A large drainage pipe health condition dynamic evaluation method based on multi-algorithm coupling

Abstract

The application discloses a kind of based on multi-algorithm coupling's large-scale drainage pipeline health condition dynamic evaluation method, specifically related to drainage pipeline health evaluation data processing technical field, including obtaining multiple period pipeline records, same pipe segment period collection is executed to pipeline record, form period sequence, output period index table and action table;Change point detection is executed to period index table, difference operation is carried out to the adjacent period value of same index in period sequence, write in direction bit according to difference symbol, and the period position that direction bit occurs switching is written into change point table;With the structure result change point in change point table as result variable point, associated change point of same pipe segment with result variable point is read along period inverse order, stop reading when encountering reverse change point;By detecting as trigger variable point, amplification variable point and result variable point to multiple period index change, and change point after maintenance action is reproduced as action challenge, write back Markov transition matrix according to, then health state is recursively by Markov model.

Description

Technical Field

[0001] This invention relates to the field of drainage pipeline health assessment data processing technology, and more specifically, to a method for dynamic assessment of the health status of large-scale drainage pipelines based on multi-algorithm coupling. Background Technology

[0002] In the dynamic assessment of the health status of large drainage pipelines, existing treatments mainly revolve around the current health level of the pipe section, the trend of risk changes, and the maintenance priority. During implementation, data related to the corrosive environment, microbial activity, hydraulic operation, material properties, and structural defects are generally collected, and multi-period indicators are input into scoring rules, prediction models, or Markov state transition models. The health level of the pipe section is corrected by the changes in indicators in the next assessment period. Taking the zonal inspection and periodic maintenance of urban main drainage pipelines as an example, inlet inspection, CCTV verification, water quality sampling, water outage operations, and traffic diversion are all limited by the operation window, drainage scheduling, and budget batches. Within a maintenance cycle, the same pipe section can often only complete limited actions such as dredging, flushing, anti-corrosion, local repair, or sealing. After the actions are completed, it is still necessary to enter the next evaluation cycle. In this case, it is often verified on-site that after dredging, the sediment retention index decreases but the initial pH value of the material continues to decrease; after anti-corrosion, the corrosion characterization weakens but the cracks still expand; and after local repair, the same defects reappear in adjacent locations. This shows that the changes in the indicators after the actions do not necessarily mean that the deterioration chain has been broken. The existing dynamic assessment still tends to directly write the indicators after the actions into the next state, causing Markov transitions to misjudge the improvement of the characterization as the termination of deterioration, or to misjudge the recurrence of the result fragment as new deterioration. The technical problem this application aims to solve is: how to transform maintenance actions into verification criteria for whether trigger segments, amplification segments, and result segments have been cut off in cross-cycle health assessments following limited maintenance actions in large drainage pipelines, so as to avoid being misled by the apparent changes in indicators caused by the actions during Markov state transitions. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a dynamic assessment method for the health status of large-scale drainage pipelines based on multi-algorithm coupling. This method detects changes in multi-period indicators as trigger points, amplification points, and result points, and uses the recurrence of changes after maintenance actions as the basis for action challenges to write the Markov transition matrix. The health status is then recursively deduced from the Markov model, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic assessment of the health status of large-scale drainage pipelines based on multi-algorithm coupling, comprising: S1. Obtain multi-period pipeline records, perform same-segment period aggregation on the pipeline records to form a period sequence, and output a period index table and an action table. S2. Perform change point detection on the periodic indicator table, perform difference operation on adjacent period values ​​of the same indicator in the periodic sequence, write the difference sign into the direction position, and write the period position where the direction position changes into the change point table. S3. Using the structural result change point in the change point table as the result change point, read the associated change points in the same pipe segment as the result change point in reverse order along the cycle. Stop reading when encountering a reverse change point. Write the associated change point that is connected to the result change point before stopping as the amplified change point. Write the associated change point that is connected to the amplified change point as the trigger change point. Output the change point state table. S4. Based on the variable state table, perform Markov transition construction, write the trigger variable, amplification variable and result variable into the state column, write the transition edge from the trigger variable to the amplification variable and from the amplification variable to the result variable, count the number of transitions according to the same starting state and ending state, and generate the transition matrix by dividing by the total number of transitions for the same starting state. S5. Execute the action challenge according to the action table, pair the action-directed state with the post-action recurrence change point, write the challenge mark into the matrix item formed by the action-directed state in the transition matrix, and output the transition matrix after the challenge.

[0005] In a preferred embodiment, it further includes: S6. Using the state sequence as the state space of the Markov model, the challenge transition matrix as the state transition matrix of the Markov model, and the initial state vector generated from the initial periodic change point state, the product of the state vector and the state transition matrix is ​​recursively executed in the Markov model in periodic order, and the health level, instability direction and chain-level maintenance direction are output according to the challenge mark and the result change point direction position in the recursive vector.

[0006] In a preferred embodiment, S1 includes: S11. Read the pipe segment index and cycle index for multi-cycle pipeline records, group them into pipe segment groups according to the same pipe segment index, and write the cycle bit into the pipe segment group in ascending order of cycle index, and output the pipe segment cycle bit table. S12. Read the record with the same index name for each cycle position in the pipe segment cycle position table. If the same index record exists in the previous cycle position, write it to the successor position. If the same index record does not exist in the previous cycle position, write it to the missing position. Generate the index cycle string according to the cycle position order and output the cycle index table. S13. Read the completion cycle of maintenance actions in the multi-cycle pipeline record, write the cycle bit into which the completion cycle falls as the action bit, write the cycle bit after the action bit as the challenge read bit, and output the action table.

[0007] In a preferred embodiment, S2 includes: S21. Read the index cycle string of the same pipe segment and the same index in the cycle index table, perform the difference between the next bit and the previous bit for the index value of adjacent cycle bits, write the positive, negative and zero results of the difference value into the up bit, down bit and zero bit respectively, and output the direction bit string. S22. Perform change point detection on the direction bit string, compare the current cycle bit direction bit with the previous cycle bit direction bit, if they are different, write the current cycle bit into the candidate change point, and concatenate the previous direction bit and the current direction bit into the change type code, and output the candidate change point table. S23. Perform a continuation check on the candidate change point table. When the direction position of the next cycle position of the candidate change point has the same value as the current direction position, retain the candidate change point and write it into the change point table. When the direction position of the next cycle position of the candidate change point returns to the previous direction position, write the swing position and remove it from the change point table.

