A loess tunnel eccentric pressure support intelligent decision analysis method and system

Abstract

The application discloses a loess tunnel bias support intelligent decision analysis method and system, and particularly relates to the technical field of intelligent management of building construction, and comprises the following steps: acquiring a construction execution flow and a monitoring sampling flow formed from an excavation starting signal to a support ring completion confirmation signal in a current excavation cycle; writing an event bit according to a construction action completion time; writing a response bit according to the symbol of the difference between the adjacent sampling values of the same monitoring point; and outputting an excavation cycle dynamic modeling graph; in the excavation cycle dynamic modeling graph, reading the response bit with the first symbol change backward from the event bit completion time, and writing the response bit number back to the event bit; when the response bit number corresponds to an event bit, generating an interpretable bit; and identifying the interpretable relationship and the aliasing relationship between the event bit and the response bit by uniformly writing the construction action completion time and the monitoring sampling change into the dynamic modeling graph, and generating a target scheduling column under the ring completion time limit and the team scheduling constraint.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for building construction, and more specifically, to an intelligent decision analysis method and system for eccentric support in loess tunnels. Background Technology

[0002] In the current construction of eccentric support for loess tunnels, the construction unit generally judges the stress state of the surrounding rock based on the settlement of the arch crown, the convergence of the perimeter, the strain of the steel arch frame, the axial force of the anchor bolts, and the observation results of the tunnel face. Then, in combination with the design support parameters and the results of on-site monitoring and measurement, adjustments are made to the spacing of the steel arch frame, the arrangement of the locking anchor bolts, the range of advance support, the thickness of the shotcrete, and the follow-up distance of the invert arch. In shallow gully sections, tunnel entrance biased sections, or loess tunnels with significant changes in burial depth on the hillside, a single excavation cycle often requires the completion of steps such as upper bench excavation, initial support spraying, hillside anchor bolts, gully side anchor bolts, drainage treatment, and invert arch follow-up. On-site operations are also limited by the safety distance of the tunnel face, the overlapping of work teams, the setting time of sprayed concrete, and the time limit for ring support formation, making it difficult to suspend construction or arrange separate test procedures in order to determine the source of bias. In this situation, multiple construction actions within the same monitoring period will simultaneously change the stress path of the surrounding rock. On-site observations show phenomena such as short-term settlement of the arch crown, continued increase in strain of the arch shoulder on the mountain side, delayed change in displacement of the arch foot on the ditch side, and short-term reverse of local deformation after drainage. This makes it difficult for the monitoring curve to correspond to a single construction action or a single cause of bias pressure. Subsequent support adjustments may easily confuse support effectiveness, continued bias pressure of the surrounding rock, release of wet collapse at the arch foot, and drainage disturbance as the same risk change. The technical problem this application aims to solve is: how to identify monitoring segments that can be used for decision-making during the construction of eccentric support without suspending normal construction of loess tunnels or adding separate testing procedures, and to determine the basis for adjusting support procedures accordingly. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an intelligent decision analysis method and system for eccentric support in loess tunnels. By uniformly writing the completion time of construction actions and monitoring sampling changes into a dynamic modeling diagram, the system identifies the readable and aliasing relationships between event bits and response bits, and generates a target scheduling column under the constraints of loop time limits and shift scheduling, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent decision analysis method for eccentric support in loess tunnels, comprising: S1. Obtain the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. Write the event bit according to the completion time of the construction action, and write the response bit according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point. Output the excavation cycle dynamic modeling diagram. S2. In the dynamic modeling diagram of the excavation cycle, read the first response bit whose sign changes from the time the event bit is completed, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, generate a readable bit; when it corresponds to two or more event bits, generate an aliased bit, and output the event interpretation diagram. S3. Based on the event interpretation graph, execute the Lagrange relaxation scheduling algorithm, move the event bit by one construction order to generate candidate schedules, calculate the number of aliasing bits to split, subtract the loop completion time from the loop completion time to obtain the loop timeout, update the loop multiplier with the loop timeout, and output the candidate schedule graph with multiplier cost. S4. For the candidate scheduling graph, execute the branch pricing cut algorithm to generate a scheduling column for each candidate scheduling. Subtract the multiplier cost from the number of aliased bits to obtain the column test value. Write the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, write the same order shift relationship of the two event bits into the cut constraint and output a readable scheduling column graph. S5. Read the construction shift schedule based on the interpretable schedule chart, generate the organizational activity risk value by the number of times the event completion time in the schedule column crosses the shift handover time, and determine the target schedule column by ascending organizational activity risk value, descending number of interpretable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support.

[0005] In a preferred embodiment, S1 includes: S11. Obtain the construction execution flow. Take the acquisition time of the excavation start signal as the zero time of the cycle. Read the completion time of each construction action. Subtract the zero time of the cycle from the completion time of the construction action to obtain the event time difference. Write the event bit in ascending order according to the event time difference and output the event bit sequence. The construction execution flow refers to the sequence of construction action completion times written by construction equipment or construction record terminals in order of completion within a single excavation cycle. S12. Obtain the monitoring sampling stream, take adjacent samples of the same monitoring point as a sampling pair, subtract the previous sampling value from the next sampling value to generate a sampling difference, write a positive sign when the sampling difference is positive, write a zero sign when the sampling difference is zero, write a negative sign when the sampling difference is negative, and obtain the response time difference by subtracting the cyclic zero time from the next sampling time, and output the response bit sequence. The monitoring sampling stream refers to the sequence of sampling times and sampling values ​​written by the tunnel monitoring points in the order of sampling within the same excavation cycle. S13. Based on the event bit sequence and response bit sequence, write the same excavation cycle index in ascending order of event time difference and response time difference. Write bit edges between adjacent event bits or response bits in ascending order. The direction of the bit edges is from the event bit or response bit with smaller event time difference or response time difference value to the event bit or response bit with larger value. Output the excavation cycle dynamic modeling diagram. The excavation cycle index refers to the data assignment number determined by the excavation start signal and the support ring confirmation signal of one excavation cycle.

[0006] In a preferred embodiment, S2 includes: S21. Read the event bits in the excavation cycle dynamic modeling diagram, take each event bit as the current event bit in sequence, read the response bits along the bit edge direction whose response time difference is greater than the current event bit, and generate the event response chain after the current event bit in ascending order of response time difference. S22. For the post-event response chain, read the sign of the adjacent response bits in sequence. When the sign changes from zero to positive or negative, or when positive and negative signs are interchanged, write the number of the next response bit into the write-back field of the current event bit and output the event bit write-back table.

