Highway construction waste intelligent sorting and recycling collaborative optimization method

Abstract

The application belongs to the technical field of waste utilization, and discloses a highway construction waste intelligent sorting and recycling collaborative optimization method, which comprises the following steps: collecting waste data by a data collection module; identifying the types of highway construction waste by a multi-modal recognition unit; sorting different waste by a dynamic sorting unit; generating a recycling processing scheme for different waste by a recycling processing unit; recycling processing different waste by a modular recycling equipment; identifying abnormal conditions by an abnormality recognition unit; realizing the space-time matching of generation, processing and consumption by a supply-demand collaborative scheduling unit; and simulating and optimizing the system bottleneck by a whole-process simulation optimization unit. The application reduces the transportation link through the immediate processing mode after sorting by the modular recycling equipment, improves the stability of the performance of the recycled materials, improves the compliance rate of the recycled materials, improves the space-time matching degree of the waste generation and processing capacity based on the digital twin global resource configuration of the supply-demand collaborative scheduling unit, and virtually predicts the bottleneck through the whole-process simulation optimization unit.

Description

Technical Field

[0001] This application relates to the field of waste utilization technology, and more specifically, to a collaborative optimization method for intelligent sorting and recycling of highway construction waste. Background Technology

[0002] Highway construction waste is generated during the construction process. In order to reduce the waste of highway construction materials, staff operate a highway construction waste recycling device to process the highway construction waste for reuse.

[0003] Existing technology publication CN116474917A discloses a waste recycling device for highway construction, relating to the field of recycling equipment technology. The invention includes a U-shaped outer shell. A pump connected to the upper inner wall of the U-shaped shell has a discharge square pipe connected to its lower surface. A U-shaped feed square pipe connected to the upper surface of the pump passes through the middle of the upper surface of the U-shaped shell and is connected to a rectangular square pipe. Horizontal bars symmetrically connected to the lower surface of a screen inside the U-shaped shell are attached to the left and right inner walls of the U-shaped shell. The upper surfaces of springs symmetrically connected to the upper surfaces of the horizontal bars are connected to the corners of the lower surface of the screen. By setting up the U-shaped feed square pipe, rectangular square pipe, pump, and discharge pipe, the workload of loading highway construction waste is reduced. The eccentric wheel and fifth horizontal shaft effectively prevent screen clogging, thus facilitating the filtration of highway construction debris. The U-shaped receiving box, sixth horizontal shaft, and third motor facilitate the retrieval of all highway construction debris inside the U-shaped receiving box.

[0004] While the existing technical solutions described above can achieve the relevant beneficial effects through their structure, they still have the following drawbacks: 1. Poor performance stability of recycled materials, lacking a dynamic matching mechanism between "waste characteristics and process parameters," with process parameters mostly set based on experience, leading to large fluctuations in the performance of recycled materials. 2. Lack of a comprehensive resource scheduling mechanism, resulting in severe spatial and temporal mismatch between waste generation and processing capacity, with equipment idle or overload rates exceeding 40%; overall operating costs are high. 3. Significant environmental and safety risks, with a failure rate of over 10% for detecting hazardous components (heavy metals, organic pollutants), and some hazardous waste flowing into the recycling process causing secondary pollution; environmental accidents are prone to occur.

[0005] In view of this, we propose a collaborative optimization method for intelligent sorting and recycling of highway construction waste. Summary of the Invention

[0006] 1. Technical problems to be solved The purpose of this application is to provide a collaborative optimization method for intelligent sorting and recycling of highway construction waste, which solves the technical problems mentioned in the background. It achieves accurate identification of waste types through multi-sensor data fusion by a multi-modal identification unit, and, combined with the "value coefficient priority scheduling" logic, prioritizes high-value waste for rapid processing, thus improving recycling rates. The "immediate processing after sorting" mode of modular recycling equipment reduces transportation links, lowers material loss compared to traditional off-site processing, and increases the recycling rate of scarce resources such as steel and asphalt. The performance stability of recycled materials is improved. A blockchain quality control unit enables full lifecycle traceability, increasing the compliance rate of recycled materials and meeting the requirements for highway engineering structural components. A supply-demand collaborative scheduling unit, based on digital twin-based full-domain resource allocation, improves the spatiotemporal matching degree between waste generation and processing capacity, avoiding equipment idleness or overload. A full-process simulation optimization unit, through virtual bottleneck prediction, controls system processing efficiency fluctuations to within 10%.

[0007] 2. Technical Solution This application provides a collaborative optimization system for intelligent sorting and recycling of highway construction waste, including: Data collection module: Collects data and images of highway construction waste, and labels the data and images as reference samples; Multimodal recognition unit: Identifies the types of highway construction waste through various methods; adopts a distributed deployment approach, setting up three sets of sensing nodes at the unloading port, the starting point of the conveyor belt, and the middle section; each set includes: a high-definition industrial camera + a multispectral lens to capture the appearance texture and spectral characteristics of the waste; Laser-induced breakdown spectroscopy (LIBS) probe for simultaneous elemental analysis; Millimeter-wave radar (penetrating detection) identifies steel bars or metal components encased in concrete. Waste types are identified through data fusion algorithms, and image, spectral, and radar data are fused using a federated learning framework to address the problem of misjudgment by a single sensor.

[0008] Dynamic sorting unit: sorts different types of waste materials; Recycling and processing unit: Constructs a dynamic matching engine for waste characteristics, process parameters, and product standards to improve the performance stability of recycled materials and generate recycling and processing solutions for different wastes.

[0009] Supply and demand collaborative scheduling unit: Based on digital twin, it optimizes the allocation of global resources to achieve spatiotemporal matching of generation, processing and consumption.

[0010] Full-process simulation optimization unit: Predicts system bottlenecks through virtual simulation, enabling full-cycle management of pre-event optimization, in-event adjustment, and post-event review.

[0011] Modular recycling equipment: It recycles different waste materials and is equipped with mobile processing units that can be quickly switched. It automatically calls the corresponding modules according to the adaptation results, realizing the on-site recycling mode of "immediate processing after sorting" and turning waste into treasure.

[0012] Blockchain quality control unit: Constructing an immutable traceability chain for the entire lifecycle of recycled materials to ensure project quality and market trust.

[0013] Anomaly identification unit: Identifies abnormal situations during sorting and recycling processes using different methods; Alarm unit: includes an alarm device that promptly issues an alert when an abnormal situation is detected; PLC Control Center: Network connected to data collection module, multimodal recognition unit, dynamic sorting unit, in-process utilization unit, blockchain quality control unit, supply and demand collaborative scheduling unit, modular regeneration equipment, full-process simulation optimization unit, anomaly recognition unit, and alarm unit.

[0014] This invention provides a collaborative optimization method for intelligent sorting and recycling of highway construction waste, comprising the following steps: S1. The data collection module collects data and images of highway construction waste, and labels the data and images as reference samples. S2, the multimodal recognition unit identifies the types of highway construction waste through various methods; S3, the dynamic sorting unit sorts different waste materials; S4. The recycling and processing unit constructs a dynamic matching engine for waste characteristics, process parameters, and product standards to improve the performance stability of recycled materials and generate recycling and processing solutions for different wastes.

[0015] The S5 modular recycling equipment recycles different waste materials and is equipped with a mobile processing unit that can be quickly switched. It automatically calls the corresponding module according to the adaptation result, realizing the on-site recycling mode of "immediate processing after sorting" and turning waste into treasure.

[0016] S6. The anomaly identification unit identifies anomalies in the sorting and recycling process through different methods. S7, the supply and demand collaborative scheduling unit, is based on digital twin-based global resource optimization and allocation to achieve spatiotemporal matching of generation, processing, and consumption.

[0017] S8, the full-process simulation optimization unit, predicts system bottlenecks through virtual simulation, enabling full-cycle management of pre-event optimization, in-event adjustment, and post-event review.

[0018] S9, the blockchain quality control unit constructs an immutable traceability chain for the entire life cycle of recycled materials, ensuring project quality and market trust.

[0019] S10. When an abnormal situation is detected, the alarm unit will issue an alarm in a timely manner.

