Intelligent identification method for soil erosion by seismic response parameter inversion

Abstract

This invention discloses an intelligent method for identifying soil erosion by inverting seismic response parameters, relating to the fields of geotechnical engineering and engineering geophysics. The method includes the following steps: collecting records of soil response changes over a continuous time period and organizing them into a change sequence according to a uniform chronological order; identifying high-frequency, densely occurring segments within the change sequence; and recording the order and duration of changes within each high-frequency, densely occurring segment to construct a rhythm reference. This invention achieves effective separation of multi-source change processes through rhythm reference and temporal misalignment, avoiding mixed expression of change information and preserving the independent characteristics of different erosion initiation points. Simultaneously, based on the change evolution sequence and multi-point concurrent location markers, it enables independent determination of multiple processes, improving the ability to identify simultaneous multi-point development and the accuracy of erosion determination.

Description

Technical Field

[0001] This invention relates to the fields of geotechnical engineering and engineering geophysics, specifically to an intelligent method for identifying soil erosion by inverting seismic response parameters. Background Technology

[0002] Intelligent identification of soil erosion by seismic response parameter inversion refers to the process of collecting propagation response data of seismic waves or vibration signals in soil, extracting multi-dimensional response parameters such as amplitude, frequency, phase, and attenuation characteristics, and constructing inversion results that characterize changes in soil structure based on these parameters. A data-driven analysis mechanism is then introduced to jointly model historical samples and real-time response data. Machine learning methods are used to learn patterns and map features of response characteristics under different erosion states, thereby enabling automatic identification and classification of the degree of erosion development, location distribution, and evolution trend within the soil. This forms an intelligent identification method for underground hidden dangers. This process emphasizes establishing a nonlinear correspondence between response parameters and erosion states through data feature mining and model training, enabling the identification results to have adaptive updating and continuous optimization capabilities, which aligns with the technical scope of information processing and intelligent analysis.

[0003] The existing technology has the following shortcomings: In existing technologies, during seismic response parameter inversion and data-driven discrimination, when multiple latent erosion initiation points appear within the soil and generate densely superimposed high-frequency response features within the same time frame, the feature extraction process is prone to aliasing of change information from different sources due to the high similarity in time location and frequency distribution between adjacent change segments. This process integrates multiple independent change processes into a continuous change segment, leading machine learning discrimination results to misclassify multiple latent erosion initiation points as a single event. This problem results in an underestimation of the spatial distribution of latent erosion, making it difficult to identify the synchronous development of multiple points. During the rapid expansion phase of latent erosion, discrimination lags or even misjudgments are likely to occur, affecting the accuracy of subsequent risk identification.

[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 an intelligent method for identifying soil erosion by inverting seismic response parameters, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently identifying soil erosion by inverting seismic response parameters, comprising the following steps: Records of soil response changes over a continuous time period are collected and organized into a change sequence according to a uniform time order. High-frequency densely occurring segments are identified in the change sequence, and the order and duration of each change within the high-frequency densely occurring segments are recorded to construct a rhythm reference. Based on the rhythm reference, the corresponding time segment in the change sequence is traced back, the change process in the high-frequency densely occurring segment is split into segments, and the superimposed changes are time-staggered according to the order of appearance and duration difference in the rhythm reference, resulting in multiple mutually separated candidate change segments. The candidate change segments are used to continuously read the change sequence, and the change trajectory of each candidate change segment in the subsequent time range is continuously organized. The candidate change segments are classified according to the difference in the continuation rhythm to form mutually independent change evolution sequences. Based on the evolution sequence, a comparative analysis is performed on the corresponding time range in the evolution sequence. The change amplitude and advancement rhythm of each evolution sequence at the same time position are compared to identify the time segments of synchronous advancement and mark the corresponding multi-point concurrent positions in the evolution sequence. The change sequence is rearranged based on the concurrent locations of multiple points, and the corresponding relationship of each change evolution sequence is restored according to the corresponding time range. In the subsequent discrimination process, the change evolution sequence is used to make segment-by-segment judgments to avoid merging multiple change processes.