[0008] In a preferred embodiment, S3 includes: S31. Read the result change points from the change point table, read the candidate change points in reverse order along the same pipe section cycle, and connect each candidate change point with the candidate change points read in the next cycle to form a candidate succession chain that traces back from the result change points to the historical cycle. S32. Perform restricted path calculation on the candidate receiving chain. Accumulate the periodic position difference for each backtracking item of the path. When a candidate change point with the opposite direction of the deterioration of the result change point is encountered, stop the path and write the blocking mark. Paths without blocking marks are formed into a receiving order table according to the fixed order of the accumulated periodic position difference and the change point number. S33. Perform forward verification based on the succession order table. Read from the candidate path with the highest order to the result change point. When the direction position of the change point in the path is consistent with the reverse reading direction position, write the change point of the adjacent result change point as the amplified change point and write the change point of the adjacent amplified change point as the trigger change point, and output the change point state table. If the consistency is not met, read the next candidate path and continue the verification.

[0009] In a preferred embodiment, S4 includes: S41. Read the succession records in the variable point state table one by one, and use the reading order of the variable points in the succession record as the succession order. Concatenate the variable point number and direction bit of each variable point into a status identifier in a fixed byte order. If the status identifier is not written to the status column, add column bits in the writing order. If the status identifier has been written to the status column, reuse the original column bits and output the succession record with column bits. S42. Perform Markov transition counting on the record with column position succession. Read the first succession column, the second succession column, and the last succession column. Increment the row and column count of the first succession column pointing to the second succession column by one. Increment the row and column count of the second succession column pointing to the last succession column by one. Write the two counts into the same transition group and output the Markov transition count table. S43. Perform transition matrix calculation on the Markov transition count table. Accumulate the row and column counts row by row to obtain the row count sum. Divide the count of each row and column by the count sum of its row to obtain the matrix item. When the row count sum is zero, write zero value to the corresponding matrix item in the row and output the transition matrix.

[0010] In a preferred embodiment, S5 includes: S51. Using the action completion cycle as the starting point, write the column position of the action pointing to the state in the state column as the challenge source position. Starting from the next cycle, read back the same source change point cycle by cycle, calculate the difference between the read-back cycle position and the action completion cycle and write it as the reproduction distance, calculate the difference between the column position of the same source change point and the challenge source position and write it as the state position difference, and output the reproduction sequence. S52. Using the matrix items containing challenge source positions in the reproduction sequence and transition matrix as participants, calculate the sum of the absolute differences between the state position difference and the column position differences at both ends of the matrix item for each reproduction point. Then, accumulate the reproduction distance and the number of occupied items in the same transition group to obtain the pairing value. Write the pairing value and the reproduction point writing order in ascending order to the challenge edge sequence.

[0011] In a preferred embodiment, S5 further includes: S53. Read the challenge edges line by line according to the challenge edge sequence. When neither the matrix item nor the reappearance point is occupied, write it into the action challenge pairing table and occupy the matrix item and the reappearance point simultaneously. Then, based on the result that the reappearance point hits both ends of the matrix item, only hits the end of the matrix item, or only hits the beginning of the matrix item, write the uncut challenge flag, the existing damage challenge flag, or the temporarily inactive challenge flag into the transition matrix, and output the transition matrix after the challenge.

[0012] In a preferred embodiment, S6 includes: S61. Read the state column order, the transition matrix after the challenge and the initial periodic change point state. Use the state column order as the state order of the Markov model. Write a value to the hit column of the initial periodic change point state and write a zero value to the missing column. Output the current state vector and the corresponding matrix table. S62. Within the Markov model, perform matrix recursion on the current state vector and the corresponding matrix table. Read each non-zero vector position and multiply it with the matrix item in the same row. Accumulate the product according to the column position of the end of the matrix item. At the same time, write the source row position, the end column position, and the challenge flag of the product into the contribution list. Output the candidate recursive vector and the contribution list.

[0013] In a preferred embodiment, S6 further includes: S63. Using the contribution list and candidate recursive vector as input for the action challenge route, read back the challenge markers in the contribution list one by one, keep the product of the uncut markers in the original endpoint column, rewrite the product of the existing damage markers in the result variable column, rewrite the product of the temporarily inactive markers in the column to be reviewed, and output the challenge route vector. S64. Using the challenge routing vector and the direction position of the result change point in the same period as the direction projection input, read the state direction position corresponding to each non-zero vector position and compare it with the direction position of the result change point. When the direction positions are the same, write the non-zero vector position into the state vector of the next period. When the direction positions are reversed, write the non-zero vector position into the conflict chain position. Feed the state vector of the next period back into the Markov model and continue to recursively push to the last period. Output the health level, instability direction and chain-level maintenance direction.

[0014] The technical effects and advantages of this invention are as follows: This scheme uses action challenges to reverse the matrix terms of the variable points that are reproduced after the action, so that the improvement of representation is not directly equivalent to the termination of the transition, thus relatively reducing the risk of false breakage of the deterioration chain. By performing differential analysis and succession verification on adjacent periodic index values, short-term fluctuations are not included in the change point table, thus relatively improving the reliability of the state source. By retrospectively tracing back from the result change point and verifying the trigger change point and amplifying the change point in the forward direction, the situation of mistakenly identifying the accompanying change as a source of degradation is relatively reduced; Aggregate similar variable states into state identifiers and statistically analyze Markov transitions to make the deterioration continuation enter a unified matrix, thereby relatively improving the recursive consistency. The matrix recursion results are carried over by the contribution linked list and fed back through challenge routing and directional projection, so that the maintenance action influences the recursion process; Based on the last state vector and the result change point direction, the health level, instability direction and chain-level maintenance direction are output, so that the result has a link basis. Attached Figure Description

