A dynamic programming method for construction waste resource utilization path
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI WENDA INFORMATION ENG COLLEGE
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-24
Smart Images
Figure CN122452891A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of construction waste treatment technology, specifically a dynamic planning method for the resource recovery path of construction waste. Background Technology
[0002] In traditional construction waste recycling processes, material logistics and physical processing are often managed as two separate systems, making it difficult to achieve globally optimal planning across physical states and spatial locations. In practice, the composition and quantity of construction waste are prone to unpredictable fluctuations. Existing scheduling platforms typically respond by directly recalculating routes when receiving these external disturbances. This approach fails to consider the current physical and spatiotemporal constraints of vehicles and the progress of logistics execution, easily leading to the issuance of change orders even after transport vehicles have passed critical intersections or other irreversible nodes, resulting in scheduling failures or frequent reconfigurations and oscillations during physical execution.
[0003] Furthermore, existing macro-logistics scheduling methods typically only provide path-level guidance and cannot directly translate macro-scheduling results into specific control parameters for the underlying processing equipment. This prevents the front-end mobile crushing equipment from adjusting its mechanical operation based on real-time scheduling changes, resulting in a mismatch between the actual crushed particle size of the output material and the process requirements of the downstream recycled building materials, thus reducing the collaborative efficiency of construction waste resource recovery. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a dynamic planning method for the resource utilization path of construction waste, which solves the problems of existing construction waste scheduling methods that separate logistics and processing links, have poor anti-disturbance capabilities, cause scheduling instructions to easily fail, and cannot directly control the physical parameters of underlying equipment.
[0005] To achieve the above objectives, the present invention provides a dynamic planning method for the resource utilization path of construction waste, comprising the following steps: Obtain bill of materials data to set baseline components, and construct a directed graph network by combining multi-source heterogeneous data; the nodes of the directed graph network are three-dimensional extended nodes containing geospatial coordinates, time windows and physical states, and the connecting edges include transportation edges and processing edges; The nodes corresponding to the waste locations are taken as source nodes, and the nodes corresponding to the recycled building material application locations are taken as sink nodes. The forward dynamic programming algorithm is applied to calculate in the directed graph network to obtain the initial optimal path instructions. Monitor environmental disturbance signals that include actual material composition deviation signals; confirm valid trigger events when the actual material composition deviation signal exceeds the tolerance boundary within the spatiotemporal accumulation confirmation window and the transport vehicle has not passed the logistics irreversible execution anchor point. Based on the effective trigger event blocking the initial optimal path instruction, the three-dimensional extended node corresponding to the current geospatial coordinates is used as the starting calculation node to update the directed graph network, generate and issue a new scheduling instruction, extract the maximum material particle size critical value in the new scheduling instruction and convert it into the discharge port setting gap control quantity of the mobile crushing equipment, and issue it to the control terminal to adjust the initial crushing particle size of the mobile crushing equipment.
[0006] In constructing a directed graph network using multi-source heterogeneous data, the system receives multi-source heterogeneous data uploaded from the front-end sensing system and the capacity management system, and performs data cleaning and temporal alignment operations on the data. Abnormal jump values in the material weight sequence are removed, and a temporal alignment algorithm based on dynamic time warping is constructed. The timestamp of a vehicle entering the monitoring area is used as the alignment benchmark, and linear interpolation compensation is performed on missing data frames. The weight sequence and image features of the same batch of materials are mapped and bound to the location information of the transport vehicles using a high-dimensional mapping. Based on the cleaned and aligned multi-source heterogeneous data and benchmark components, the node topology of the directed graph network is instantiated, generating multiple three-dimensional extended nodes. Connection edges are generated between the various three-dimensional extended nodes. Transportation edges connect two three-dimensional extended nodes with the same physical state but different geographic coordinates, and processing edges connect two three-dimensional extended nodes with the same geographic coordinates but different physical states.
[0007] When generating connection edges between various 3D extended nodes, a maximum dissociation threshold parameter is set for the processing edges; the acquired visual fragmentation feature values are weighted and superimposed with the proportion of the historical crushing energy consumption of the material in the previous process to the rated baseline energy consumption of the target equipment to calculate the actual dissociation coefficient; if the actual dissociation coefficient is lower than the maximum dissociation threshold parameter, the processing edges are generated and retained in the directed graph network; if the actual dissociation coefficient is higher than the maximum dissociation threshold parameter, topology pruning is performed in the directed graph network and the corresponding processing edges are disconnected.
[0008] When applying the forward dynamic programming algorithm to obtain the initial optimal path instruction, a carbon value collaborative cost assessment model is constructed to obtain the logistics transportation cost assessment value, physical processing cost assessment value, carbon emission converted cost assessment value, and market transaction value generated by each connecting edge in the feasible path. First weight coefficients, second weight coefficients, third weight coefficients, and fourth weight coefficients are assigned to the logistics transportation cost assessment value, physical processing cost assessment value, carbon emission converted cost assessment value, and market transaction value, respectively. A multidimensional historical cost index matrix is obtained and standardized. The variance of each column of characteristic indicators is detected, and a perturbation bias term is added if the variance is lower than the tolerance constant. The covariance matrix of the standardized matrix is calculated and eigenvalue decomposition is performed to extract principal components with a cumulative variance contribution rate greater than a preset threshold. The variance contribution rates of each principal component are weighted and normalized to derive the first weight coefficient, second weight coefficient, third weight coefficient, and fourth weight coefficient. Based on this, the product of the logistics transportation cost assessment value and the first weight coefficient, the product of the physical processing cost assessment value and the second weight coefficient, and the product of the carbon emission converted cost assessment value and the third weight coefficient are added together to obtain the superimposed sum. The product of the market transaction value and the fourth weight coefficient is subtracted from the superimposed sum to obtain the total agency value. The path that obtains the minimum total agency value is taken as the initial optimal path instruction.
[0009] In the step of monitoring environmental disturbance signals, the system receives material actual composition deviation signals uploaded by the visual acquisition device, local capacity fluctuation signals of the resource recovery plant uploaded by the capacity management system, and market demand change signals of recycled building materials synchronized by the external interface. The system determines whether the material actual composition deviation signal exceeds the tolerance boundary within the spatiotemporal cumulative confirmation window using the following methods: Calculate the cumulative deviation index within the spatiotemporal cumulative confirmation window by multiplying the difference between the actual impurity percentage and the baseline impurity percentage, the actual material quality measurement value aligned with the image timestamp, and the environmental perception confidence level; sum the multiplication results of each sampling frame within the spatiotemporal cumulative confirmation window period to obtain the cumulative deviation index; derive the environmental perception confidence level by adding the equipment's basic noise bias constant to the environmental particulate matter concentration value as the denominator; and determine whether the material actual composition deviation signal exceeds the tolerance boundary if the cumulative deviation index exceeds the tolerance boundary.