[0007] Preferably, the rhythm reference construction steps are as follows: The response change records formed within a continuous time range are collected moment by moment and arranged in the order of timestamps to form a change sequence. The changes at each time point in the change sequence are expressed in a unified format while maintaining the continuity of time intervals. Based on the change sequence, each time segment is scanned segment by segment. The number of changes is counted and the distribution is compared using time windows. The continuous time segments where the number of changes is concentrated are identified and high-frequency dense occurrence segments are formed. At the same time, the time range corresponding to the segments is recorded. For high-frequency and densely occurring segments, the changes within the segment are sorted by time position and the start and end times of each change are tracked to form a set of change descriptions containing sequential numbers and duration intervals. Based on the set of change descriptions, the sequence number is bound to the duration interval and arranged in chronological order to construct a rhythm reference structure, and the rhythm reference is associated with the time segment corresponding to the change sequence.

[0008] Preferably, the steps for obtaining candidate change fragments are as follows: Based on the sequence and duration information in the rhythm reference, the high-frequency densely occurring segments in the change sequence are back-tracked and located, the time range corresponding to the rhythm unit is mapped to the time position of the change sequence, and the change location relationship is established. Based on the relationship of change location, the high-frequency densely occurring segments in the change sequence are segmented one by one, and the continuous changes are grouped according to the sequence information in the rhythm reference to form an initial change segment with a start time and an end time. For the initial change segment, based on the duration interval in the rhythm reference, the superimposed changes are time-staggered, and changes at the same time position are separated and arranged according to sequential numbering, while maintaining the correlation with the time position of the change sequence; Based on the time-staggered processing results, the changes are continuously spliced ​​and the time range is re-marked to form multiple candidate change segments, and the candidate change segments are associated with the corresponding time segments of the change sequence.

[0009] Preferably, based on the sequential numbering and duration interval in the rhythm reference, the changes within the initial change segment are arranged in a time-expanded manner, and the different changes are sequentially numbered and distributed in a staggered manner along the time axis. The time range corresponding to each change is independently marked, thereby forming candidate change segments with continuous time-expanded relationships.

[0010] Preferably, the steps for forming the change evolution sequence are as follows: Starting with candidate change segments, the time range after the end time of the corresponding candidate change segment in the change sequence is continuously read, and change information is extracted along the time axis to form a continuous reading path. Around the continuous reading path, the change information in the change sequence is continuously organized, and the candidate change segments are connected with the changes within the continuous reading range in chronological order to form a change trajectory. Based on the temporal continuity and distribution of changes between the trajectories, a difference analysis of the continuity rhythm of the trajectories is conducted. Trajectories with consistent temporal progression are classified into the same category, while trajectories with different temporal progression are classified into different categories. Based on the classification results of change trajectories, change trajectories of the same category are arranged in chronological order and associated with the start time of candidate change segments and the subsequent change time range to form a change evolution sequence.

[0011] Preferably, when performing a continuous rhythm difference analysis on the change trajectory, the change time intervals and durations in the change trajectory are arranged accordingly, and the change trajectory is compared segment by segment in combination with the time axis progression order. The change trajectory categories are formed based on the consistency between the time progression relationship and the change distribution.

[0012] The preferred multi-point concurrent location marking process is as follows: Based on the time range corresponding to the change and evolution sequence, the corresponding time segments in the change sequence are extracted and a time range mapping relationship is established so that the change and evolution sequence and the change sequence form a time correspondence. Based on the temporal correspondence, the magnitude of change in each evolution sequence at the same time position is extracted and arranged to form a comparative relationship of the magnitude of change; Based on the comparison of the magnitude of change, the progress of each change evolution sequence on the time axis is organized, and the magnitude of change is associated with the order of progress to form an expression of the progress rhythm; Based on the expression of the advancement rhythm, the change amplitude and advancement rhythm of each change evolution sequence at the same time position are compared to identify the time segments of synchronous advancement and mark the concurrent positions of multiple points in the change sequence.

[0013] Preferably, the change sequence is rearranged based on the concurrent locations of multiple points, and the change sequence is organized and its corresponding relationship is restored according to the time range corresponding to the change evolution sequence. The segment-by-segment judgment steps based on the change evolution sequence during the discrimination process are as follows: Based on the time points corresponding to the concurrent positions of multiple points in the change sequence, the change sequence is segmented along the time axis to form multiple time segments with start and end times, while maintaining the chronological order of changes within each segment. Around time segments, the change information in the change sequence is classified and organized, and the changes within the time segment are matched according to the time range corresponding to the change evolution sequence, forming a change set corresponding to the change evolution sequence. Based on the set of changes, the changes of the same evolution sequence in different time periods are connected, and the changes in each time period are spliced ​​together in chronological order to restore the continuous change path. Based on the sequence of changes, the change path is determined segment by segment, and different change sequences are kept separate from each other within the same time range, so as to achieve independent expression of the change process.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a rhythm reference and structurally decomposes the change sequence, enabling multiple change processes that originally overlapped within the same time range to form a staggered unfolding relationship on the time axis. This separates change information from different sources from a continuous mixed state, forming mutually independent candidate change fragments at the change expression level. This effectively avoids the integration of change information into a single change process during the extraction stage, allowing the change characteristics at different starting points during the latent development process to be fully preserved, thereby improving the ability to express multi-source change structures.