[0015] Figure 1 This is a flowchart outlining the method steps of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1The present invention provides a dynamic assessment method for the health status of large-scale drainage pipelines based on multi-algorithm coupling, comprising: S1. Obtain multi-period pipeline records, perform same-segment period aggregation on the pipeline records to form a period sequence, and output a period index table and an action table. This implementation method is used to organize multi-period pipeline records into a periodic data foundation that can be jointly read by change point detection and action challenges. Its mechanism is as follows: first, records from different sources are grouped into the same pipe segment using a pipe segment index; then, a period index is used to form period positions within the same pipe segment. This allows health indicator records to form a periodic string of indicators in cyclical order, while simultaneously enabling maintenance actions to be located at the action position and challenge reading position. Multi-period pipeline records include at least a pipe segment index, a period index, a collection time, an indicator name, an indicator value, an indicator category, a maintenance action name, and an action completion cycle. Health indicator records and maintenance action records share the pipe segment index and period index for inclusion in the same pipe segment periodic aggregation. This implementation process includes the following steps: The purpose of S11 is to form stable pipe segment cycle positions, so that subsequent indicator continuation and action positioning are all executed based on the same cycle sequence. Specifically, it reads the pipe segment index and cycle index from multi-cycle pipe records, and writes records with the same pipe segment index into the same pipe segment group. For records within the same pipe segment group, they are written into cycle positions in ascending order of cycle index. Records with the same cycle index are merged into the same cycle position, and no new cycle positions are added due to different collection times. Within the same cycle position, the record reading order is preserved in ascending order of collection time to form a same cycle record group, and the pipe segment cycle position table is output. If the pipe segment index is empty, the corresponding record is written into the missing pipe segment index record table and does not participate in the generation of pipe segment cycle positions. If the cycle index is empty, the corresponding record is written into the missing cycle index record table and does not participate in the cycle position sorting. If records with the same cycle index but different collection times appear under the same pipe segment index, they are all kept in the same cycle position for S12 to read the last collection time record by indicator name, avoiding the same evaluation cycle from being split into multiple cycle positions. The purpose of S12 is to convert discrete records of the same indicator in the same pipe segment into an indicator periodic string capable of performing differential operations, while simultaneously marking missing values ​​to prevent subsequent change point detection from directly calculating across missing values. Specifically, it reads the health indicator records in each periodic position of the pipe segment periodic position table and groups the same-name indicators by indicator name. When there are multiple records in the same pipe segment, the same periodic position, and the same indicator name, they are read in ascending order of acquisition time, retaining the indicator value recorded at the last acquisition time and entering it into the indicator periodic string. The remaining records with the same name are written into the duplicate indicator record table and the source record index is retained. For the same indicator name, it is read sequentially along the periodic positions. If there is a record of the same indicator in the previous periodic position, the continuation position is written in the current periodic position. If there is no record of the same indicator in the previous periodic position, the missing value is written in the current periodic position, and the periodic indicator table is output. Missing values ​​do not participate in the differential calculation of subsequent adjacent periods. When there are indicator values ​​on both sides of the missing value, a differential relationship is not directly formed across the missing value. Instead, they wait for the subsequent period to be continued before entering the change point detection, thereby ensuring that the direction position only comes from the actual adjacent periodic indicator values. The purpose of S13 is to embed maintenance actions into the same cycle position system, so that action challenges can be read from the cycle with a record basis after the action is completed. Specifically, the maintenance action name and action completion cycle are read from the multi-cycle pipeline records, and the cycle position in which the action completion cycle falls is written as the action position. If there is a cycle record after the action position, the next position after the action position is written as the challenge read position. If there is no position after the action position, the next position with a cycle record is read in ascending order of cycle positions and written as the challenge read position. If there is no cycle record after the action position, the corresponding maintenance action is written into the unchallenged action table and does not enter the subsequent action challenge pairing. When outputting the action table, the action table must retain at least the action name, action position, challenge read position and source record index, so that S5 can read the action pointing status and the change points reproduced after the action. When the action completion cycle falls into a missing position, the action position is still written into the action table according to the action completion cycle, and the challenge read position is determined from the first position with a cycle record after the action position, so as to avoid discarding the implemented maintenance actions due to the lack of cycle indicators. Through the above implementation process, the periodic index table provides S2 with the periodic sequence, continuation position, missing position and index value of the same index in the same pipe segment, and the action table provides S5 with the action position, challenge reading position and action source. Both are formed based on the same pipe segment periodic position, ensuring that subsequent change point detection, Markov transfer construction and action challenge rewriting can be executed along the same periodic coordinate. In practical applications: A main drainage pipeline has water quality and CCTV records in the first, second, third, and fifth assessment cycles. In the fourth assessment cycle, no indicator records were generated due to water outage operations. In this case, S11 still writes the corresponding cycle bit according to the cycle index, and S12 writes the missing bit in the fourth cycle bit and prohibits direct difference between the third cycle bit and the fifth cycle bit. If the dredging action is completed in the third cycle, the detection record is missing in the fourth cycle, and the detection is resumed in the fifth cycle, then S13 writes the third cycle as the action bit and the fifth cycle as the challenge read bit, so that the subsequent action challenge reads the change point from the fifth cycle, instead of directly judging that the dredging action has no challenge result due to the missing fourth cycle.

[0018] S2. Perform change point detection on the periodic indicator table, perform difference operation on adjacent period values ​​of the same indicator in the periodic sequence, write the difference sign into the direction position, and write the period position where the direction position changes into the change point table. This implementation method is used to transform the index period string in the periodic index table into a change point table that can enter the change point state generation. Its working mechanism is as follows: First, the adjacent periodic index values ​​of the same index in the same pipe section are subjected to a difference between the previous and subsequent positions, forming a direction position string composed of an upward position, a downward position, and a zero position. Then, candidate positions of actual direction position switching in the direction position string are extracted. Subsequently, short-term fluctuations are eliminated through subsequent period position continuation verification, so that the change point table only retains change positions that can be continued by the next period. This implementation process includes the following steps: The purpose of S21 is to convert the numerical changes in the index period string into a direction bit string, so that subsequent change point detection does not directly read the original index value, but reads the differential direction. Specifically, it reads the index period string of the same index for the same pipe segment in the period index table, and reads the index values ​​of two adjacent period positions in the order of period positions. When both the previous period position and the next period position have index values, it performs a difference operation by subtracting the index value of the previous period position from the index value of the next period position. When the difference value is positive, it writes an up position in the next period position; when the difference value is negative, it writes a down position in the next period position; when the difference value is zero, it writes a zero position in the next period position and outputs the direction bit string. When there is a missing position in the previous period position or the next period position, it does not perform a difference operation and does not write a direction position in the next period position. When there are index values ​​on both sides of the missing position, it does not perform a direct difference operation across the missing position. The missing position is retained in the period index table for subsequent missing position records to read, avoiding the formation of a direction position by non-adjacent real acquisition results. The purpose of S22 is to locate the position where the direction change occurs from the direction bit string and write the direction change into a change type code that can be looked up. Specifically, it reads the current cycle bit direction bit and the previous cycle bit direction bit from the direction bit string and performs a bit-by-bit comparison. If the current cycle bit direction bit is the same as the previous cycle bit direction bit, it is not written into the candidate change point. If the current cycle bit direction bit is different from the previous cycle bit direction bit, the current cycle bit is written into the candidate change point, and the previous and current direction bits are concatenated into a change type code in the order of reading, and the candidate change point table is output. The direction bit switching includes zero bit to up bit, zero bit to down bit, up bit to zero bit, down bit to zero bit, up bit to down bit, and down bit to up bit. When the direction bit is empty or missing, it does not participate in the comparison, and the current cycle bit is written into the missing direction record table for the cycle index table to look up. The purpose of S23 is to use the next cycle position of the candidate change point to perform a continuity check on the change trend, avoiding single-cycle jumps from being written into the change point table. Specifically, it reads the candidate change point, the current cycle position direction position, and the next cycle position direction position from the candidate change point table. When the next cycle position direction position of the candidate change point has the same value as the current cycle position direction position, the candidate change point is retained and written into the change point table. When the next cycle position direction position of the candidate change point returns to the previous direction position, the candidate change point is written into the swing position and removed from the change point table. When the next cycle position of the candidate change point does not exist, the candidate change point is written into the change point table to be continued, and the continuity check is performed after the record of the next evaluation cycle arrives. The swing position is written into the swing record table. The swing record table retains the pipe segment index, indicator name, previous direction position, current direction position, next direction position, and cycle position. It does not participate in the generation of the S3 change point state, but only serves as supporting evidence of data quality for subsequent review and reading. Through the above implementation process, the direction bit string provides a unified differential direction basis for the extraction of candidate change points, the candidate change point table provides objects to be judged for the continuation verification, and the change point table provides S3 with the change positions that have passed the subsequent cycle bit continuation verification, thereby ensuring that the subsequent result change points, amplified change points and triggered change points all come from the direction changes of real adjacent cycles. In practical applications: When the pipe wall erosion index of the same pipe section rises from the first to the second cycle and continues to rise from the second to the third cycle, the direction of the index changes in the second cycle and is then carried over to the change point table in the third cycle. If the index rises from the first to the second cycle and returns to zero or declines from the second to the third cycle, the candidate change point of the second cycle is written into the swing record table and does not enter the change point generation. If there is no detection record in the third cycle, the candidate change point of the second cycle enters the change point table to be continued, and the continued verification is completed after the fourth cycle has a record of the same index.