[0010] In the step where the transport vehicle has not passed the irreversible execution anchor point, the real-time latitude and longitude coordinates of the transport vehicle are obtained and mapped to the high-precision road network model; the actual distance traveled by the vehicle on the current transport edge is calculated, and the ratio of the actual distance traveled to the total mileage of the current transport edge is determined as the logistics execution progress parameter; if the logistics execution progress parameter is less than the preset execution anchor point threshold, it is confirmed that the vehicle has not passed the preset irreversible execution anchor point; if the logistics execution progress parameter is greater than or equal to the execution anchor point threshold, environmental disturbance signals are shielded and the vehicle continues to drive towards the original target node.
[0011] Upon confirming a valid trigger event and updating the directed graph network, a transition hold signal is sent to the controlled machinery to maintain its current safe operating posture; the geospatial coordinates of the transport vehicle carrying the corresponding batch of materials are locked, and the real-time 3D extended node corresponding to the geospatial coordinates is used as the starting calculation node; based on the latest road congestion index fed back by the front-end perception system and the capacity load parameters of each resource recovery plant, the impedance weights of the unexecuted connection edges in the directed graph network are rewritten.
[0012] When converting the discharge port clearance setting control value for a mobile crushing equipment, the following parameters are obtained: crushing chamber type correction coefficient, material brittleness damping coefficient, estimated average Mohs hardness of the current batch of material, and jaw plate physical wear compensation margin. The crushing chamber type correction coefficient, raised to the power of the maximum material particle size critical value, is multiplied by the exponent of the natural constant, where the exponent is the product of the material brittleness damping coefficient and the estimated average Mohs hardness. The jaw plate physical wear compensation margin is then subtracted from the multiplication result to obtain the discharge port clearance setting control value.
[0013] This invention provides a dynamic planning method for the resource recovery path of construction waste. It has the following beneficial effects: 1. This invention constructs a three-dimensional extended node directed graph network that includes geographic spatial coordinates, time windows, and physical states, and integrates transportation edges and processing edges into the network model for calculation using a forward dynamic programming algorithm. This mechanism jointly models the logistics transportation and circulation process of construction waste with the physical processing process, avoiding the local suboptimal problem caused by the independence of logistics scheduling and processing links in traditional scheduling methods, and improving the global optimization capability of resource recovery paths.
[0014] 2. This invention employs a spatiotemporal cumulative confirmation window and a logistics irreversible execution anchor point mechanism to effectively determine the triggering events of environmental disturbance signals. By performing spatiotemporal cumulative calculations on the actual material composition deviation signals, it can filter out short-term data jumps and noise interference. At the same time, combined with the current logistics execution progress of the vehicle, it prevents invalid rescheduling from being triggered after irreversible nodes, thereby avoiding the waste of computing resources caused by frequent system recalculations and the physical execution oscillation of scheduling instructions.
[0015] 3. This invention extracts the critical value of the maximum material particle size in the new scheduling instruction and combines it with physical parameters such as the crushing chamber shape correction coefficient of the mobile crushing equipment, the material brittle damping coefficient, and the physical wear compensation margin of the jaw plate to directly calculate the set gap control quantity of the discharge port. This process establishes a feedforward control mechanism for the macro scheduling path planning on the operating status of the underlying equipment, so that the scheduling result of the algorithm can be directly converted into the action command of the mechanical control end, improving the matching accuracy between the initial crushed particle size of construction waste and the subsequent application requirements of recycled building materials. Attached Figure Description
[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system architecture diagram of the present invention; Figure 3 This is a comparison chart of the breakdown costs and benefits of the present invention; Figure 4 This is a final comparison chart of the total value of the present invention. Detailed Implementation
[0017] The technical solutions in 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.
[0018] See attached document Figure 2 In a specific embodiment of the present invention, a dynamic planning system for the resource utilization path of construction waste is provided, the system specifically comprising: The front-end sensing system is deployed at the demolition site and along the logistics route. The front-end sensing system includes a weighbridge that collects material weight data, a visual acquisition device that acquires material surface texture for composition determination, a mechanical status sensor that collects data on the gap between the discharge ports of the mobile crushing equipment, and a vehicle positioning terminal configured on the transport vehicles.
[0019] The cloud-based scheduling server serves as the central hub for data processing and logic control. In this embodiment, the cloud-based scheduling server includes a graph network construction module, a path optimization module, an event filtering module, and a status feedback module. Each module acquires data uploaded by the front-end sensing system through the industrial communication network and performs model calculations.
[0020] The terminal execution system receives and responds to operation instructions issued by the cloud-based dispatch server. Specifically, the terminal execution system includes an on-board receiving terminal that displays the dispatch route to the driver, a mechanical control terminal that adjusts the operation parameters of the dismantling site equipment, and a capacity management system that uploads queuing and load status to the cloud-based dispatch server in real time.
