A non-destructive testing method for evaluating the fatigue properties of long-life pavement materials

Abstract

This invention discloses a non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials, belonging to the field of pavement material performance testing technology. The method includes the following steps: mapping the detection data collected by each sensing unit in an intelligent sensing array to the same time reference, constructing a continuous and consistent time-stamped acquisition chain, and generating a time-stamped consensus band at the end of the time-stamped acquisition chain. By constructing the time-stamped acquisition chain and consensus band, this invention identifies the dominance switching moments of multi-source detection data, divides fatigue evolution stages, rearranges the data output order, constructs a relay sequence, and introduces a dynamic control mechanism to release the dominant signal segment by segment, achieving continuous expression and jump suppression of the fatigue state during the dominance switching process.

Description

Technical Field

[0001] This invention relates to the field of pavement material performance testing technology, specifically to a non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials. Background Technology

[0002] Non-destructive testing for evaluating the fatigue characteristics of long-life pavement materials refers to a technical approach that continuously senses and analyzes the mechanical response of long-life pavement materials under long-term traffic loads using intelligent sensing systems, without core sampling, damage to the pavement structure, or impact on normal road service. This allows for the determination of the fatigue damage evolution state and remaining load-bearing capacity. This method involves deploying multiple types of intelligent sensing units on the pavement or test area to collect dynamic information such as strain, vibration, displacement, response spectrum, or energy attenuation in real time. Combined with data fusion and state recognition mechanisms, it indirectly characterizes the initiation, propagation, and cumulative fatigue behavior of microcracks within the material. This enables long-term, online, and traceable evaluation of the fatigue performance of pavement materials, avoiding the interference of traditional destructive testing on structural integrity and service status. It is suitable for performance monitoring and evaluation scenarios throughout the entire life cycle of long-life pavements.

[0003] The existing technology has the following shortcomings: In existing technologies, when multiple types of sensor data are simultaneously introduced for nondestructive testing and evaluation of the fatigue characteristics of long-life pavement materials, response information from different sensor sources is typically directly superimposed or included in parallel within the same time series for analysis. However, the dominant role of various sensor data in different fatigue evolution stages is not effectively constrained. During the service life of pavement materials, different sensor data exhibit significant differences in their sensitivity ranges and time scales to fatigue changes. This leads to a situation where, in actual evaluation, different data sources alternately dominate the interpretation of the time series, causing fatigue indicators to exhibit directional reversals within adjacent time windows. This reversal does not stem from actual changes in the material's fatigue state but is caused by the switching of dominance among multiple data sources. Consequently, the fatigue state output by the system nears the critical threshold stage of fatigue shows abrupt changes, making it difficult to form a continuous and interpretable fatigue evolution trajectory, thus affecting the stability and reliability of the fatigue assessment results.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials, comprising the following steps: The detection data collected by each sensing unit in the intelligent sensing array is uniformly mapped to the same time base, a continuous and consistent time-stamped acquisition chain is constructed, and a time-stamped consensus band for subsequent processing is generated at the end of the time-stamped acquisition chain. The detection data in the time-stamped acquisition chain is analyzed segment by segment according to the time-stamped consensus band. The phase change trend and energy decay trend of each detection data are compared and judged. The time position with pointing conflict is marked in the time-stamped consensus band, and the dominance switching prompt frame corresponding to the time-stamped consensus band is generated. The time-stamped acquisition chain is divided into stages according to the time sequence of the dominance switching prompt frame. Within each stage, the dominant signal source and subordinate signal source of the detection data are determined, and the handover time of the dominant signal source is clearly marked in the dominance switching prompt frame, forming a weight traction list corresponding to each stage. The output order of detection data in the time-stamped acquisition chain is rearranged according to the weighted traction list. A continuous time window relay sequence is constructed before and after the handover time of the dominant signal source, so that the time position pointing to the conflict can be smoothly transitioned in the relay sequence, and a sequence index corresponding to the time-stamped consensus band is generated. A dynamic control mechanism is introduced based on sequence indexing. The output rhythm of the detection data is guided by the reverse time-series breathing phase-locking method. In the time window relay sequence, the time occupation of the subordinate signal source is temporarily suspended, and the dominant signal source is released segment by segment according to the relay order, thereby eliminating the fatigue state jump caused by the switching of dominance in the time dimension.

[0007] Preferably, the steps for generating the time-stamped consensus band are as follows: Deploy an intelligent sensor array within the non-destructive testing area, set the acquisition timing parameters for each sensor unit, and initialize the acquisition time reference to the start time consistent with an external high-stability clock source. Use equidistant time trigger logic to anchor the data time information. After the sensing unit completes the acquisition, the original timestamp information is mapped according to a unified time reference, and the converted time information is written into the acquisition chain. Time alignment of different sensing data is achieved through time interval calibration. After completing time standardization, data nodes are arranged in chronological order and divided into non-overlapping time periods. At the end of each period, data with unique time identifiers and signal coverage are selected to build consensus nodes, forming a time-stamped consensus band. After the time-stamped consensus band is constructed, the consensus nodes are bound to the source data segments and their logical positions are recorded. The end information is then synchronously written to the time output port to complete the organization of the unified time series structure.

[0008] Preferably, the consensus node is constructed using an equally spaced division method. At the end of each time period, the time position that simultaneously contains data collected by multiple sensor units is selected as the node reference, and the node is bound to the logical position of its corresponding source data in the collection chain.