[0019] S3. Using the structural result change point in the change point table as the result change point, read the associated change points in the same pipe segment as the result change point in reverse order along the cycle. Stop reading when encountering a reverse change point. Write the associated change point that is connected to the result change point before stopping as the amplified change point. Write the associated change point that is connected to the amplified change point as the trigger change point. Output the change point state table. This implementation method is used to organize the structural result change points in the change point table into a change point state table. Its mechanism is as follows: first, the structural result change points are used as the backtracking endpoints; then, candidate change points capable of accepting structural result changes are read in reverse order along the same pipe segment cycle. Through reverse blocking, path sorting, and forward verification, the candidate change points are stably written into amplified change points and trigger change points. Structural result change points are read from the change points corresponding to the degree of structural defects, concrete strength, and initial pH value of the material. When multiple structural result change points exist in the same pipe segment and the same cycle, result change points are formed according to a fixed reading order of structural defect degree, concrete strength, and initial pH value of the material. Each result change point independently enters the backtracking process. This implementation process includes the following steps: The purpose of S31 is to form a candidate succession chain with a clear source for each result change point, so that subsequent path calculations are only performed within the same pipe segment and the same period sequence system. Specifically, after reading the result change point from the change point table, candidate change points with period positions earlier than the result change point are read in reverse order along the same pipe segment period. When the index category of the candidate change point does not belong to the structural result category, or belongs to the material performance category and the direction position matches the deterioration direction position of the corresponding index name, it is written into the candidate reading sequence. Each time a candidate change point is read, it is connected with the candidate change point read in the next period. The first candidate change point is connected with the result change point to form a candidate succession chain that traces back from the result change point to the historical period. If the same index name appears repeatedly in the candidate succession chain, the candidate change point read first is retained in reverse order according to the period and enters the current candidate succession chain. The remaining candidate change points with the same name are written into the same-name side branch table for re-chaining and reading after the current candidate succession chain fails to be verified. If there are no candidate change points before the result change point, the result change point is written into the no-succession result table and does not enter S32. The purpose of S32 is to eliminate candidate succession chains that are truncated by reverse change points through restricted path calculation, and to form a definite reading order for the remaining candidate paths. Specifically, the candidate succession chains are backtracked item by item from the result change point to the historical cycle. For each item backtracked, the difference between the current candidate change point cycle position and the result change point cycle position is calculated, and the difference is accumulated into the cycle position difference of the current path. The reverse change point is not based solely on the result change point direction position, but on the deterioration direction position corresponding to the candidate change point's own indicator name. When the candidate change point direction position is opposite to its own deterioration direction position, the current path is stopped and a blocking mark is written. Candidate paths without a blocking mark are entered into the succession order table. The change point number is generated in a fixed byte order of pipe segment index, cycle position, indicator name, and direction position. The succession order table first reads the candidate paths without a blocking mark, then reads them in ascending order according to the accumulated cycle position difference. When the accumulated cycle position differences are the same, they are read in ascending order according to the change point number, so that the path reading result does not depend on manual selection. The purpose of S33 is to re-verify the candidate paths obtained by reverse calculation in forward time, ensuring that the triggering change points, amplification change points, and result change points written into the change point state table are consistent in direction. Specifically, according to the succession order table, the candidate paths are read starting from the first in order, and the candidate paths are replayed in ascending order of periodic position to the result change point. The forward replay direction position is compared with the reverse reading direction position in S31 item by item. When each item is consistent, the candidate change point of the adjacent result change point is written as the amplification change point, and the candidate change point of the adjacent amplification change point is written as the trigger change point, and the change point state table is output. When each item is inconsistent, the candidate path is written into the conflict path table, and the next candidate path is read and the verification continues. If the candidate path contains only one candidate change point before the result change point, the candidate change point is written as the amplification change point, and the trigger change point is written into the empty trigger mark. The change point state record with the empty trigger mark does not enter the subsequent front-end Markov transition, but only enters the rear-end Markov transition count from the amplification change point to the result change point. Through the above implementation process, the result change point, amplification change point and trigger change point all come from the change point table of the same pipe segment, and are processed by three layers of reverse backtracking, reverse blocking and forward verification. This can avoid mistakenly writing the accompanying changes after the structural result change as the trigger source, and can also provide a clear change point state record with a clear succession order for the subsequent Markov transfer construction. In practical applications: When a section of the pipe experiences a downward change in concrete strength during the sixth cycle, the system writes this change point as the result change point and then reads the downward change points in material properties during the fifth cycle, the upward change points in the corrosion environment during the fourth cycle, and the upward change points in hydraulic operation during the third cycle in reverse order. If the upward position of the corrosion environment during the fourth cycle matches the deterioration direction of the corrosion environment index, and the downward position of material properties during the fifth cycle matches the deterioration direction of the material performance index, then the candidate path passes the blocking calculation and enters the positive verification. Finally, the downward change point in material properties during the fifth cycle is written as the amplified change point, and the upward change point in the corrosion environment during the fourth cycle is written as the trigger change point. If the direction of the hydraulic operation change point during the third cycle is opposite to its own deterioration direction, then the hydraulic operation change point during the third cycle is written as the blocking marker and does not enter the change point state table.