[0020] 3. Beneficial effects One or more technical solutions provided in this application have at least the following technical effects or advantages: 1. This invention maximizes resource utilization efficiency: The multimodal recognition unit achieves accurate identification of waste types through multi-sensor data fusion (image + spectrum + radar). Combined with the "value coefficient priority scheduling" logic, high-value waste (steel, asphalt) is given priority to enter the fast processing channel, and the recycling rate is increased to more than 85%. 2. The modular recycling equipment's "immediate processing after sorting" mode reduces transportation links, lowers material loss compared to the traditional off-site processing mode, and increases the recycling rate of scarce resources such as steel and asphalt.

[0021] 3. Improved stability of recycled material performance: The dynamic matching engine of "waste characteristics-process parameters-product standards" in the recycling processing unit, combined with the random forest algorithm prediction model, can adjust key parameters such as the amount of asphalt recycling agent and concrete mix ratio in advance; the blockchain quality control unit realizes full life cycle traceability, improves the compliance rate of recycled materials, meets the requirements of highway engineering structural components, and breaks the industry prejudice that "recycled materials have poor performance".

[0022] 4. The supply and demand collaborative scheduling unit, based on digital twins, enables full-domain resource allocation, thereby improving the spatiotemporal matching degree between waste generation and processing capacity and avoiding equipment idleness or overload. 5. The full-process simulation optimization unit can control the fluctuation of system processing efficiency within 10% by virtually predicting bottlenecks; the modular equipment quick-disassembly and parallel operation mode can increase the daily waste processing capacity to 1.5 times that of traditional fixed production lines, and reduce the overall operating cost.

[0023] 6. Environmental and safety risks are controllable: The anomaly identification unit uses multi-device collaborative sensing (LIBS / XRF spectroscopy + vision + radar) to achieve real-time monitoring of harmful components and abnormal physical properties. Combined with a multi-level alarm mechanism, it prevents harmful waste from flowing into the recycling process. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the overall optimization system for intelligent sorting and recycling of highway construction waste disclosed in a preferred embodiment of this application; Figure 2 This is a flowchart illustrating the recycling processing unit of the intelligent sorting and recycling synergistic optimization method for highway construction waste disclosed in a preferred embodiment of this application, which generates recycling processing schemes for different waste materials. Detailed Implementation

[0025] The present application will be further described in detail below with reference to the accompanying drawings.

[0026] Reference Figure 1 This application provides a collaborative optimization system for intelligent sorting and recycling of highway construction waste, comprising: Data collection module: Collects data and images of highway construction waste, and labels the data and images as reference samples; Multimodal recognition unit: Identifies the types of highway construction waste through various methods; adopts a distributed deployment method, setting up three sets of sensing nodes at the unloading port, the starting point of the conveyor belt, and the middle section. Each set includes: a high-definition industrial camera (12 million pixels, 30fps) + a multispectral lens (covering the 400-1000nm band) to capture the appearance texture and spectral characteristics of the waste. Laser-induced breakdown spectroscopy (LIBS) probe (response time < 0.5s) for simultaneous analysis of elemental composition (such as the carbon-hydrogen ratio in asphalt and the Ca / Si ratio in concrete). Millimeter-wave radar (penetrating detection) identifies steel bars or metal components encased in concrete. A data fusion algorithm identifies waste types, and a federated learning framework integrates image, spectral, and radar data to address the problem of misclassification by a single sensor (e.g., improving the accuracy of distinguishing between wet asphalt blocks and dark concrete to 99.2%). The priority scheduling logic dynamically adjusts channel allocation based on waste value coefficients (steel 1.0 > asphalt 0.8 > concrete 0.6 > earthwork 0.3), with high-value waste receiving priority access to the fast processing channel.

[0027] Dynamic sorting unit: sorts different types of waste materials; Recycling and Processing Unit: A dynamic matching engine is constructed to integrate waste characteristics, process parameters, and product standards, improving the performance stability of recycled materials and generating recycling and processing solutions for different wastes. An AI performance prediction model is introduced, using a random forest algorithm to predict the performance of recycled materials and adjust process parameters in advance.

[0028] Supply and demand collaborative scheduling unit: Based on digital twin, it optimizes the allocation of global resources to achieve spatiotemporal matching of generation, processing and consumption.

[0029] Full-process simulation optimization unit: Predicts system bottlenecks through virtual simulation, enabling full-cycle management of pre-event optimization, in-event adjustment, and post-event review.

[0030] Modular recycling equipment: It recycles different waste materials and is equipped with mobile processing units that can be quickly switched (such as vehicle-mounted asphalt recycling machines and mobile concrete crushing stations). The corresponding modules are automatically called according to the adaptation results to realize the on-site recycling mode of "immediate processing after sorting" and turn waste into treasure.

[0031] Blockchain quality control unit: Constructing an immutable traceability chain for the entire lifecycle of recycled materials to ensure project quality and market trust.

[0032] Anomaly identification unit: Identifies anomalies during the sorting and recycling process using different methods; Alarm unit: includes an alarm device that promptly issues an alert when an abnormal situation is detected; The PLC control center is networked with the data collection module, multimodal recognition unit, dynamic sorting unit, waste recycling processing unit, blockchain quality control unit, supply and demand coordination scheduling unit, modular recycling equipment, full-process simulation optimization unit, anomaly identification unit, and alarm unit. It adopts a redundant PLC architecture (dual CPU synchronous operation) to ensure uninterrupted operation in the event of a single point of failure, with a control cycle ≤10ms. It supports edge-cloud collaboration: local PLC handles real-time control (e.g., sorting actions), while the cloud handles big data analysis (e.g., monthly simulation optimization), achieving data interaction via the MQTT protocol. It also interfaces with highway project management systems (e.g., progress management, cost accounting) to achieve linked control of "waste disposal - project progress".

[0033] Furthermore, the multimodal recognition unit identifies the types of highway construction waste through various methods, including the following steps: 1. Deployment of sensing devices: Three sets of distributed sensing devices are set up at the unloading port, the starting point of the conveyor belt, and the middle section. Each set of devices integrates a high-definition industrial camera (12 million pixels, 30fps), a multispectral lens (covering the 400-1000nm band), a laser-induced breakdown spectroscopy (LIBS) probe (response time <0.5s) and millimeter-wave radar.

[0034] 2. Multi-source data acquisition: High-definition industrial cameras and multispectral lenses simultaneously capture the appearance, texture, color, and spectral characteristics of waste materials; The laser-induced breakdown spectroscopy (LIBS) probe analyzes the elemental composition of waste materials in real time and outputs key parameters such as the carbon-hydrogen ratio of asphalt and the Ca / Si ratio of concrete. Millimeter-wave radar can penetrate the surface of waste materials and identify the distribution of steel bars or metal components encased in concrete.

[0035] 3. Local data preprocessing: Denoising (wavelet transform) and format standardization are performed on the acquired image, spectral and radar data, and feature values ​​(such as spectral peak intensity, image texture entropy, and radar echo intensity) are extracted.

[0036] 4. Federated learning local training: Based on the ResNet-50 architecture, the recognition model is trained with preprocessed data, and only the model parameters are uploaded (the original data is not transmitted) to protect data privacy.

[0037] 5. Global Model Aggregation: The PLC control center receives parameters from each model and aggregates them to generate a global recognition model every hour, dynamically optimizing feature weights (e.g., the weight of the metal magnetic feature is set to 0.8).

[0038] 6. Multimodal data fusion: By fusing image, spectral, and radar data through a federated learning framework and cross-validating the recognition results, the accuracy of distinguishing between wet asphalt blocks and dark concrete is improved to 99.2%.

[0039] 7. Waste type determination: Based on the fused data, output the waste type (steel, asphalt, concrete, earthwork, etc.) and confidence level (≥95% is considered valid identification).

[0040] 8. Priority scheduling application: Based on the waste value coefficient (steel 1.0 > asphalt 0.8 > concrete 0.6 > earthwork 0.3), the identification results are transmitted to the dynamic sorting unit, and high-value waste is preferentially allocated to the fast processing channel.

