A method for quantitatively predicting a particle breakage process of a rock-filled roadbed

Abstract

The present application relates to geotechnical engineering calculation and simulation technical field, disclose a kind of quantitative prediction method of rock-filled roadbed particle crushing process, the method includes: build a double-diagram structure including activation map and dormancy map, to characterize the actual force chain and potential contact of particle system;Under the driving of external load, the propagation, dissipation and accumulation of energy are simulated on the activation map, and the energy storage of each particle node is updated;By comparing the node energy storage and the critical breaking energy, determine and handle point dissipation event (particle crushing), and distribute the energy shock wave released by crushing to the adjacent nodes;Using the energy shock wave, determine and handle topological plasticity events;Finally, statistical analysis of the dissipation events in the whole process, generate quantitative prediction results about damage evolution and failure path.The present application greatly improves the calculation efficiency through the energy evolution mechanism driven by events, while truly reproducing the core physical processes of particle crushing and force chain reconstruction.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering calculation and simulation technology, specifically a quantitative prediction method for the particle crushing process of rockfill roadbeds. Background Technology

[0002] As a key load-bearing structure for modern major infrastructure projects such as high-speed railways and heavy-haul highways, the long-term stability and service performance of rockfill subgrades are of paramount importance. Composed of numerous irregularly shaped stone particles, these subgrades experience stress concentration under long-term, high-frequency cyclic loading from trains or vehicles. When this stress exceeds the strength limit of the granular material, it can lead to edge spalling and even overall fragmentation. This microscopic particle fragmentation gradually accumulates and evolves into macroscopic uneven subgrade settlement, track bed voids, and deterioration of overall load-bearing capacity, posing a potential threat to track smoothness and traffic safety. Therefore, accurately and efficiently predicting the particle fragmentation evolution process of rockfill subgrades during service is of significant engineering value for structural design optimization, life assessment, and disaster early warning.

[0003] Currently, research on this problem mainly relies on two methods: indoor physical experiments and numerical simulation. While physical testing methods such as large-scale triaxial tests can provide intuitive macroscopic mechanical responses, they are costly, time-consuming, and difficult to observe in real time the dynamic evolution of the force chain structure and the spatial distribution of damage within the particle system, exhibiting a black box effect. Therefore, the academic and engineering communities have increasingly turned to numerical simulation methods. Among them, methods based on continuum mechanics (such as the finite element method) treat the stone-filled body as a homogeneous material. Although this method can efficiently analyze the overall stress-strain field, its basic assumptions determine that it cannot fundamentally capture the microscopic damage mechanism dominated by discrete behaviors such as the breakage and rearrangement of individual particles.

[0004] In contrast, discontinuous medium methods, represented by the Discrete Element Method (DEM), offer a more physically accurate approach to studying such problems by tracking the motion and interactions of each individual particle. However, the DEM faces inherent and irreconcilable contradictions in practical applications. On the one hand, to accurately simulate particle breakage, complex particle replacement or cohesive breakage algorithms need to be introduced into the model, making model setup cumbersome. Furthermore, the occurrence of breakage events leads to the generation of new particles and a sharp increase in contact relationships, resulting in an explosive increase in computational load. On the other hand, a real-world engineering-scale rockfill subgrade can contain hundreds of millions of particles. To perform full-scale, long-term dynamic simulations of such a massive system, even using the most advanced supercomputers, would require a time cost far exceeding the acceptable range of conventional engineering design. Therefore, existing numerical simulation methods either sacrifice accurate descriptions of core physical mechanisms or are limited by high computational costs, making them difficult to promote and apply in engineering practice. A solution that effectively balances computational efficiency and physical realism is lacking. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a quantitative prediction method for the particle crushing process of rock-filled roadbeds, which solves the problem that existing numerical simulation methods struggle to balance the realism of physical mechanisms with the economic efficiency of large-scale engineering calculations when predicting particle crushing in rock-filled roadbeds.