[0020] S4. Based on the variable state table, perform Markov transition construction, write the trigger variable, amplification variable and result variable into the state column, write the transition edge from the trigger variable to the amplification variable and from the amplification variable to the result variable, count the number of transitions according to the same starting state and ending state, and generate the transition matrix by dividing by the total number of transitions for the same starting state. This implementation transforms triggering variables, amplifying variables, and result variables in the variable state table into state columns and transition matrices that can participate in the Markov model recursion. Its mechanism involves first converting the succession relationships within the variable state records into aggregable state identifiers, then writing adjacent state columns into Markov transition counts, and finally calculating matrix items based on the transition distribution of states starting from the same point. To avoid splitting similar states by event serial numbers, the variable point number in the state identifier uses a state class number, and the event serial number of the source change point is written separately as the source record number, used only for retrieving the original change point and not participating in transition matrix aggregation. This implementation process includes the following steps: The purpose of S41 is to form a state list that can be aggregated and read by the transition matrix, so that similar change point states enter the same state position in different pipe segment cycles. Specifically, the receiving records in the change point state table are read one by one, and the reading order of trigger change points, amplification change points and result change points in the receiving records is written as the receiving sequence. The state class number of each change point is formed by concatenating the receiving role, indicator category and direction bit in a fixed byte order. Among them, the receiving role is written into the role byte according to trigger, amplification and result, the indicator category is written into the category byte, and the direction bit is written into the direction. Byte; The status class number and direction bit are concatenated together to form the status identifier. The source change point number and source record index are written to the source field and do not participate in the status identifier concatenation. When the status identifier is not written to the status column, the column bits are appended in the order of first writing. When the status identifier has been written to the status column, the original column bits are reused and the corresponding source field is written to the record with column bits. If the change point state record contains a null trigger flag, the null trigger flag is not written to the status column. The record with column bits only retains the amplified change point column bits and the result change point column bits for subsequent Markov transfer reading. The purpose of S42 is to convert the succession order in the column-based succession record into a Markov transition count, so that the state sequence from triggering to amplification and from amplification to result can enter the same matrix statistical system. Specifically, it reads the first succession column, the second succession column, and the last succession column in the column-based succession record; when the first succession column exists, it takes the first succession column as the row position and the second succession column as the column position, and performs an increment operation on the corresponding row and column in the Markov transition count table; at the same time, it takes the second succession column as the row position and the last succession column as the column position. As a column position, the corresponding row and column cells are incremented by one; the two counts are written to the same transfer group; the transfer group number is generated by concatenating the result change point state class number, the result change point source record index, and the change point state record writing order, and is used for subsequent actions to challenge the same deteriorated inheritance relationship; if the column position inheritance record contains a null trigger flag, only the increment operation of the second inheritance column position pointing to the last inheritance column position is performed, and the first inheritance column position pointing to the second inheritance column position is not generated, thereby avoiding the null trigger flag from forming a false Markov transfer; The purpose of S43 is to normalize the Markov transition count table into a transition matrix, so that the subsequent state direction corresponding to each starting state can be read by both the S5 action challenge and the S6 Markov model recursion. Specifically, the Markov transition count table is read row by row, and the counts of all rows and columns in the same row are accumulated to obtain the row count sum. For rows where the row count sum is not zero, the count of each row and column is divided by the row count sum of the row in which it is located to obtain the matrix item, and it is written into the transition matrix according to the row and column positions. For rows where the row count sum is zero, the matrix item in the corresponding row is written with a zero value, and the row is written as the termination state column. The termination state column does not contribute to the next cycle in the S6 matrix recursion, but its source state column, source transition group, and source change state record continue to be written into the termination contribution record for the last cycle result to read, so as to avoid the link source being lost due to the accidental deletion of the termination state. Through the above implementation process, the state column is formed by aggregateable state identifiers, the Markov transition count table is formed by adjacent state column positions in the succession order, and the transition matrix is ​​formed by the normalization of transition counts of the same starting state, thereby ensuring that subsequent action challenges can accurately locate the challenged matrix items, and the Markov model can perform recursion according to a unified state position order. In practical applications: the connection between the upward triggering change point of the same type of corrosive environment and the downward amplification change point of material properties may appear in multiple pipe sections and multiple evaluation cycles. S41 will not directly use the change point numbers from different sources as status identifiers, but will form the same status identifier by the connection role, index category and direction position, so that these records enter the same status column; S42 will increment the connection between the positions of the same status column one by one; S43 will count and calculate the matrix items by row, thereby obtaining a transition matrix that can reflect the distribution of this type of deterioration connection relationship.