[0041] Furthermore, the dynamic sorting unit sorts different types of waste materials, including the following steps: 1. Magnetic Separation (Dedicated Steel Channel): A strong magnetic roller (300mm in diameter, with the same width as the conveyor belt) is installed at a 30° angle at the beginning of the conveyor belt. The magnetic strength is stable at 15000Gs (monitored in real time by a Hall sensor, with a fluctuation error of <5%). A wear-resistant rubber scraper (Shore 85A hardness) is installed below the roller to prevent the steel from sticking together.

[0042] The intelligent adsorption logic is as follows: For steel bars with a diameter > 8mm and an area > 0.1m² 2 For large steel materials such as steel formwork, the magnetic rollers maintain full power operation with an adsorption response time of <0.5s; For small items such as broken steel bars and steel strands with a diameter of <8mm, the density is predicted by multispectral image recognition, and the roller speed is automatically reduced (from 60r / min to 30r / min) to avoid stacking and clogging caused by excessive adsorption force. Non-magnetic waste (concrete, asphalt, etc.) enters the next level of sorting via a conveyor belt (speed 1.5m / s), while steel is introduced into the "metal recycling channel" via an inclined slide at the top (slope 45°). An infrared counter is installed at the end of the slide to count the steel flow rate in real time (error <2%).

[0043] 2. Particle size classification (separation of concrete and earthwork): A three-layer vibrating screen is used, combined with multispectral identification of particle size prediction data, and the vibration frequency is dynamically adjusted (30-60Hz). Upper screen (50mm aperture, high manganese steel, 10mm thickness): intercepts large pieces of concrete / brick (>50mm) and allows them to fall into the "coarse crushing channel" by gravity. The screen vibrates at a frequency of 60Hz (5mm amplitude) to ensure it does not clog. Middle layer screen (20mm aperture, polyurethane material, elastic modulus 200MPa): Separates 20-50mm gravel / concrete blocks, vibration frequency 45Hz, combined with ultrasonic screen cleaning device (500W power, triggered once every 30 seconds) to prevent adhesion. Lower screen (5mm aperture, stainless steel woven mesh, 4 mesh): screens stone / soil particles of 5-20mm, vibrates at 30Hz, and is equipped with a weighing sensor at the bottom (accuracy ±0.1kg) to provide real-time feedback on material flow.

[0044] A dynamic frequency adjustment algorithm is adopted: based on the particle size distribution prediction data of the multimodal recognition unit (e.g., waste material larger than 50mm accounts for >30%), the vibration frequency is automatically adjusted by the PLC. The dynamic frequency adjustment algorithm model is as follows:

[0045] Σw i =1; F=1.732×U×I×cosΦ; where f is the adjusted real-time operating frequency of the vibrating screen (unit: Hz), with a value range of 30-60Hz; f0 is the reference vibration frequency (unit: Hz), with a default value of 45Hz, which can be manually calibrated according to the initial material characteristics; i is the index; w i It is the weighting factor for each particle size range, reflecting the degree of influence of different particles on the screening efficiency; ΔP i ΔP is the deviation rate (unitless) of the proportion of particles exceeding the threshold in the i-th interval; if the actual proportion is lower than the threshold, ΔP i A negative value reduces the frequency adjustment amplitude. Q is the real-time material flow rate, collected by the conveyor belt weighing sensor; Q max F is the maximum allowable flow rate of the equipment; F is the real-time equipment load (unit: kW), calculated from the motor current sensor; F max It is the maximum load of the equipment (unit: kW), with a default value of 15kW; U is the line voltage of the motor, I is the line current of the motor, and Φ refers to the phase difference angle between voltage and current in a three-phase AC circuit (unit: degrees or radians).

[0046] The routing path is as follows: Large concrete / brick pieces >50mm: fall into the "coarse crushing channel" for recycling coarse aggregate (equipped with a jaw crusher with a processing capacity of 50 tons / hour). 20-50mm crushed stone / concrete blocks: enter the "medium crushing channel" (cone crusher, discharge particle size controlled at 10-20mm) for use in the subgrade layer; 5-20mm stone chips / soil particles: enter the "fine sieve channel" for graded crushed stone base course; Soil with a thickness of less than 5mm: enters the "soil treatment channel" (equipped with a drying chamber, and is reused after the moisture content drops to below 15%), and is used for covering green belts or solidifying fill.

[0047] 3. Airflow separation (separation of asphalt and concrete): For mixed waste materials (easily confused asphalt blocks and concrete blocks) of 20-50mm, a combination device of "airflow separation + hardness testing" is used. Airflow sorting chamber: 2m long × 1m wide × 1.5m high, with a high-pressure centrifugal fan (3000Pa pressure, adjustable 20m / s) installed on the top, forming a horizontal directional airflow through guide vanes; a density sensor (accuracy ±0.01g / cm³) is installed inside the chamber. 3 ), real-time detection of waste density.

[0048] Hardness testing station: Located below the outlet of the airflow sorting chamber, it is equipped with a piezoelectric hardness sensor (measurement range 1-10 Mohs hardness, error ±0.2), with a detection point spacing of 50mm, and can detect 10 points per second.

[0049] The two-parameter sorting logic is as follows: Airflow sorting stage: based on density threshold (1.2 g / cm³) 3 Automatically adjusts wind speed: Low density (1.0-1.2 g / cm³) 3 The asphalt blocks were blown by the airflow to the right-hand "asphalt pre-diversion channel"; high density (2.3-2.5 g / cm³) 3 The concrete block fell naturally onto the hardness testing platform.

[0050] Secondary hardness verification: If a Mohs hardness of 1-2 (asphalt characteristic value) is detected: trigger the pneumatic gate (response time < 0.3s) and import into the "asphalt recycling channel"; If a Mohs hardness of 3-4 (characteristic value of concrete) is detected: keep the gate closed and enter the "concrete recycling channel".

[0051] A false judgment prevention mechanism is adopted. When the density and hardness test results conflict (such as low density but high hardness), the system automatically pauses the batch of waste materials and calls up the multispectral image for backtracking analysis (the judgment is completed within 10 seconds).

[0052] 4. Irregularly Shaped Waste Capture (Separation of Templates and Miscellaneous Materials): A 6-axis robotic arm with 3D vision guidance is installed at the end of the conveyor belt to identify irregularly shaped waste such as twisted steel bars, broken templates, and plastic signboards. This includes an intelligent grasping system and a classification and grasping logic control module. Intelligent grasping system: includes a visual positioning unit and an actuator; Visual positioning unit: A 3D vision camera (point cloud resolution 0.5mm, frame rate 15fps) is deployed above the end of the conveyor belt. Combined with a deep learning model (YOLOv8-nano), it identifies irregularly shaped waste materials with a classification accuracy of ≥98%. Actuator: 6-axis robotic arm (arm span 2.5m, repeatability ±0.1mm), equipped with a quick gripper change device (magnetic gripper / vacuum chuck / mechanical gripper), gripper change time <5 seconds.

[0053] Classification and grabbing logic control module: This module controls the classification and grabbing logic for different types of waste materials. For irregularly shaped metal parts (such as twisted steel bars and broken anchors): use the magnetic claw (magnetic force 8000Gs) to grab them and place them into the "metal repair channel" (equipped with a straightening machine and rust removal device). Wooden / bamboo template: Use vacuum suction cup (50N suction) to introduce into the "biomass processing channel" (after being crushed, it can be used as fuel with a calorific value of ≥4000kcal / kg). Plastic waste (such as signboards and damaged guardrails): Mechanical grippers are used to send it to the "plastic recycling channel" (melt granulation, purity ≥95%).

[0054] We analyze failed capture cases (such as severely deformed steel bars) weekly and update the capture angle and force parameters through reinforcement learning, keeping the failure rate below 1%.

[0055] 5. Chemical property auxiliary verification: LIBS detection point (laser power 100mJ, sampling frequency 1 time / second) is set at the end of each sorting channel to analyze the elemental composition of waste materials in real time; Establish an element feature library: Asphalt: Carbon content > 60%, hydrogen content > 8%, sulfur content < 1%; Concrete: Calcium content >30%, silicon content 15-25%, aluminum content <10%; Concrete with excessive chloride ions: Cl - Content > 0.06% (converted to water-soluble chloride ions).

