Bridge construction intelligent scheduling risk assessment method and system

Abstract

This invention provides a method and system for intelligent scheduling risk assessment in bridge construction. The method includes: reading construction task package data, resource occupancy data, work surface boundary data, time sequence constraint data, and responsibility handover data from the bridge construction site; generating an original set of work packages after field unification, time alignment, work surface number verification, responsibility entity mapping, and resource identifier verification; generating an interface work package set based on the pre- and post-positional relationships, shared transportation channels, and responsibility handover relationships between work packages; verifying the interface integrity of the interface work packages, calculating the interface breakage risk value, and generating an interface integrity record; and generating a list of propagation objects and interface risk propagation results based on the interface integrity record. This invention can identify the interface breakage risk formed by the combination of multiple normal work packages at the interface, improving the accuracy of bridge construction scheduling risk assessment and the continuity of scheduling and handling.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for bridge construction, and in particular to a method and system for intelligent scheduling and risk assessment of bridge construction. Background Technology

[0002] Bridge engineering projects are typically characterized by long construction periods, numerous participating entities, dispersed work areas, and frequent overlapping of resources. Existing intelligent scheduling systems for bridge construction usually manage risks and adjust plans based on individual construction tasks or single resources. However, when dealing with the continuous construction of complex projects, existing technologies have the following shortcomings: First, the existing system focuses on verifying the status of a single task, neglecting the time continuity, space transfer, resource release, responsibility handover, and release relationships required when multiple normal work packages are interconnected. Due to the lack of an evaluation mechanism that quantitatively verifies "interfaces" as independent objects across multiple dimensions (time, space, resources, responsibility, and status), the scheduling process is prone to overlooking the risks of delays and disruptions that occur at the handover points after task combinations. Secondly, the risk of interface failure during actual construction is highly contagious. When a delay occurs at an interface, existing technologies cannot effectively trace the propagation path of the risk along chains such as preceding and following relationships, continuous resource occupation, or responsibility handover. Because it is difficult to accurately retrieve affected subsequent objects, schedulers cannot predict in advance the chain reaction caused by local interface problems leading to restrictions on the initiation of subsequent work packages, large-scale resource freezes, and delayed opening of work areas. Therefore, this invention proposes a method and system for intelligent scheduling risk assessment in bridge construction. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of existing technologies by providing a method and system for intelligent scheduling and risk assessment of bridge construction, thereby resolving the technical problems mentioned in the background section.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent scheduling and risk assessment of bridge construction includes the following steps: S1. Read the construction task package data, resource occupation data, work surface boundary data, time sequence constraint data and responsibility handover data of the bridge construction site. After field unification, time alignment, work surface number verification, responsibility subject mapping and resource identifier verification, generate the original work package set. S2. Generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships among work packages in the original work package set. Match the output conditions of the previous work package with the receiving conditions of the next work package, and generate an interface work package set through imitation learning correction. S3. Generate interface baseline condition records based on the interface job package set, perform interface integrity verification, and generate interface integrity records containing interface breakage risk values. S4. Generate a list of propagation objects based on the interface integrity record and the original job package set, calculate the propagation impact value of the interface breakage risk, and generate an interface priority scheduling sequence. S5. Generate the bridge construction intelligent scheduling risk assessment result based on the interface priority scheduling sequence, interface integrity record and interface risk propagation result, output and conflict-check the scheduling adjustment instruction, write it back to the original work package set, and generate the updated bridge construction intelligent scheduling scheme.

[0005] S1 includes: reading construction task package data, resource occupancy data, work surface boundary data, time sequence constraint data and responsibility handover data at the bridge construction site, and generating an initial construction data table according to the rule that a single construction task forms a record; The initial construction data table is standardized in terms of fields, time alignment, work area number verification, responsibility mapping, and resource identification verification. Records lacking necessary fields are deleted and abnormal records are marked to generate a standard construction data table. The work packages are encapsulated based on the standard construction data table to generate an original set of work packages. Each work package carries a work package identifier, input conditions, output conditions, resource usage field, boundary usage field, timing constraint field, and responsibility handover field.