[0021] S5. Perform the action challenge according to the action table, pair the action-pointing state with the post-action reproduction change point, write the challenge mark into the matrix item formed by the action-pointing state in the transition matrix, and output the transition matrix after the challenge. This implementation method transforms maintenance actions into action challenges that test whether matrix items in the transition matrix still hold true. Its mechanism involves first organizing the recurring change points after the action completion cycle into a reproduction sequence, then calculating the pairing cost using the reproduction sequence and the matrix items containing the challenge source position, and finally writing the pairing results determined by the cost order back into the transition matrix. The action pointing state is mapped from the action name in the action table. The action name first corresponds to the action pointing category, and then the state column reads the state where the index category matches the action pointing category and the direction position matches the deterioration direction position as the action pointing state. When the same action hits multiple action pointing states, they are written in order according to the state column to form action challenge records. This implementation process includes the following steps: The purpose of S51 is to determine whether a change point originating from the same source as the action's target state appears after the maintenance action is completed, and to convert this originating change point into a computable reproducible sequence. Specifically, the action completion cycle and the action target state in the action table are read, and the column position of the action target state in the state column is written as the challenge source position. Starting from the cycle after the action completion cycle, the originating change points in the change point table that are in the same pipe segment, the same index category, and the same direction as the challenge source position are read back in ascending order of cycle position. For each originating change point, the difference between the cycle position of the originating change point and the corresponding cycle position of the action completion cycle is calculated, and this difference is written as the reproducible distance. At the same time, the difference between the column position of the originating change point and the challenge source position is calculated, and this difference is written as the state position difference, and the reproducible sequence is output. If there is no originating change point after the action completion cycle, the action challenge record is marked as no reproducible and does not participate in the pairing in S52. If there is a missing period after the action completion cycle, the originating change points are read back from the challenge reading position to avoid the action challenge being interrupted prematurely due to the missing period. The purpose of S52 is to perform cost pairing between the reproduction sequence and the matrix items in the transition matrix affected by the maintenance action, so that the action challenge can point to a specific matrix item. Specifically, the reproduction sequence and the matrix items containing the challenge source position in the transition matrix are read. For each reproduction point, the column position difference from the challenge source position to the first column position of the matrix item and the column position difference from the challenge source position to the last column position of the matrix item are calculated respectively. Then, the absolute difference between the state position difference of the reproduction point and the first column position difference of the matrix item and the absolute difference between the state position difference of the reproduction point and the last column position difference of the matrix item are calculated. The two absolute differences, the reproduction distance and the number of occupied items in the same transition group are added to obtain the pairing cost value, and written into the challenge edge sequence in ascending order of pairing cost value and in ascending order of reproduction point writing order. The number of occupied items in the same transition group is written to zero before the challenge edge sequence is calculated. After each matrix item is locked in S53, the number of occupied items in the corresponding transition group is incremented by one. When there is a subsequent action challenge record, the challenge edge sequence of the subsequent action challenge record is recalculated according to the updated number of occupied items in the same transition group. The purpose of S53 is to complete one-to-one pairing according to the challenge edge sequence and write the pairing results back to the transition matrix to form the post-challenge transition matrix. Specifically, challenge edges are read line by line according to the challenge edge sequence. When neither the matrix item nor the reappearance point is occupied, the challenge edge is written into the action challenge pairing table, and the matrix item and reappearance point are occupied simultaneously. Two adjacent reappearance points in the reappearance sequence form a reappearance pair. The first column of the reappearance pair is used to hit the beginning of the matrix item, and the last column of the reappearance pair is used to hit the end of the matrix item. Only when the first column and the last column of the reappearance pair hit the matrix item respectively... When both ends are hit, write an uncut challenge flag to the matrix item; when only the last column hits the end of the matrix item, write an existing damage challenge flag to the matrix item; when only the first column hits the beginning of the matrix item, write a temporary inactivation challenge flag to the matrix item and output the challenge-transfer matrix; if the same matrix item has already been written with a challenge flag, retain the challenge flag written earlier in ascending order of action completion cycle; if the action completion cycles are the same, retain the challenge flag read earlier in ascending order of paired value, and the challenge flag written later enters the challenge conflict table and retains the source action record; Through the above implementation process, the maintenance actions in the action table are no longer used as ordinary input fields in the health level calculation. Instead, they are directly rewritten by the matrix items pointed to by the actions in the transition matrix through challenge source position, reproduction sequence, pairing cost value and challenge mark, so that the subsequent Markov model recursion can read the state transition results after the action challenge. In practical applications: When a pipe section implements anti-corrosion action in the third cycle, the anti-corrosion action is mapped to the state columns of corrosion environment and material performance. In the fifth cycle, an upward change point in the corrosion environment occurs again, and in the sixth cycle, a downward change point in the material performance occurs. The system writes the common change points of the fifth and sixth cycles into the reproduction sequence, calculates the reproduction distance and state difference respectively, and then calculates the pairing cost with the matrix item from the corrosion environment trigger state to the material performance amplification state. If the reproduction point of the fifth cycle hits the beginning of the matrix item and the reproduction point of the sixth cycle hits the end of the matrix item, the matrix item is marked with an unbroken challenge, indicating that the anti-corrosion action has not broken the corresponding degradation transmission relationship.