[0009] Preferably, the steps for generating the control switch notification frame are as follows: According to the time period sequence of the time-stamped consensus band, the time-stamped acquisition chain is divided into continuous analysis segments, and detection data segments containing the original arrangement order and time reference information are extracted. Based on the detection data segments of each analysis section, phase change trend structure analysis and energy decay trend evolution analysis are performed respectively to generate phase trend description structure and energy trend description structure. Within each analysis segment, based on the relative relationship between the phase trend description structure and the energy trend description structure, the time location of directional conflict is identified and labeled. After completing the marking of all conflict times, the master control switching prompt frames are compiled and constructed to correspond one-to-one with the time-stamp consensus bands, while maintaining their original logical connection with the time-stamp acquisition chain.

[0010] Preferably, the phase change trend structure analysis includes classifying the response fluctuation direction, persistence and abrupt change frequency of the detection data within the analysis section, and the energy decay trend evolution analysis includes constructing the energy dissipation path based on the trajectory of response value amplitude change, and labeling the trend type according to the decay rate, plateau period length and later collapse degree.

[0011] Preferably, the steps for generating the weighted traction list are as follows: According to the time position of the control switching prompt frame, the time-stamped acquisition chain is divided into multiple continuous time stages, and the stage boundaries and handover nodes are kept in a one-to-one correspondence. At each time stage, all detection data are retrieved, and the dominant signal source and subordinate signal source are determined by comparing the response change trajectory. After identifying the dominant signal source and the subordinate signal source, trace back the dominant power switching prompt frame node, insert a handover tag in the prompt frame and record the dominant signal handover time point; After confirming the master-slave relationship and writing the handover tags at each stage, the data is summarized and organized into a weighted traction list, which serves as the basis for subsequent output rearrangement and control.

[0012] Preferably, when forming the weighted traction list, the weighted traction list is arranged in chronological order according to the fatigue evolution stage. The source of the dominant signal in each stage remains consistent within the stage, and the handover time point of the dominant signal is fixedly recorded as the stage boundary to constrain the continuous succession relationship of the output order of subsequent detection data.

[0013] Preferably, the sequence index generation steps are as follows: According to the dominant signal source and handover time recorded in the weighted traction list, the detection data in the time-stamped acquisition chain are rearranged in stages to form rearrangement units corresponding to the fatigue evolution stage. After completing the phased rearrangement, a time window relay sequence is constructed between adjacent phases around the handover time of the dominant signal source, and the output order of subordinate signals is delayed in the relay sequence. After forming the time window relay sequence, each rearranged unit is integrated with the corresponding relay sequence in chronological order to generate a new detection data output sequence; After the output sequence is generated, a sequence index is established based on the consensus nodes in the time-stamped consensus band, so that the time positions pointing to the conflict are smoothly transitioned in the relay sequence and the time correspondence is maintained.

[0014] Preferably, the output order of the dominant signal in the relay sequence is arranged according to the stage affiliation relationship of the dominant signal in the weighted traction list, the time period occupied by the dominant signal in the output order is continuous, and the rhythm remains consistent before and after the handover moment.

[0015] Preferably, a dynamic control mechanism is introduced according to the sequence index, and the output rhythm of the detection data is adjusted by using a reverse-time breathing phase-locking method. The time occupation of the subordinate signal source is delayed within the time window relay sequence, and the dominant signal source is output segment by segment according to the relay order. The steps are as follows: After completing the sequence index construction, extract the handover position and relay structure marker fields, set the start interval of the dominant signal handover, and establish the basic rhythm anchor point; A reverse-time breathing phase-locking method is introduced within the initial interval for regulation, compressing the unreleased dominant signal forward and extending the release time window of the dominant signal in the new stage backward; Based on completing the rhythm traction, the subordinate signals in the receiving section are sorted and delayed to retain the original time markers and control the time occupation ratio of the dominant signal. After the regulation process is completed, the rhythm structure and time period structure are bound and recorded to generate a continuous release output mapping trajectory and establish a corresponding relationship with the time nodes.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention achieves precise labeling and dynamic perception of signal dominance switching moments by constructing a time-stamped acquisition chain and a time-stamped consensus band, and by combining phase changes and energy decay trends of multi-source detection data for conflict identification. By introducing a relay-style structure and rhythmic control mechanism in the output stage, the dominant signal is continuously released during critical handover periods, avoiding non-physical jumps in the fatigue state trajectory caused by abrupt changes in the signal interpretation dominance relationship. This improves the continuity and interpretability of fatigue assessment results in the critical state stage.

[0017] This invention dynamically identifies the master-slave relationship of each detection data point throughout the entire fatigue evolution process, and performs rhythmic regulation based on a weighted traction mechanism and sequence indexing, enabling fatigue state assessment to have autonomous smoothing capabilities at the time series level. The output rhythm of the detection data is no longer limited by static rules, but actively adjusts the signal release order according to stage changes, realizing flexible control and consistent expression of the state transition process, effectively enhancing the tracking stability of complex fatigue behavior and the response accuracy of the monitoring system. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of a non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1 The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials, as shown, includes the following steps: The detection data collected by each sensing unit in the intelligent sensing array is uniformly mapped to the same time base, a continuous and consistent time-stamped acquisition chain is constructed, and a time-stamped consensus band for subsequent processing is generated at the end of the time-stamped acquisition chain. To meet the need for non-destructive assessment of the fatigue characteristics of long-life pavement materials, it is necessary to first deploy an intelligent sensor array within the target pavement area. The detection data collected by multiple sensor units in the array will then be fully mapped and synchronously reconstructed based on a unified time reference to form a structurally continuous and time-consistent time-stamped acquisition chain. Finally, a time-stamped consensus band for subsequent fatigue assessment processing will be established at the end of the acquisition chain. The specific implementation steps are as follows: The intelligent sensor array is deployed within a pre-defined non-destructive testing area, and the acquisition timing parameters of each sensor unit are set. This allows each sensor unit to record its original acquisition time point information when sensing different mechanical response parameters of the road material. To ensure consistent alignment of the timing data in subsequent steps, the acquisition time reference of each sensor unit is initialized to a start time consistent with an external high-stability clock source during the initial deployment phase. Equidistant time triggering logic is then used to anchor the time of all subsequent data segments. During this process, the data acquired by all sensor units is not directly fused or compressed; instead, the integrity of their original timestamps and response information is maintained to provide a high-fidelity time structure foundation for the next step.