[0021] See attached document Figure 1 This invention provides a dynamic planning method for the resource utilization path of construction waste, specifically including the following steps: S10, the graph network construction module obtains the estimated dismantling bill of materials data for building engineering to set the baseline components, and combines the data uploaded in real time by the front-end perception system and the capacity management system to construct a directed graph network; the nodes of the directed graph network are defined as three-dimensional extended nodes, and the internal parameters include geographic spatial coordinates, time windows and physical states of materials; the connecting edges of the directed graph network are divided into transportation edges that change the geographic spatial location of materials and processing edges that change the physical state attributes of materials. S20, the path optimization module uses the three-dimensional extended nodes corresponding to the spatial location of construction waste as source nodes and the three-dimensional extended nodes corresponding to the application location of recycled building materials as sink nodes. It applies the forward dynamic programming algorithm to perform path optimization calculation in the directed graph network. In this embodiment, the optimization calculation aims to minimize the total cost value of the path. The total cost value is calculated by weighting and positively superimposing the logistics transportation cost assessment value generated by each transportation edge, the physical processing cost assessment value generated by each processing edge, and the carbon emission conversion cost assessment value generated by each edge operation in the path sequence. Then, the estimated market transaction value of the material arriving at the sink node in a specific physical state is subtracted from the superimposed sum to generate the initial optimal path instruction. The instruction includes the cross-plant flow route and the processing technology flow route. S30, during physical transportation and processing, the event filtering module monitors the actual material composition deviation signal, the capacity fluctuation signal of the resource recovery plant, and the demand change signal of the recycled building materials market, and sets the spatiotemporal cumulative confirmation window parameter and the logistics irreversible execution anchor point parameter; further, when the material composition deviation signal continuously detected by the visual acquisition device exceeds the set tolerance boundary range within the set spatiotemporal cumulative confirmation window, and the transport vehicle carrying the corresponding batch of materials has not passed the preset logistics irreversible execution anchor point according to the data fed back by the vehicle positioning terminal, the event filtering module determines that the current signal is a valid triggering event; S40, after a valid trigger event is generated, the status feedback module blocks the execution of the original initial optimal path instruction, updates the directed graph network parameters with the real-time three-dimensional extended node where the material is currently located as the new starting calculation node, recalculates and generates a new scheduling instruction, and sends it to the corresponding vehicle-mounted receiving terminal to change the logistics endpoint or adjust the processing procedure; at the same time, the status feedback module extracts the initial physical state boundary parameters required for the material to enter the next stage of processing in the new scheduling instruction, converts the initial physical state boundary parameters into the equipment's underlying control quantity, and sends it to the mechanical control terminal in the source area through reverse communication; the mechanical control terminal adjusts the source material digging and sorting accuracy or adjusts the initial crushing particle size of the mobile device according to the equipment's underlying control quantity.
[0022] See attached document Figure 1 The dynamic planning method for resource recovery of construction waste provided by this invention includes the following specific implementation steps for step S10: In step S101, in this embodiment, the graph network construction module obtains the estimated bill of materials for the demolition of the building project and establishes initial baseline composition parameters for the construction waste accordingly. As a preferred method, the graph network construction module parses building information model files or receives manually entered project cost estimates to extract the volumetric characteristics and material distribution information of the building to be demolished. The specific content of the aforementioned initial baseline composition parameters includes the estimated mass proportions and spatial distribution of concrete, steel bars, bricks, and slag. Based on the extracted physical attributes of the building entities, a priori data base is constructed, enabling the system to provide a basic material attribute reference for the site selection and initial path planning of the resource recovery plant in the early stages of demolition operations.
[0023] In step S102, the graph network construction module receives multi-source heterogeneous data uploaded in real time from the front-end sensing system and the capacity management system, and then performs data cleaning and time-series alignment operations. The multi-source heterogeneous data specifically includes material weight sequences uploaded by the weighbridge equipment, material surface texture image features uploaded by the visual acquisition equipment, and the idle rate and number of queuing vehicles in the resource recovery plant uploaded by the capacity management system. It should be noted that, to address the asynchronous issues caused by inconsistent sampling frequencies of multiple sensors and network latency in real industrial environments, during the data cleaning process, the graph network construction module not only removes abnormal jump values in the weight sequences but also constructs a time-series alignment algorithm based on dynamic time warping. The graph network construction module uses the timestamp of a vehicle entering a specific monitoring area as the alignment benchmark, performs linear interpolation compensation for missing data frames, and then performs high-dimensional mapping and binding of the weight sequences and image features of the same batch of materials with the positioning information of the transport vehicles.
[0024] To address the issue of feature extraction from visual acquisition devices, simply extracting features is insufficient to support subsequent component calculations. This embodiment constructs and deploys a lightweight convolutional neural network model for component determination. The model's network structure, from bottom to top, includes an input layer, alternating convolutional and max-pooling layers, a fully connected layer, and a Softmax classification output layer. The input data consists of RGB three-channel images of the surface of materials at a set resolution. Before entering the network, the image data undergoes preprocessing for illumination equalization and Gaussian denoising. The network classification output layer outputs the probability distribution of the proportion of brick and concrete in the waste within the current field of view. During model construction and training, the system pre-collects a large number of on-site material images with real component classification labels to construct a sample set. Cross-entropy is used as the loss function, and the network weights are updated through backpropagation until the model's accuracy on the validation set converges.
[0025] Step S103: Based on the cleaned and aligned multi-source heterogeneous data and initial baseline component parameters, the graph network construction module instantiates the node topology of the directed graph network. In this embodiment, the nodes of the directed graph network are configured as three-dimensional extended nodes. The data structure of this three-dimensional extended node is divided into a geospatial coordinate dimension, a time window dimension, and a material physical state dimension. The geospatial coordinate dimension represents the latitude and longitude information of the demolition site, the resource recovery plant, and the final building material demand location. The time window dimension limits the earliest start time and the latest end time that materials are allowed to reach specific geospatial coordinates; its specific application scenarios mainly include avoiding urban truck traffic restrictions and matching factory scheduling plans. Furthermore, the material physical state dimension represents the macroscopic morphological characteristics of construction waste at different processing stages. The specific manifestations of this physical state dimension cover the original mixed slag state, the state of primary coarsely crushed concrete blocks, and the state of recycled aggregate after grading and screening. Upgrading the spatial nodes to three-dimensional extended nodes that include time and physical state allows for a complete and multi-dimensional mapping of the spatiotemporal displacement and morphological evolution of construction waste during the resource recovery process.
[0026] In step S104, the graph network construction module generates directed graph network connections between each 3D extended node. These connections consist of transportation edges that change the geographic location of the material and processing edges that change the physical state properties of the material. Transportation edges connect two 3D extended nodes with the same physical state dimension but different geographic coordinate dimensions. Considering the complexity of actual road conditions, the graph network construction module calculates the travel time impedance of transportation edges based on the driving trajectory fed back by the vehicle positioning terminal and the road network congestion index. Processing edges connect two 3D extended nodes with the same geographic coordinate dimension but different physical state dimensions, representing the mechanical process of waste transforming from a preceding physical state to a subsequent physical state within the same resource recovery plant.
[0027] To ensure the graph network model accurately reflects the actual processing capacity of the physical equipment and prevents the system from allocating materials to production lines unable to handle such impurities, this embodiment configures a maximum dissociation threshold parameter for the processing edge. The maximum dissociation threshold parameter limits the upper limit of the material adhesion complexity that a specific physical sorting device can handle. The maximum dissociation threshold parameter is set to a range of [0,1], specifically determined by the critical value safety redundancy coefficient for jamming or shutdown failures in the target equipment's historical operating data.