[0022] S6. Using the state column as the state space of the Markov model, the challenge transition matrix as the state transition matrix of the Markov model, the initial state vector is generated from the initial periodic change point state. The product of the state vector and the state transition matrix is ​​recursively executed in the Markov model in a periodic order. Based on the challenge mark and the result change point direction in the recursive vector, the health level, instability direction and chain-level maintenance direction are output. This implementation method incorporates the post-challenge transition matrix into the Markov model recursion and transforms the action challenge marker into health level, instability direction, and chain-level maintenance direction during the recursion process. Its mechanism involves first aligning the state column order with the row and column order of the post-challenge transition matrix; then calculating candidate recursion contributions using the current state vector and matrix items; subsequently rewriting the contribution direction according to the challenge marker; and finally verifying the recursion direction using the change point direction of the same period's results and feeding back the state vector for the next period. This implementation process includes the following steps: The purpose of S61 is to establish a unified state order for the Markov model, so that the state vector and the post-challenge transition matrix can participate in the product recursion according to the same column and row positions. Specifically, the state column order, the post-challenge transition matrix, and the initial period change state are read, and the state column order is used as the state order of the Markov model. When there is a change state in the initial period, the column position that is hit by the change state in the initial period is written with a value, and the column position that is not hit is written with a value of zero, forming the current state vector. When there is no change state in the initial period, the first period with a change state is read along the period sequence as the model starting period, and the periods before the model starting period are written into the missing change state record table. Before reading the post-challenge transition matrix, the row and column positions are rearranged according to the state column order, and the missing row and column cells are written with a value of zero. The current state vector and the corresponding matrix table are output for S62 to perform matrix recursion. The purpose of S62 is to calculate the recursive contribution of each state's destination within the Markov model and retain the source of the contribution for action challenge routing. Specifically, it reads the non-zero vector bits in the current state vector and multiplies the values ​​of the non-zero vector bits with the matrix entries in the same row of the same position in the matrix table. Each product is accumulated in the corresponding column of the candidate recursive vector according to the end column of the matrix entry. At the same time, it reads back the source matrix entries of the product. If the source matrix entry has a challenge flag, the challenge flag is written to the contribution list. If the source matrix entry has not a challenge flag, the ordinary transition flag is written to the contribution list. The contribution list is synchronously written with the source row of the product, the end column, the product value, and the transition group number. The candidate recursive vector and the contribution list are output. If all the matrix rows corresponding to a certain non-zero vector bit are zero, the non-zero vector bit is written to the termination contribution record. No candidate recursive contribution is generated for the next cycle, but the source state column and the transition group number are retained for the last cycle result to be read. The purpose of S63 is to rewrite the candidate recursive contributions according to the action challenge results, so that the influence of maintenance actions on Markov transfers is incorporated into the recursive process itself. Specifically, the contribution list and candidate recursive vectors are used as the input for action challenge routing. The challenge markers and product values ​​in the contribution list are read one by one. The products corresponding to ordinary transfer markers and uncut markers are retained in the original endpoint column. The product corresponding to the existing damage marker is written to the result change point column of the corresponding transfer group. If the transfer group lacks a result change point column, the result status column with the same direction as the original endpoint column is written. If there is still no corresponding result status column, the result position to be located is written. The product corresponding to the temporary inactivation marker is written to the column to be reviewed. If there is no column to be reviewed in the status column, the column to be reviewed is appended to the end of the status column, and the rows and columns of the same position matrix table are expanded simultaneously. For multiple products of the same destination column, the column is accumulated and the challenge routing vector is output for S64 to perform directional projection. The purpose of S64 is to verify the challenge routing vector using the result change point direction bits of the same period, so that the recursive state only enters the next period when the direction is consistent, and retains the conflict direction as the basis for subsequent chain-level maintenance. Specifically, the challenge routing vector and the result change point direction bits of the same period are used as direction projection inputs. The non-zero vector bits in the challenge routing vector are read one by one, and the corresponding state direction bits are read. When the result change point direction bits of the same period are empty, the challenge routing vector does not perform direction projection and is directly written into the state vector of the next period. When there are multiple result change point direction bits in the same period, they are compared item by item according to the order of writing the result change points. When the state direction bit has the same value as any result change point direction bit, ... Write the non-zero vector bit into the state vector of the next cycle. When the state direction bit is opposite to the result change point direction bit, write the non-zero vector bit into the conflict chain bit. When the state direction bit is zero, write the non-zero vector bit into the vector bit to be verified. The state vector of the next cycle is fed back into the Markov model to continue participating in the recursion from S62 to S64 until the last cycle is read. The number of non-zero result state columns in the last cycle is written into the health level. The structural result class state direction bit hit in the last cycle is written into the instability direction. The uncut mark, the existing damage mark, and the temporary inactivation mark are written into the original action uncut chain, the structural result processing chain, and the verification chain, respectively. Output the health level, the instability direction, and the chain-level maintenance direction. Through the above implementation process, the Markov model not only reads the transition matrix to perform state recursion, but also reads the challenge marker, the result change point direction position and the conflict chain position in each cycle of recursion. This allows the rewriting of the direction of the action challenge's contribution to the transition to directly participate in the state vector evolution, thereby transforming whether the reproduced change point after the action proves that the original transition still holds into a recursive state result. In practical applications: After the anti-corrosion action is implemented in the third cycle of a certain pipe section, the upward state of the corrosion environment in the fifth cycle and the downward state of material properties in the sixth cycle both re-enter the state column. S62 first calculates the recursive contribution from the upward state of the corrosion environment to the downward state of material properties based on the challenge-transition matrix. S63 reads the uncut mark of the matrix item and retains the product in the column position of the downward state of material properties. S64 then uses the direction position of the change point of the downward result of concrete strength in the sixth cycle to perform directional projection. If the direction position of the downward state of material properties is the same as the direction position of the change point of the downward result of concrete strength, then the recursive contribution enters the state vector of the next cycle, and the chain-level maintenance direction of the corresponding deterioration chain that was not cut off by the anti-corrosion action is output in the last cycle.