[0015] This invention independently organizes the evolutionary sequence and introduces multi-point concurrent location markers, enabling multiple change processes to maintain their independent evolutionary paths within the same time frame. In subsequent judgment processes, the evolutionary sequence is processed segment by segment, thus preventing different change processes from intersecting and merging during the judgment stage. This allows for the accurate identification of multi-point synchronous development trends and improves the accuracy of judging the spatial distribution and temporal evolution of the erosion development process, thereby enhancing the reliability of underground hazard identification. Attached Figure Description

[0016] 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.

[0017] Figure 1 This is a flowchart of the intelligent soil erosion discrimination method based on seismic response parameter inversion according to the present invention. Detailed Implementation

[0018] 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.

[0019] This invention provides, for example Figure 1 The intelligent method for identifying soil erosion based on seismic response parameter inversion, as shown, includes the following steps: Records of soil response changes over a continuous time period are collected and organized into a change sequence according to a uniform time order. High-frequency densely occurring segments are identified in the change sequence, and the order and duration of each change within the high-frequency densely occurring segments are recorded to construct a rhythm reference. By organizing and reconstructing the response change records within a continuous time range, the temporal relationships, distribution density, and persistence characteristics between changes are fully expressed, thus forming a rhythmic reference that can be used to distinguish changes from multiple sources. The specific implementation steps are as follows: The system collects response change records formed within a continuous time range on a time-by-time basis. During the collection process, a preset time interval is used as the collection benchmark to record the response change state at each time point. The change performance corresponding to each time point is marked with a timestamp, and all records are sorted according to the order of the timestamps. The change information at discrete time points is arranged one by one in the order of the time axis to form a continuously unfolding change sequence. In the process of constructing the change sequence, the change record at each time point is expressed in a unified format so that each change contains a clear time position and corresponding change performance. At the same time, the time intervals between adjacent time points are continuously connected to ensure that the change sequence presents a continuously unfolding state in the time dimension, so that the temporal sequence relationship and interval relationship between changes are fully reflected in the change sequence. After the change sequence is formed, the occurrence of changes in each time segment of the change sequence is scanned segment by segment. A fixed-length time window slides along the time axis sequentially, and the number of changes contained in each time window is counted. The statistical results are compared with the number of changes in adjacent time windows. By continuously comparing the distribution of the number of changes, continuous time segments in the change sequence where the number of changes is consistently concentrated are identified. These continuous time segments are marked as high-frequency dense occurrence segments. In the process of identifying high-frequency dense occurrence segments, the start and end positions of the time windows are continuously recorded, and the time ranges formed by multiple consecutive time windows are merged to obtain the segment range with a clear start and end time. At the same time, the original change sequence is referenced during the segment identification process, so that a one-to-one correspondence is formed between the high-frequency dense occurrence segments and the specific time positions in the change sequence. After obtaining the high-frequency densely occurring segments, the changes within each segment are extracted one by one, and each change is sorted according to its temporal position in the change sequence. All changes are arranged from earliest to latest according to their occurrence time. At the same time, the start and end times of each change are continuously tracked. By marking the time range in which the change occurs continuously in the change sequence, the duration of each change is determined. The duration is then associated with the order of occurrence of the corresponding changes, so that each change corresponds to a unique sequence number and a corresponding duration interval. In this process, by continuously referencing the time markers in the change sequence, the change order and time position are kept consistent. By continuously recording the start and end times of the changes, the duration expression has a complete time span, thus forming a change description set that includes sequential relationships and continuous characteristics. After organizing the sequence and duration of changes within high-frequency, densely occurring segments, the sequence and duration information in the change description set are combined. Each change is taken as a basic unit, and its corresponding sequence number is bound to the duration interval. All changes are arranged in chronological order to construct a continuous rhythmic reference structure. In the process of forming the rhythmic reference, the sequence position and duration interval of each change are organized as rhythmic units, so that each rhythmic unit forms a continuous arrangement relationship on the time axis. At the same time, the rhythmic reference as a whole is associated with the corresponding time segment in the change sequence. This makes the rhythmic reference not only reflect the sequential relationship between changes within high-frequency, densely occurring segments, but also retain the time position mapping relationship of each change in the change sequence. Thus, the rhythmic reference can be used to distinguish different sources of change in subsequent processing and provide a time structure basis for change decomposition and evolution analysis.