[0007] In a preferred embodiment, S2 further includes: S23. Based on the event bit write-back table, count the number of event bits written to the same response bit number using the response bit number as the index, and write the number of event bits to the write-back count value of the response bit number, and output the response bit write-back count table. S24. Read the response bit write-back count table. When the write-back count value is equal to one, write the event bit written to the same response bit number as a readable bit. When the write-back count value is greater than one, write the event bit written to the same response bit number as an aliased bit. Then write the event bits in ascending order of event time difference into the construction sequence and write them into the event interpretation diagram.

[0008] In a preferred embodiment, S3 includes: S31. Read the event interpretation diagram, take the event bit written as the aliased bit as the event bit to be moved, copy the construction order in the event interpretation diagram to generate the baseline schedule, and under the construction constraints that the excavation start event bit maintains the first construction order and the support ring confirmation event bit maintains the tail construction order, swap the construction order of the event bit to be moved one position forward or one position backward, and output the candidate schedule. S32. Re-execute the response bit number write-back based on the candidate schedule, record the number of event bits with the same response bit number written in the baseline schedule and different response bit numbers written in the candidate schedule as the number of aliasing bit splits, and subtract the loop completion time from the loop completion time of the candidate schedule to obtain the loop timeout, and output the candidate schedule evaluation table.

[0009] In a preferred embodiment, S3 further includes: S33. Perform the Lagrange relaxation scheduling algorithm on the candidate scheduling evaluation table. Multiply the ring multiplier by the ring timeout to obtain the ring penalty. Subtract the number of aliasing bit splits from the ring penalty to obtain the multiplier cost. Add the ring timeout to the ring multiplier to generate the updated ring multiplier and output the multiplier cost table. The ring multiplier is a numerical term used to calculate the ring penalty of the candidate schedule. The initial value of the ring multiplier is one. In each round, the ring multiplier of the previous round is added to the ring timeout of the candidate schedule in the current round to obtain the ring multiplier of the next round. The ring multiplier of the next round is multiplied by the ring timeout of the next round to obtain the ring penalty of the next round. S34. Based on the multiplier cost table, select candidate schedules in ascending order of multiplier cost. The calculation ends when the candidate schedules selected in two adjacent rounds have the same construction order and the loop timeout is equal to zero. Write the candidate schedule, the number of overlapping positions, the loop timeout, and the multiplier cost into the same candidate schedule graph node and output the candidate schedule graph with multiplier cost.

[0010] In a preferred embodiment, S4 includes: S41. Read the candidate scheduling diagram with multiplier cost, generate a scheduling column for each candidate scheduling, concatenate the event bit numbers according to the construction priority of the event bits in the candidate scheduling to generate a construction priority string, and concatenate the response bit numbers according to the event bit write-back field to generate a response bit number string, and output the scheduling column encoding table. S42. Calculate the pricing using the branch pricing algorithm based on the scheduling column encoding table. Subtract the multiplier cost from the number of aliased bit splits to obtain the column check value. Write the scheduling columns with positive column check values ​​into the main problem scheduling column set and output them.

[0011] In a preferred embodiment, S4 further includes: S43. Read the response bit number string from the main problem scheduling column set. When two event bit numbers in the response bit number string correspond to the same response bit number, generate a prohibited combination according to the event bit numbers of the two event bits and the direction of movement of the construction sequence by exchanging the two event bits one position forward or one position backward. Write the prohibited combination into the cut constraint table and output it. S44. Read the candidate scheduling graph with multiplier cost according to the cut constraint table, delete the candidate schedule that simultaneously contains two event bit numbers, movement directions and the same response bit number in the prohibited combination, recalculate the column test value for the remaining candidate schedules, and end the branch pricing cut algorithm when all the column test values ​​of the remaining candidate schedules are not positive, and write the main problem scheduling column set and the cut constraint table into the readable scheduling column graph.

[0012] In a preferred embodiment, S5 includes: S51. Read the interpretable schedule chart and construction shift schedule, sort the shift handover times in ascending order of time, take the handover times of two adjacent shifts as the start and end times of the shift time respectively, and assign the event completion time to the shift time where the start time of the shift time is not later than the event completion time and the end time of the shift time is later than the event completion time. Output the event position shift affiliation table. S52. Based on the event position group affiliation table, set the initial value of the cross-group count to zero. Starting from the second event position in the scheduling column, read the current group time period to which the current event position is assigned and the previous group time period to which the previous event position is assigned. If the two are inconsistent, increment the cross-group count by one. After reading each position, write the cross-group count as the risk value of the organization activity and output the risk scheduling column chart. Cross-shift counting refers to the cumulative number of shift time changes obtained by adding one when adjacent event bits within the same schedule column are assigned to different shift time periods. S53. Read the risk scheduling column chart, subtract the number of readable positions from zero to obtain the inverse number of readable positions, and concatenate the organizational activity risk value, the inverse number of readable positions, the loop timeout amount and the scheduling column number in sequence to generate the scheduling column sorting key. Select the first-order scheduling column as the target scheduling column according to the ascending order of the scheduling column sorting key, and output the intelligent decision analysis results of the bias support of loess tunnel. The number of readable bits refers to the total number of event bits in the same schedule column whose write-back count is equal to the number of event bits that have been written as readable bits.

[0013] A smart decision-making and analysis system for eccentric support in loess tunnels, the system comprising a modeling module, an interpretation module, a scheduling module, a pricing module, and a decision-making module: The modeling module is used to acquire the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. It writes the event bits according to the completion time of the construction action and writes the response bits according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point, and outputs the dynamic modeling diagram of the excavation cycle. The interpretation module is used to read the first sign-changed response bit from the event bit completion time in the dynamic modeling diagram of the excavation cycle, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, a readable bit is generated; when it corresponds to two or more event bits, an aliased bit is generated, and the event interpretation diagram is output. The scheduling module executes the Lagrange relaxation scheduling algorithm based on the event interpretation graph, generates candidate schedules by moving one construction order according to the event bit, calculates the number of aliasing bits to be split, obtains the ring formation timeout by subtracting the ring formation deadline from the ring formation completion time, updates the ring formation multiplier with the ring formation timeout, and outputs a candidate schedule graph with multiplier cost. The pricing module is used to perform a branch-pricing cut algorithm on the candidate scheduling graph. It generates a scheduling column for each candidate scheduling, obtains the column test value by subtracting the multiplier cost from the number of aliased bit splits, and writes the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, the same order shift relationship of the two event bits is written into the cut constraint, and the output is a readable scheduling column graph. The decision module is used to read the construction shift schedule based on the readable schedule list, generate the organizational activity risk value by the number of times the completion time of the event position in the schedule list crosses the shift handover time, and determine the target schedule list by ascending organizational activity risk value, descending number of readable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support.