[0056] A dynamic correction mechanism is adopted: When a system detects that the carbon content is greater than 60% but the material is located in a concrete channel, it automatically triggers a pneumatic deflector (switching time < 0.5s) to guide the waste material into the asphalt channel. If concrete with excessive chloride ions is found, immediately close the regular passage and open the "special treatment passage" (equipped with cement curing device, after curing, the chloride ion leaching rate is <0.01mg / L).

[0057] Reference Figure 2The recycling and processing unit constructs a dynamic matching engine for waste characteristics, process parameters, and product standards to improve the performance stability of recycled materials and generate recycling and processing solutions for different wastes. This includes the following steps: 1. Build a dynamic matching engine: 1.1 Data Integration: Collect three types of core data: waste characteristics (physical parameters such as particle size and density; chemical parameters such as asphalt content and concrete strength grade), process parameters (such as crushing speed, heating temperature, and additive ratio), and product standards (such as the crushing value of recycled aggregate and the Marshall stability of recycled asphalt mixture). Construct a four-dimensional correlation matrix; horizontally, divide into four major waste categories: "steel, asphalt, concrete, and earthwork"; vertically, stratify into three process stages: "basic treatment, performance enhancement, and finished product forming," and deeply correlate with the corresponding industry standards for the products (such as the "Technical Specification for the Application of Recycled Aggregates in Highway Engineering").

[0058] 1.2 Establishing a rule base and adaptation logic: Based on historical engineering data and expert experience, formulate matching rules: for example, "When the penetration of asphalt milled aggregate (RAP) is >80 (0.1mm), hot recycling process shall be given priority" and "When the crushing value of recycled concrete aggregate is >16%, it shall be restricted to use in structures above the base layer".

[0059] Develop a dynamic adjustment mechanism: When the characteristics of the input waste material exceed the preset range (such as extreme aging RAP), the system automatically triggers the process parameter candidate pool and generates 3 sets of adaptation schemes for decision-making (such as adjusting the type of recycling agent or increasing the amount of new asphalt).

[0060] 1.3 Real-time Interaction: In real-time linkage with the front-end sorting unit, it acquires characteristic data of each batch of waste material (such as the corrosion grade of sorted steel and the strength test results of concrete blocks), and automatically calls matching rules to output process parameter suggestions. After processing every 100 tons of waste material, it collects performance test data of the actual recycled products (such as the yield strength of recycled steel bars and the compaction degree of recycled base layer), and reverse-optimizes the matching rules to improve the accuracy of adaptation.

[0061] 2. Application of AI performance prediction models: 2.1 Training data preprocessing: Collect historical recycling process data, including the original characteristics of waste materials (such as the age of concrete blocks and the aging degree of asphalt materials), process parameters (such as the number of crushing times and heating time), environmental factors (temperature and humidity) and corresponding product performance results (such as compressive strength and abrasion resistance).

[0062] Data cleaning: Remove outliers (such as invalid data caused by equipment failure) and unify parameter magnitudes through normalization (such as converting intensity indicators of different units into relative compliance rates).

[0063] 2.2 Model Training and Optimization: A random forest algorithm is used to construct the prediction model, with waste characteristics and process parameters as input variables and product performance as the output variable. A performance prediction model is generated through ensemble learning of multiple decision trees. New engineering data is incorporated monthly for retraining. The prediction accuracy is verified using a confusion matrix (target error < 5%). When the prediction deviation for a certain type of waste exceeds 8%, the weight of that category of samples is increased separately for retraining.

[0064] 2.3 Predictive Application: The model outputs predicted product performance results based on real-time input of the characteristics of the current batch of waste materials and the proposed process parameters (e.g., "The 28-day strength of recycled concrete is expected to reach 92% of the C30 standard"). The 28-day compressive strength prediction model for recycled concrete is as follows:

[0065] In the formula, f cu This represents the 28-day cubic compressive strength of recycled concrete (predicted by the random forest model); k1, k2, and k3 are model coefficients (regression parameters obtained through training with the random forest algorithm), which are strongly correlated with the training dataset (such as regional recycled aggregate characteristics and construction technology), and need to be calibrated with local data, and have no fixed values; R is the recycled aggregate replacement rate (the percentage of recycled aggregate by mass in the total aggregate); C is the cement content (the mass of cement in each cubic meter of recycled concrete); T is the standard curing temperature (the average temperature of the curing environment); w 杂质 It is the impurity content in recycled aggregate (the percentage of impurities such as soil and organic matter in the mass of recycled aggregate).

[0066] If the predicted result is lower than the standard threshold, the system will automatically push parameter adjustment suggestions, such as "It is recommended to increase the cement content by 3%" or "Extend the carbonation treatment time by 1 hour", until the predicted result meets the standard.

[0067] 3. Waste recycling by type: 3.1 Steel Waste Disposal: Pre-treatment: Remove surface rust and concrete residue using rust removal equipment, and sort and classify (divided into two categories according to diameter and degree of rust: "can be directly reused" and "requires smelting and regeneration").

[0068] Recycling process: Reusable steel (such as lightly rusted rebar) is cut to the designed length and welded into rebar mesh for small structures (such as ditch covers). Non-reusable steel (such as twisted rebar ends) is sent to a smelting furnace for remelting, with controlled smelting temperature and deoxidizer addition to produce new rebar or steel components that meet the strength grade.

[0069] 3.2 Asphalt Pavement Milling Material (RAP): Pretreatment: After crushing and screening, the asphalt content and aging degree are tested, and the particles are graded and stored according to particle size.

[0070] Recycling process: Hot recycling: RAP is heated to 160-180℃, and new asphalt and recycling agent are added in proportion (the amount of additive is increased when the aging is severe). After stirring, it is used for the lower layer or base layer of asphalt pavement.

[0071] Cold recycling: Mixed with emulsified asphalt and water, no heating required, for use in low-traffic road bases or temporary access roads.

[0072] 3.3 Concrete waste: Pretreatment: After crushing, coarse aggregate (≥5mm) and fine aggregate (<5mm) are obtained by three-stage screening. The surface mortar is removed by washing.

[0073] Recycling process: Coarse aggregate: Mixed with new aggregate (recycled aggregate content ≤30%), used to prepare precast components such as curbs and drainage ditches.

[0074] Fine aggregate: mixed with lime, cement and other hardening agents, used as a leveling layer for roadbeds or as edging soil for embankments.

[0075] Extra-large concrete blocks: After being crushed twice to meet the requirements, they are used for roadbed filling or retaining structure foundations.

[0076] 3.4 Earthwork and Auxiliary Waste: Earthwork: After screening to remove humus and impurities, it is used for site leveling, covering green belts with soil, or mixing with crushed stone to form earth and stone embankments.

[0077] Wooden formwork: After being crushed and dried, it can be used as biomass fuel or pressed into fiberboard for temporary projects.

[0078] Plastic waste (such as signboards): can be melted and reshaped into small components (such as cable protection pipes), or crushed and used as lightweight filler in the roadbed.

[0079] 3.5 Special Waste Treatment: For concrete with excessive chloride ions and road marking paint containing heavy metals, a curing and stabilization process (such as adding chelating agents) should be used separately. After testing and confirming compliance, the materials should be safely landfilled according to regulations to avoid secondary pollution.

[0080] Furthermore, the supply and demand collaborative scheduling unit, based on digital twin-based global resource optimization and allocation, achieves spatiotemporal matching of generation, processing, and consumption. This includes the following steps: 1. Digital Twin Model Construction: Integrate the BIM model (LOD500 accuracy) of the construction area with GIS geographic information to construct a three-dimensional digital twin scene that includes waste sources (milling section, demolition area), processing station (sorting and recycling equipment), and demand points (roadbed filling section, base paving area), mapping the core attributes such as entity location, production capacity, and progress.

[0081] Embedded equipment parameter library: Input basic data such as the waste processing capacity of the processing station (e.g., 200 tons of asphalt material per day), the load capacity of transport vehicles (15-30 tons), and the material consumption at the demand point (e.g., 150 tons of recycled aggregate required for the roadbed per day).

[0082] 2. Real-time data acquisition and synchronization: Waste source end: Real-time waste generation (statistics by type: asphalt material, concrete blocks, etc.) and generation time (accurate to the hour) are collected through weighing sensors at the unloading port and the construction progress APP.