[0028] Let the physical state of the source 3D extended node be... The physical state of the target 3D extended node is The processing edge between the two is denoted as The graph network construction module calculates the actual dissociation coefficient of the current material in real time based on the data uploaded by the front-end sensing system. To avoid the failure of a single sensor due to environmental interference such as dust obstruction, the system comprehensively considers the visual morphological characteristics of the material and the historical processing equipment load for weighted judgment. Specifically, the system weights and superimposes the visual fragmentation feature values extracted by the visual acquisition device with the ratio of the historical crushing energy consumption of the corresponding material in the previous process to the rated baseline energy consumption of the target equipment, thereby calculating the actual dissociation coefficient. The sum of the weighted coefficients is set to 1. To avoid system crashes due to division by zero at the underlying algorithm logic, the rated baseline energy consumption of the target equipment is determined by the equipment's factory calibration parameters and is physically constrained to remain a value greater than 0.
[0029] When the actual dissociation coefficient of the material is lower than the maximum dissociation threshold parameter corresponding to the processing edge, the graph network construction module generates and retains the processing edge in the graph topology. If the actual dissociation coefficient of the material is higher than the maximum dissociation threshold parameter, the system directly performs topology pruning in the graph network, disconnecting the corresponding processing edge topology association. This constraint control mechanism eliminates process flow routes that exceed the equipment's processing capacity, ensuring effective matching between the underlying directed graph network structure and the actual engineering environment.
[0030] See attached document Figure 1 The dynamic planning method for resource recovery of construction waste provided by this invention includes the following specific implementation steps for step S20: In step S201, in this embodiment, the path optimization module determines the set of source nodes and sink nodes required to execute the forward search algorithm in the constructed directed graph network. The geospatial coordinate dimension of the source nodes is set to the latitude and longitude of the building demolition site, the time window dimension corresponds to the estimated waste loading time, and the physical state dimension is initialized to the state of the raw mixed waste without prior processing. The sink nodes correspond to the physical mapping of the final application of recycled building materials, and their geospatial coordinates point to the specific concrete mixing plant or roadbed backfilling project site. Based on actual engineering needs, since the physical form required for the same batch of waste varies in different application scenarios, the physical state dimension of the sink nodes is set to specific gradation parameters that meet the target engineering standards, such as the state of recycled coarse aggregate with a particle size between 5 and 10 mm. Based on the node boundary conditions set by the above endpoint constraints, a clear calculation endpoint is set for the subsequent system to traverse each potential process and logistics combination.
[0031] Step S202: The path optimization module constructs a carbon-value collaborative cost assessment model to evaluate the comprehensive benefits of multi-segment combined paths. While traditional logistics planning focuses solely on minimizing transportation distance or freight costs, this embodiment introduces a comprehensive assessment of physical energy consumption and carbon emission reduction. To achieve quantitative comparison of cross-domain indicators at the algorithm's underlying level, the system constructs a global single-objective function encompassing economic costs, environmental penalties, and market benefits. As a preferred approach, for any set of feasible paths in a directed graph network, the calculation logic for its total cost value is implemented according to the following formula: ; In the above formula, Indicates the selected full-process combination path Total agency value; Representing a path Any connecting edge contained therein. This represents the estimated logistics transportation cost generated by the vehicle on the corresponding connecting edge. The system obtains this value by calling a preset vehicle fuel consumption rate benchmark table and combining it with the travel distance through textual calculations; when the connecting edge is a processing edge used to change the state of materials, The value is 0. This represents the estimated physical processing cost of the material at the corresponding connecting edge. This value is derived from the product of the rated power of the corresponding resource recovery plant's processing equipment and the estimated processing time. When the connecting edge is a transport edge used to change the material's position... The value is 0.
[0032] This represents the estimated absolute value of carbon dioxide emissions corresponding to the flow or processing of waste at the connection point. This represents the real-time standard guidance price per unit of carbon emission allowances in the external carbon trading market. The product of these two items transforms the environmental impact into an economically comparable impedance term. This indicates that waste reaches the final sink node and is transformed into the target physical state. The estimated market transaction value below.
[0033] The first, second, third, and fourth weighting coefficients represent the four cost and benefit items mentioned above, respectively. All weighting coefficients are normalized. The value range of each weighting coefficient is (0,1), and their sum is set to 1. To avoid the subjective bias of manual weighting, this embodiment uses principal component analysis (PCA) to adaptively determine the weighting coefficients based on the historical engineering data matrix. Before performing PCA, the system performs Z-score standardization on the input multidimensional historical cost index matrix. To prevent division overflow errors caused by the denominator approaching zero during standardization, the system pre-detects the variance of each column of feature indicators. When the variance of a certain indicator data is lower than the set tolerance constant, the system adds a perturbation bias term. Subsequently, the system calculates the covariance matrix of the standardized matrix and performs eigenvalue decomposition, extracting principal components with a cumulative variance contribution rate greater than a threshold. By weighted normalizing the variance contribution rates of each principal component, the system derives... to The specific numerical values are specified. This processing logic ensures that the cost function can be objectively and adaptively adjusted as the cost center of gravity of real-world engineering projects shifts.
[0034] In step S203, based on the determined start and end nodes and the carbon value collaborative cost evaluation model, the path optimization module applies a forward dynamic programming algorithm to perform a global path sequence search in the directed graph network. Considering that the time dimension in the real physical world has an irreversible unidirectional increasing characteristic, the system assigns a monotonically increasing constraint to the time attribute during the directed graph network construction stage. This unidirectional time constraint effectively reduces the dimensionality of the complex network structure into a directed acyclic graph, effectively avoiding the singularity problem in matrix operations from a theoretical perspective, and preventing the algorithm from jumping back and forth between nodes and getting stuck in a logical infinite loop.
[0035] During the specific algorithm execution, the system utilizes the Bellman optimality principle to establish state transition logic. The path optimization module starts from the source node and extends the search range layer by layer along feasible connection edges. To ensure state independence during the dynamic programming solution process, at each intermediate three-dimensional expansion node, the module accumulates all transportation costs, processing costs, and carbon emission costs incurred at that node, and stores the path with the minimum cumulative cost to achieve the current state of that node. To control the scale of computational resource consumption and prevent the state space from expanding rapidly, the system implements boundary condition pruning operations during the search. When the current cumulative time attribute of a sub-path exceeds the latest deadline window specified by the sink node, the system immediately interrupts further in-depth expansion of that branch and discards the entire array containing that sub-path.