[0014] The technical effects and advantages of this invention are as follows: This scheme incorporates the completion time of construction actions and changes in monitoring sampling into the same dynamic modeling graph, enabling the monitoring response to be written back to the event location. This allows for the identification of monitoring segments that can be used for support decision-making without suspending construction or adding testing procedures. The response bit number is used to write back the statistics to distinguish between readable bits and overlapping bits, so that the situation where multiple construction actions point to the same monitoring response is explicitly marked, which relatively reduces the confusion between support effectiveness and the continued development of bias. Lagrange relaxation scheduling is performed on the event bits corresponding to the aliasing positions, and loop timeout and loop multiplier are introduced to make the adjustment of support procedures subject to the loop time limit, thereby improving the on-site executability of candidate scheduling. In the branch pricing cut calculation, the scheduling columns with positive test values ​​are retained, and the cut constraints are used to exclude the move combinations that still cause response aliasing, which enhances the readability of the candidate schedules. Calculate the risk value of organizational activities based on the handover time of the work group, so that the selection of target scheduling columns takes into account monitoring and interpretation, loop timeout and work group handover relationship, thereby relatively reducing the organizational risk brought about by cross-work group execution. Attached Figure Description

[0015] Figure 1 This is a flowchart of the dynamic modeling and event interpretation process of the present invention; Figure 2 This is a flowchart of the Lagrange relaxation scheduling process of the present invention; Figure 3 This is a flowchart of the scheduling column filtering and target scheduling determination process of the present invention; Figure 4 This is a system module execution diagram 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 1-4 The present invention provides an intelligent decision analysis method for eccentric support in loess tunnels, comprising: S1. Obtain the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. Write the event bit according to the completion time of the construction action, and write the response bit according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point. Output the excavation cycle dynamic modeling diagram. In this embodiment, S1 is used to place the completion time of construction actions and the changes in monitoring point sampling within one excavation cycle of a loess tunnel into the same time sequence, so that the support actions and monitoring responses in construction management have a calculable sequential relationship; the excavation start signal comes from the first excavation confirmation record of this cycle written to the construction record terminal, and the support loop confirmation signal comes from the support completion acceptance record. If either signal is missing, the excavation cycle dynamic modeling diagram will not be generated for this excavation cycle, and the excavation cycle index corresponding to the missing signal will be written into the anomaly record; this implementation process includes the following steps: The purpose of S11 is to generate an event bit sequence, so that the construction execution flow can be transformed from the original construction records into sortable event bit data; The input quantities are the construction execution flow, the acquisition time of the excavation start signal, and the completion time of each construction action. During processing, the acquisition time of the excavation start signal is used as the zero time of the cycle. The completion time of each construction action is read one by one. The event time difference is obtained by subtracting the zero time of the cycle from the completion time of the construction action. Then, the event time difference is written into the event bit in ascending order. The event bit must be written with at least the event bit number, construction action completion time, event time difference, construction sequence and write-back field. If the event time differences are the same, the event bit number is generated according to the writing order of the construction execution flow. The event bit sequence is output and read by S13. When the construction execution flow is empty, the event bit sequence is not written, and the current excavation cycle index is written to the missing record of the construction execution flow. The purpose of S12 is to generate a response bit sequence, so that the monitoring sampling stream is transformed from the changes in sampled values ​​into response bit data composed of positive signs, zero signs and negative signs; The input quantities are the monitoring sampling stream, the cycle zero time, the sampling time of the monitoring point, and the sampled value. During processing, the sampled value is first retained according to the recording accuracy of the monitoring equipment. Then, adjacent samples at the same monitoring point are taken as a sampling pair. The sampling difference is generated by subtracting the previous sampled value from the subsequent sampled value. If the sampling difference is greater than zero, a positive sign is written. If the sampling difference is equal to zero, a zero sign is written. If the sampling difference is less than zero, a negative sign is written. The response time difference is obtained by subtracting the cycle zero time from the subsequent sampling time. The response bit must be written with at least the response bit number, monitoring point number, next sampling time, response time difference and sign. If the response time differences are the same, the response bit number is generated in ascending order of the monitoring point number. The response bit sequence is output and read by S13. If there are fewer than two sampled values ​​at the same monitoring point in the current excavation cycle, no sample pair will be generated and no response bit will be written. The purpose of S13 is to generate a dynamic modeling diagram of the excavation cycle, so that the event bit sequence and the response bit sequence form a traversable time bit map according to the same excavation cycle index. The input consists of an event bit sequence, a response bit sequence, an excavation start signal, and a support loop confirmation signal. During processing, the excavation cycle index is determined by the excavation start signal and the support loop confirmation signal of one excavation cycle. The event bits and response bits are written into the excavation cycle index in ascending order of event time difference and response time difference. When the event time difference and response time difference are equal, the event bits are written first and then the response bits. When the time differences of two event bits are equal, they are written in ascending order of event bit number. When the time differences of two response bits are equal, they are written in ascending order of response bit number. Then, bit edges are written between adjacent event bits or response bits in ascending order, with the direction of the bit edges pointing from the event bit or response bit with smaller event time difference or response time difference value to the event bit or response bit with larger value, and the excavation cycle dynamic modeling diagram is output for subsequent event interpretation. When the event bit sequence or response bit sequence is empty, the excavation cycle dynamic modeling diagram only writes the bit data corresponding to the non-empty sequence and writes a sequence missing marker under the excavation cycle index; Through the above implementation process, the completion time of construction actions, the sampling time of monitoring points, and the changes in sampling values ​​are converted into event bits, response bits, and bit edges under the same excavation cycle index. Subsequently, the response bits after the event bits can be read along the bit edge direction, avoiding the inability to directly calculate the sequential relationship caused by storing construction records and monitoring records separately. In practical applications: After a shallow-buried, biased loess tunnel completes one step of excavation, the construction record terminal writes the excavation start signal, the support completion acceptance record writes the support loop confirmation signal, the completion time of shotcrete support and the completion time of anchor support are converted into event time difference and written into the event bit, the subsequent value of adjacent sampling values ​​of the arch shoulder monitoring point is subtracted from the previous value to form a positive, zero, or negative sign and written into the response bit, the system then connects the event bit and response bit in ascending order of time difference to form an excavation cycle dynamic modeling diagram for intelligent decision analysis of biased support.