[0006] To achieve the above objectives, the present invention provides a method for quantitatively predicting the particle crushing process of rock-filled roadbeds, comprising the following steps: S1. The stone-filled roadbed is abstracted into a dual-graph structure containing nodes and edges. The dual-graph structure includes an active graph and a dormant graph. The nodes are given energy storage and critical fracture energy attributes, the edges of the active graph are given frictional energy barrier attributes, and the edges of the dormant graph are given activation potential energy threshold attributes. S2. Based on the external load, energy is propagated between nodes through the edges of the activation graph and the energy storage of the nodes is updated. The external load is at least one of static load, cyclic load and dynamic impact load. S3. Based on the updated node energy storage and the node's preset critical breakage energy, determine and process point dissipation events as particle breakage; wherein, processing the point dissipation event includes splitting the node's energy storage to generate an energy shock wave. S4. Using the energy shock wave generated in the point dissipation event, determine and process the topological plasticity event that activates the edge in the dormant graph, and update the structure of the activated graph and the dormant graph based on the occurrence of the point dissipation event and the topological plasticity event. S5. Record each occurrence of the point dissipation event and the topological plasticity event, and accumulate or spatially visualize the recorded events to generate a quantitative prediction result of the particle crushing process of the rockfill subgrade.

[0007] Preferably, in step S1, the critical breaking energy of the node is assigned independently to each node using a Weber distribution.

[0008] Preferably, when performing step S2, a marginal dissipation step is also included, wherein if the propagated energy exceeds the frictional energy barrier of the corresponding edge, a portion of the energy is dissipated and the remaining energy is transferred to the target node.

[0009] Preferably, in step S2, updating the energy storage of the node specifically involves adding the energy storage of any node to be updated at the previous moment to all the remaining energy transferred to this node to be updated after the marginal dissipation step.

[0010] Preferably, in step S3, processing the point dissipation event further includes: taking the energy equal to the critical breaking energy of the node where the point dissipation event occurs as the breaking dissipation energy, and taking the energy storage portion of the node where the point dissipation event occurs that exceeds the critical breaking energy of the node where the point dissipation event occurs as the energy shock wave.

[0011] Preferably, in step S4, the specific condition for determining the topological plasticity event is as follows: for two neighboring nodes sharing a broken node, if there is an edge of a dormant graph between the two neighboring nodes, and the energy shock wave acting on the two neighboring nodes satisfies the activation potential energy threshold of the edge between the two neighboring nodes, then the topological plasticity event is determined to have occurred.

[0012] Preferably, the quantitative prediction results generated in step S5 include at least one of the following: the total number of nodes where the point dissipation event occurred, the total breakage dissipation energy, and the spatial distribution of failure paths formed by the edges of the removed activation graph.

[0013] Preferably, the quantitative prediction result generated in step S5 also includes the energy dissipated in the marginal dissipation step.

[0014] Preferably, in step S4, updating the structure of the activation graph and the dormant graph specifically includes: When the point dissipation event occurs, remove the edge connected to the node where the point dissipation event occurred from the activation graph; When the topological plasticity event occurs, the activated edge is removed from the dormant graph and added to the active graph.

[0015] Preferably, processing the point dissipation event further includes weighting the energy shockwave according to the preset attributes of the activated edges between the node and each neighboring node.

[0016] This invention provides a quantitative prediction method for the particle crushing process of rock-filled roadbeds. It has the following beneficial effects: 1. This invention abstracts complex rock-filled roadbeds into a dual-graph structure and uses graph theory algorithms in subsequent steps to handle energy evolution and event determination, thereby reducing the computational complexity when simulating large-scale particle systems. It avoids the huge computational overhead caused by contact search and motion equation solving for each particle in the traditional discrete element method, thus enabling efficient simulation of the gradual failure process of roadbeds over long time scales.

[0017] 2. This invention introduces a dual-graph structure of activation and dormancy graphs, and couples point dissipation events and topological plasticity events, which can simulate the core microscopic physical mechanism of the failure of stone-filled materials with high fidelity. In this invention, the force chain structure represented by the activation graph breaks after the point dissipation event, while the dormancy graph provides a potential path for the formation of new force chains triggered by energy shock waves. This dynamic evolution process accurately reproduces the physical phenomenon of stress redistribution and adaptive reconstruction of the load-bearing structure caused by local fracture within the material.