[0020] Based on the rhythm reference, the corresponding time segment in the change sequence is traced back, the change process in the high-frequency densely occurring segment is split into segments, and the superimposed changes are time-staggered according to the order of appearance and duration difference in the rhythm reference, resulting in multiple mutually separated candidate change segments. Based on the established rhythm reference, by backtracking the corresponding time segments in the change sequence and combining the change order and duration relationship reflected by the rhythm reference, the superimposed changes in the high-frequency densely occurring segments are structurally decomposed and temporally staggered, so that the originally overlapping change process can be transformed into mutually independent candidate change segments. The specific implementation steps are as follows: Based on the established sequence and duration information in the rhythm reference, the high-frequency densely occurring segments in the change sequence are back-tracked and located. The time range corresponding to each rhythm unit in the rhythm reference is mapped to the specific time position in the change sequence. During the mapping process, the sequence number recorded in the rhythm reference is used as an index to match the changes in the change sequence one by one, so that each rhythm unit can find the corresponding change segment in the change sequence. At the same time, the duration interval in the rhythm reference is associated with the actual time span of the change in the change sequence, thereby forming a change positioning relationship guided by the rhythm reference in the change sequence. Through this positioning relationship, the change process that was originally continuously superimposed on the time axis is initially partitioned according to the rhythm reference, so that the time position of each change in the change sequence has a clear belonging mark. After establishing the location correspondence between the rhythm reference and the change sequence, the change process in the change sequence is segmented around the marked high-frequency densely occurring segments. During the segmentation process, based on the sequence information in the rhythm reference, the continuous changes in the change sequence are sequentially grouped, and changes with sequential continuity in adjacent time positions are divided into the same initial change segment. At the same time, the time range of each initial change segment is recorded, so that each initial change segment corresponds to a clear start time and end time. During the segmentation process, by continuously referencing the density of changes in the change sequence, the initial change segment is always within the range of high-frequency densely occurring segments, thereby ensuring the correspondence between the segmentation result and the original change distribution. After the initial change segment is formed, based on the duration differences of each change in the rhythm reference, the changes with superimposed relationships within the initial change segment are time-staggered. During the processing, the duration interval corresponding to each rhythm unit in the rhythm reference is used as the basis for time expansion. The changes within the initial change segment are processed one by one according to their sequential numbers. Changes that originally overlapped at the same time position are separated and arranged according to their duration intervals, so that different changes form a staggered distribution relationship on the time axis. During the staggering process, by referencing the original time position in the change sequence, the staggered changes still maintain the association with the original time segment. At the same time, by expressing the differences in duration intervals, the distribution of each change on the time axis is separated, thereby eliminating the overlapping relationship between superimposed changes and making each change form an independent time expansion path. After completing the time staggering process, the staggered changes are reorganized. The changes that have undergone time staggering are then continuously spliced ​​together according to their new time distribution relationship, and the time range corresponding to each change is re-marked, thus forming multiple candidate change segments that are separated from each other on the time axis. During the formation of candidate change segments, the sequential information in the rhythm reference is continuously incorporated, so that each candidate change segment maintains its correspondence with the rhythm reference. At the same time, the candidate change segments are associated with the original time segments in the change sequence, so that the candidate change segments have both an independent time unfolding structure and retain the mapping relationship with the original change sequence. This provides a clear and independent data foundation for subsequent change trajectory organization and evolution analysis based on candidate change segments.