[0018] S2. In the dynamic modeling diagram of the excavation cycle, read the first response bit whose sign changes from the time the event bit is completed, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, generate a readable bit; when it corresponds to two or more event bits, generate an aliased bit, and output the event interpretation diagram. In this embodiment, S2 is used to identify the first response bit with directional change significance after each event bit in the excavation cycle dynamic modeling diagram, and to determine whether the relationship between the construction action and the monitoring response is one-to-one or many-to-one by writing back the response bit number; wherein, the conversion of a positive or negative sign to a zero sign is not considered as the first sign change, and when the response chain after the event is empty, the write-back field of the current event bit is written as a null value, and null values ​​are not included in the response bit write-back count table statistics; the implementation process includes the following steps: The purpose of S21 is to generate a corresponding post-event response chain for each event bit, so that the response bits after the event bit form a readable object in the time direction; The input quantities are the event position, response position, position edge, event time difference and response time difference in the excavation cycle dynamic modeling diagram. During processing, each event position is taken as the current event position in sequence, and the data is read from the current event position along the position edge direction to the position before the support ring confirmation event position. When an event bit is read, skip it and continue reading along the bit edge direction. When a response bit with a response time difference greater than the event time difference of the current event bit is read, write it into the event response chain of the current event bit, and then arrange the event response chain in ascending order of response time difference. The output is the response chain following the current event bit, which is then read by S22; if there is no response bit following the current event bit, the response chain following the current event bit is written as an empty chain. The purpose of S22 is to determine the first sign-changed response bit from the post-event response chain, so that the response bit number can be written back to the event bit that triggered the judgment. The input consists of the event response chain, the response bit number, and the sign. During processing, the response bit is read bit by bit starting from the second response bit in the event response chain, and the sign of the current response bit is compared with the sign of the previous response bit. When the sign changes from zero to positive or negative, the current response bit number is written to the write-back field of the current event bit. When the sign changes from positive to negative or vice versa, the current response bit number is written to the write-back field of the current event bit. After writing, the reading of the post-event response chain of the current event bit is stopped. The output is the event bit write-back table, which is read by S23; if the response chain after the event is empty, or if the above symbol change does not occur in the response chain after the event, the write-back field of the current event bit is written to null. The purpose of S23 is to count how many event bits write back the same response bit number, so that the one-to-one and many-to-one relationships between event bits and response bits can be quantified. The input consists of the event bit number and write-back field in the event bit write-back table. During processing, event bit records with null values ​​in the write-back field are deleted. A response bit number statistical item is created with the response bit number as the index, and non-null write-back fields are read one by one. For each response bit number read, the number of event bits corresponding to the response bit number is incremented by one, and the event bit number is written into the event bit number group corresponding to the response bit number. The output is the response bit write-back count table, which is read by S24; if all write-back fields in the event bit write-back table are empty, the response bit write-back count table is written as an empty table. The purpose of S24 is to generate readable bits and aliasing bits based on the write-back count value, and to form an event interpretation graph for subsequent scheduling to read; The inputs are the response bit write-back count table, the event bit number group, and the event time difference. During processing, the write-back count value corresponding to each response bit number is read one by one. When the write-back count value is equal to one, the event bits in the event bit number group are written as readable bits. When the write-back count value is greater than one, all event bits in the event bit number group are written as aliased bits. Then, the event bits are written into the event interpretation diagram in ascending order according to the event time difference to generate the construction sequence. The output is an event interpretation graph, which is then read by the subsequent Lagrange relaxation scheduling algorithm. If the response bit write-back counter table is empty, an empty interpretation flag is written to the event interpretation graph, and no interpretable bits or aliased bits are generated. Through the above implementation process, the time bit edge in the excavation cycle dynamic modeling diagram is transformed into a write-back relationship between event bit and response bit number. When the response bit number is written back by one event bit, it forms a readable bit. When the response bit number is written back by two or more event bits, it forms an aliased bit, thus providing clear splitting objects for subsequent candidate scheduling generation. In practical applications: After the shotcrete support completion event is completed within a certain excavation cycle, the response bit of the arch shoulder monitoring point changes from zero to positive, and the response bit number is written back to the shotcrete support completion event. When the lock foot support completion event is also written back to the same response bit number, the two event bits are written as overlapping bits. Subsequent scheduling separates the event bit relationship corresponding to the same response bit number by moving the construction sequence.

