:

[0125] S501. Construct a digital mirror model containing 12 dynamic parameters, covering torque fluctuation, vibration spectrum, and temperature gradient. The digital twin model is injected with 23 sets of boundary test cases per work cycle to simulate the crushing scenario of ultra-long materials with an aspect ratio of 20:1 and to construct an extreme working condition model with tool wear of up to 80%.

[0126] S502. Update the control strategy network parameters using a deep learning algorithm, with ≥500 training iterations per cycle;

[0127] S503 automatically migrates the optimized PID parameters to the physical controller, and the migration process performs a dual verification mechanism.

[0128] Example 3

[0129] To enable those skilled in the art to better understand and implement this invention, the implementation principles of this invention will be further described in detail below with reference to a specific application scenario. In actual operation, when a batch of waste plastic bottles enters the crushing chamber, they are detected by a multimodal material feature sensing module. The linear array photoelectric sensor group calculates the projected area and aspect ratio of the material by receiving the light curtain obstruction rate formed by the light beam, with an accuracy of ±0.5mm. The millimeter-wave radar obtains the surface roughness and dielectric constant of the material by analyzing the Doppler frequency shift. The signal acquisition unit fuses the data collected by each sensor to generate a three-dimensional feature vector of [aspect ratio, roughness, dielectric constant], which is then sent to the pre-cutting dynamic planning module via the industrial bus. After receiving the above three-dimensional feature vector, the model prediction controller in the pre-cutting dynamic planning module uses a rolling time-domain optimization algorithm to determine the cutting head motion trajectory planning instruction with the optimization objectives of minimizing energy consumption and reducing entanglement risk. After receiving the three-dimensional feature vector, the model prediction controller assigns a weight coefficient W = 0.78 ± 0.05 according to the entanglement risk, and establishes an objective function for optimizing the cutting parameters. The blade cutting angle setting is updated every 200ms and sent to the blade angle adjustment mechanism via the servo control system. The blade angle adjustment mechanism is driven by a stepper motor to adjust the blade cutting angle to 55 degrees. At the same time, the adjustable radial bearing cutter head assembly swings to the designated position under the control of the hydraulic drive device to ensure optimal contact between the cutter head and the material.

[0130] A three-level closed-loop control module participates in the real-time operation of the entire process. The execution-level closed-loop unit uses a camera to capture images of the cut surface and compares them with the desired shape. If the flatness of the cut surface deviates by more than 15%, the photoelectric sensor sensitivity compensation program will be activated to recalibrate the sensitivity of the linear array photoelectric sensor group and improve its accuracy. The early warning-level closed-loop unit continuously monitors readings from the harmonic monitoring module to determine whether graded response measures should be activated. The system-level closed-loop unit comprehensively analyzes historical and current data and status information to adjust the relative priorities of different modules. When a sudden increase in material density is detected, the system-level closed-loop unit adjusts the weight ratio to make subsequent processing more accurate. Several high-precision current sensors within the harmonic monitoring module are installed in the power supply circuit of the main crusher to extract the 3rd, 5th, and 7th harmonic components. The graded response execution unit activates different graded response measures based on the harmonic distortion rate. When the harmonic distortion rate reaches the Level 1 threshold, the axial swing device drives the conveyor belt to swing horizontally back and forth by ±10mm and reduces the feed speed by 15%. When the harmonic distortion rate rises to the Level 2 threshold, the speed coupler synchronously adjusts the conveyor belt speed down by 15% and the cutter head speed up by 20% to alleviate the jamming of the main crusher. When the harmonic distortion rate rises to the Level 3 threshold, the emergency reversing mechanism drives the main crusher to rotate in the opposite direction to clear the obstruction.

[0131] The digital twin optimization module continuously optimizes the control strategy throughout the entire operation. The virtual calibration engine injects multiple boundary test cases each job cycle, performs deep learning on the strategy, and updates the control strategy network parameters. The physical system mirror simulates the real system operation, verifying whether the optimized PID meets performance requirements. If so, the new parameters are moved to the physical controller, achieving self-evolution of system parameters.