[0083] Processing station: IoT sensors collect data on equipment operating status (e.g., crusher load rate, recycled material inventory), processing efficiency (tons / hour), and recycled material type (e.g., recycled coarse aggregate, RAP material). Demand point: Combining construction logs and on-site sensors, record material demand type (e.g., graded crushed stone for base course), demand quantity, and latest arrival time.

[0084] Data is synchronized to the digital twin platform every 5 minutes to ensure that the virtual scene is consistent with the physical state.

[0085] 3. Multi-dimensional data fusion: Collected data is categorized and integrated, and a related database is established according to "waste type - generation time - location", "processing capacity - inventory - equipment status", and "demand type - quantity - time window". A dynamic dashboard is generated on the digital twin interface to intuitively display the waste accumulation in each area (red warning zone > 50 tons), the idle capacity of the processing station (green zone > 30%), and the waiting status of demand points (yellow zone requires replenishment within 24 hours).

[0086] 4. Intelligent supply and demand matching: 4.1 Based on waste type matching rules, automatically match similar waste materials with corresponding treatment processes (e.g., asphalt material is matched with RAP recycling station, and concrete blocks are matched with aggregate recycling line).

[0087] 4.2. Using a spatiotemporal collaborative algorithm, prioritizing "on-site treatment within 3 kilometers and nearby utilization within 5 kilometers", calculate the shortest path and time from the point of generation to the treatment station and then to the point of demand for waste (considering factors such as road restrictions and construction congestion).

[0088] 4.3. Adjust capacity balance: When the load rate of a certain processing station is >80%, some waste materials will be dispatched to idle processing stations; when the material shortage at the demand point is >20%, a cross-regional allocation warning will be triggered.

[0089] 5. Dynamic scheduling plan generation: Automatically generate a full-chain plan of "processing-transportation-delivery": specify that a certain batch of waste material (such as 100 tons of asphalt material in milling section A) will be received by processing station X, and the RAP material generated after processing will be transported by vehicle Y and delivered to base paving area B before 8:00 the next day.

[0090] Additional motivational explanations: The basis for the plan is indicated (e.g., "X processing station is 1.2 kilometers away from section A, with a current load rate of 60%, and can be given priority for acceptance" and "B area needs RAP material for cold recycling base layer, which matches the type of material after processing").

[0091] 6. Real-time adjustments and emergency response: Abnormal Trigger Adjustment: When the processing station equipment fails (such as the screening machine stops), the demand point is ahead of schedule (materials need to arrive 6 hours in advance), or there is a rainstorm (transportation efficiency decreases by 30%), the digital twin platform will automatically trigger a rescheduling.

[0092] Alternative solutions are generated: for example, if the originally planned transport vehicle breaks down, an available vehicle in the same area is automatically matched; if the processing station's capacity is insufficient, waste batches are split and sent to two surrounding processing stations to ensure that the demand points are supplied with materials on time.

[0093] 7. Cross-project resource collaboration: When a local processing station cannot process excess waste (e.g., a project's RAP material inventory exceeds 500 tons), the system automatically searches for the needs of other highway or municipal projects within the region (e.g., surrounding rural roads require cold-recycled base materials) and generates a cross-project allocation plan. Through smart contract confirmation, the system completes online verification of recycled material quality (linked to blockchain test reports), allocation of transportation responsibilities, and settlement of fees, enabling cross-entity resource transfer.

[0094] 8. Dispatch Performance Evaluation and Optimization: Generate daily dispatch reports, and statistically analyze indicators such as waste disposal timeliness rate (target ≥95%), transportation empty load rate (control <10%), and number of stockouts at demand points. Train optimization models based on historical data: For example, analyze the patterns of transportation delays during the rainy season and adjust dispatch plans 24 hours in advance; optimize transportation times (avoiding construction peak periods) for high-frequency congested road sections.

[0095] Furthermore, the full-process simulation optimization unit uses virtual simulation to predict system bottlenecks, enabling full-cycle management from pre-event optimization to in-event adjustment and post-event review. This includes the following steps: 1. Pre-emptive optimization: Predict bottlenecks through virtual simulation; 1.1 Constructing a Digital Twin Model: A full-process digital twin is built based on BIM+GIS technology, integrating basic data such as equipment parameters (e.g., crusher processing capacity, transport vehicle load capacity), waste characteristics (e.g., average daily output, type proportion), and geographical information of the construction area (e.g., material yard location, transport routes). The model accuracy reaches the millimeter level. Historical operation and maintenance data (equipment failure rate and sorting misclassification rate of similar projects in the past 3 years) are imported to construct equipment performance degradation curves and waste fluctuation models.

[0096] 1.2 Multi-scenario parameter configuration: Set up 8 typical working conditions, including the baseline scenario (based on the designed capacity of 100 tons / hour), the peak scenario (waste production increases by 30%), and the extreme weather scenario (rainfall causes a 20% decrease in open-air processing efficiency). Define the constraints for each scenario (such as maximum equipment load and environmental emission limits).

[0097] 1.3 Simulation and Bottleneck Identification: Initiate system dynamics simulation to simulate 72 hours of full-process operation, outputting dynamic curves of key indicators (such as waste accumulation in each channel, equipment load rate, and transportation waiting time). Automatically mark bottleneck nodes, such as "when the proportion of waste larger than 50mm exceeds 30%, the processing delay in the coarse crushing channel reaches 15 minutes" and "the queuing time of transport vehicles at the recycling station exceeds 20 minutes," and generate bottleneck levels (severe / moderate / minor).

[0098] 1.4 Optimization Solution Generation: Three optimization solutions are pushed to bottleneck nodes, such as "adding one temporary screening machine during peak hours" and "adjusting the frequency of transport vehicle departures from 30 minutes / shift to 20 minutes / shift," with a visual representation of the impact of the solutions on processing efficiency (improvement of 8-12%) and cost (increase of 3-5%). 2. Real-time Adjustment: Dynamic optimization in real time; 2.1 Real-time data access: Real-time data is synchronized with the digital twin model through 5G IoT: equipment operating status (such as crusher vibration frequency, conveyor belt speed), real-time waste output (updated every 5 minutes), meteorological data (rainfall, wind speed), etc., and the deviation rate is controlled within 5%.

[0099] 2.2 Deviation Warning and Root Cause Analysis: When the deviation between the actual index and the simulation prediction exceeds 10% (such as the recycled material inventory being lower than the warning value), the system automatically triggers analysis: the root cause is located through the causal chain model (such as "the milling section was completed ahead of schedule, resulting in a concentrated influx of waste" or "the screening machine malfunction caused a delay in processing").

[0100] 2.3 Dynamic adjustment instruction generation: Output adjustment instructions based on root causes: For example, for "concentrated influx of waste", temporarily increase the load of downstream recycling equipment to 110% of the design value, and dispatch standby transport vehicles to divert the waste; for "equipment failure", activate the standby channel and push maintenance work orders (including 3D location map of the faulty component).

[0101] 2.4 Effect Verification and Optimization: After adjustment, continuously track the changes in indicators (evaluate every 10 minutes). If the optimization effect does not meet expectations (e.g., inventory continues to decline), initiate a second optimization (e.g., further increase the number of vehicles transferred out) until the deviation rate returns to within 5%.

[0102] 3. Post-event review: Summarize and iterate throughout the entire cycle; 3.1 Full-cycle data archiving: After the project is completed, the full-process data is automatically summarized: cumulative waste processed amount, running time of each equipment, number of times the optimization plan is executed, cost and carbon emission data (such as energy consumption per ton of waste processed), etc., to form a visual report (including timeline and key node annotations).

[0103] 3.2 Multi-dimensional evaluation: The system performance is evaluated from three dimensions: efficiency (processing tons / hour), economy (unit cost), and environmental protection (carbon emission reduction rate). The difference between the simulation prediction value and the actual value is compared. For example, "the actual transportation cost is 8% higher than the simulation value due to the premium of temporary dispatch vehicles."