[0006] S2 includes: reading the original set of work packages, taking work packages with temporal or on-site overlapping relationships as work package pairs to be judged, verifying the pre- and post-positional relationships, adjacent relationships of work surfaces, resource sharing relationships, shared transportation channels, and responsibility handover relationships, and generating a candidate interface relationship table; Read the candidate interface relationship table and the original job package set, match the output conditions of the previous job package with the receiving conditions of the next job package, determine the interface type, interface triggering conditions and interface acceptance conditions, and generate the initial set of interface job packages; Based on historical bridge construction scheduling and handling samples, we perform imitation learning to generate interface discrimination weight records, and perform deduplication, completion and priority screening on the initial set of interface operation packages to generate a set of interface operation packages.

[0007] S3 includes: reading the interface job packet set, extracting the output conditions of the previous job packet and the receiving conditions of the next job packet in each interface job packet, and generating interface baseline condition records according to time connection, space occupation, resource release, responsibility handover and release status; Read the interface baseline condition record, perform interface integrity verification on each interface job package, determine time connection deviation, space occupation overlap, resource release deviation, responsibility handover gap and release status gap, and generate interface deviation record; Read the interface deviation record and interface discrimination weight record, calculate the interface breakage risk value, generate the interface integrity record, and use the interface integrity record for subsequent interface risk propagation analysis.

[0008] S4 includes: reading the interface integrity record and the original work package set, starting from the interface work package identifier, interface breakage risk value and deviation source field, retrieving the affected subsequent work packages, resources, transportation channels, work surfaces and time intervals, and generating a list of propagation objects; Based on the list of objects to be propagated and the integrity record of the interfaces, the risk propagation results of the interfaces are generated along the relationships of preceding and subsequent relationships, continuous relationships of work surfaces, continuous relationships of resource occupation, relationships of transportation channel occupation, and relationships of responsibility handover, and the propagation impact value is calculated; Read the interface risk propagation results and historical scheduling and handling samples, generate scheduling ranking correction records through imitation learning, and generate interface priority scheduling sequence based on interface breakage risk value, propagation impact value and ranking correction coefficient.

[0009] S5 includes: reading the interface priority scheduling sequence, interface integrity record and interface risk propagation result; determining the risk handling type based on the deviation source field, interface breakage risk value, propagation impact value and scheduling priority value; and generating intelligent scheduling risk assessment results for bridge construction. Based on the risk assessment results of intelligent scheduling for bridge construction, scheduling adjustment instructions are generated, and conflict verification is performed on scheduling adjustment instructions that have overlapping execution time intervals and affect the same resource, work surface, or responsible entity. The scheduling adjustment instructions after conflict verification are written back to the original work package set, generating an updated work package set and an updated intelligent scheduling scheme for bridge construction, and triggering the next round of risk assessment based on field changes.

[0010] A bridge construction intelligent scheduling and risk assessment system includes: The construction data parsing and work package construction module is used to read construction task package data, resource occupancy data, work surface boundary data, time sequence constraint data and responsibility handover data from the bridge construction site. After field unification, time alignment, work surface number verification, responsibility entity mapping and resource identifier verification, the module generates the original work package set. The construction interface relationship matching module is used to generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships between work packages in the original work package set. It matches the output conditions of the previous work package with the receiving conditions of the next work package and generates an interface work package set through imitation learning correction. The interface baseline condition verification module is used to generate interface baseline condition records based on the interface job package set, perform interface integrity verification on each interface job package, determine time connection deviation, space occupation overlap, resource release deviation, responsibility handover gap and release status gap, and generate an interface integrity record containing interface breakage risk value. The risk propagation chain analysis and priority calculation module is used to generate a list of propagation objects based on the interface integrity record and the original work package set, calculate the propagation impact value of the interface breakage risk, and generate an interface priority scheduling sequence in combination with historical bridge construction scheduling and disposal samples. The scheduling instruction generation and scheme update module is used to generate intelligent scheduling risk assessment results for bridge construction based on the interface priority scheduling sequence, interface integrity record and interface risk propagation results, output and conflict-check scheduling adjustment instructions, write back the conflict-checked scheduling adjustment instructions to the original work package set, and generate an updated work package set and an updated intelligent scheduling scheme for bridge construction.