[0019] S3. Based on the event interpretation graph, execute the Lagrange relaxation scheduling algorithm, move the event bit by one construction order to generate candidate schedules, calculate the number of aliasing bits to split, subtract the loop completion time from the loop completion time to obtain the loop timeout, update the loop multiplier with the loop timeout, and output the candidate schedule graph with multiplier cost. In this embodiment, S3 is used to convert the aliased bits in the event interpretation graph into construction scheduling adjustment objects, and to generate a candidate scheduling graph with multiplier cost under the constraint of support loop formation time limit using the Lagrange relaxation scheduling algorithm; the loop formation cutoff time comes from the construction scheduling record corresponding to the current excavation cycle, and is taken as the excavation start signal acquisition time plus the allowable support loop formation time registered in this cycle; the completion time of the event bits in the candidate schedule is recalculated according to the interval between adjacent events in the baseline schedule; this implementation process includes the following steps: The purpose of S31 is to generate candidate schedules so that the event positions corresponding to the overlapping positions can change the construction sequence without disrupting the excavation start and the support ring boundary. The input quantities are the event bits, aliasing bits, and construction sequence in the event interpretation diagram. During processing, the event bits written as aliasing bits are used as the event bits to be moved. The construction sequence in the event interpretation diagram is copied to generate the baseline schedule, and the excavation start event bit is kept in the first construction sequence and the support ring confirmation event bit is kept in the last construction sequence. For each event position to be moved, if exchanging the construction order with the next position will not occupy the first construction order of the excavation start event position, then a forward candidate schedule is generated. If exchanging the construction order with the next position will not occupy the tail construction order of the support ring confirmation event position, then a backward candidate schedule is generated, and the event position number to be moved and the direction of movement are written in the candidate schedule. The output is a candidate schedule and is read by S32; if there is no aliasing bit in the event judgment graph, the candidate schedule is written as an empty schedule; The purpose of S32 is to re-determine whether the candidate schedule should be split into overlapping positions and to calculate the occupancy result of the candidate schedule on the support ring formation time limit; The input quantities are the baseline schedule, candidate schedule, event position completion time, response position number, and loop termination time. During processing, the difference in completion time of event positions of adjacent construction positions in the baseline schedule is first used as the event interval. Then, the event interval is accumulated item by item from the excavation start event position according to the construction order of the candidate schedule, and the completion time of each event position in the candidate schedule is recalculated. Then, the response bit number of the candidate schedule is regenerated according to the response bit number write-back rule of S2. The number of event bits that are written with the same response bit number in the baseline schedule and different response bit numbers in the candidate schedule is recorded as the number of aliasing bit splits. Next, subtract the loop formation deadline from the loop formation completion time of the candidate schedule to obtain the loop formation difference. If the loop formation difference is less than zero, the loop formation timeout is written to zero. If the loop formation difference is equal to or greater than zero, the loop formation timeout is written to the loop formation difference. The output is the candidate schedule evaluation table, which is read by S33. When the candidate schedule is an empty schedule, the candidate schedule evaluation table is written to an empty table. The purpose of S33 is to combine the aliasing bit splitting benefit and the support ring timeout penalty into a multiplier cost using the Lagrange relaxation scheduling algorithm. The input quantities are the candidate scheduling evaluation table, the number of aliasing bit splits, the ring formation timeout, and the ring formation multiplier. During processing, the initial value of the ring formation multiplier is set to 1, and one candidate scheduling record is read from the candidate scheduling evaluation table in each round. Multiply the cyclic multiplier by the cyclic timeout to obtain the cyclic penalty. Subtract the number of aliased bits from the cyclic penalty to obtain the multiplier cost. Add the cyclic timeout to the cyclic multiplier to generate the updated cyclic multiplier. The output is the multiplier cost table, which is read by S34. When the candidate scheduling evaluation table is empty, the multiplier cost table is written as empty, and the cyclic multiplier retains the value before the current reading. The purpose of S34 is to converge and select candidate schedules from the multiplier cost table and form a candidate schedule graph with multiplier costs for the branch-pricing cut algorithm to read. The input quantities are multiplier cost table, candidate schedule, number of aliasing bit splits, ring timeout and multiplier cost. During processing, candidate schedules are selected in ascending order of multiplier cost, and the event bit numbers in the candidate schedules are concatenated into a construction sequence string according to the construction order. The calculation ends when the candidate schedules selected in two adjacent rounds have the same construction priority string and the loop timeout is equal to zero. If the construction priority strings in two adjacent rounds are different, the loop multiplier updated in the previous round continues to execute S33. After the calculation is completed, the candidate schedule, construction sequence string, number of aliased positions split, ring timeout and multiplier cost are written into the same candidate schedule graph node, and the candidate schedule graph with multiplier cost is output; when the multiplier cost table is empty, the candidate schedule graph with multiplier cost is marked as empty candidate. Through the above implementation process, the event bits corresponding to the aliasing bits are no longer directly output as risk results, but are converted into candidate schedules constrained by the support looping time limit. The multiplier cost of the candidate schedule is determined by the number of aliasing bit splits and the looping timeout, so that the candidate schedules read by the subsequent branch pricing cut algorithm have clear construction order, response splitting results and looping constraint results. In practical applications: After the shotcrete support completion event and the toe support completion event are written back to the same response bit number within a certain excavation cycle, the two event bits are written as overlapping bits. S3 swaps the construction order of the toe support completion event bit one position forward and one position backward, recalculates the completion time of the event bits in the candidate schedule, and then re-executes the response bit number writing back. If the two event bits are written back to different response bit numbers after swapping the construction order one position backward and the completion time of the loop does not exceed the loop cutoff time, the candidate schedule obtains a lower multiplier cost and is written into the candidate schedule chart with multiplier cost for subsequent schedule column generation and reading.

[0020] S4. For the candidate scheduling graph, execute the branch pricing cut algorithm to generate a scheduling column for each candidate scheduling. Subtract the multiplier cost from the number of aliased bits to obtain the column test value. Write the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, write the same order shift relationship of the two event bits into the cut constraint and output a readable scheduling column graph. In this embodiment, S4 is used to convert the candidate scheduling graph with multiplier cost into a readable scheduling column graph, so that candidate schedules that can split aliased bits and meet the multiplier cost requirements are added to the main problem scheduling column set. At the same time, the movement relationship that still causes two event bits to be written back to the same response bit number is written into the cut constraint table. The candidate scheduling graph read by S4 is the candidate scheduling graph with multiplier cost output by S3. The scheduling column number is generated incrementally according to the order in which the candidate schedules are written into the scheduling column encoding table. This implementation process includes the following steps: The purpose of S41 is to convert candidate schedules into schedule column codes so that the construction order and response bit write-back relationship have comparable string fields; The input consists of candidate schedules, event bit numbers, construction priority, and event bit write-back fields from a candidate scheduling diagram with multiplier costs. During processing, a schedule column is generated for each candidate schedule. Event bit numbers are read sequentially according to the construction priority of the event bits within the candidate schedule. The fields are then padded with the total number of digits of the event bit numbers and concatenated to generate a construction priority string. Then, response bit numbers in the event bit write-back fields are read according to the same construction priority. The fields are then padded with the total number of digits of the response bit numbers and concatenated to generate a response bit number string. The output is a scheduling column encoding table. The scheduling column encoding table contains the scheduling column number, construction order string, response bit number string, number of aliased bit splits and multiplier cost, and is available for S42 to read. When the candidate schedule is empty, the scheduling column encoding table is marked with an empty column. The purpose of S42 is to perform the pricing calculation in the branch-and-price cut algorithm so that the scheduling column with positive test results enters the main problem scheduling column set; The input consists of the scheduling columns, the number of aliasing bits to be split, and the multiplier cost from the scheduling column coding table. During processing, the scheduling columns are read one by one, and the number of aliasing bits to be split is subtracted from the multiplier cost to obtain the column check value. If the column check value is greater than zero, the scheduling column number, the construction order string, the response bit number string, and the column check value are written into the main problem scheduling column set. The output is the main problem scheduling column set, which is read by S43. When the scheduling column coding table is marked as an empty column, the main problem scheduling column set is marked as an empty set. The purpose of S43 is to identify the response bit number conflicts that still exist in the main problem scheduling list set and to convert the conflict movement relationship into a cut constraint. The input consists of the response bit number string, the construction priority string, and the movement direction in the candidate schedule from the main problem scheduling list. During processing, the response bit number string is read one by one. If two event bit numbers point to the same response bit number in the response bit number string, the event bit numbers of the two event bits are written in ascending order of value. At the same time, the movement direction of the construction priority and the same response bit number are swapped one position forward or backward between the two event bits to generate a prohibited combination. The output is a cut constraint table, which is read by S44. When the main problem scheduling list is marked as empty, the cut constraint table is written with an empty constraint mark. The purpose of S44 is to use the cut constraint table to delete candidate schedules that still retain conflicting response bit numbers, and to continue to perform pricing calculations on the remaining candidate schedules until the column test value is no longer positive. The input consists of a cut constraint table, a candidate scheduling graph with multiplier costs, and a main problem scheduling list. During processing, each candidate scheduling is read sequentially. If a candidate scheduling contains two event bit numbers, the movement direction of the two event bits, and the same response bit number from a prohibited combination, the candidate scheduling is deleted. Recalculate the column test values ​​for the candidate schedules that have not been deleted. If there are candidate schedules with column test values ​​greater than zero, return to S42 to continue writing to the main problem schedule column set. If there are no candidate schedules with column test values ​​greater than zero, end the branch pricing cut algorithm. The output is a readable scheduling list diagram, which is written into the main problem scheduling list set and the cut constraint table. When the cut constraint table is marked as empty, no deletion process is performed, and the readable scheduling list diagram is directly generated based on the main problem scheduling list set. Through the above implementation process, the candidate schedules not only participate in pricing with multiplier cost, but also undergo cut constraint verification through the construction sequence string and response bit number string, so that the candidate schedules that still cause two event bits to be written back to the same response bit number are excluded, and the remaining schedule column can be used as a readable object for subsequent target schedule column selection. In practical applications: When the lock foot support completion event position and the drainage treatment completion event position are still written back to the same response position number in a certain candidate schedule, S43 writes the two event position numbers, two movement directions and the same response position number as a prohibited combination, S44 deletes the candidate schedule that hits the prohibited combination again, and recalculates the column test value for the remaining candidate schedules until all the column test values ​​of the remaining candidate schedules are not positive. The system writes the retained main problem schedule column set and cut constraint table into the readable schedule column diagram.