[0018] 3. This invention achieves multi-dimensional quantitative prediction of roadbed failure process by systematically recording and analyzing point dissipation events and topological plasticity events that occur during the simulation process. In addition, this method can not only provide macroscopic statistics such as the cumulative number of broken particles and total dissipated energy, but also output microscopic information such as the spatial distribution map of the failure zone composed of failure activation edges. Thus, it provides comprehensive and detailed quantitative data support for evaluating the stability of rockfill roadbeds, predicting their service life and identifying potential failure modes. Attached Figure Description

[0019] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a physical model and a schematic diagram of the dual-graph structure of the rock-filled roadbed of the present invention; Figure 3 This is a schematic diagram of the energy propagation and dissipation process on the activation diagram according to the present invention; Figure 4 This is a schematic diagram of the point dissipation event determination and processing flow of the present invention; Figure 5 This is a schematic diagram of the topological plasticity event determination and dual-graph structure update of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] See attached document Figure 1 The method can be executed by a computing device. The computing device includes a processor, a memory communicatively connected to the processor, and an input / output interface. The memory stores computer program instructions that, when executed by the processor, cause the computing device to implement the method described in this invention.

[0022] In one embodiment, the computer program stored in the memory may be organized into multiple functional modules to collaboratively execute the methods described in this invention. These functional modules include: a model building module, an energy evolution module, an event processing module, and a prediction generation module.

[0023] When the model building module is executed by the processor, it is used to perform step S1, which transforms a physical rock-filled roadbed system into a computable bi-graph structure data with preset physical properties.

[0024] When the energy evolution module is executed by the processor, it performs step S2, which simulates the propagation, dissipation and accumulation of energy on the active graph of the dual-graph structure based on the external load input, and updates the energy storage status of each node in real time.

[0025] When the event processing module is executed by the processor, it is used to execute steps S3 and S4. After each calculation step, based on the node energy storage updated by the energy evolution module, it determines and processes point dissipation events (particle breakage) and topological plasticity events (force chain reconstruction), and updates the topology of the bi-graph structure based on the event results.

[0026] When the prediction generation module is executed by the processor, it performs step S5, which records, accumulates, and analyzes all events processed by the event processing module throughout the simulation process, and finally generates quantitative prediction results.

[0027] See attached document Figure 2 Step S1 of the present invention is executed by the model building module, which aims to transform a physical rock-fill roadbed system into a dual-graph structure with physical property definitions that can be processed in a computing device.

[0028] This step first abstracts the rock-fill roadbed into a bigraph structure containing nodes and edges. In a specific implementation, tools such as the Particle Flow 3D (PFC3D) program can be used to generate a digital model containing a large number of discrete particles as the initial geometric configuration of the rock-fill roadbed. Each independent particle in the model is defined as a node in the bigraph structure. In some embodiments, to reduce computational load, multiple closely packed particles in the initial model whose relative positions remain essentially unchanged during loading can be preprocessed into a particle cluster, which is then abstracted as a node.

[0029] After nodes are defined, the model building module establishes edges for the active and dormant graphs based on the contact and proximity relationships between nodes. For any two nodes, if the particles they represent have a contact force greater than zero in the initial configuration, an edge is established between these two nodes in the active graph, representing a force chain path currently bearing a load. For any two nodes without contact forces, the shortest distance between the surfaces of the particles they represent is calculated. If this distance is less than a preset proximity threshold... If the two nodes are connected, then an edge of the dormant graph is created between them, representing a potential path that may form a connection in the future.

[0030] Subsequently, the model building module assigns physical properties to the nodes and edges of the established bi-graph structure. For each node... Assigning energy storage properties and critical fracture energy properties. Node energy storage. Representative particles The internal storage provides elastic energy; all nodes store energy before loading begins. The initial value was set to 0.

[0031] Critical breaking energy of the node This represents the energy threshold required for the particle corresponding to that node to break. Considering the heterogeneity of the material, the critical breakage energy is... Values ​​are assigned independently by sampling from a Weiber distribution. Its probability density function... for: ; in, The critical breaking energy of the node; Weber modulus is a quantitative characterization of the dispersion or heterogeneity of material strength. The characteristic crushing energy of the material; The normalized energy term is obtained by using the critical breaking energy. Its eigenvalues Division can transform a dimensional energy problem into a dimensionless mathematical problem; This describes the cumulative effect of failure risk with increasing normalized energy, the magnitude of which is determined by the Weber modulus. control; For reliability or survival function, it gives the condition when energy reaches The probability that the node has not yet broken at that time; The normalization coefficient is used to ensure the normalization of the entire probability density function. The integral from 0 to infinity results in 1, making it a mathematically complete probability distribution. The term is an exponential term, which is generated mathematically when differentiating the reliability function to obtain the probability density function.