[0022] After each sensing unit completes its initial data acquisition, its output data undergoes a unified time reference mapping process. This process uses an external, highly stable clock as a unified reference, standardizing the raw timestamp information recorded by each sensing unit according to a predefined mapping function. The transformed time information is then written into the acquisition chain as the unique time identifier of the data frame. In this step, to ensure the continuity and stability of the time reference mapping, for detection data with high sampling frequency and wide dynamic range, the time intervals in the mapping results need to be equidistantly adjusted to eliminate the micro-delays generated by different sensing units during actual sensing triggering, thereby achieving unified alignment of data frames in the time dimension. Through this mapping process, it is ensured that all response data from different types of sensing units are synchronously incorporated into a structured chain with a unified time axis, forming a time-stamped acquisition chain with horizontal data consistency and vertical time traceability.

[0023] After time standardization of all sensor data, based on the constructed time-stamped acquisition chain, the data nodes are arranged and numbered chronologically, and a data reference set for subsequent processing is constructed at the end of the acquisition chain. This data reference set is the time-stamped consensus band, which contains multiple time node segments, each corresponding to a complete sensor data convergence window. In this step, to improve the representational capability of the time-stamped consensus band, the entire time-stamped acquisition chain is divided into continuous, non-overlapping time periods using an equally spaced partitioning method. At the end of each time period, a set of detection data with unique time identifiers and comprehensive signal source coverage is selected to construct consensus nodes. These consensus nodes will serve as the core components of the time-stamped consensus band, providing a unified time reference for subsequent fatigue evolution identification and state interpretation.

[0024] After the time-stamped consensus band is constructed, it is indexed and registered in the data structure, and its end information is synchronously written to the time output port of the acquisition chain. This ensures that subsequent processing steps can perceive the transition between different fatigue states based on the time position of each node in the consensus band. During this process, to maintain structural consistency between the time-stamped consensus band and the time-stamped acquisition chain, the generation of each consensus node must be bound to its corresponding source data segment, and its logical position in the original acquisition chain must be recorded to form a one-to-one time mapping relationship. Through this unified end-to-end processing method from data acquisition to time output, the synchronization, continuity, and structured organization of the detection data time reference are achieved in a multi-source sensor data environment, laying a unified time series foundation for subsequent multi-stage fatigue state identification, dominance switching labeling, and dynamic state control.

[0025] The detection data in the time-stamped acquisition chain is analyzed segment by segment according to the time-stamped consensus band. The phase change trend and energy decay trend of each detection data are compared and judged. The time position with pointing conflict is marked in the time-stamped consensus band, and the dominance switching prompt frame corresponding to the time-stamped consensus band is generated. Based on the unified time benchmark established by the time-stamp consensus band, segmented processing of the detection data in the time-stamp acquisition chain is carried out. By analyzing the correlation between the phase change trends and energy decay trends of various detection data, time segments with conflicting interpretation directions are identified, and conflict markers are added at positions strictly corresponding to the time-stamp consensus band, ultimately forming a dominant switching prompt frame for the closed-loop structure. The specific steps are as follows: Based on the established time-stamped acquisition chain and time-stamped consensus band structure, and using the time period sequences already constructed within the time-stamped consensus band, the time-stamped acquisition chain is divided into several consecutive analysis segments according to their chronological order. Each analysis segment is centered on a consensus node, with a fixed-length time window extending before and after it. All types of detection data fragments within the time window are fully extracted according to their source, response type, and original timestamp. To ensure sufficient analysis coverage and consistency of boundary conditions, the original arrangement order of various detection data within the acquisition chain is preserved during data fragment extraction, and the time reference information attached to each data fragment is synchronously indexed, thereby ensuring that all analyzed data maintains a consistent pairing relationship with the time-stamped consensus band in terms of time structure.

[0026] Based on the segmented extraction results, structural analysis of phase change trends and evolutionary analysis of energy decay trends are performed on the detection data segments within each analysis segment. The structural analysis of phase change trends, with time as the main axis, comprehensively classifies the response fluctuation direction, persistence, and abrupt change frequency of each detection data segment within the analysis segment. By observing the arrangement order of response peaks, periodic recurrence trends, and waveform extension morphology, the dynamic directionality exhibited within the corresponding time period is determined. Simultaneously, the evolutionary analysis of energy decay trends constructs corresponding energy dissipation paths based on the amplitude change trajectory of the detection data response values ​​along the time axis, and calibrates the decay trend type according to the decay rate, plateau length, and subsequent collapse degree. These two analytical directions are performed independently, ultimately generating phase trend description structures and energy trend description structures within each analysis segment, serving as the basis for subsequent comparison and judgment.