[0132] As described above, this invention solves the problem of knotting, entanglement, and jamming in the crushing process of complex materials by using multimodal sensing technology, dynamic programming methods, harmonic monitoring, and digital twin optimization. Each module is designed based on the connection, position, and fit relationships of the components, thereby achieving the safe and efficient operation of the entire system.

[0133] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A resource recovery and crushing system, comprising: Comprise: Multimodal material feature perception module, including staggered arrangement of linear array photosensor group and millimeter wave radar, for obtaining material aspect ratio, surface roughness and dielectric constant parameters; Pre-cut dynamic planning module, connected with feature perception module and integrated model predictive controller, outputting cutter head motion trajectory planning instructions; Three-level closed-loop control module, including execution layer closed-loop unit, early warning layer closed-loop unit and system layer closed-loop unit, data interaction between units through industrial bus; Harmonic monitoring module, including high-precision current sensor installed on the main crusher motor, real-time acquisition of 3, 5, 7 harmonic components; Digital twin optimization module, including virtual tuning engine and physical system mirror model, outputting parameter optimization instructions to each actuator; The pre-cut dynamic planning module comprises: Adjustable angular bearing cutter head assembly with axial ±15mm swing travel; Blade angle adjusting mechanism to achieve dynamic adjustment of 50-60 degree cutting angle; Model predictive control algorithm with minimum energy consumption and reduced entanglement risk as optimization target.

2. A resource recovery crushing method, characterized by, Comprise the following steps: S1, obtain material feature vector through multimodal perception technology; S2, generate pre-cut dynamic planning parameters based on model predictive control; S3, implement three-level closed-loop control to realize the collaborative optimization of execution-early warning-system; S4, monitor current harmonic components and trigger a hierarchical response mechanism; S5, use digital twin model to complete system parameter self-evolution; The step S3 comprises the following steps: S301, first-level execution closed-loop: real-time adjustment loop based on cutting section morphology feedback; S3011, acquire actual cutting section image through high-speed industrial camera; S3012, calculate the deviation value of section flatness from the preset target; S3013, when the deviation value is >15%, adjust the photosensor sensitivity coefficient in reverse by ±0.5-3.2%; S302, second-level early warning closed-loop: hierarchical response loop based on harmonic features; S3021, continuously monitor the time sequence changes of 3, 5, 7 harmonic distortion rates; S3022, construct a regression prediction model of harmonic components-winding probability; S3023, trigger three-level actions of axial swing, speed coupling and emergency reverse when the predicted probability is >40%; S303, third-level system closed-loop: parameter self-optimization loop based on digital twin; S3031, synchronize physical system operation data to virtual mirror model; S3032, iteratively optimize control strategy through reinforcement learning algorithm; S3033, automatically migrate optimized parameters to physical actuators.

3. The method of claim 2, wherein, The step S1 comprises the following steps: S101, calculate material projection aspect ratio through light curtain blockage rate; S102, obtain surface roughness parameters through millimeter wave radar Doppler shift; S103, construct a three-dimensional feature vector of [aspect ratio, roughness, dielectric constant].

4. The method of claim 2, wherein, In step S2: When establishing the cutting parameter optimization objective function, assign a winding risk weight coefficient of 0.78±0.05; Use rolling horizon optimization algorithm to generate cutter head motion trajectory; Update blade cutting angle set value every 200ms.

5. The method of claim 2, wherein, The step S4 comprises the following steps: S401, construct a regression model of harmonic distortion rate-winding probability; S402, activate the third level response when the predicted probability > 40%: Level1 trigger the axial swing of the cutterhead and reduce the feed speed by 15%; Level2 synchronously increase the cutterhead speed by 20% and reduce the conveyor speed by 15%; Level3 execute the emergency reverse of the main crusher and start the audible and visual alarm.

6. The method of claim 5, wherein, The hierarchical threshold determination method of the third level response in the step S402 includes the following steps: S402a, establish a Weibull probability distribution model based on historical failure data; S402b, determine the threshold confidence interval through kernel density estimation; S402c, dynamically scale the threshold band range combined with the material density.

7. The method of claim 2, wherein, The step S5 includes the following steps: S501, inject 23 groups of boundary test cases into the digital twin model every job cycle; S502, update the control strategy network parameters through deep learning algorithm; S503, automatically migrate the optimized PID parameters to the physical controller.

Images