[0032] The model building module assigns a frictional energy barrier property to each edge of the activation graph. . The threshold value for energy dissipation due to friction during energy transfer between two contacting particles can be set to be proportional to the normal force and the coefficient of friction on the contact surface.

[0033] The model building module assigns an activation potential threshold attribute to each edge of the dormant graph. . Characterizes the minimum energy required to activate a potential contact, and its value can be set to be related to the geometric barrier or tiny gaps between particles that need to be overcome to form the contact.

[0034] After completing the above operations, a complete bi-graph structure data containing node sets, active graph edge sets, dormant graph edge sets, and the physical attributes corresponding to each element is generated and stored in the memory of the computing device for subsequent steps.

[0035] In a specific embodiment, the physical property parameters assigned in step S1 can be determined through experiments or empirical formulas.

[0036] For example, the parameters of the Weiber distribution. and The data can be obtained by conducting a series of uniaxial compression tests on single-particle samples of the same material as the rock used in the riprap roadbed, statistically analyzing their fracture strength data, and then fitting the statistical data to a Weiper distribution.

[0037] Activated edge frictional energy barrier It can be calculated using the following formula: in The dynamic friction coefficient is obtained through rock shearing experiments. This refers to the initial normal force of the contact surface read from the PFC3D model.

[0038] Activation potential threshold of dormant edge It can interact with the geometric gap between two particles. Association, for example, set as ,in This is a proportionality coefficient related to the material contact stiffness.

[0039] Proximity threshold for sleep edge determination It can be set to a specific percentage of the average particle size, such as 5% to 10%.

[0040] To enable those skilled in the art to reproduce this method, the following provides a set of validated model parameter ranges in simulations of typical rockfill roadbed materials (such as limestone crushed stone): Weiber distribution parameters: Weiber modulus The value range is typically from 1.5 to 5.0, characteristic fracture energy. The range of values ​​is J to J (depends on the average particle size; in this example, a particle size of 40-60 mm is used).

[0041] Triboelectric barrier parameter: dynamic friction coefficient The value ranges from 0.3 to 0.6; Activation potential threshold coefficient: proportional coefficient Related to the contact stiffness of the material, the recommended value range is: N / m to N / m; Proximity threshold: Set as average particle radius 0.1 to 0.2 times; Time step: To ensure the convergence of energy propagation calculations, the calculation time step is... Recommended value s to s.

[0042] See attached document Figure 3 In this invention, step S2 is executed by the energy evolution module. Its core task is to simulate the propagation, dissipation and accumulation of energy on the activation graph established in step S1 based on external loads, and update the energy storage state of each node accordingly.

[0043] This step begins with the application of external loads. In a loading step... Internally, external loads (such as pressure or vibration caused by road traffic) are converted into a total input energy. This energy is applied to the boundary nodes in the dual-graph structure that represent direct contact with the externally loaded region. For example, in an embodiment simulating uniaxial compression, It is evenly distributed to all boundary nodes located at the top of the model as their initial energy input.

[0044] Specifically, the method for calculating the energy input to the system by external loads depends on the type of load.

[0045] In this embodiment, for cyclic loads (such as simulated train wheel axle pressure), it is assumed that they are applied to the set of boundary nodes. The total normal force on is: ; in, For dynamic load amplitude, For frequency, For static load.

[0046] At a time step Within, the average displacement increment generated by the boundary nodes is Then the total energy input at that time step. The calculation formula is: ; Total energy Distribute to the set either on average or according to the contact area weight of the boundary nodes. Each node in the process completes the energy injection.

[0047] The energy evolution module then processes the propagation of energy across the activation graph. This process is performed iteratively. In one iteration, one node... Store its energy Assign to all its neighboring nodes connected by active edges. Assign to a specific neighboring node. The energy portion, denoted as .

[0048] In a preferred embodiment, energy is not distributed equally, but rather weighted. It is allocated to neighboring nodes. energy portion It can be calculated based on the properties of the active edge, such as being proportional to contact stiffness or contact area. The calculation formula is: ; in, For connecting nodes and The contact stiffness of the activated edge, For nodes The set of all neighboring nodes on the activation graph; To represent the node at the current moment The total energy stored and available for distribution; Let $i$ represent the sum of the weights of all outgoing edges (connections to all neighbors) of node $i$. The overall channel capacity for transferring energy outward. In this way, energy tends to propagate along stronger force chain paths.