[0021] The candidate change segments are used to continuously read the change sequence, and the change trajectory of each candidate change segment in the subsequent time range is continuously organized. The candidate change segments are classified according to the difference in the continuation rhythm to form mutually independent change evolution sequences. Based on the formation of candidate change segments, the change sequence is continuously read and organized along the extension process of the candidate change segments on the time axis. This allows each candidate change segment to form a change trajectory with continuous characteristics in the subsequent time range. Furthermore, the candidate change segments are classified according to the differences in their continuation rhythm, thereby constructing mutually independent change evolution sequences. The specific implementation steps are as follows: Starting with candidate change segments as the initial unit, the time range after the end time of the corresponding candidate change segment in the change sequence is continuously read. During the continuous reading process, the end time position of each candidate change segment is used as the starting point, and the change information in the change sequence is extracted step by step along the time axis. The changes that occur at each time position are recorded, so that the change information within the continuous reading range is continuously unfolded in chronological order. In this process, the start time and end time of the candidate change segment are marked as the basic time range, and the correspondence with the original time position in the change sequence is maintained during the continuous reading process, so that a continuous time connection is formed between the candidate change segment and its subsequent changes. At the same time, each candidate change segment is processed independently during the continuous reading process, so that different candidate change segments can form their own corresponding continuous reading paths in the change sequence. After completing the continuous reading, the change information in the change sequence is continuously organized around the continuous reading path corresponding to each candidate change segment. During the organization process, the candidate change segments and their subsequent time ranges are connected in series based on the time order. Changes belonging to the same continuous path are arranged in chronological order to form a continuous change trajectory. During the construction of the change trajectory, the original time range of the candidate change segment and the continuous reading range are merged as a whole, so that the change trajectory covers the entire change process from the start time of the candidate change segment to the continuous time range. At the same time, the time position of each change in the change trajectory is marked, so that the change trajectory presents a continuous unfolding state on the time axis, so that each candidate change segment corresponds to a complete change trajectory expression. After obtaining the change trajectories corresponding to each candidate change segment, the change trajectories are analyzed for rhythmic differences based on their temporal continuity and distribution. During the analysis, the changes in each trajectory are compared one by one in chronological order. The rhythmic performance of different change trajectories is distinguished based on the time interval between changes, the duration of changes, and the progression of changes on the time axis. Change trajectories with consistent temporal progression and distribution patterns are classified into the same category, while change trajectories with different temporal progression are classified into different categories. In this process, the rhythmic characteristics of the change trajectories are kept consistent with the sequence and duration information in the rhythmic reference by continuously referencing the original rhythmic reference information, thereby ensuring that the division of rhythmic differences has continuity and consistency. After classifying the change trajectories, those belonging to the same category are integrated and arranged chronologically. The start times of corresponding candidate change segments and their subsequent time ranges are then correlated to form a change evolution sequence covering the entire time range. During the formation of this sequence, candidate change segments are used as the starting point, and the continuously read change trajectories are used as the evolution process. This ensures that each change evolution sequence has a clear starting point and a continuous path of change development. Simultaneously, the change evolution sequence maintains its correspondence with the original time position in the change sequence, giving it both an independent temporal structure and reflecting its actual distribution within the change sequence. This allows for the independent evolution of multiple candidate change segments on the time axis, providing a clear foundation for subsequent comparative analysis and multi-point concurrent identification of the change evolution sequences.