[0104] 3.3 Experience Extraction and Model Iteration: Extract effective optimization solutions (such as "adjusting the open-air processing plan 2 hours in advance during rainy weather can reduce downtime losses by 15%) and incorporate them into the knowledge base; retrain the simulation model with actual data, update parameters such as equipment failure probability and waste fluctuation coefficient, so as to improve the simulation prediction accuracy of subsequent projects by 10-15%.

[0105] 3.4 Case Study Library: The bottleneck handling cases, optimization solutions and effects of this project will be entered into the case study library and categorized by scenario tags (such as "equipment failure" and "weather impact") to support subsequent project retrieval and reuse (including the function of generating optimization plans for similar scenarios with one click).

[0106] Furthermore, the blockchain-based quality control unit constructs an immutable traceability chain for the entire lifecycle of recycled materials, ensuring project quality and market trust. This includes the following steps: 1. Distributed node deployment: 15 blockchain consensus nodes are deployed at key locations such as waste sorting points, recycling stations, engineering demand points, third-party testing agencies, and supervision units. The consortium blockchain architecture (Hyperledger Fabric) is adopted, and each node is authorized to access through digital certificates to ensure controllable data write permissions.

[0107] 2. Full lifecycle data collection and on-chain: Raw material layer: Real-time collection of GPS coordinates of waste source, initial component detection report (such as LIBS elemental analysis results), and Beidou trajectory of transport vehicles, generating a unique raw material ID and associating it with a hash value to upload to the blockchain; Processing layer: Synchronously records regeneration process parameters (such as heating temperature and stirring time) and equipment operation logs (vibration frequency and energy consumption), generating a data block every 30 seconds and binding it to the raw material ID; Product Layer: Upload the CMA certification report (indicators such as strength and durability) issued by a third-party testing agency, the product factory acceptance certificate, and associate it with the final road section information (e.g., K1+200-K1+500 roadbed section). 3. Automatic Execution of Smart Contracts: Three types of pre-set contract rules: Raw material acceptance contract: When the hazardous components of waste exceed the standard (such as heavy metals > 0.1%), it will be automatically rejected for storage and an early warning will be triggered; Process control contract: If the recycling process parameters deviate from the standard range (e.g., asphalt heating temperature > 180℃), the processing procedure shall be suspended and the technical personnel shall be notified; Product delivery contract: When the on-site sampling value of recycled material deviates from the blockchain-stored value by more than 3%, the transaction will be frozen and the traceability process will be initiated.

[0108] 4. Full-process traceability query: A unique traceability QR code is generated for each batch of recycled materials, which includes a blockchain index address; Authorized users (owners, supervisors, regulatory authorities) can view the entire process data from "waste generation - sorting - recycling - transportation - use" by scanning a code or entering the batch number in a blockchain browser. All records are tamper-proof and timestamped.

[0109] 5. Credit Incentive and Penalty Mechanism: Each month, credit scores for each participant are automatically calculated based on node data: 5 points for recycling rate ≥ 90%, 3 points for quality pass rate ≥ 95%, and 2 points deducted for each data upload delay. Credit scores are linked to bidding eligibility (e.g., a credit score of ≥90 points can earn additional points for bidding) and are also associated with carbon emission allowance redemption (1 point can be redeemed for 1 ton of CO2 allowance).

[0110] 6. Cross-entity collaborative verification: When a quality dispute occurs, multi-node consensus verification is initiated: the supervision unit, testing agency, and construction party nodes jointly retrieve the blockchain evidence data and reach a consensus conclusion through the Byzantine Fault Tolerance (BFT) algorithm to ensure the credibility of the traceability results.

[0111] Furthermore, the anomaly identification unit monitors unmanageable scenarios in the sorting and regeneration process in real time, triggering emergency responses through a multi-level alarm mechanism to prevent system paralysis or environmental risks. This includes the following steps: 1. Real-time monitoring startup: Multiple types of monitoring equipment are deployed on the sorting line (unloading port, conveyor belt, screening equipment) and the recycling line (crushing unit, heating module, mixing device), including: spectral detection equipment (LIBS, XRF), high-definition camera, millimeter-wave radar (monitoring waste characteristics), vibration sensor, temperature sensor, current sensor (monitoring equipment status), flow counter, and level gauge (monitoring material accumulation).

[0112] Data is collected in real time, with each device collecting data at a frequency of 100ms / time and transmitting it to the PLC control center via industrial Ethernet to form a dynamic monitoring data stream.

[0113] 2. Abnormal scenario determination: 2.1 Judgment of excessive waste characteristics: Excessive levels of hazardous components: When LIBS / XRF detects heavy metal content >0.1%, or GC-MS confirms the presence of pollutants such as benzene series compounds, and high-definition images match pollution characteristics (such as oil stains, discolored paint), it is determined to be "hazardous waste"; Abnormal physical properties: If the stereo vision camera detects waste material size >1.5m, or the millimeter-wave radar identifies a bundle of tangled steel bars (magnetic attraction force drops by 50%+), it is determined to be "abnormal physical properties". High-definition industrial cameras simultaneously capture images of the appearance of waste materials (such as oil-stained asphalt blocks and fragments of colored road marking paint), and anomalies are identified through an image semantic segmentation model (training samples include 100,000+ images of contaminated waste materials).

[0114] Oversized waste: Two sets of high-definition stereo vision cameras (binocular vision, ranging accuracy ±2mm) are deployed at the unloading port and the starting point of the conveyor belt. The size of the waste is calculated in real time through 3D point cloud reconstruction. When any dimension is >1.5m, the "oversized block" identification signal is triggered. Bundled steel bars: Combining the magnetic force sensor of the magnetic roller (monitoring changes in adsorption force) with millimeter-wave radar scanning, when the steel bar bundle is not separated by magnetic attraction and the radar echo shows the characteristics of "dense metal clumps", it is determined to be "wrapped waste". Mixed polluting waste: Multispectral cameras capture the infrared spectral characteristics of the waste (asphalt has a strong absorption peak at 1700nm, and plastic has a characteristic peak at 3300nm), combined with the texture recognition of high-definition cameras (the granular texture of concrete and the smooth surface of plastic). When the system simultaneously identifies more than 3 different material characteristics, it is determined to be "composite polluting waste".

[0115] Severe Mixed Pollution Identification: The "multispectral imaging + AI classification" mode is adopted. The waste images are classified by material through a trained deep convolutional neural network (CNN). When the feature probability of asphalt, concrete, plastic and chemicals (such as oil) appearing simultaneously in a single frame image is all greater than 60%, it is judged as mixed pollution and the identification signal is triggered.

[0116] 2.2 Equipment and Process Fault Judgment: Sorting equipment failure: If the robotic arm fails to grasp 5 times in a row (visual positioning deviation > 50mm), or the magnetic force of the magnetic roller decreases by > 30% (monitored by Hall sensor), it is judged as "sorting failure"; Recycling process adaptation failure: If the intelligent process library fails to match waste parameters (such as RAP bitumen content <1%) for 3 consecutive times, it is judged as "process adaptation failure". Safety threshold breach: Heating module temperature > 200℃ (infrared thermal imager monitoring), or crushing equipment vibration frequency > 80Hz (vibration sensor), is judged as "safety risk".

[0117] 3. Multi-level alarm triggering: Yellow alert: The trigger condition is a minor anomaly (such as the amount of waste in a certain channel reaching 80% of the threshold, or a single sensor signal fluctuation); the response action is a pop-up window in the system background + on-site audible and visual alarm (80dB buzzer), and push processing suggestions to the operation and maintenance personnel's APP. If it is not handled within 15 minutes, it will automatically escalate.

[0118] Orange Alert: Triggering conditions are: partial function failure (such as blockage of a single screening channel or sensor failure); the response actions are to automatically block the abnormal channel, activate the backup path (such as activating the second screening unit), notify the technical supervisor via SMS, and record the abnormal location (three-dimensional coordinates) simultaneously.

[0119] Red Emergency Alert: Triggered by systemic risk (hazardous waste entering a normal passage, equipment fire, excessive vibration triggering structural risk); Response actions are: Equipment shutdown: Emergency braking of conveyor belt and crusher unit (response time < 1s); Isolation and treatment: Abnormal waste is transferred to a seepage-proof emergency storage bin, and the bin door is automatically closed and sealed; Multi-level notification: Call the project manager (3 retries), send an email to the emergency command center, and upload blockchain evidence (including time, location, and anomaly type).