[0027] After independently describing the phase and energy trends, a comparative judgment is made based on the relative relationship between the two trend descriptions within each analysis segment to identify potential areas of conflict in interpretation. Specifically, if the gradient directions of the phase change trend and the energy decay trend on the time axis diverge in a certain segment, or if their trends undergo a short-term crossover within the time window, then the segment is considered to have a potential conflict. Such conflicts may originate from a shift in the dominant response between different detection data, causing the originally consistent fatigue evolution description to appear as an interpretation divergence within that segment. For time segments identified as having conflicts, their start and end time ranges are determined according to their logical position in the time-stamped acquisition chain. These time ranges are then labeled as time segments with dominance contest characteristics and marked and written into the prompt structure associated with the corresponding consensus node.

[0028] After marking all time segments as conflict-prone, the identified time positions of dominance struggles in each segment are summarized and arranged according to the time sequence of the time-stamped consensus bands to construct dominance switching prompt frames corresponding one-to-one with the time-stamped consensus bands. The dominance switching prompt frame uses the consensus node sequence as a framework, with each frame unit bound to a consensus time node. Its corresponding conflict time segment, detection data source type, trend conflict intensity level, and other information are packaged and encapsulated to form a structured identifier that can be referenced in subsequent stages. During the prompt frame generation process, all time positions marked as conflicting maintain their original logical connection with the time-stamped acquisition chain, ensuring that the prompt frame can accurately guide the identification and release rhythm adjustment of the dominance signal in subsequent weight traction and output control stages. Thus, the data trend conflict identification and prompt frame generation process based on the time-stamped consensus bands is completed, providing clear time basis and data support for the division of fatigue evolution stages and the control of dominance handover.

[0029] The time-stamped acquisition chain is divided into stages according to the time sequence of the dominance switching prompt frame. Within each stage, the dominant signal source and subordinate signal source of the detection data are determined, and the handover time of the dominant signal source is clearly marked in the dominance switching prompt frame, forming a weight traction list corresponding to each stage. Based on the established dominance switching prompt frames, it is necessary to perform stage-by-stage decomposition and signal dominance relationship determination operations, combined with the time sequence of the time-stamped acquisition chain, to extract the dominant and subordinate signal sources in each fatigue evolution stage. Furthermore, based on the dominance switching positions recorded in the prompt frames, a data weight index structure, i.e., a weighted traction list, is established for subsequent output reconstruction and dynamic control. The specific steps are as follows: Based on the time segment covered by the dominance handover prompt frame, the corresponding time points of each dominance signal handover event on the time-stamped acquisition chain are determined according to its internal time position markings. Using these dominance signal handover time points as boundaries, the entire time-stamped acquisition chain is divided into multiple continuous but non-overlapping time stages. Each time stage is defined by two adjacent dominance handover time points, forming a complete fatigue evolution sub-interval. To ensure the continuity and contextual relevance of the stage division structure, when setting stage boundaries, the starting point of each stage must be strictly aligned with the ending point of the previous stage, and the start and end time information within each stage must maintain a one-to-one correspondence with the handover nodes in the dominance handover prompt frame. In this way, a fatigue evolution stage sequence arranged chronologically is formed, with each stage having a closed structure, clear boundaries, and well-defined indexes, providing a stable time container for subsequent dominance determination operations.

[0030] Within each defined time phase, all detection data within that phase are centrally retrieved. Using the conflict time segments marked in the dominance switching prompt frame as a reference, the persistence and dynamic dominance of each detection data source's response characteristics within the current phase are identified. Specifically, the response change trajectory of each type of detection data in the acquisition chain within the phase, including the response intensity change range, energy contribution density, and phase proportion, serves as a crucial basis for determining its dominance. The time span and response activity of all detection data within the phase are compared. Detection data sources with high-frequency response, strong trend guidance, and stable decay paths are identified as the dominant signal sources within that phase; while other detection data sources in a low-activity state or passively following the dominant trend are classified as subordinate signal sources. The number of dominant signal sources can be a single source or multiple sources co-dominant, and this master-slave structure must remain unchanged within the current phase. This step ensures that the signal interpretation rights within each phase are clearly defined, thus physically corresponding to the main control factors of fatigue state evolution.

[0031] After identifying the dominant and subordinate signal sources within each stage, the dominance handover prompt frame node associated with the initial boundary of that stage is traced back. A clear label indicating the dominance handover time is inserted into this node, recording the type of dominant signal source taking over in the current stage, the handover method, and the starting time of the relay mechanism. To ensure the clarity and resolvability of the handover time, the handover label must include three dimensions: first, the time identifier of the handover time point in the time-stamped acquisition chain; second, the category comparison relationship of the dominant signal sources in the two stages; and third, the response cross-window interval of various signals within the handover section. By embedding such handover labels in the dominance handover prompt frame, the structural information of the prompt frame is further enriched, enabling it not only to have conflict prompting functions but also to carry a structural description of the dominance change path, thereby providing stronger control guidance for subsequent output regulation.

[0032] After confirming the master-slave relationship and writing the handover information for all fatigue evolution stages, the dominant signal source, subordinate signal source, and handover time information for each stage are summarized in chronological order and organized into a structured data mapping set, forming a weighted traction list. This weighted traction list uses fatigue evolution stages as index units, with each stage corresponding to a list entry. Each entry sequentially lists the dominant signal source type, subordinate signal source type, duration range of the master control signal within the stage, the specific handover time, and its logical position in the acquisition chain. As the hub linking the acquisition and output structures, the weighted traction list not only guides the execution of subsequent data rearrangement and release strategies but also determines whether the signal connection logic in the stage transition region meets the continuous output requirements. In subsequent stages, the weighted traction list serves as the execution blueprint for time window relay construction and rhythm regulation; its structural integrity and information accuracy directly affect the quality of the continuous expression of the fatigue state sequence.