[0022] Based on the evolution sequence, a comparative analysis is performed on the corresponding time range in the evolution sequence. The change amplitude and advancement rhythm of each evolution sequence at the same time position are compared to identify the time segments of synchronous advancement and mark the corresponding multi-point concurrent positions in the evolution sequence. Based on the established evolutionary sequence and its temporal correspondence with the change sequence, by comparing and analyzing the evolutionary sequence with the corresponding time ranges within the change sequence, the changes in different evolutionary sequences at the same time position can be uniformly expressed and compared horizontally. This allows for the identification of time segments with synchronous progression relationships, and further, the marking of multiple concurrent positions in the change sequence. The specific implementation steps are as follows: Based on the time range corresponding to each evolutionary sequence in the evolutionary sequence, the same time range in the evolutionary sequence is extracted and processed. The time segments in the evolutionary sequence corresponding to each evolutionary sequence are mapped segment by segment, so that each evolutionary sequence has a corresponding time interval expression in the evolutionary sequence. In the mapping process, the time axis in the evolutionary sequence is segmented and extracted with the start time and end time of the evolutionary sequence as the boundary, and the extracted time segments are bound with the corresponding evolutionary sequences, so that the evolutionary sequences form a consistent correspondence in time range between the evolutionary sequences. At the same time, the time ranges between multiple evolutionary sequences are uniformly organized, so that different evolutionary sequences have an alignable time position expression on the time axis, thereby providing a unified time benchmark for subsequent comparative analysis at the same time position. After aligning the time ranges, the magnitude of change in each evolutionary sequence at the corresponding time position within the evolutionary sequence is extracted and arranged point by point, based on the corresponding time position of the evolutionary sequence. During the processing, a unified time base is used as an index to synchronously record the magnitude of change corresponding to each evolutionary sequence at each time position, and multiple magnitudes at the same time position are arranged side by side, so that the changes in different evolutionary sequences at the same time position form a horizontal comparison relationship. At the same time, during the magnitude extraction process, the original time sequence information in the evolutionary sequence is continuously associated with the time position in the evolutionary sequence, so that the expression of magnitude of change not only reflects the internal development of the evolutionary sequence, but also reflects its time distribution position in the evolutionary sequence, thus forming a magnitude of change sequence that can be used for comparative analysis. After aligning the magnitudes of change, the progression rhythm is further organized around the timeline of each evolution sequence. During this process, the progression order of each change in the evolution sequence is extracted segment by segment, and the progression process is expressed temporally by combining the distribution intervals of the changes on the timeline. This ensures that each evolution sequence forms a complete description of the progression rhythm. In organizing the progression rhythm, the magnitude of change is correlated with the progression order, so that the progression rhythm includes not only the temporal progression relationship but also the magnitude of change at the corresponding time position. This forms a comprehensive rhythmic expression structure that includes both temporal progression relationship and change performance. At the same time, the progression rhythms are arranged side by side between different evolution sequences, allowing for horizontal comparison of the progression of each evolution sequence within the same time range. After comparing and organizing the magnitude of change and the pace of advancement, a comprehensive comparison is made of the magnitude of change and the pace of advancement of each change evolution sequence at the same time position. During the comparison, the magnitude of change of each change evolution sequence at the same time position is analyzed one by one with the corresponding pace of advancement. When multiple change evolution sequences show a consistent time advancement relationship within a continuous time range and the magnitude of change at the corresponding time position maintains a synchronous trend, the continuous time range is identified as a synchronous advancement time segment. The position corresponding to this time segment is marked in the change sequence to form a multi-point concurrent position identifier. During the marking process, the multi-point concurrent position is bound to the time position in the change sequence, so that the multi-point concurrent position can be clearly indicated in the change sequence, while maintaining the correspondence with each change evolution sequence. Thus, the multi-point concurrent position can not only reflect the synchronous development state of multiple change evolution sequences, but also accurately locate them in the change sequence, providing a direct basis for subsequent change sequence recombination and discrimination processing based on the multi-point concurrent position.

[0023] The change sequence is rearranged based on the concurrent locations of multiple points, and the change evolution sequence is organized according to the corresponding time range and the corresponding relationship is restored. In the subsequent discrimination process, the change evolution sequence is used to make segment-by-segment judgments to avoid merging multiple change processes. After the concurrent locations at multiple points have been marked in the change sequence, the change sequence is reorganized based on time position, and the correspondence between changes is reconstructed by combining the time range and development path of the change evolution sequence. This ensures that each change process remains independently expressed in subsequent judgment processes. The specific implementation steps are as follows: Using the time point corresponding to each marked multi-point concurrent position in the change sequence as the basis for division, the change sequence is segmented along the time axis. During the segmentation process, starting from the starting time position of the change sequence, the change information of each time point is read one by one in chronological order. When the time point corresponding to a certain multi-point concurrent position is read, that time point is taken as the end boundary of the current time segment, and all the change information contained in that time segment is collected and organized to form an independent time segment. Then, the time point corresponding to the multi-point concurrent position is taken as the starting boundary of the next time segment, and the change information is read backward along the time axis until the time point corresponding to the next multi-point concurrent position is encountered. The above segmentation process is repeated until the end of the change sequence, thereby dividing the complete change sequence into multiple continuous time segments with clear start and end times. In each time segment, the change records of each time point are arranged in the original chronological order, and the time position corresponding to each change is recorded, so that the internal structure of the time segment maintains a complete temporal continuity. After dividing the time into time segments, the change information within each time segment is assigned and processed. During the processing, the change information of each time point within the time segment is read sequentially, and the changes within the time segment are matched one by one with the start and end time ranges corresponding to each change evolution sequence. When the time position of a change falls within the time range of a change evolution sequence, the change is assigned to the change set corresponding to that change evolution sequence, and the change is inserted into the corresponding position of that change evolution sequence within the time segment according to the time order. Thus, multiple change sets corresponding to different change evolution sequences are formed within each time segment. During the processing, each change is uniquely assigned, so that the same change corresponds to only one change evolution sequence. At the same time, the change sets of different change evolution sequences within the same time segment are recorded independently, so that different change evolution sequences form separate change path expressions within the time segment. After completing the classification of changes within each time segment, the change sets of the same evolution sequence in multiple time segments are connected across time segments. During the process, the evolution sequence is used as the main line, and the change sets in each time segment belonging to the same evolution sequence are spliced ​​together in chronological order. The end of the change set in the previous time segment is sequentially connected to the beginning of the change set in the next time segment, so that the scattered changes in each time segment can form a continuous evolution path. At the same time, the time position of each change is continuously recorded during the connection process, so that the spliced ​​change path maintains a complete chronological order on the time axis. Through this cross-segment connection process, the changes that were originally scattered in multiple time segments due to the division of multiple concurrent positions are restored into a continuous evolution path, thereby realizing the restoration of the correspondence of the evolution sequence within the overall time range. After rearranging the change sequences and restoring the correspondence of the change evolution sequences, the change paths in the change evolution sequences are judged segment by segment based on the change evolution sequences as the basic unit. During the judgment process, the change paths are divided into several continuous time segments according to the time unfolding order of the change evolution sequences, and the change information is processed independently in each time segment. This keeps the changes of different change evolution sequences in the same time range separate. During the segment-by-segment judgment process, there is no cross-reference between different change evolution sequences, so that each change evolution sequence participates in subsequent processing as an independent object. This avoids the situation where multiple change processes in the change sequence are integrated into a single change process under the condition of time overlap. It enables multiple change processes formed by multiple points to participate in subsequent judgment in the form of independent paths, realizing the separate expression of multi-source change processes.