[0120] 4. Emergency Response Coordination: Digital twin-assisted decision-making: After an alarm is triggered, the digital twin platform generates a 3D model of the anomaly point, marks the waste type and equipment status, and recommends treatment options (such as "manually sorting oversized pieces" or "contacting a hazardous waste treatment agency"). Manual intervention: The operations and maintenance team handles the situation according to the recommended solutions (such as removing the wrapped steel bars or replacing the faulty sensor), and the results are entered into the system; Resumption of operation verification: After confirming that the anomaly has been eliminated, the technical supervisor unlocks the equipment through the system and starts a no-load trial run (5 minutes). If there are no anomalies, normal production is resumed.

[0121] 5. Recording and Optimization: All process data (response time, handling measures, results) are written into the blockchain as a basis for environmental compliance traceability; abnormal data are analyzed weekly to optimize the identification model (such as supplementing image samples of new types of polluting waste) and reduce the recurrence rate of similar abnormalities (target is a 10% reduction per month).

[0122] Furthermore, modular recycling equipment recycles different waste materials and is equipped with quickly switchable mobile processing units (such as vehicle-mounted asphalt recycling machines and mobile concrete crushing stations). Based on the adaptation results, the corresponding module is automatically called to achieve an on-site recycling mode of "immediate processing after sorting," including the following steps: 1. Sorting and waste material classification and docking: The dynamic sorting unit transports the sorted waste materials (steel, asphalt, concrete blocks, earthwork, etc.) to the modular processing area via a dedicated conveyor belt. Each category corresponds to an independent inlet (e.g., the steel inlet is connected to the metal recycling module, and the asphalt inlet is connected to the asphalt recycling module). The radio frequency identification (RFID) tag reader at the feed inlet automatically identifies the waste category information and transmits it synchronously to the PLC control center, triggering the start command of the corresponding mobile processing unit.

[0123] 2. Intelligent scheduling of mobile processing units: The control center allocates mobile units through an IoT scheduling platform based on the real-time waste flow rate (e.g., 50 tons of asphalt per hour, 80 tons of concrete blocks per hour) and equipment load status. Call in a vehicle-mounted asphalt recycling machine (equipped with a heated drum and a recycling agent spraying system) to connect to the asphalt material channel; Dispatch mobile concrete crushing plants (including jaw crushers and impact crushers) to the concrete block channel; The metal baling and pressing machine is connected to the steel channel, and the small soil remediation vehicle is connected to the earthwork channel. The mobile unit automatically drives to the designated work station through the automatic driving system (centimeter-level positioning) and completes the mechanical connection with the feed port (connection error ≤5cm), without the need for manual intervention.

[0124] 3. Type-based regeneration treatment: 3.1 Asphalt Recycling: The vehicle-mounted recycling machine feeds asphalt blocks into a heated drum (temperature controlled at 160-180℃), and monitors the material temperature in real time using an infrared thermometer to prevent overheating and aging. Add a composite recycling agent (the dosage is automatically adjusted according to the aging degree of the asphalt; for example, increase the dosage by 3% when the penetration is <40mm), mix for 120 seconds in a twin-shaft mixer to form a recycled asphalt mixture; output to an insulated storage silo for use in the paving of the road surface subbase.

[0125] 3.2 Concrete block recycling: The mobile crushing station first crushes concrete blocks >50mm to 20-50mm using a jaw crusher, and then crushes them a second time to 5-20mm using an impact crusher. The vibrating screen is used for grading and screening (5mm / 20mm aperture). The separated coarse aggregate (20-50mm) is used for roadbed filling, while the fine aggregate (5-20mm) enters the sand washing machine to remove surface dust (mud content reduced to <3%). The clean aggregate is mixed with new cement and water in a certain proportion (recycled aggregate replacement rate 60%) to generate recycled concrete, which is used for casting small precast components.

[0126] 3.3 Steel Recycling: Metal baling and briquetting machines compress loose steel bars and steel components into rectangular blocks of 1m × 0.5m × 0.3m (density ≥ 3t / m³).3 ), reducing transportation volume; The straightening module (hydraulic straightening accuracy ±2°) can be used for bent reinforcing bars and can be directly applied to temporary support structures. Severely corroded steel is transported to a medium-frequency induction heating furnace (temperature 1200℃), melted, and cast into steel ingots for recycling.

[0127] 3.4 Soil and rock regeneration: The soil remediation vehicle adjusts the soil pH value (to 6.5-7.5) and removes acidic pollutants through a chemical spraying system (such as adding quicklime); Adding straw fiber (3% admixture) improves soil structure and increases compaction (≥93%). The treated soil is used for backfilling green belts or leveling temporary sites.

[0128] 4. Real-time monitoring of regeneration quality: Each mobile unit is equipped with an online detection device. The stability (≥8kN) and flow value (20-40mm) of the asphalt mixture were tested in real time using a Marshall stability meter. The workability of recycled concrete is monitored by a slump meter (180±20mm), and compressive strength test blocks are automatically formed (one set is made for every 50 tons). The test data is uploaded to the blockchain system in real time, and unqualified products are automatically diverted to the rework channel (such as asphalt material being reheated and mixed).

[0129] 5. Rapid equipment switching and collaborative operation: When the waste material type is changed (e.g., from concrete blocks to asphalt material), the control center issues a switching command, and the mobile unit completes the separation of pipelines and circuits within 10 minutes through the quick-release interface (hydraulic latch); The "multi-unit parallel operation" mode is adopted (such as 1 asphalt recycling machine + 2 concrete crushing stations operating at the same time). The number of equipment is dynamically adjusted according to the waste generation rhythm to ensure that the processing capacity matches the feed rate (fluctuation ≤10%).

[0130] In this technical solution, through modular design, the system achieves seamless integration of recycling and on-site construction, increasing the on-site utilization rate of recycled materials to over 85%, reducing transportation costs by 60%, and lowering carbon emissions by 40%.

[0131] This invention provides a collaborative optimization method for intelligent sorting and recycling of highway construction waste, comprising the following steps: S1. The data collection module collects data and images of highway construction waste, and labels the data and images as reference samples. S2, the multimodal recognition unit identifies the types of highway construction waste through various methods; S3, the dynamic sorting unit sorts different waste materials; S4. The recycling and processing unit constructs a dynamic matching engine for waste characteristics, process parameters, and product standards to improve the performance stability of recycled materials and generate recycling and processing solutions for different wastes.

[0132] The S5 modular recycling equipment recycles different waste materials and is equipped with a mobile processing unit that can be quickly switched. It automatically calls the corresponding module according to the adaptation result, realizing the on-site recycling mode of "immediate processing after sorting" and turning waste into treasure.

[0133] S6. The anomaly identification unit identifies anomalies in the sorting and recycling process through different methods. S7, the supply and demand collaborative scheduling unit, is based on digital twin-based global resource optimization and allocation to achieve spatiotemporal matching of generation, processing, and consumption.

[0134] S8, the full-process simulation optimization unit, predicts system bottlenecks through virtual simulation, enabling full-cycle management of pre-event optimization, in-event adjustment, and post-event review.

[0135] S9, the blockchain quality control unit constructs an immutable traceability chain for the entire life cycle of recycled materials, ensuring project quality and market trust.

[0136] S10. When an abnormal situation is detected, the alarm unit will issue an alarm in a timely manner.