[0033] The output order of detection data in the time-stamped acquisition chain is rearranged according to the weighted traction list. A continuous time window relay sequence is constructed before and after the handover time of the dominant signal source, so that the time position pointing to the conflict can be smoothly transitioned in the relay sequence, and a sequence index corresponding to the time-stamped consensus band is generated. Based on the information on the stage-specific dominant signal sources and handover times provided by the weighted traction list, to ensure the continuity and interpretability of the fatigue state output time series during the signal dominance change process, the output order of the detection data in the original time-stamped acquisition chain needs to be reconstructed holistically. On this basis, by constructing time-window relay sequences with stage succession relationships before and after the dominant signal source handover time, the data structure before and after the dominant switching node is softened, and a sequence index structure strictly aligned with the time-stamped consensus band is generated to provide precise support for subsequent dynamic control. The specific steps are as follows: Based on the dominant signal sources and transition times of each stage in the established weighted traction list, all detection data in the original time-scaled acquisition chain are reorganized into primary rearrangement units based on stages. Each primary rearrangement unit corresponds to a fatigue evolution stage in the weighted traction list. Within this stage, the dominant signal source is used as the primary sorting reference, and the data is sorted in a secondary manner based on time order. This ensures that the dominant signal data in this stage remains relatively continuous in the time dimension and is output earlier, while subordinate signal data is orderly filled between the dominant data, maintaining the integrity of the data structure without interrupting the output rhythm of the dominant signal. During the data processing, to ensure a smooth transition between subsequent stages, a buffer interface is established at the end of each rearrangement unit with the beginning of its next stage. Using the time span of the data frame as the connection benchmark, the dominant signal at the end of the stage is extended into the data window of the beginning of the next stage, serving as a time transition channel for the relay structure.

[0034] Building upon the initial rearrangement unit, a relay sequence of time windows, continuously connected across stages, is constructed around the handover moments of the dominant signal sources. This sequence uses each handover moment recorded in the weighted traction list as its core, extending fixed-duration data windows forward and backward, and forming overlapping intervals between two adjacent stages. This ensures that the dominant signal response data of the previous stage overlaps with that of the next stage in the time dimension. These overlapping intervals serve as the core buffer region of the relay sequence, coordinating the continuity of signal interpretation during the transition of dominance. During the construction of the relay sequence, to avoid excessive interference from non-dominant signals within the overlapping windows, subordinate signal data within the overlapping intervals are delayed in the output sorting. Only continuous transition segments between dominant signals are retained, and their time markers are used as binding references to the dominant handover moments to determine the order of dominance and subordination within the relay sequence. Through the bidirectional extension of this relay structure, a bridging channel with time buffering capabilities and data dominance consistency is constructed, laying the structural foundation for a natural transition of fatigue state expression at stage boundaries.

[0035] After the relay sequence is generated, all rearranged stage units and their corresponding time-window relay sequences are integrated in chronological order to generate a complete new output data sequence, and a sequence index structure is established accordingly. This sequence index uses consensus nodes in the time-stamped consensus band as the basic unit, with each consensus node corresponding to an index entry. Each entry includes multiple fields such as the start and end times of the rearranged data segment corresponding to the node, the category of the detected data source, the stage number to which the dominant signal belongs, and whether it contains a relay segment identifier. To maintain the bidirectional retrieval capability of the index structure, the index also needs to record the handover times of the previous and next dominant signals associated with the node, so that the generation source of the relay sequence can be traced back during dynamic control. Furthermore, for consensus nodes containing pointers to conflict time positions, their sequence index needs to include the dominant signal weight state at that position, as well as a descriptive field describing the range of signal interpretation changes before and after the rearrangement, so that subsequent identification of abrupt change risk areas based on the index structure can prioritize the execution of buffering mechanisms.

[0036] After the sequence index is generated, the rearranged output data sequence is expanded over time according to the structural relationship of its internal records to form an executable output structure for fatigue state assessment. Each data segment in this output structure clearly corresponds to the contribution range of the dominant signal at a certain stage. Before entering the dominant handover section, it automatically completes the transition and docking with the relay sequence, and within the relay section, it maintains the data output order consistent with the priority of the dominant signal, without signal master-slave role interchange. This data output method ensures that there is no output jump at the dominant switching position, and the time position pointing to conflict is smoothly absorbed by the buffering effect of the relay sequence, presenting a continuous and non-abrupt response path on the output trajectory, thus providing steady-state support in the time dimension for the expression of the entire fatigue evolution process. At this point, the rearrangement and relay mechanism is completed. The sequence index, as an intermediary structure, connects the entire path mapping relationship between the acquisition structure and the output structure, providing a foundation for subsequent dynamic control and state lock equal processing operations.