[0049] Energy from the node propagation to nodes During this process, a marginal dissipation step is triggered. This step simulates frictional dissipation at the particle contact surface. Specifically, when the energy portion Through connection and When an active edge is encountered, the energy evolution module applies a frictional energy barrier to that edge. Compare the data. Pass the data to the target node. Remaining energy Calculated according to the following rules: ; in, To be passed to the node after dissipation The remaining energy. If Then energy It is dissipated, and its value is recorded as marginal dissipation energy; the remaining part Continue to spread. If In this case, all energy is dissipated, and no energy remains to be transferred to the node. .

[0050] After all nodes have completed one round of energy propagation and marginal dissipation calculations, the energy evolution module updates each node. Energy storage. Nodes At the present moment Energy storage For its previous moment Energy storage The sum of the remaining energy received from all its neighboring nodes. This update process is described by the following equation: ; in, Representative node The set of all neighboring nodes on the activation graph; For nodes At the current time step, follow all its neighboring nodes. The sum of the received energy increments.

[0051] By executing this formula, a loading step is completed. Update the energy storage status of all nodes. The updated node energy storage data will be used as input for the next step.

[0052] See attached document Figure 4In this invention, step S3 is executed by the event processing module. After the energy evolution module completes the update of the energy storage of all nodes, the event processing module traverses the system to determine and process point dissipation events, i.e., particle breakage events, caused by the energy storage of nodes exceeding their carrying capacity limits.

[0053] The decision-making process is executed at the end of each calculation step. The event handling module checks each node in the system one by one. For nodes To put it at the current moment Energy storage The critical breaking energy assigned to it in step S1 Compare the nodes. If the following conditions are met, then the node is determined. A point dissipation event occurred: ; When a node After a point dissipation event is detected, the node is marked as broken. The event handling module immediately checks the node's energy storage. The process involves processing and splitting the nodes. This process will reduce the total energy storage of the nodes. It is divided into two parts: one part is the energy dissipated to complete the crushing process itself, and the other part is released outward in the form of energy shock waves.

[0054] The specific energy splitting process is as follows: Energy dissipation from breakage This will be equal to the critical fracturing energy at that node. The energy value is recorded as the breakup dissipation energy. This portion of energy represents the work done in forming new particle surfaces and is considered to have been permanently dissipated from the system.

[0055] Energy shockwave The portion of a node's total energy storage that exceeds its critical fragmentation energy is defined as an energy shock wave. This is the excess energy released during a fragmentation event that can affect surrounding particles.

[0056] After completing the energy splitting calculations, the broken node Its own energy storage A value reset to 0 indicates that the particle has become ineffective and no longer stores energy.

[0057] Subsequently, the event handling module processes the generated energy shockwave. The energy shockwave did not radiate uniformly, but rather spread along the fracture nodes. Connected active edges are assigned to their neighboring nodes. In one specific embodiment, this assignment process is weighted. Each broken node is connected... Its neighboring nodes The activated edges all have a preset weight attribute. Assigned to neighboring nodes Energy shock wave part Calculated by the following formula: ; in, It is a broken node The set of all neighboring nodes on the activation graph; As the source node The sum of the weights of all outgoing edges (i.e., connections to all neighbors).

[0058] Weight It can be set based on the geometric or physical properties of the contact surface, such as being proportional to the contact stiffness. The calculated energy shock wave portion... This will serve as a key input for the next step in determining topological plasticity events.

[0059] See attached document Figure 5 Step S4 of the present invention follows immediately after step S3 and is executed by the event processing module. Its purpose is to use the energy shock wave released by the broken node in step S3 to determine and process the topological plasticity event as a force chain reconstruction mechanism, and update the topological connections of the bi-graph structure accordingly.

[0060] This step first addresses the activation graph structure update caused by node fragmentation. When a node... After being identified as a broken node in step S3, the event handling module will remove all nodes associated with it from the active graph. Connected active edges. This operation represents breaking particles. Failure and breakage of the local force chain structure with the core as the core.

[0061] Subsequently, the event handling module uses the energy shockwaves distributed to neighboring nodes to determine topology plasticity events. For broken nodes... any pair of neighboring nodes and The module first checks the hibernation graph to see if a connection exists. and dormant side .