[0137] This invention maximizes resource utilization efficiency: The multimodal identification unit achieves accurate identification of waste types through multi-sensor data fusion, and combined with the "value coefficient priority scheduling" logic, high-value waste is given priority to enter the fast processing channel, increasing the recycling rate to over 85%; the "immediate processing after sorting" mode of the modular recycling equipment reduces transportation links, lowers material loss compared to traditional off-site processing modes, and improves the recycling rate of scarce resources such as steel and asphalt. The performance stability of recycled materials is improved: The dynamic matching engine of the recycling processing unit, which combines "waste characteristics - process parameters - product standards" with a random forest algorithm prediction model, can adjust key parameters such as asphalt recycling agent dosage and concrete mix proportions in advance; the blockchain quality control unit enables full life-cycle traceability, improving the compliance rate of recycled materials, meeting the requirements for use in highway engineering structural components, and breaking the industry prejudice that "recycled materials have poor performance." The supply and demand coordination scheduling unit, based on digital twins, enables the spatiotemporal matching of waste generation and processing capacity, thus avoiding equipment idleness or overload. The full-process simulation optimization unit, through virtual bottleneck prediction, controls the fluctuation of system processing efficiency to within 10%. The modular equipment quick-disassembly and parallel operation mode increases the daily waste processing capacity to 1.5 times that of traditional fixed production lines, while reducing overall operating costs.

[0138] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A collaborative optimization method for intelligent sorting and recycling of highway construction waste, characterized in that, Includes the following steps: S1, the data collection module collects data and images of highway construction waste; S2, the multimodal recognition unit identifies the types of highway construction waste through various methods; S3, the dynamic sorting unit sorts different waste materials; S4. The recycling and processing unit constructs a dynamic matching engine for waste characteristics, process parameters, and product standards to generate recycling and processing solutions for different wastes. S5, a modular recycling equipment, recycles different waste materials; S6. The anomaly identification unit identifies abnormal situations during the sorting and recycling process; S7. The supply and demand collaborative scheduling unit optimizes resource allocation across the entire domain based on digital twins, achieving spatiotemporal matching of generation, processing, and consumption. S8, the full-process simulation optimization unit predicts system bottlenecks through virtual simulation; S9. The blockchain quality control unit constructs an immutable traceability chain for the entire life cycle of recycled materials; S10. When an abnormal situation is detected, the alarm unit will issue an alarm in a timely manner.

2. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S2 includes the following steps: S21. Deployment of sensing devices: Three sets of distributed sensing devices are set up at the unloading port, the starting point of the conveyor belt, and the middle section of the conveyor belt, respectively. S22, Multi-source data acquisition; S23. Local data preprocessing: Preprocessing the acquired image, spectral, and radar data; S24. Federated Learning Local Training: Based on the ResNet-50 architecture, the recognition model is trained using preprocessed data. S25, Global Model Aggregation: The PLC control center receives parameters from each model, aggregates them hourly to generate a global recognition model, and dynamically optimizes feature weights. S26. Multimodal data fusion: Image, spectral, and radar data are fused through a federated learning framework, and the recognition results are cross-validated. S27. Waste type determination: Output waste type and confidence level based on fused data; S28. Priority scheduling application: Based on the waste value coefficient, the identification results are transmitted to the dynamic sorting unit, and high-value waste is preferentially allocated to the fast processing channel.

3. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S3 includes the following steps: S31. Magnetic Separation: A strong magnetic roller is installed at the beginning of the conveyor belt, and a wear-resistant rubber scraper is set below the roller; steel parts are separated by magnetic force. S32. Particle size classification: Separating concrete from earthwork, using a three-layer vibrating screen, combined with multi-spectral particle size prediction data, and dynamically adjusting the vibration frequency; S33, Airflow separation: Separating asphalt material from concrete; S34. Irregularly Shaped Waste Capture: Separate templates from debris; a 6-axis robotic arm with 3D vision guidance is installed at the end of the conveyor belt to identify irregularly shaped waste. S35. Chemical property auxiliary verification: Set up LIBS detection points at the end of each sorting channel to analyze the elemental composition of waste materials in real time; auxiliary verification.

4. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S4 includes the following steps: S41. Build a dynamic matching engine; S42, Application of AI performance prediction model; S43. Different types of waste recycling treatment; steel waste, asphalt pavement milling material, concrete waste, earthwork and related waste, and special waste are recycled in different ways.

5. The method for intelligent sorting and recycling of highway construction waste according to claim 4, characterized in that: Step S42 includes the following steps: S42.1 Training data preprocessing: Collect historical regeneration process data and clean the data; S42.2 Model Training and Optimization: A prediction model is constructed using the random forest algorithm, with waste characteristics and process parameters as input variables and product performance as output variables. Through ensemble learning of multiple decision trees, a performance prediction model is generated. S42.3 Predictive Application: Input the characteristics of the current batch of waste and the proposed process parameters in real time, and the model will output the product performance prediction results; if the prediction results are lower than the standard threshold, the model will automatically push parameter adjustment suggestions.

6. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S7 includes the following steps: S71. Digital Twin Model Construction: Integrate the BIM model and GIS geographic information of the construction area to construct a three-dimensional digital twin scene that includes waste sources, treatment stations, and demand points; S72, Real-time data acquisition and synchronization; S73. Multi-dimensional data fusion: Classify and integrate the collected data to establish a related database; S74, Intelligent supply and demand matching; S75, Dynamic Scheduling Plan Generation: Automatically generates plans for the entire processing, transportation, and delivery chain; S76. Real-time adjustment and emergency response; S77, Cross-project resource collaboration; S78. Optimization of scheduling performance evaluation.

7. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S8 includes the following steps: S81. Pre-emptive optimization: Predict bottlenecks through virtual simulation; includes building digital twin models, simulation and bottleneck identification, multi-scenario parameter configuration, and optimization scheme generation. S82. In-process adjustment: Real-time dynamic optimization; including real-time data access, deviation warning and root cause analysis, dynamic adjustment instruction generation and effect verification optimization; S83. Post-event review: full-cycle summary and iteration.

8. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S6 includes the following steps: S61. Real-time monitoring starts: Collects data in real time to form a dynamic monitoring data stream; S62. Abnormal scenario determination: including determination of waste material characteristics exceeding standards and determination of equipment and process failures; S63, Multi-level alarm triggering; S64. Emergency Response Coordination; S65. Recording and Optimization: All process data is written to the blockchain as a basis for environmental compliance traceability; abnormal data is analyzed weekly to optimize the identification model.

9. The method for intelligent sorting and recycling of highway construction waste according to claim 1, characterized in that: Step S5 includes the following steps: S51. Sorted waste material classification and docking: The sorted waste material is transported to the modular processing area via a dedicated conveyor belt, with each category having an independent inlet. S52. Intelligent scheduling of mobile processing units: Based on real-time waste flow and equipment load status, mobile units are allocated through an IoT scheduling platform. S53. Type-based recycling: including asphalt recycling, concrete block recycling, steel recycling, and earthwork recycling; S54. Real-time monitoring of regeneration quality: Each mobile unit is equipped with an online detection device to monitor quality. S55. Rapid equipment switching and collaborative operation.

10. A collaborative optimization system for intelligent sorting and recycling of highway construction waste, comprising: The system comprises a PLC control center, a data collection module, a multimodal recognition unit, a dynamic sorting unit, a waste utilization processing unit, a blockchain quality control unit, a supply and demand collaborative scheduling unit, modular recycling equipment, a full-process simulation optimization unit, an anomaly identification unit, and an alarm unit; its features include: Data collection module: Collects data and images of highway construction waste, and annotates the data and images; Multimodal recognition unit: Identifies the types of highway construction waste through multiple different methods; Dynamic sorting unit: sorts different types of waste materials; Recycling and processing unit: Construct a dynamic matching engine for waste characteristics, process parameters, and product standards to improve the performance stability of recycled materials and generate recycling and processing solutions for different wastes; Supply and demand coordinated scheduling unit: Based on digital twin, global resource optimization and allocation is achieved to realize spatiotemporal matching of generation, processing and consumption; Full-process simulation optimization unit: Predicts system bottlenecks through virtual simulation; Modular recycling equipment: recycles different types of waste materials; Blockchain-based quality control unit: Constructing an immutable traceability chain for the entire lifecycle of recycled materials; Anomaly identification unit: Identifies abnormal situations during sorting and recycling processes using different methods; Alarm unit: includes an alarm device that promptly issues an alert when an abnormal situation is detected; PLC Control Center: Network connected to data collection module, multimodal recognition unit, dynamic sorting unit, in-process utilization unit, blockchain quality control unit, supply and demand collaborative scheduling unit, modular regeneration equipment, full-process simulation optimization unit, anomaly recognition unit, and alarm unit.

Images