[0036] In step S204, the path optimization module traverses all feasible paths to the sink node and outputs an initial globally optimal path sequence that minimizes the total agency value. After obtaining this optimal mathematical sequence, the path optimization module performs reverse semantic decoding on the connecting edges of the sequence. The system transforms the transportation edges in the sequence into specific license plate binding instructions, route guidance, and target unloading plant coordinates; simultaneously, it transforms the processing edges in the sequence into process scheduling plans and recommended gear settings for the corresponding resource-based plant production lines. The decoded cross-plant flow routes and processing flow routes are then distributed to the corresponding terminal control nodes, thus completing the initial overall coordination and allocation before the physical entity flow.
[0037] See attached document Figure 1 The dynamic planning method for resource recovery of construction waste provided by this invention includes the following specific implementation steps for step S30: In step S301, during the physical transportation and processing, the event filtering module continuously monitors environmental disturbance signals in the engineering execution chain. These environmental disturbance signals specifically include material composition deviation signals uploaded by the visual acquisition device, local capacity fluctuation signals from the resource recovery plant uploaded by the capacity management system, and market demand change signals for recycled building materials synchronized by external interfaces. Due to interference factors such as dust obstruction, mechanical vibration, and network latency at industrial demolition sites and along transportation routes, the front-end sensing system is prone to outputting abnormal data with instantaneous jumps. If the system directly responds to a single instantaneous abnormal data and frequently reconstructs the directed graph network, it will often cause repeated jumps in scheduling instructions and system scheduling oscillations in the physical logistics flow. To address the technical difficulties encountered in the aforementioned practical engineering, this embodiment introduces an anti-oscillation mechanism composed of dual constraints in the time and spatial domains in the underlying control logic.
[0038] Step S302: To address the issue of transient noise contamination in time-series sensing data, the event filtering module performs anti-oscillation filtering in the time domain. The event filtering module sets spatiotemporal cumulative confirmation window parameters to shield high-frequency sensing and recognition noise. Before specific implementation and judgment, the system needs to aggregate the multi-dimensional sensing data into time-series sequences to eliminate the randomness of single-point data. When the visual acquisition device continuously detects that the material composition deviates from the estimated baseline, the event filtering module does not immediately trigger scheduling changes, but instead calculates the cumulative deviation index within the set time window based on the following model: ; In the above formula, This represents the cumulative deviation index within the spatiotemporal cumulative confirmation window. This represents the total number of sampled frames within the spatiotemporal cumulative confirmation window period. Indicates the first The actual percentage of impurities in the frame is identified and output by the visual acquisition device. This indicates the percentage of baseline component impurities for the corresponding batch of materials as set in the estimated dismantling bill of materials. Indicates the relationship with the first The actual measurement of material mass, timestamped and synchronously collected by the weighbridge equipment, is a frame-by-frame visual image. This indicates the environmental perception confidence level corresponding to the current data collection environment.
[0039] As a preferred approach, the range of the environmental perception confidence score is strictly defined as (0,1], and its specific value is inversely proportional to the particulate matter concentration value uploaded by the on-site dust concentration sensor. To prevent division overflow errors caused by the denominator approaching 0 in an ideal dust-free environment (i.e., when the particulate matter concentration detection value is 0) from the underlying algorithm, the system rigidly introduces a device fundamental noise bias constant that is always greater than 0 for the corresponding concentration denominator term when calculating the confidence score.
[0040] By introducing the aforementioned multivariate weighted calculation logic, the system transforms the visually singular two-dimensional area deviation into a three-dimensional actual deviation quality with physical quality and environmental tolerance. Furthermore, when the cumulative deviation index exceeds the set tolerance boundary range, the event filtering module initially determines that the component deviation signal enters the valid candidate pool. Here, the specific upper limit of this tolerance boundary range is not arbitrarily specified, but rather deduced in reverse based on the maximum impurity handling redundancy capacity that the downstream resource recovery plant production line originally intended for this batch of materials can withstand, ensuring the engineering objectivity of the judgment standard.
[0041] Step S303: Based on the temporal filtering feature extraction, the event filtering module further performs spatial anti-vibration filtering based on irreversible logistics execution anchor points. Heavy transport vehicles have significant physical inertia and sunk costs. Frequently changing scheduling instructions can easily increase the risk of traffic congestion and wasted vehicle energy. To ensure the continuity of traffic flow and overall economy, the event filtering module dynamically calculates and sets irreversible logistics execution anchor point parameters for each batch of materials in transit.
[0042] Specifically, the event filtering module obtains the real-time latitude and longitude coordinates of the transport vehicles carrying the corresponding batch of materials based on feedback from the vehicle positioning terminal. The system maps these real-time coordinates to a high-precision road network model and calculates the actual distance traveled by the vehicle along the transport edge using textual calculations. To eliminate inconsistencies in judgment criteria caused by differences in the absolute distances of different transport edges, the system defines the ratio of the actual distance traveled to the total mileage of that transport edge as the logistics execution progress parameter. During this calculation, if there are extreme cases such as waiting in place that cause the total mileage of the transport edge to be recorded as 0, the system proactively sets the logistics execution progress parameter to 0 to avoid division crash errors. Subsequently, the system extracts a preset execution anchor point threshold and compares it numerically with the logistics execution progress parameter. The value range of the execution anchor point threshold is set to [0.6, 0.9], and its specific determination is based on the economic break-even point between the sunk costs (including fuel consumption and default time penalties) incurred by the vehicle performing a turnaround and lane change action on the current specific road segment and the estimated processing time cost that can be saved by going to the new alternative factory.
[0043] Step S304: Based on the aforementioned multi-dimensional evaluation logic in the time and spatial domains, the event filtering module performs final arbitration of the signal. When the material composition deviation signal continuously detected by the visual acquisition device exceeds the set tolerance boundary range after weighted calculation within the set spatiotemporal cumulative confirmation window, and simultaneously, the logistics execution progress parameter of the transport vehicle carrying the corresponding batch of materials is less than the preset execution anchor point threshold according to the data fed back by the vehicle positioning terminal, i.e., it is confirmed that the vehicle has not passed the preset irreversible logistics execution anchor point, the event filtering module determines that the current signal is a valid triggering event.