[0021] S5. Read the construction shift schedule based on the readable schedule list, generate the organizational activity risk value by the number of times the event position completion time in the schedule list crosses the shift handover time, and determine the target schedule list by ascending organizational activity risk value, descending number of readable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support. In this implementation, S5 is used to select the target schedule column that can be implemented in the readable schedule column diagram. Its processing logic is as follows: First, the completion time of the event position in the schedule column is assigned to the shift time period formed by the construction schedule. Then, the organizational activity risk value is generated by the number of times adjacent event positions cross the shift time period. Finally, the organizational activity risk value, the number of readable positions, the loop timeout amount, and the schedule column number are converted into the schedule column sorting key, and the intelligent decision analysis result of the loess tunnel bias support is output with a unified sorting result. The construction schedule comes from the construction schedule record corresponding to the current excavation cycle. If the excavation start signal acquisition time is earlier than the handover time of the first shift, the excavation start signal acquisition time is written as the start time of the first shift time period. If the support loop confirmation time is later than the handover time of the last shift, the support loop confirmation time is written as the end time of the last shift time period. This implementation process includes the following steps: The purpose of S51 is to map the completion time of the event in the scheduling column to a specific work group time period, so that the scheduling column can be included in the calculation of the risk value of organizational activities. The input consists of a readable schedule chart, construction shift schedule time, schedule column number, and event completion time. During processing, the shift handover times are sorted in ascending order of time. The handover times of two adjacent shifts are used as the start and end times of the shift time, respectively. Then, the event completion times in the schedule column are read one by one, and the event completion times are assigned to shift time periods where the start time of the shift time is no later than the event completion time and the end time of the shift time is later than the event completion time. When the completion time of an event bit equals the handover time of the middle shift, it is assigned to the next shift's time period; when the completion time of an event bit equals the end time period of the last shift, it is assigned to the last shift's time period. The output is the event bit shift assignment table, which is read by S52. If the construction schedule is missing, the event bit shift assignment table is marked as missing and is not selected in the target schedule column. The purpose of S52 is to convert the changes in work group time periods into organizational activity risk values, so that the number of construction organization handovers caused by cross-work group execution can be formed into a calculable field. The input consists of the event position group affiliation table, the schedule column number, the event position number, and the group time period. During processing, the initial value of the cross-group count for each schedule column is written to zero. Starting from the second event position according to the construction sequence, the current group time period to which the current event position is assigned and the previous group time period to which the previous event position is assigned are read position by position. When the current shift's time period is different from the previous shift's time period, increment the cross-shift count by one. When the current shift's time period is the same as the previous shift's time period, keep the cross-shift count unchanged. After reading each digit, write the cross-shift count as the risk value of the organized activity. Cross-shift count refers to the cumulative number of shift time period changes obtained by adding one when adjacent event positions within the same schedule column are assigned to different shift time periods; the output is a risk schedule column chart, which is read by S53; when there is only one event position in the schedule column, the cross-shift count remains zero, and the risk value of the organized activity is written to zero; The purpose of S53 is to determine the target scheduling series based on the risks, interpretability and cyclic constraints of construction organization, so that the final output retains both the executability of construction organization and the value of monitoring and interpretation. The input quantities are the risk schedule chart, organizational activity risk value, number of readable bits, loop timeout quantity and schedule column number. During processing, the number of readable bits is counted based on the response bit write-back count table corresponding to the schedule column. Event bits with a write-back count value of one are included in the number of readable bits. Then, the number of readable bits is subtracted from zero to obtain the inverse number of readable bits. Then, the organizational activity risk value, the inverse of the readable bit, the loop timeout amount, and the scheduling column number are written as four fixed-length numerical fields, and the scheduling column sorting key is generated according to the above field order. When sorting, the organizational activity risk value is compared first. If the organizational activity risk values ​​are the same, the inverse of the readable bit is compared. If the inverse of the readable bit is the same, the loop timeout amount is compared. If the loop timeout amount is the same, the scheduling column number is compared. The output is the first-ordered scheduling column, which is written as the target scheduling column, and forms the intelligent decision analysis result for the bias support of loess tunnels; if the risk scheduling column diagram is empty, the output will be marked as no target scheduling column. Through the above implementation process, the scheduling columns in the interpretable scheduling column chart are no longer selected solely based on the ability to split overlapping positions, but also simultaneously include the work group time period affiliation, organizational activity risk value, number of interpretable positions, and loop overtime amount, so that the target scheduling column has a clear data source, sorting criteria, and construction execution constraints. In practical applications: Within the same excavation cycle, there are two candidate scheduling columns. The first scheduling column can generate more readable positions, but the completion time of the event position spans two shift handover times. The second scheduling column generates one less readable position, but the completion time of the event position only spans one shift handover time. The system first generates a shift attribution table for the event position, then obtains the organizational activity risk value, and compares them in ascending order according to the scheduling column sorting key. If the organizational activity risk value of the second scheduling column is lower, the second scheduling column is selected as the target scheduling column and is output as the intelligent decision analysis result of the loess tunnel bias support.