[0062] If such a dormant edge exists The event handling module then further determines whether the energy applied to the dormant edge is sufficient to activate it. The total energy applied to the dormant edge is defined as the energy of the neighboring nodes. and From the broken node The sum of received energy shock waves. The criterion for the occurrence of a topologically plastic event is: ; in, and Assigned to nodes respectively and The energy shockwave component; It is a dormant edge The assigned activation potential threshold.

[0063] If the criteria are met, a topological plasticity event is determined to have occurred. The event handling module immediately performs a topological update operation on the bigraph structure to reflect the formation of this new force chain connection. This update operation specifically includes: Dormant edge Remove from the edge set of the dormant graph.

[0064] Side Add to the edge set of the activated graph.

[0065] Through the above operations, a previously potential contact is activated into an actual load-bearing contact, completing the bypass-style reconstruction of the new force chain after the old force chain fails due to particle breakage. After this step is completed, the system's topology and energy state have been updated, preparing for the calculation of the next external load loading step.

[0066] In another embodiment, the determination of topological plasticity events may also consider factors other than broken nodes. direct neighbors and Other than that. For example, when a node From the broken node Received energy shock wave Then, this energy can further trigger nodes. With one of its non neighboring nodes dormant edges between Activation. Its determination criteria can be adjusted as follows: ; in For dormant edge The activation potential threshold. This mechanism simulates the chain reaction triggered by an energy shock wave in a local area, thereby activating new force chain paths.

[0067] To verify the effectiveness of this method, a specific simulation example was constructed.

[0068] Model settings: Generate a cuboid sample with dimensions of 0.5m × 0.5m × 1.0m, containing 5000 randomly generated non-spherical particles. The average particle size is 0.05m. The initial porosity is 0.3.

[0069] Parameter assignment: Set Weber modulus = 3.0, characteristic crushing energy = 0.01J, friction coefficient = 0.5, activation coefficient = 105N / m.

[0070] Loading process: A sinusoidal cyclic load with an amplitude of 100 kPa and a frequency of 5 Hz was applied to the top of the specimen for a total of 10,000 cycles.

[0071] Prediction results: Damage evolution: Simulation results show that in the first 1000 cycles, point dissipation events (particle breakage) increase rapidly, with a cumulative number of breakage nodes reaching 150; then the growth rate slows down, and by the end of 10,000 cycles, the cumulative number of breakage nodes reaches 320.

[0072] Energy dissipation: The cumulative energy dissipated by crushing is 4.5 J, and the cumulative energy dissipated by friction is 120.0 J.

[0073] Spatial distribution: The system outputs a three-dimensional distribution map of the failure path, clearly showing that the fracture nodes are mainly concentrated in the lower part of the sample, forming a continuous shear band. This is consistent with the macroscopic failure mode observed in real large-scale indoor triaxial fatigue tests. This example demonstrates that the proposed method can quantitatively predict the progressive failure process of rockfill subgrades within a reasonable computation time (approximately two orders of magnitude more efficient than the traditional DEM method).

[0074] Step S5 of the present invention is executed by the prediction generation module, which runs continuously throughout the simulation process, is responsible for recording key event data, and processes and outputs these data as quantitative prediction results of the particle crushing process of the rock-filled roadbed after the simulation ends.

[0075] The core of this step lies in the systematic recording of various events occurring during the simulation. The prediction generation module maintains multiple data loggers in the computing device's memory. In each computational step of the simulation: When the event handling module determines in step S3 that a node has experienced a point dissipation event, the prediction generation module performs a recording operation: storing the identifier of the broken node, the time of occurrence, and the calculated breakage dissipation energy into an event log.

[0076] When the event handling module removes all active edges of node i in step S4 due to node i being broken, the prediction generation module adds the endpoint coordinates or identifiers of these removed edges to a failed path dataset.

[0077] When the energy evolution module calculates the marginal dissipation energy in step S2, the dissipation value is accumulated into a total friction dissipation energy counter.

[0078] After the entire simulation process (e.g., after applying a preset number of cyclic loads or reaching a preset strain level) is completed, the prediction generation module performs final processing and integration of the recorded data to generate quantitative prediction results in various forms.

[0079] One generated quantitative prediction result is a time history curve of system damage evolution. The prediction generation module processes the event log to calculate and output: The curve showing the cumulative total number of nodes experiencing point dissipation events over time or the number of loading cycles. This curve visually reflects the rate of accumulation of internal damage within the material.

[0080] The cumulative fragmentation dissipation energy curve over time is used to quantify the degree of material damage from an energy perspective.