[0044] Conversely, if the logistics execution progress parameter is greater than or equal to the execution anchor point threshold, even if the system detects a local capacity fluctuation signal caused by queuing at the target resource processing plant, the event filtering module will directly block that local capacity fluctuation signal. In this case, the system rejects the lane change request and allows the vehicle to continue driving towards the original target node, thereby effectively avoiding disorderly oscillations of the global system caused by minor local disturbances through objective constraints at the physical space level.
[0045] See attached document Figure 1 The dynamic planning method for resource recovery of construction waste provided by this invention includes the following specific implementation steps for step S40: Step S401: In this embodiment, after the event filtering module outputs a valid trigger event confirmation command, the state feedback module blocks the execution of the original initial optimal path command. Industrial heavy equipment typically has significant mechanical inertia during operation. To prevent the running equipment from entering a runaway shutdown or hydraulic overload state due to a sudden loss of control commands, the system, while blocking the original command, prioritizes issuing a transitional hold signal to the controlled machinery to maintain its current safe operating posture. Based on the above safety assurance premise, the state feedback module locks the current geographic spatial coordinates of the transport vehicle carrying the corresponding batch of materials. Furthermore, the system uses the real-time three-dimensional extended node corresponding to the current location of the materials as the new starting calculation node, and based on the latest road congestion index fed back by the front-end sensing system and the capacity load parameters of each resource recovery plant, rewrites the impedance weights of the unexecuted connection edges in the directed graph network in the cloud, thereby completing the dynamic update of the underlying graph theory network environment.
[0046] In step S402, based on the updated directed graph network parameters, the state feedback module re-executes the forward optimization calculation to generate new scheduling instructions. This optimization calculation continues to use the aforementioned carbon value collaborative cost assessment model, generating a strategy combination for changing the logistics endpoint or adjusting subsequent production line processing steps by solving for the global minimum cost from the new starting point to each candidate sink node. As a preferred method, after generation, the system encapsulates the new scheduling instructions into a standard industrial communication message and sends it to the corresponding vehicle-mounted receiving terminal via the wireless network. The vehicle-mounted receiving terminal parses the message content and redraws the navigation route on the driver's visual interactive interface, thereby guiding the vehicle to change lanes.
[0047] Step S403: Traditional logistics scheduling systems are often unidirectional control flows, passively changing the transportation destination of materials to adapt to downstream factories, easily leading to resource waste due to disconnect between dismantling and utilization. To solve the above problems, this embodiment constructs a cross-physical stage state entanglement feedback closed loop, realizing feedforward control from passive allocation and consumption to active source intervention. Specifically, the state feedback module parses the new scheduling instruction and extracts the initial physical state boundary parameters required for the material to enter the next stage of processing. The specific content of the above initial physical state boundary parameters includes the maximum allowable material particle size critical value, moisture content upper limit, and impurity tolerance index for the target processing line. These boundary parameters objectively reflect the constraints of the newly matched downstream equipment on the upstream material form.
[0048] In step S404, the state feedback module converts the extracted initial physical state boundary parameters into equipment-level control quantities and sends them back to the mechanical control terminal in the source area via reverse communication. When the target equipment for intervention is a tracked mobile crusher at the demolition site, simply sending the downstream size requirements directly as control commands to the front-end machine often fails to achieve the expected control accuracy. The system needs to establish a reverse mapping system from physical space to mechanical structure. To accurately control the size of the output material, the system needs to map the extracted maximum material particle size critical value to the crusher's discharge port set gap control quantity.
[0049] Because construction waste of different compositions exhibits significant differences in compressive strength and cleavage properties, deriving the discharge port gap using a simple linear proportional relationship can lead to severe over-sized crushed products or blockage of the crushing chamber. Therefore, this embodiment constructs a nonlinear gap mapping model that integrates material mechanical characteristics and equipment mechanical losses. The state feedback module calculates the underlying control parameters of the equipment based on the following formula: ; In the above formula, This represents the deduced set gap control value for the discharge port of the target mobile crushing equipment. This represents the critical value of the maximum material particle size extracted by the reverse constraint of the next stage of processing. This represents the estimated average Mohs hardness of the current batch of materials. and These represent the crushing chamber type correction coefficient and the material brittleness damping coefficient, respectively, which are determined by the equipment's factory calibration. Both are calibrated to be constants that are always greater than 0. This represents the current physical wear compensation margin of the jaw plate, calculated based on historical load data recorded over a long period by mechanical condition sensors.
[0050] It is important to emphasize that in the above calculation logic, since the hardness parameter participates in the exponential term calculation, and both the coefficient and the characteristics of the exponential function ensure that the denominator remains greater than 0, this avoids the risk of data overflow from the underlying calculation logic. Furthermore, when the discharge port clearance control value calculated by the formula is less than the minimum clearance allowed by the mechanical structure of this type of crusher, the system actively clamps the control value to the mechanical safety setting baseline, thereby protecting the equipment's hydraulic drive system from overload damage.
[0051] After parameter mapping is completed, the mechanical control unit adjusts the initial crushing particle size of the mobile device according to the underlying control variables of the equipment. In specific implementation, the mechanical control unit directly drives the underlying electro-hydraulic proportional valve to adjust the stroke of the discharge hydraulic push rod of the mobile crusher. Similarly, if the reverse controlled object is a digging and sorting machine, the system modifies the torque overload limit parameters of its servo motor and the grab closing speed through the underlying control variables of the equipment, thereby adjusting the digging and sorting accuracy of the source material, and ultimately forming an adaptive closed-loop adjustment response to environmental disturbances in the source physical space.
[0052] Specific application examples: Project Background: A city's urban renewal project requires the demolition of an old commercial building, which is expected to generate 1,000 tons of mixed construction waste (including concrete, bricks, etc.). There are three resource recovery plants, A, B, and C, in the vicinity of the city. Plant A is the closest but has outdated equipment (high energy consumption and high carbon emissions), Plant B is at a moderate distance and has advanced equipment, and Plant C is farther away but has stricter requirements for the particle size of the incoming material (recycled coarse aggregate in the form of 5-10mm).
[0053] Implementation steps: Step 1: Initial Data Awareness and Baseline Graph Network Construction The graph network construction module obtains the project's estimated breakdown list (estimated mass percentage and spatial distribution), and performs data cleaning and temporal alignment based on data uploaded from the front-end sensing system (weighbridge equipment, visual acquisition equipment, vehicle positioning terminal) to construct a directed graph network. Network nodes are instantiated as three-dimensional extended nodes containing geospatial coordinate dimensions, time window dimensions, and physical state dimensions.