[0022] Furthermore, the present invention also includes an intelligent decision analysis system for eccentric support in loess tunnels, the system comprising a modeling module, an interpretation module, a scheduling module, a pricing module, and a decision module: The modeling module is used to acquire the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. It writes the event bits according to the completion time of the construction action and writes the response bits according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point, and outputs the dynamic modeling diagram of the excavation cycle. The interpretation module is used to read the first sign-changed response bit from the event bit completion time in the dynamic modeling diagram of the excavation cycle, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, a readable bit is generated; when it corresponds to two or more event bits, an aliased bit is generated, and the event interpretation diagram is output. The scheduling module executes the Lagrange relaxation scheduling algorithm based on the event interpretation graph, generates candidate schedules by moving one construction order according to the event bit, calculates the number of aliasing bits to be split, obtains the ring formation timeout by subtracting the ring formation deadline from the ring formation completion time, updates the ring formation multiplier with the ring formation timeout, and outputs a candidate schedule graph with multiplier cost. The pricing module is used to perform a branch-pricing cut algorithm on the candidate scheduling graph. It generates a scheduling column for each candidate scheduling, obtains the column test value by subtracting the multiplier cost from the number of aliased bit splits, and writes the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, the same order shift relationship of the two event bits is written into the cut constraint, and the output is a readable scheduling column graph. The decision module is used to read the construction shift schedule based on the readable schedule list, generate the organizational activity risk value by the number of times the completion time of the event position in the schedule list crosses the shift handover time, and determine the target schedule list by ascending organizational activity risk value, descending number of readable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support.

[0023] Working Principle: This scheme first places the completion time of construction actions and monitoring sampling changes within an excavation cycle into the same time sequence, forming a dynamic modeling diagram of the excavation cycle. Then, it searches along the time sequence for the first monitoring response with a sign change after each construction action and writes the response bit number back to the corresponding event bit, thereby distinguishing between construction actions that can be judged individually and overlapping actions where multiple construction actions point to the same monitoring response. For overlapping actions, the system attempts to move the construction order using the Lagrange relaxation scheduling algorithm, generating candidate schedules within the support loop formation time limit constraint. Then, it uses the branch-pricing cutting algorithm to retain the schedule columns that can split the overlapping responses and eliminates the movement combinations that will still cause response overlap. Finally, it calculates the risk value of organizational activities based on the construction shift schedule, selects the target schedule column from the readable capability, loop formation timeout, and shift handover risk, and outputs the intelligent decision analysis results of bias support. For example, in the construction of shallow-buried biased sections of loess tunnels, after the shotcrete support and anchor support are completed in succession, the deformation direction of the arch shoulder monitoring point changes. However, traditional records make it difficult to determine which construction action the change mainly comes from. This solution first writes the completion time of shotcrete support, the completion time of anchor support, and the arch shoulder sampling change into a unified dynamic modeling diagram. When it is found that two construction actions are written back to the same response position number, it is identified as an overlapping position. Then, the system attempts to adjust one of the support actions forward or backward by one construction sequence, recalculate the response attribution and the loop timeout. If the two construction actions correspond to different monitoring responses after adjustment and do not exceed the loop cutoff time, the schedule with the lower risk value of the organizational activity is selected based on the number of shift handovers as the basis for adjusting the support schedule for the next excavation cycle.

[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 smart decision-making analysis method for eccentric support in loess tunnels, characterized in that, include: S1. Obtain the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. Write the event bit according to the completion time of the construction action, and write the response bit according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point. Output the excavation cycle dynamic modeling diagram. S2. In the dynamic modeling diagram of the excavation cycle, read the first response bit whose sign changes from the time the event bit is completed, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, generate a readable bit; when it corresponds to two or more event bits, generate an aliased bit, and output the event interpretation diagram. S3. Based on the event interpretation graph, execute the Lagrange relaxation scheduling algorithm, move the event bit by one construction order to generate candidate schedules, calculate the number of aliasing bits to split, subtract the loop completion time from the loop completion time to obtain the loop timeout, update the loop multiplier with the loop timeout, and output the candidate schedule graph with multiplier cost. S4. For the candidate scheduling graph, execute the branch pricing cut algorithm to generate a scheduling column for each candidate scheduling. Subtract the multiplier cost from the number of aliased bits to obtain the column test value. Write the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, write the same order shift relationship of the two event bits into the cut constraint and output a readable scheduling column graph. S5. Read the construction shift schedule based on the interpretable schedule chart, generate the organizational activity risk value by the number of times the event completion time in the schedule column crosses the shift handover time, and determine the target schedule column by ascending organizational activity risk value, descending number of interpretable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support.

2. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 1, characterized in that: S1 includes: S11. Obtain the construction execution flow. Take the acquisition time of the excavation start signal as the zero time of the cycle. Read the completion time of each construction action. Subtract the zero time of the cycle from the completion time of the construction action to obtain the event time difference. Write the event bit in ascending order according to the event time difference and output the event bit sequence. S12. Obtain the monitoring sampling stream, take adjacent samples of the same monitoring point as a sampling pair, subtract the previous sampling value from the next sampling value to generate a sampling difference, write a positive sign when the sampling difference is positive, write a zero sign when the sampling difference is zero, write a negative sign when the sampling difference is negative, and obtain the response time difference by subtracting the cyclic zero time from the next sampling time, and output the response bit sequence. S13. Based on the event bit sequence and response bit sequence, write them into the same excavation cycle index in ascending order of event time difference and response time difference. Write bit edges between adjacent event bits or response bits in ascending order. The direction of the bit edges is from the event bit or response bit with smaller event time difference or response time difference value to the event bit or response bit with larger value. Output the excavation cycle dynamic modeling diagram.

3. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 2, characterized in that: S2 includes: S21. Read the event bits in the excavation cycle dynamic modeling diagram, take each event bit as the current event bit in sequence, read the response bits along the bit edge direction whose response time difference is greater than the current event bit, and generate the event response chain after the current event bit in ascending order of response time difference. S22. For the post-event response chain, read the sign of the adjacent response bits in sequence. When the sign changes from zero to positive or negative, or when positive and negative signs are interchanged, write the number of the next response bit into the write-back field of the current event bit and output the event bit write-back table.

4. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 3, characterized in that: S2 also includes: S23. Based on the event bit write-back table, count the number of event bits written to the same response bit number using the response bit number as the index, and write the number of event bits to the write-back count value of the response bit number, and output the response bit write-back count table. S24. Read the response bit write-back count table. When the write-back count value is equal to one, write the event bit written to the same response bit number as a readable bit. When the write-back count value is greater than one, write the event bit written to the same response bit number as an aliased bit. Then write the event bits in ascending order of event time difference into the construction sequence and write them into the event interpretation diagram.

5. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 4, characterized in that: S3 includes: S31. Read the event interpretation diagram, take the event bit written as the aliased bit as the event bit to be moved, copy the construction order in the event interpretation diagram to generate the baseline schedule, and under the construction constraints that the excavation start event bit maintains the first construction order and the support ring confirmation event bit maintains the tail construction order, swap the construction order of the event bit to be moved one position forward or one position backward, and output the candidate schedule. S32. Re-execute the response bit number write-back based on the candidate schedule, record the number of event bits with the same response bit number written in the baseline schedule and different response bit numbers written in the candidate schedule as the number of aliasing bit splits, and subtract the loop completion time from the loop completion time of the candidate schedule to obtain the loop timeout, and output the candidate schedule evaluation table.

6. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 5, characterized in that: S3 also includes: S33. Perform the Lagrange relaxation scheduling algorithm on the candidate scheduling evaluation table. Multiply the ring multiplier by the ring timeout to obtain the ring penalty. Subtract the number of aliasing bit splits from the ring penalty to obtain the multiplier cost. Add the ring timeout to the ring multiplier to generate the updated ring multiplier and output the multiplier cost table. S34. Based on the multiplier cost table, select candidate schedules in ascending order of multiplier cost. The calculation ends when the candidate schedules selected in two adjacent rounds have the same construction order and the loop timeout is equal to zero. Write the candidate schedule, the number of overlapping positions, the loop timeout, and the multiplier cost into the same candidate schedule graph node and output the candidate schedule graph with multiplier cost.

7. The intelligent decision analysis method for eccentric support in loess tunnels according to claim 6, characterized in that: S4 includes: S41. Read the candidate scheduling diagram with multiplier cost, generate a scheduling column for each candidate scheduling, concatenate the event bit numbers according to the construction priority of the event bits in the candidate scheduling to generate a construction priority string, and concatenate the response bit numbers according to the event bit write-back field to generate a response bit number string, and output the scheduling column encoding table. S42. Calculate the pricing using the branch pricing algorithm based on the scheduling column encoding table. Subtract the multiplier cost from the number of aliased bit splits to obtain the column check value. Write the scheduling columns with positive column check values ​​into the main problem scheduling column set and output them.

8. The intelligent decision analysis method for eccentric support in loess tunnels according to claim 7, characterized in that: S4 also includes: S43. Read the response bit number string from the main problem scheduling column set. When two event bit numbers in the response bit number string correspond to the same response bit number, generate a prohibited combination according to the event bit numbers of the two event bits and the direction of movement of the construction sequence by exchanging the two event bits one position forward or one position backward. Write the prohibited combination into the cut constraint table and output it. S44. Read the candidate scheduling graph with multiplier cost according to the cut constraint table, delete the candidate schedule that simultaneously contains two event bit numbers, movement directions and the same response bit number in the prohibited combination, recalculate the column test value for the remaining candidate schedules, and end the branch pricing cut algorithm when all the column test values ​​of the remaining candidate schedules are not positive, and write the main problem scheduling column set and the cut constraint table into the readable scheduling column graph.

9. The intelligent decision-making analysis method for eccentric support in loess tunnels according to claim 8, characterized in that: S5 includes: S51. Read the interpretable schedule chart and construction shift schedule, sort the shift handover times in ascending order of time, take the handover times of two adjacent shifts as the start and end times of the shift time respectively, and assign the event completion time to the shift time where the start time of the shift time is not later than the event completion time and the end time of the shift time is later than the event completion time. Output the event position shift affiliation table. S52. Based on the event position group affiliation table, set the initial value of the cross-group count to zero. Starting from the second event position in the scheduling column, read the current group time period to which the current event position is assigned and the previous group time period to which the previous event position is assigned. If the two are inconsistent, increment the cross-group count by one. After reading each position, write the cross-group count as the risk value of the organization activity and output the risk scheduling column chart. S53. Read the risk scheduling column chart, subtract the number of readable positions from zero to obtain the inverse of the readable positions, and concatenate the organizational activity risk value, the inverse of the readable positions, the loop timeout amount and the scheduling column number in sequence to generate the scheduling column sorting key. Select the first-ranked scheduling column as the target scheduling column according to the ascending order of the scheduling column sorting key, and output the intelligent decision analysis results of the loess tunnel bias support.

10. A smart decision-making analysis system for eccentric support in loess tunnels, used to implement the smart decision-making analysis method for eccentric support in loess tunnels as described in any one of claims 1-9, the system comprising a modeling module, an interpretation module, a scheduling module, a pricing module, and a decision-making module, characterized in that: The modeling module is used to acquire the construction execution flow and monitoring sampling flow formed from the excavation start signal to the support ring confirmation signal within the current excavation cycle. It writes the event bits according to the completion time of the construction action and writes the response bits according to the sign of the subsequent value minus the previous value of adjacent sampling values ​​at the same monitoring point, and outputs the dynamic modeling diagram of the excavation cycle. The interpretation module is used to read the first sign-changed response bit from the event bit completion time in the dynamic modeling diagram of the excavation cycle, and write the response bit number back to the event bit; when the response bit number corresponds to one event bit, a readable bit is generated; when it corresponds to two or more event bits, an aliased bit is generated, and the event interpretation diagram is output. The scheduling module executes the Lagrange relaxation scheduling algorithm based on the event interpretation graph, generates candidate schedules by moving one construction order according to the event bit, calculates the number of aliasing bits to be split, obtains the ring formation timeout by subtracting the ring formation deadline from the ring formation completion time, updates the ring formation multiplier with the ring formation timeout, and outputs a candidate schedule graph with multiplier cost. The pricing module is used to perform a branch-pricing cut algorithm on the candidate scheduling graph. It generates a scheduling column for each candidate scheduling, obtains the column test value by subtracting the multiplier cost from the number of aliased bit splits, and writes the scheduling column with a positive column test value into the main problem. When two event bits are still written back to the same response bit number, the same order shift relationship of the two event bits is written into the cut constraint, and the output is a readable scheduling column graph. The decision module is used to read the construction shift schedule based on the readable schedule list, generate the organizational activity risk value by the number of times the completion time of the event position in the schedule list crosses the shift handover time, and determine the target schedule list by ascending organizational activity risk value, descending number of readable positions, and ascending number of loop overtime, and output the intelligent decision analysis results of loess tunnel bias support.

Images