[0023] Working principle: This scheme first organizes the inspection records and maintenance action records of large drainage pipelines in multiple evaluation cycles into a periodic sequence according to the same pipe section. Then, it performs difference analysis on the adjacent period values ​​of the same index to identify the locations where the direction changes, forming a change point table. Subsequently, starting from the change points related to structural defects, concrete strength, and initial pH value of materials, it reads back the change points that can bear the result in reverse order along the cycle, forming trigger change points, amplification change points, and result change points. On this basis, the three types of change points are written into the status column, and the Markov transitions from trigger to amplification and from amplification to result are statistically analyzed to form a transition matrix. After the dredging, flushing, anti-corrosion, local repair, or sealing actions are completed, it reads whether the same source change points appear after the action, and rewrites the corresponding matrix items with the reproduction situation. Finally, the health level, instability direction, and chain-level maintenance direction are recursively output by the Markov model. In practical applications, for example, after the third cycle of corrosion protection for a main drainage pipeline, the corrosive environment deteriorates again in the fifth cycle, and the material performance and concrete strength decrease in the sixth cycle. This solution will not directly treat the changes in indicators after corrosion protection as ordinary health level input. Instead, it will first confirm the state indicated by the corrosion protection action, and then check whether the same source change point reappears after the action. If the change point of the corrosive environment and the change point of the material performance still continue along the original path to the change point of the concrete strength result, the system will mark the corresponding Markov transition as an uncut transition, and output the instability direction and maintenance direction of the corresponding deterioration chain still exist in the pipeline segment in the subsequent recursion, thereby avoiding misjudging the short-term characterization improvement as the termination of the deterioration chain.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic assessment method for the health status of large-scale drainage pipelines based on multi-algorithm coupling, characterized in that, include: S1. Obtain multi-period pipeline records, perform same-segment period aggregation on the pipeline records to form a period sequence, and output a period index table and an action table. S2. Perform change point detection on the periodic indicator table, perform difference operation on adjacent period values ​​of the same indicator in the periodic sequence, write the difference sign into the direction position, and write the period position where the direction position changes into the change point table. S3. Using the structural result change point in the change point table as the result change point, read the associated change points in the same pipe segment as the result change point in reverse order along the cycle. Stop reading when encountering a reverse change point. Write the associated change point that is connected to the result change point before stopping as the amplified change point. Write the associated change point that is connected to the amplified change point as the trigger change point. Output the change point state table. S4. Based on the variable state table, perform Markov transition construction, write the trigger variable, amplification variable and result variable into the state column, write the transition edge from the trigger variable to the amplification variable and from the amplification variable to the result variable, count the number of transitions according to the same starting state and ending state, and generate the transition matrix by dividing by the total number of transitions for the same starting state. S5. Perform the action challenge according to the action table, pair the action-pointing state with the post-action reproduction change point, write the challenge mark into the matrix item formed by the action-pointing state in the transition matrix, and output the transition matrix after the challenge. S6. Using the state column as the state space of the Markov model, the challenge transition matrix as the state transition matrix of the Markov model, the initial state vector is generated from the initial periodic change point state. The product of the state vector and the state transition matrix is ​​recursively executed in the Markov model in a periodic order. Based on the challenge mark and the result change point direction in the recursive vector, the health level, instability direction and chain-level maintenance direction are output. S4 includes: S41. Read the succession records in the variable point state table one by one, and use the reading order of the variable points in the succession record as the succession order. Concatenate the variable point number and direction bit of each variable point into a status identifier in a fixed byte order. If the status identifier is not written to the status column, add column bits in the writing order. If the status identifier has been written to the status column, reuse the original column bits and output the succession record with column bits. S42. Perform Markov transition counting on the record with column position succession. Read the first succession column, the second succession column, and the last succession column. Increment the row and column count of the first succession column pointing to the second succession column by one. Increment the row and column count of the second succession column pointing to the last succession column by one. Write the two counts into the same transition group and output the Markov transition count table. S43. Perform transition matrix calculation on the Markov transition count table. Accumulate the row and column counts row by row to obtain the row count sum. Divide the count of each row and column by the count sum of its row to obtain the matrix item. When the row count sum is zero, write the zero value to the corresponding matrix item in the row and output the transition matrix. S5 includes: S51. Using the action completion cycle as the starting point, write the column position of the action pointing to the state in the state column as the challenge source position. Starting from the next cycle, read back the same source change point cycle by cycle, calculate the difference between the read-back cycle position and the action completion cycle and write it as the reproduction distance, calculate the difference between the column position of the same source change point and the challenge source position and write it as the state position difference, and output the reproduction sequence. S52. Using the matrix items containing challenge source positions in the reproduction sequence and transition matrix as the input, calculate the sum of the absolute differences between the state position difference and the column position differences at both ends of the matrix item for each reproduction point, and then accumulate the reproduction distance and the number of occupied items in the same transition group to obtain the pairing value. Write the pairing value in ascending order and the reproduction point writing order into the challenge edge sequence. S53. Read the challenge edges line by line according to the challenge edge sequence. When neither the matrix item nor the reappearance point is occupied, write it into the action challenge pairing table and occupy the matrix item and the reappearance point simultaneously. Then, based on the result that the reappearance point column hits both ends of the matrix item, only hits the end of the matrix item, or only hits the beginning of the matrix item, write the uncut challenge flag, the existing damage challenge flag, or the temporarily inactive challenge flag into the transition matrix, and output the transition matrix after the challenge. S6 includes: S61. Read the state column order, the transition matrix after the challenge and the initial periodic change point state. Use the state column order as the state order of the Markov model. Write a value to the hit column of the initial periodic change point state and write a zero value to the missing column. Output the current state vector and the corresponding matrix table. S62. In the Markov model, perform matrix recursion on the current state vector and the same-position matrix table. Read the non-zero vector position one by one and multiply it with the matrix item in the same row. Accumulate the product according to the end column of the matrix item. At the same time, write the source row position, end column position and challenge flag of the product into the contribution list. Output the candidate recursive vector and the contribution list. S63. Using the contribution list and candidate recursive vector as input for the action challenge route, read back the challenge markers in the contribution list one by one, keep the product of the uncut markers in the original endpoint column, rewrite the product of the existing damage markers in the result variable column, rewrite the product of the temporarily inactive markers in the column to be reviewed, and output the challenge route vector. S64. Using the challenge routing vector and the direction position of the result change point in the same period as the direction projection input, read the state direction position corresponding to each non-zero vector position and compare it with the direction position of the result change point. When the direction positions are the same, write the non-zero vector position into the state vector of the next period. When the direction positions are reversed, write the non-zero vector position into the conflict chain position. Feed the state vector of the next period back into the Markov model and continue to recursively push to the last period. Output the health level, instability direction and chain-level maintenance direction.

2. The method for dynamic assessment of the health status of large-scale drainage pipelines based on multi-algorithm coupling according to claim 1, characterized in that: S1 includes: S11. Read the pipe segment index and cycle index for multi-cycle pipeline records, group them into pipe segment groups according to the same pipe segment index, and write the cycle bit into the pipe segment group in ascending order of cycle index, and output the pipe segment cycle bit table. S12. Read the record with the same index name for each cycle position in the pipe segment cycle position table. If the same index record exists in the previous cycle position, write it to the successor position. If the same index record does not exist in the previous cycle position, write it to the missing position. Generate the index cycle string according to the cycle position order and output the cycle index table. S13. Read the completion cycle of maintenance actions in the multi-cycle pipeline record, write the cycle bit into which the completion cycle falls as the action bit, write the cycle bit after the action bit as the challenge read bit, and output the action table.

3. The method for dynamic assessment of the health status of large-scale drainage pipelines based on multi-algorithm coupling according to claim 2, characterized in that: S2 includes: S21. Read the index cycle string of the same pipe segment and the same index in the cycle index table, perform the difference between the next bit and the previous bit for the index value of adjacent cycle bits, write the positive, negative and zero results of the difference value into the up bit, down bit and zero bit respectively, and output the direction bit string. S22. Perform change point detection on the direction bit string, compare the current cycle bit direction bit with the previous cycle bit direction bit, if they are different, write the current cycle bit into the candidate change point, and concatenate the previous direction bit and the current direction bit into the change type code, and output the candidate change point table. S23. Perform a continuation check on the candidate change point table. When the direction position of the next cycle position of the candidate change point has the same value as the current direction position, retain the candidate change point and write it into the change point table. When the direction position of the next cycle position of the candidate change point returns to the previous direction position, write the swing position and remove it from the change point table.

4. The method for dynamic assessment of the health status of large-scale drainage pipelines based on multi-algorithm coupling according to claim 3, characterized in that: S3 includes: S31. Read the result change points from the change point table, read the candidate change points in reverse order along the same pipe section cycle, and connect each candidate change point with the candidate change points read in the next cycle to form a candidate succession chain that traces back from the result change points to the historical cycle. S32. Perform restricted path calculation on the candidate receiving chain. Accumulate the periodic position difference for each backtracking item of the path. When a candidate change point with the opposite direction of the deterioration of the result change point is encountered, stop the path and write the blocking mark. Paths without blocking marks are formed into a receiving order table according to the fixed order of the accumulated periodic position difference and the change point number. S33. Perform forward verification based on the succession order table. Read from the candidate path with the highest order to the result change point. When the direction position of the change point in the path is consistent with the reverse reading direction position, write the change point of the adjacent result change point as the amplified change point and write the change point of the adjacent amplified change point as the trigger change point, and output the change point state table. If the consistency is not met, read the next candidate path and continue the verification.

Images