[0081] The cumulative marginal dissipation energy over time is used to assess the total energy loss of the system due to friction.

[0082] Another type of generated quantitative prediction is a visualization of the spatial distribution of damage. The prediction generation module uses data from the failure path dataset to render the edges of the removed activation graph in a three-dimensional coordinate system. These rendered edges spatially cluster to form one or more continuous band-like regions, i.e., macroscopic shear bands or failure paths. This image can clearly reveal the most vulnerable locations within the subgrade and the potential overall instability modes.

[0083] These generated quantitative predictions can be output as data files, reports, or graphical interfaces through the input / output interfaces of computing devices, providing data support for engineering evaluation and design optimization.

Claims

1. A quantitative prediction method for the particle crushing process of rock-filled roadbeds, characterized in that, Includes the following steps: S1. The stone-filled roadbed is abstracted into a dual-graph structure containing nodes and edges. The dual-graph structure includes an active graph and a dormant graph. The nodes are given energy storage and critical fracture energy attributes, the edges of the active graph are given frictional energy barrier attributes, and the edges of the dormant graph are given activation potential energy threshold attributes. S2. Based on the external load, energy is propagated between nodes through the edges of the activation graph and the energy storage of the nodes is updated. The external load is at least one of static load, cyclic load and dynamic impact load. S3. Based on the updated node energy storage and the node's preset critical breakage energy, determine and process point dissipation events as particle breakage; wherein, processing the point dissipation event includes splitting the node's energy storage to generate an energy shock wave. S4. Using the energy shock wave generated in the point dissipation event, determine and process the topological plasticity event that activates the edge in the dormant graph, and update the structure of the activated graph and the dormant graph based on the occurrence of the point dissipation event and the topological plasticity event. S5. Record each occurrence of the point dissipation event and the topological plasticity event, and accumulate or spatially visualize the recorded events to generate a quantitative prediction result of the particle crushing process of the rockfill subgrade.

2. The quantitative prediction method for the particle crushing process of rock-filled roadbed according to claim 1, characterized in that, In step S1, the critical breaking energy of the node is assigned independently to each node using a Weber distribution.

3. The quantitative prediction method for the particle crushing process of rock-filled roadbed according to claim 1, characterized in that, When performing step S2, a marginal dissipation step is also included, which is: if the propagated energy exceeds the frictional energy barrier of the corresponding edge, then some of the energy is dissipated and the remaining energy is transferred to the target node.

4. The quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 3, characterized in that, In step S2, updating the energy storage of the node specifically involves adding the energy storage of any node to be updated at the previous moment to all the remaining energy transferred to this node after the marginal dissipation step.

5. The quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 1, characterized in that, In step S3, processing the point dissipation event further includes: taking the energy equal to the critical breaking energy of the node where the point dissipation event occurs as the breaking dissipation energy, and taking the energy storage portion of the node where the point dissipation event occurs that exceeds the critical breaking energy of the node where the point dissipation event occurs as the energy shock wave.

6. The quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 1, characterized in that, In step S4, the specific conditions for determining the topological plasticity event are as follows: for two neighboring nodes of a shared broken node, if there is an edge of a dormant graph between the two neighboring nodes, and the energy shock wave acting on the two neighboring nodes satisfies the activation potential energy threshold of the edge between the two neighboring nodes, then the topological plasticity event is determined to have occurred.

7. The quantitative prediction method for the particle crushing process of rock-filled roadbed according to claim 1, characterized in that, The quantitative prediction results generated in step S5 include at least one of the following: the total number of nodes where the point dissipation event occurred, the total breakage dissipation energy, and the spatial distribution of failure paths formed by the edges of the removed activation graph.

8. The quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 3, characterized in that, The quantitative prediction results generated in step S5 also include the energy dissipated in the marginal dissipation step.

9. The quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 1, characterized in that, In step S4, updating the structure of the activation graph and the dormant graph specifically includes: When the point dissipation event occurs, remove the edge connected to the node where the point dissipation event occurred from the activation graph; When the topological plasticity event occurs, the activated edge is removed from the dormant graph and added to the active graph.

10. A quantitative prediction method for the particle crushing process of a rock-filled roadbed according to claim 5, characterized in that, The processing of the point dissipation event further includes weighting the energy shock wave according to the preset attributes of the active edges between the node and each of its neighboring nodes.

Images