[0024] This invention constructs a rhythm reference and structurally decomposes the change sequence, enabling multiple change processes that originally overlapped within the same time range to form a staggered unfolding relationship on the time axis. This separates change information from different sources from a continuous mixed state, forming mutually independent candidate change fragments at the change expression level. This effectively avoids the integration of change information into a single change process during the extraction stage, allowing the change characteristics at different starting points during the latent development process to be fully preserved, thereby improving the ability to express multi-source change structures.

[0025] This invention independently organizes the evolutionary sequence and introduces multi-point concurrent location markers, enabling multiple change processes to maintain their independent evolutionary paths within the same time frame. In subsequent judgment processes, the evolutionary sequence is processed segment by segment, thus preventing different change processes from intersecting and merging during the judgment stage. This allows for the accurate identification of multi-point synchronous development trends and improves the accuracy of judging the spatial distribution and temporal evolution of the erosion development process, thereby enhancing the reliability of underground hazard identification.

[0026] 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 method for intelligently identifying soil erosion by inverting seismic response parameters, characterized in that, Includes the following steps: Records of soil response changes over a continuous time period are collected and organized into a change sequence according to a uniform time order. High-frequency densely occurring segments are identified in the change sequence, and the order and duration of each change within the high-frequency densely occurring segments are recorded to construct a rhythm reference. Based on the rhythm reference, the corresponding time segment in the change sequence is traced back, the change process in the high-frequency densely occurring segment is split into segments, and the superimposed changes are time-staggered according to the order of appearance and duration difference in the rhythm reference, resulting in multiple mutually separated candidate change segments. The candidate change segments are used to continuously read the change sequence, and the change trajectory of each candidate change segment in the subsequent time range is continuously organized. The candidate change segments are classified according to the difference in the continuation rhythm to form mutually independent change evolution sequences. Based on the evolution sequence, a comparative analysis is performed on the corresponding time range in the evolution sequence. The change amplitude and advancement rhythm of each evolution sequence at the same time position are compared to identify the time segments of synchronous advancement and mark the corresponding multi-point concurrent positions in the evolution sequence. The change sequence is rearranged based on the concurrent locations of multiple points, and the corresponding relationships of each change evolution sequence are restored according to the corresponding time range. In the subsequent discrimination process, the change evolution sequence is used to make segment-by-segment judgments.

2. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 1, characterized in that, The steps for constructing a rhythm reference are as follows: The response change records formed within a continuous time range are collected moment by moment and arranged in the order of timestamps to form a change sequence. The changes at each time point in the change sequence are expressed in a unified format while maintaining the continuity of time intervals. Based on the change sequence, each time segment is scanned segment by segment. The number of changes is counted and the distribution is compared using time windows. The continuous time segments where the number of changes is concentrated are identified and high-frequency dense occurrence segments are formed. At the same time, the time range corresponding to the segments is recorded. For high-frequency and densely occurring segments, the changes within the segment are sorted by time position and the start and end times of each change are tracked to form a set of change descriptions containing sequential numbers and duration intervals. Based on the set of change descriptions, the sequence number is bound to the duration interval and arranged in chronological order to construct a rhythm reference structure, and the rhythm reference is associated with the time segment corresponding to the change sequence.

3. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 2, characterized in that, The steps for obtaining candidate change fragments are as follows: Based on the sequence and duration information in the rhythm reference, the high-frequency densely occurring segments in the change sequence are back-tracked and located, the time range corresponding to the rhythm unit is mapped to the time position of the change sequence, and the change location relationship is established. Based on the relationship of change location, the high-frequency densely occurring segments in the change sequence are segmented one by one, and the continuous changes are grouped according to the sequence information in the rhythm reference to form an initial change segment with a start time and an end time. For the initial change segment, based on the duration interval in the rhythm reference, the superimposed changes are time-staggered, and changes at the same time position are separated and arranged according to sequential numbering, while maintaining the correlation with the time position of the change sequence; Based on the time-staggered processing results, the changes are continuously spliced ​​and the time range is re-marked to form multiple candidate change segments, and the candidate change segments are associated with the corresponding time segments of the change sequence.

4. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 3, characterized in that, Based on the sequential numbering and duration interval in the rhythm reference, the changes within the initial change segment are arranged in a time-expanded manner. The different changes are sequentially numbered and distributed in a staggered manner along the time axis, and the time range corresponding to each change is independently marked, thereby forming candidate change segments with continuous time expansion relationship.

5. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 3, characterized in that, The steps for forming the change and evolution sequence are as follows: Starting with candidate change segments, the time range after the end time of the corresponding candidate change segment in the change sequence is continuously read, and change information is extracted along the time axis to form a continuous reading path. Around the continuous reading path, the change information in the change sequence is continuously organized, and the candidate change segments are connected with the changes within the continuous reading range in chronological order to form a change trajectory. Based on the temporal continuity and distribution of changes between the trajectories, a difference analysis of the continuity rhythm of the trajectories is conducted. Trajectories with consistent temporal progression are classified into the same category, while trajectories with different temporal progression are classified into different categories. Based on the classification results of change trajectories, change trajectories of the same category are arranged in chronological order and associated with the start time of candidate change segments and the subsequent change time range to form a change evolution sequence.

6. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 5, characterized in that, When performing a continuous rhythm difference analysis on the change trajectory, the time intervals and durations of the change trajectory are arranged in a corresponding manner, and the change trajectory is compared segment by segment in combination with the time axis progression order. The change trajectory categories are formed based on the consistency between the time progression relationship and the change distribution.

7. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 5, characterized in that, The multi-point concurrent location marking process is as follows: Based on the time range corresponding to the change and evolution sequence, the corresponding time segments in the change sequence are extracted and a time range mapping relationship is established so that the change and evolution sequence and the change sequence form a time correspondence. Based on the temporal correspondence, the magnitude of change in each evolution sequence at the same time position is extracted and arranged to form a comparative relationship of the magnitude of change; Based on the comparison of the magnitude of change, the progress of each change evolution sequence on the time axis is organized, and the magnitude of change is associated with the order of progress to form an expression of the progress rhythm; Based on the expression of the advancement rhythm, the change amplitude and advancement rhythm of each change evolution sequence at the same time position are compared to identify the time segments of synchronous advancement and mark the concurrent positions of multiple points in the change sequence.

8. The intelligent method for identifying soil erosion by seismic response parameter inversion according to claim 7, characterized in that, The change sequence is rearranged based on the concurrent locations of multiple points, and the change sequence is organized and its corresponding relationship is restored according to the time range corresponding to the change evolution sequence. The segment-by-segment judgment steps based on the change evolution sequence are as follows: Based on the time points corresponding to the concurrent positions of multiple points in the change sequence, the change sequence is segmented along the time axis to form multiple time segments with start and end times, while maintaining the chronological order of changes within each segment. Around time segments, the change information in the change sequence is classified and organized, and the changes within the time segment are matched according to the time range corresponding to the change evolution sequence, forming a change set corresponding to the change evolution sequence. Based on the set of changes, the changes of the same evolution sequence in different time periods are connected, and the changes in each time period are spliced ​​together in chronological order to restore the continuous change path. Based on the sequence of changes, the change path is determined segment by segment, and different change sequences are kept separate from each other within the same time range, so as to achieve independent expression of the change process.

Images