[0054] Step 2: Initial path optimization for carbon number coordination: The path optimization module uses the demolition site as the source node and applies a forward dynamic programming algorithm to perform optimization calculations. Based on the carbon value collaborative cost assessment model provided in this embodiment, the formula for calculating the total cost is as follows: ; Traditional static scheduling scheme: Without considering carbon synergy, materials are dispatched to Plant A. Based on the extracted impedance values of relevant connection sides, the logistics and transportation costs are assessed. The physical processing cost is 15,000 yuan. The carbon emission cost assessment value is 20,000 yuan. The estimated market transaction value is 10,000 yuan. It costs 8,000 yuan.
[0055] The initial solution of this invention: After comprehensive optimization, the initial decision of this system is to send the materials to Plant B, which has higher energy efficiency. The estimated value is: transportation cost of 18,000 yuan, processing cost of 16,000 yuan, carbon emission conversion cost of 6,000 yuan, and estimated market revenue of 9,000 yuan.
[0056] Step 3: Event Filtering and Dynamic Monitoring During physical transportation, the event filtering module detected a sudden congestion and queue at Plant B. Simultaneously, the visual acquisition equipment continuously detected deviations in the actual composition of the materials. Based on the spatiotemporal cumulative confirmation window parameters, the system calculates the cumulative deviation index, with the calculation formula strictly corresponding to the following: ; When calculated If the system calculates the logistics execution progress parameter as 0.4 based on the data from the vehicle positioning terminal (which is less than the preset execution anchor point threshold between [0.6, 0.9], i.e., the vehicle has not passed the irreversible logistics execution anchor point), the system determines the current signal as a valid triggering event.
[0057] Step 4: Status Feedback, Low-Level Reverse Control, and Itemized Cost Assessment The status feedback module blocks the command to plant B and reschedules the vehicle to plant C. It then extracts the initial physical state boundary parameters (maximum particle size critical value) required by plant C. The system converts this into low-level equipment control quantities and sends them to the mechanical control terminal via reverse communication. The calculation formula for the discharge port gap control quantity strictly corresponds to the following: ; According to the calculation The mobile device automatically adjusts the discharge port gap to meet Plant C's requirement for a particle size of 5-10 mm.
[0058] After the aforementioned dynamic lane changing and underlying control intervention, various consumption and benefits are combined and attached. Figure 3 To provide a more in-depth explanation. Figure 3 The horizontal axis distinguishes between the traditional static scheduling scheme and the dynamic programming scheme of this invention. The vertical axis represents the economic value (unit: yuan). The legend box at the top of the chart uses five gray-scale squares to correspond to five parallel bars below (the additional sunk cost caused by congestion is extracted as a separate item for comparison and monitoring): The darkest gray column (transportation cost assessment value) The invention changes the route to Plant C, increasing transportation costs to 20,000 yuan; the traditional solution, stopping at Plant A, costs 15,000 yuan.
[0059] Dark gray column (physical processing cost assessment value) The price of this invention is reduced to 17,000 yuan in Factory C; the traditional solution costs as much as 20,000 yuan in Factory A.
[0060] Medium gray column (carbon emission cost assessment value) The present invention is priced at 6,500 yuan, while the traditional solution is priced at 10,000 yuan.
[0061] Lighter gray column (with additional congestion penalty): Traditional solutions fail to prevent oscillations and lane changes, resulting in an additional time penalty of 8,000 yuan for waiting; the solution of this invention successfully avoids this by changing lanes, and the height of this column is 0 yuan.
[0062] Nearly white column (estimated market transaction value) The present invention ensures high aggregate quality by regulating the underlying machinery through formula, with an estimated market transaction value of 9,000 yuan, which is better than the traditional solution of 8,000 yuan.
[0063] Comparison of conclusions and overall evaluation: According to the embodiments of the present invention The general agent value formula (which incorporates additional penalties into cost considerations) is weighted and positively superimposed with revenue deduction: The total value of the traditional static scheduling scheme = 15,000 yuan + 20,000 yuan + 10,000 yuan + 8,000 yuan (default surcharge) - 8,000 yuan = 45,000 yuan.
[0064] The total value of the dynamic programming scheme of this invention = 20,000 yuan + 17,000 yuan + 6,500 yuan + 0 yuan - 9,000 yuan = 34,500 yuan.
[0065] The global optimization conclusion is intuitively mapped to the appendix. Figure 4 , Figure 4 The horizontal axis also distinguishes the objects being compared, while the vertical axis represents... Total Value (RMB). The dark gray bars on the left of the diagram represent the total value calculated using the traditional method, with a relatively high height; the light gray bars on the right represent the total value calculated using the present invention, with a significantly reduced height. The numerical labels floating at the top of the bars display the precise calculation results of the objective function. The left side shows 45000, and the right side shows 34500. This demonstrates that when faced with effective triggering events, the system, through network reconstruction and underlying closed-loop control, reduces the overall energy consumption and economic cost of the project by 23.3%, significantly improving the collaborative operational efficiency of waste resource utilization.
Claims
1. A dynamic planning method for the resource recovery path of construction waste, characterized in that, Includes the following steps: Obtain bill of materials data to set baseline components, and construct a directed graph network by combining multi-source heterogeneous data; the nodes of the directed graph network are three-dimensional extended nodes containing geospatial coordinates, time windows and physical states, and the connecting edges include transportation edges and processing edges; The nodes corresponding to the waste locations are taken as source nodes, and the nodes corresponding to the recycled building material application locations are taken as sink nodes. The forward dynamic programming algorithm is applied to the directed graph network to calculate and obtain the initial optimal path instructions. Monitor environmental disturbance signals that include actual material composition deviation signals; when the actual material composition deviation signal exceeds the tolerance boundary within the spatiotemporal accumulation confirmation window, and the transport vehicle has not passed the logistics irreversible execution anchor point, confirm a valid trigger event. Based on the effective triggering event that blocks the initial optimal path instruction, the three-dimensional extended node corresponding to the current geospatial coordinates is used as the starting calculation node to update the directed graph network, generate and issue a new scheduling instruction, extract the maximum material particle size critical value in the new scheduling instruction and convert it into the discharge port setting gap control quantity of the mobile crushing equipment, and issue it to the control terminal to adjust the initial crushing particle size of the mobile crushing equipment.
2. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The construction of a directed graph network by combining multi-source heterogeneous data includes: Receive multi-source heterogeneous data uploaded by the front-end sensing system and the capacity management system, and perform data cleaning and time-series alignment operations on the multi-source heterogeneous data; Based on the cleaned and aligned multi-source heterogeneous data and the reference components, the node topology of the directed graph network is instantiated to generate multiple three-dimensional extended nodes. The connecting edges are generated between each of the three-dimensional extended nodes. The transportation edge connects two three-dimensional extended nodes that have the same physical state but different geographic coordinates, and the processing edge connects two three-dimensional extended nodes that have the same geographic coordinates but different physical states.
3. The dynamic planning method for resource recovery paths of construction waste according to claim 2, characterized in that, The data cleaning and time-series alignment operations performed on the multi-source heterogeneous data include: By removing anomalous jump values in the material weight sequence, a time-series alignment algorithm based on dynamic time warping is constructed. The timestamp of the vehicle entering the monitoring area is used as the alignment reference, and linear interpolation is performed to compensate for the missing data frames. The weight sequence and image features of the same batch of materials are mapped and bound to the location information of the transport vehicle in a high-dimensional manner.
4. The dynamic planning method for resource recovery paths of construction waste according to claim 2, characterized in that, The step of generating the connection edges between each of the three-dimensional extended nodes includes: Set a maximum dissociation threshold parameter for the processed edge; The obtained visual fragmentation feature values are weighted and superimposed with the proportion of the material's historical crushing energy consumption in the previous process to the rated baseline energy consumption of the target equipment to calculate the actual degree of dissociation coefficient. When the actual dissociation coefficient is lower than the maximum dissociation threshold parameter, the processing edge is generated and retained in the directed graph network; If the actual dissociation coefficient is higher than the maximum dissociation threshold parameter, topology pruning is performed in the directed graph network and the corresponding processing edge is disconnected.
5. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The application of the forward dynamic programming algorithm in the directed graph network to obtain the initial optimal path instruction includes: A carbon value collaborative cost assessment model is constructed to obtain the logistics transportation cost assessment value, physical processing cost assessment value, carbon emission conversion cost assessment value, and market transaction value generated by each connecting edge in the feasible path; The first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient are respectively assigned to the logistics transportation cost assessment value, the physical processing cost assessment value, the carbon emission converted cost assessment value, and the market transaction value; The total agency value is obtained by adding the product of the logistics transportation cost assessment value and the first weighting coefficient, the product of the physical processing cost assessment value and the second weighting coefficient, and the product of the carbon emission converted cost assessment value and the third weighting coefficient. The total agency value is obtained by subtracting the product of the market transaction value and the fourth weighting coefficient from the total agency value. The path that minimizes the total value is taken as the initial optimal path instruction.
6. The dynamic planning method for resource recovery paths of construction waste according to claim 5, characterized in that, The allocation of the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, and the fourth weighting coefficient includes: Obtain a multidimensional historical cost index matrix and perform standardization processing; The variance of each column of characteristic indicators is detected, and a perturbation bias term is added if the variance is lower than the tolerance constant. Calculate the covariance matrix of the standardized matrix and perform eigenvalue decomposition to extract principal components whose cumulative variance contribution rate is greater than a preset threshold; The variance contribution rates of each principal component are weighted and normalized to derive the first weight coefficient, the second weight coefficient, the third weight coefficient, and the fourth weight coefficient.
7. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The monitored environmental disturbance signals include: Receives material actual composition deviation signals uploaded by visual acquisition equipment, local capacity fluctuation signals of resource recovery processing plants uploaded by capacity management system, and market demand change signals of recycled building materials synchronized by external interface; If the actual component deviation signal of the material exceeds the tolerance boundary range within the spatiotemporal cumulative confirmation window, it will be determined by the following method: The cumulative deviation index within the spatiotemporal cumulative confirmation window is calculated by multiplying the difference between the actual impurity percentage and the baseline component impurity percentage, the actual material quality measurement value aligned with the image timestamp, and the environmental perception confidence level, and summing the multiplication results of each sampling frame within the spatiotemporal cumulative confirmation window period to obtain the cumulative deviation index. The environmental perception confidence level is derived by adding the equipment's basic noise bias constant to the environmental particulate matter concentration value as the denominator. If the cumulative deviation index exceeds the tolerance boundary range, it is determined that the actual composition deviation signal of the material exceeds the tolerance boundary range.
8. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The transport vehicle did not pass through the logistics irreversible execution anchor point, including: The real-time latitude and longitude coordinates of the transport vehicle are obtained and mapped to a high-precision road network model; Calculate the actual distance the vehicle has traveled on the current transport edge, and determine the ratio of the actual distance traveled to the total mileage of the current transport edge as the logistics execution progress parameter; If the logistics execution progress parameter is less than the preset execution anchor point threshold, it is confirmed that the vehicle has not passed the preset irreversible logistics execution anchor point. If the logistics execution progress parameter is greater than or equal to the execution anchor point threshold, the environmental disturbance signal is shielded and the vehicle continues to drive toward the original target node.
9. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The step of updating the directed graph network by using the three-dimensional extended node corresponding to the current geospatial coordinates as the starting computation node includes: Send a transition hold signal to the controlled machinery to maintain its current safe operating posture; Lock the current geospatial coordinates of the transport vehicle carrying the corresponding batch of materials, and use the real-time three-dimensional extended node corresponding to the geospatial coordinates as the starting calculation node; Based on the latest traffic congestion index fed back by the front-end perception system and the capacity load parameters of each resource recovery plant, the impedance weights of the unconnected edges in the directed graph network are rewritten.
10. The dynamic planning method for resource recovery paths of construction waste according to claim 1, characterized in that, The conversion to the set gap control amount of the discharge port of the mobile crushing equipment includes: Obtain the crushing chamber shape correction coefficient, material brittleness damping coefficient, estimated average Mohs hardness of the current batch of material, and jaw plate physical wear compensation margin of the mobile crushing equipment; The maximum material particle size critical value of the crushing chamber type correction coefficient is raised to the power of the coefficient and multiplied by the exponent of the natural constant, where the exponent is the product of the material brittle damping coefficient and the estimated value of the average Mohs hardness. Subtract the jaw plate physical wear compensation margin from the multiplication result to obtain the set gap control amount of the discharge port.