[0037] A dynamic control mechanism is introduced based on the sequence index. The output rhythm of the detection data is guided by the reverse time-series breathing phase-locking method. The time occupation of the subordinate signal source is temporarily suspended in the time window relay sequence, and the dominant signal source is released segment by segment according to the relay order, thereby eliminating the fatigue state jump caused by the switching of dominance in the time dimension. Based on the constructed relay sequence structure and sequence index content, a dynamic control mechanism is introduced to orderly guide the output rhythm of the detection data using a reverse-time breathing phase-locking method. This forms a slow-release signal release path in the time dimension, achieving rhythmic coordination and priority hierarchical management of the dominant and subordinate signals during the output process, ultimately achieving the goal of continuous state expression during the switching of dominance. The specific steps are as follows: After constructing the sequence index, all fields related to the handover position of the dominant signal source, stage affiliation, and relay structure markers need to be extracted from the sequence index. Based on this information, a dynamic control starting interval is constructed. This starting interval is located within the edge transition segment of each time window relay sequence. Its time start point corresponds to the handover time of the dominant signal source, extended forward by a set buffer length, and its time end point is the same time extension backward from the handover time. The entire starting interval is used as the basis for carrying the execution of the initial control command. Within this interval, according to the dominant signal stage affiliation recorded in the index fields, the type of dominant signal to appear is locked, and its priority status as the output traction target within the current control segment is established. At the same time, to coordinate with the rhythmic scheduling of the output sequence, the response rhythm of the dominant signal of the previous stage needs to be projected backward by a certain duration within the current starting interval to form an artificially elongated signal release tail, forming a soft connection structure with the leading segment of the dominant signal of the new stage, and constructing the basic rhythmic anchor point for subsequent breathing phase-locking actions.

[0038] Based on the definition of the initial interval, a reverse-time breathing phase-locking method is introduced within the relay sequence to formally trigger the output regulation action. This phase-locking method uses the dominant signal handover moment as the core anchor point, scanning backward from the handover point to gradually identify the dominant signal response segments that have not yet been released within that time period. The output sequence and time density of these segments are then compressed in stages, transforming the release rhythm from a fixed interval to a variable interval with slow-release characteristics, resulting in a breathing-like output pattern in terms of time sequence. Simultaneously, moving backward from the handover point, the release time window of the new stage's dominant signal is gradually expanded, and the time proportion of the new dominant signal in the output sequence is gradually increased according to a set growth rate, achieving a smooth transition of the dominant signal rhythm. This reverse-forward bidirectional phase-locking operation ensures that the dominant signals of the two stages complete rhythmic connection in the relay segment, while simultaneously forming staggered release bands on the time axis, suppressing abrupt behavior caused by the switch of dominance. Based on this, the constructed breathing phase-locking structure will present a release path with a gradual shift in center of gravity on the output time axis, contributing to a smooth transition of signal dominance.

[0039] After establishing a basic rhythmic framework during the breathing phase-locking process, a deferred operation is performed on subordinate signal source data located within the relay segment. Specifically, during the relay sequence execution phase, the output time occupation of subordinate signals is preferentially suppressed to avoid interpretive interference with the dominant signal. This operation uses the signal role identifier field in the sequence index as the criterion, sorting and delaying the time segment containing the subordinate signal data, migrating its release time to the later part of the relay sequence, while retaining its original time identifier. This ensures that the data content remains traceable without impacting the rhythm of the dominant signal. To enhance the supplementary value of subordinate signal release in expressing the system state, subordinate signal content can be gradually introduced after the relay segment ends, reducing output frequency to achieve a natural transition between master and subordinate roles. Furthermore, throughout the entire relay segment, to prevent the output rhythm from being dominated by non-dominant signals, a time occupation sorting strategy should be used to maintain the data dominance rate of the dominant signal within a unit of time at no less than a preset threshold, thereby providing rhythmic assurance for the continuous expression of the dominant signal.

[0040] After completing the segmented release of the dominant signal and the delayed scheduling of the subordinate signals, the rhythmic structure of the entire control process on the output time axis is bound and recorded with the corresponding time period structure to form the final output mapping trajectory. This output trajectory will exhibit a rhythmic characteristic of early slow release, transition, and later enhancement. All dominant signal sources are released in a non-abrupt manner at the transition nodes, and their response paths are consistent with the structure of the relay sequence, presenting a continuous signal band that expands in both directions. The subordinate signals that are temporarily delayed are inserted at a low frequency in the later part of the trajectory, which not only preserves the diversity of information sources but also avoids them from becoming dominant interference in the state judgment process. The entire output trajectory structure will eventually correspond one-to-one with the time nodes in the original time-stamped consensus band and be bound through the position index field, enabling precise time-level retrieval of the trajectory in subsequent fatigue evolution analysis or system state backtracking. At this point, the dynamic control mechanism has been executed, the output trajectory structure has been assembled, and the temporal dimension of the dominant switching jump problem has been effectively resolved through the anti-time-series breathing phase-locking method.

[0041] This invention achieves precise labeling and dynamic perception of signal dominance switching moments by constructing a time-stamped acquisition chain and a time-stamped consensus band, and by combining phase changes and energy decay trends of multi-source detection data for conflict identification. By introducing a relay-style structure and rhythmic control mechanism in the output stage, the dominant signal is continuously released during critical handover periods, avoiding non-physical jumps in the fatigue state trajectory caused by abrupt changes in the signal interpretation dominance relationship. This improves the continuity and interpretability of fatigue assessment results in the critical state stage.

[0042] Experimental verification and measured data analysis To verify the engineering applicability, data stability, and reproducibility of this method, a field test was conducted on the downstream overtaking lane of the G18 Rongwu Expressway (Shanping section) from K1248+000 to K1250+000. This pavement has been in service for 11 years and is a typical long-life asphalt pavement. It has been subjected to heavy traffic for a long time, and the base layer exhibits progressive fatigue damage such as poor interlayer bonding, structural cracks, and internal loosening. The disease evolution is gradual, with no sudden structural failure, making it suitable for fatigue evaluation tests of long-life pavement base layers. The test used a three-dimensional ground-penetrating radar multi-source sensor array to collect pavement structure data. The entire process was standardized, and original data such as equipment parameters and disease records were completely preserved. The detection process is traceable, and the results are reproducible.

[0043] This experiment employed three types of sensing units to collaboratively acquire data and complete unified timing calibration: a 1.5GHz three-dimensional ground-penetrating radar, using a high-precision grid scanning mode, acquired data on the dielectric properties of the base layer, interlayer conditions, and damage and loosening; strain and vibration sensing units acquired data on fatigue deformation and load vibration attenuation of the pavement structure, respectively. All sensing devices were uniformly connected to a 0.01ms high-precision external clock source, and timing alignment was achieved through equidistant triggering to eliminate time differences in data acquisition from multiple devices and ensure the consistency of timing among multi-source data.

[0044] The experiment used the area of ​​subgrade disease and the internal condition index of ISCI structure as the core evaluation indicators, and fully implemented the complete set of technical processes of multi-source data time-series calibration, signal identification, stage division, sequence reconstruction and dynamic rhythm control of the present invention, to verify the ability of the present method to identify and quantitatively evaluate the hidden fatigue damage of pavement subgrade.

[0045] For the three types of multi-source data collected on-site—radar, strain, and vibration—standardization processing is completed according to the specifications of this invention: unifying the timing reference, eliminating micro-latency acquisition, constructing a time-stamped acquisition chain and a time-stamped consensus band; analyzing the phase and energy attenuation characteristics of each sensor data, identifying sensor response differences, and generating a signal dominance switching prompt frame.

[0046] Based on the switching prompt frames, the road fatigue evolution stages are divided, the master-slave attributes of the sensor signals in each stage and the handover time are determined, and a weighted traction list is generated. By reconstructing the data output order, building a time window relay sequence, and combining it with a reverse-time breathing phase-locked loop control mechanism, the handover of the dominant signal is smoothed, the interference of the subordinate signal is suppressed, the problem of non-physical data jumps is completely avoided, and the entire process log is retained to ensure that the data and evaluation results are traceable and reproducible.

[0047] The ISCI (Internal Condition Index) effectively characterizes the structural integrity and fatigue deterioration of pavement base layers. A lower index indicates more severe base material deterioration, interlayer failure, and internal damage, making it suitable for progressive fatigue accumulation evaluation of long-life pavements. This experiment completed a 2km pavement full-coverage non-destructive test using 100-meter sections as units. The area and corresponding ISCI index of three types of defects—poor base layer adhesion, structural cracks, and internal loosening—were statistically analyzed to obtain complete quantitative parameters of base layer fatigue. The measured data are shown in the table below. Table 1. Road surface inspection data of 100-meter segments on the test section of G18 Rongwu Expressway

[0048] This field test utilizes the multi-source data temporal calibration, signal smoothing, and dynamic rhythm control algorithm of this invention to conduct a quantitative evaluation of pavement base fatigue. This method, by constructing a time-scaled acquisition chain, a time-window relay sequence, and a reverse-time breathing phase-locked loop mechanism, accurately identifies multi-source sensor signal conflicts and dominance switching nodes, resolving data fluctuations and non-physical jumps caused by sensor unit response differences. This experiment selected an in-service highway pavement with 11 years of service and gradually developing damage, covering multiple levels of base fatigue deterioration conditions through segmented, refined testing. Traditional testing suffers from large fluctuations and low accuracy in critical fatigue state evaluation due to frequent switching of multi-source signal dominance. This invention can dynamically adapt to the signal control logic of each fatigue stage, eliminating evaluation disturbances caused by signal switching while adhering to the inherent correlation between defects and the ISCI index, accurately distinguishing the different degrees of hidden fatigue deterioration in the base layer under the same defect conditions. The experimental results show that the fatigue evaluation trajectory output by this invention is smooth and stable, and highly matches the service law of steady accumulation of damage and gradual performance decline of long-life pavement. It can effectively remove surface disease interference, accurately identify hidden damage such as base material deterioration and interlayer bonding failure, and make up for the defects of poor stability and distortion of critical state identification of traditional evaluation methods. It can quantify the progressive fatigue deterioration state of base layer of long-life pavement with high precision and stability, and is suitable for non-destructive evaluation engineering scenarios of long-life pavement.

[0049] This invention dynamically identifies the master-slave relationship of each detection data point throughout the entire fatigue evolution process, and performs rhythmic regulation based on a weighted traction mechanism and sequence indexing, enabling fatigue state assessment to have autonomous smoothing capabilities at the time series level. The output rhythm of the detection data is no longer limited by static rules, but actively adjusts the signal release order according to stage changes, realizing flexible control and consistent expression of the state transition process, effectively enhancing the tracking stability of complex fatigue behavior and the response accuracy of the monitoring system.

[0050] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials, characterized in that, Includes the following steps: The detection data collected by each sensing unit in the intelligent sensing array is uniformly mapped to the same time base, a continuous and consistent time-stamped acquisition chain is constructed, and a time-stamped consensus band is generated at the end of the time-stamped acquisition chain. The detection data in the time-stamped acquisition chain is analyzed segment by segment according to the time-stamped consensus band. The phase change trend and energy decay trend of each detection data are compared and judged. The time position with pointing conflict is marked in the time-stamped consensus band, and the dominance switching prompt frame corresponding to the time-stamped consensus band is generated. The time-stamped acquisition chain is divided into stages according to the time sequence of the dominance switching prompt frame. Within each stage, the dominant signal source and subordinate signal source of the detection data are determined, and the handover time of the dominant signal source is clearly marked in the dominance switching prompt frame, forming a weight traction list corresponding to each stage. The output order of detection data in the time-stamped acquisition chain is rearranged according to the weighted traction list. A continuous time window relay sequence is constructed before and after the handover time of the dominant signal source, so that the time position pointing to the conflict can be smoothly transitioned in the relay sequence, and a sequence index corresponding to the time-stamped consensus band is generated. A dynamic control mechanism is introduced based on sequence indexing. The output rhythm of the detection data is guided by the reverse time-series breathing phase-locking method. The time occupation of the subordinate signal source is temporarily suspended in the time window relay sequence, and the dominant signal source is released segment by segment according to the relay order.

2. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 1, characterized in that, The steps for generating the time-stamped consensus band are as follows: Deploy an intelligent sensor array within the non-destructive testing area, set the acquisition timing parameters for each sensor unit, and initialize the acquisition time reference to the start time consistent with an external high-stability clock source. Use equidistant time trigger logic to anchor the data time information. After the sensing unit completes the acquisition, the original timestamp information is mapped according to a unified time reference, and the converted time information is written into the acquisition chain. Time alignment of different sensing data is achieved through time interval calibration. After completing time standardization, data nodes are arranged in chronological order and divided into non-overlapping time periods. At the end of each period, data with unique time identifiers and signal coverage are selected to build consensus nodes, forming a time-stamped consensus band. After the time-stamped consensus band is constructed, the consensus nodes are bound to the source data segments and their logical positions are recorded. The end information is then synchronously written to the time output port to complete the organization of the unified time series structure.

3. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 2, characterized in that, The consensus nodes are constructed using an equally spaced division method. At the end of each time period, the time position that simultaneously contains data collected by multiple sensor units is selected as the node reference, and the logical position of the node and its corresponding source data in the collection chain is established as a binding relationship.

4. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 2, characterized in that, The steps for generating the control switch notification frame are as follows: According to the time period sequence of the time-stamped consensus band, the time-stamped acquisition chain is divided into continuous analysis segments, and detection data segments containing the original arrangement order and time reference information are extracted. Based on the detection data segments of each analysis section, phase change trend structure analysis and energy decay trend evolution analysis are performed respectively to generate phase trend description structure and energy trend description structure. Within each analysis segment, based on the relative relationship between the phase trend description structure and the energy trend description structure, the time location of directional conflict is identified and labeled. After completing the marking of all conflict times, the master control switching prompt frames are compiled and constructed to correspond one-to-one with the time-stamp consensus bands, while maintaining their original logical connection with the time-stamp acquisition chain.

5. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 4, characterized in that, Phase change trend structure analysis includes classifying the response fluctuation direction, persistence and abrupt change frequency of the detection data within the analysis section. Energy decay trend evolution analysis includes constructing energy dissipation paths based on the trajectory of response value amplitude changes, and labeling the trend type according to the decay rate, plateau period length and later collapse degree.

6. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 4, characterized in that, The steps for generating the weighted traction list are as follows: According to the time position of the control switching prompt frame, the time-stamped acquisition chain is divided into multiple continuous time stages, and the stage boundaries and handover nodes are kept in a one-to-one correspondence. At each time stage, all detection data are retrieved, and the dominant signal source and subordinate signal source are determined by comparing the response change trajectory. After identifying the dominant signal source and the subordinate signal source, trace back the dominant power switching prompt frame node, insert a handover tag in the prompt frame and record the dominant signal handover time point; After confirming the master-slave relationship and writing the handover tags at each stage, the data is summarized and organized into a weighted traction list.

7. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 6, characterized in that, When forming the weighted traction list, the weighted traction list is arranged in chronological order according to the fatigue evolution stage. The source of the dominant signal in each stage remains consistent within the stage, and the handover time of the dominant signal is fixedly recorded as the stage boundary.

8. The non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 6, characterized in that, The steps for generating a sequence index are as follows: According to the dominant signal source and handover time recorded in the weighted traction list, the detection data in the time-stamped acquisition chain are rearranged in stages to form rearrangement units corresponding to the fatigue evolution stage. After completing the phased rearrangement, a time window relay sequence is constructed between adjacent phases around the handover time of the dominant signal source, and the output order of subordinate signals is delayed in the relay sequence. After forming the time window relay sequence, each rearranged unit is integrated with the corresponding relay sequence in chronological order to generate a new detection data output sequence; After the output sequence is generated, a sequence index is established based on the consensus nodes in the time-stamped consensus band, so that the time positions pointing to the conflict are smoothly transitioned in the relay sequence and the time correspondence is maintained.

9. A non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 8, characterized in that, The output order of the dominant signal in the relay sequence is arranged according to the stage affiliation relationship of the dominant signal in the weighted traction list. The time period occupied by the dominant signal in the output order is continuous, and the rhythm remains consistent before and after the handover moment.

10. A non-destructive testing method for evaluating the fatigue characteristics of long-life pavement materials according to claim 8, characterized in that, A dynamic control mechanism is introduced based on sequence indexing. The output rhythm of the detection data is adjusted using a reverse-time breathing phase-locking method. Within the time window relay sequence, the time occupation of the subordinate signal source is delayed, and the dominant signal source is output segment by segment according to the relay order. The steps are as follows: After completing the sequence index construction, extract the handover position and relay structure marker fields, set the start interval of the dominant signal handover, and establish the basic rhythm anchor point; A reverse-time breathing phase-locking method is introduced within the initial interval for regulation, compressing the unreleased dominant signal forward and extending the release time window of the dominant signal in the new stage backward; Based on completing the rhythm traction, the subordinate signals in the receiving section are sorted and delayed to retain the original time markers and control the time occupation ratio of the dominant signal. After the regulation process is completed, the rhythm structure and time period structure are bound and recorded to generate a continuous release output mapping trajectory and establish a corresponding relationship with the time nodes.

Images