[0011] The beneficial effects of this invention are as follows: This invention encapsulates bridge construction tasks into a set of original work packages and further generates a set of interface work packages, thereby expanding the risk assessment object from a single construction task to the interface relationship between work packages. It can identify the risk of interface breakage formed when multiple normal work packages are combined at the connection point, thus improving the accuracy of bridge construction scheduling risk identification.

[0012] This invention generates interface baseline condition records based on time connection, space occupation, resource release, responsibility handover and release status, and calculates the interface breakage risk value. It can transform interface blockage, overlap, release lag, handover gap and release gap into verifiable data results, reducing the error of relying solely on human experience to judge interface risk.

[0013] This invention determines the impact range of interface failure risk on subsequent construction task packages, resource occupation periods, and work site opening periods by disseminating a list of objects and the results of interface risk propagation. It can detect the trend of a single interface problem spreading to subsequent work chains in advance, reducing the probability of restricted initiation of subsequent work packages, resource freezes, and delayed opening of work sites.

[0014] This invention combines historical bridge construction scheduling and handling samples for imitation learning to generate interface discrimination weight records and scheduling order correction records, so that interface identification and interface priority scheduling sequences can absorb historical manual handling experience and improve the consistency between interface risk ranking and on-site handling priority.

[0015] This invention generates scheduling adjustment instructions based on the interface priority scheduling sequence, and writes the scheduling adjustment instructions back to the original work package set after conflict verification, forming an updated work package set and an updated intelligent scheduling scheme for bridge construction. It can realize a closed loop of interface identification, risk assessment, scheduling handling and scheme updating, and improve the continuity and executability of construction scheduling. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a bridge construction intelligent scheduling risk assessment method according to the present invention; Figure 2 This is a schematic diagram of the framework of a bridge construction intelligent scheduling risk assessment system according to the present invention. Detailed Implementation

[0017] 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.

[0018] Example 1: As Figure 1 As shown in the figure, this embodiment provides a method for intelligent scheduling risk assessment of bridge construction, including the following steps: S1. Read the construction task package data, resource occupation data, work surface boundary data, time sequence constraint data and responsibility handover data of the bridge construction site. After field unification, time alignment, work surface number verification, responsibility subject mapping and resource identifier verification, generate the original work package set. S2. Generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships among work packages in the original work package set. Match the output conditions of the previous work package with the receiving conditions of the next work package, and generate an interface work package set through imitation learning correction. S3. Generate interface baseline condition records based on the interface job package set, perform interface integrity verification, and generate interface integrity records containing interface breakage risk values. S4. Generate a list of propagation objects based on the interface integrity record and the original job package set, calculate the propagation impact value of the interface breakage risk, and generate an interface priority scheduling sequence. S5. Generate the bridge construction intelligent scheduling risk assessment result based on the interface priority scheduling sequence, interface integrity record and interface risk propagation result, output and conflict-check the scheduling adjustment instruction, write it back to the original work package set, and generate the updated bridge construction intelligent scheduling scheme.

[0019] S1 specifically includes the following sub-steps: S110. Read the construction task package data, resource occupation data, work surface boundary data, time sequence constraint data and responsibility handover data of the bridge construction site, and generate an initial construction data table according to the rule that a single construction task forms a record.

[0020] Construction task package data is read from the construction planning system, project progress ledger or daily site report, and includes at least the process name, planned start time, planned end time, contract section, work surface and preceding / following relationship; resource occupancy data is read from equipment schedule, personnel schedule, material arrival record and transportation channel occupancy record, and includes at least the equipment resource identifier, personnel team identifier, material batch identifier, transportation channel identifier and resource occupancy period. Work area boundary data is read from the construction plan division record or work area handover record, and includes at least the work area location, adjacent work area identification, boundary occupancy conditions, and boundary release conditions; time sequence constraint data is read from the construction organization plan, road closure plan, navigation closure plan, and process waiting requirements, and includes at least the earliest start time, latest completion time, waiting time, and closed construction window; responsibility handover data is read from the acceptance record, release record, and handover confirmation record, and includes at least the output responsible entity, the receiving responsible entity, acceptance conditions, and release conditions.

[0021] Each record in the initial construction data table is bound to a contract section, a work surface, and a planned time interval, which is determined by the planned start and end times. When the same construction task corresponds to multiple resources or multiple responsible parties, a resource occupancy field and a responsibility handover field are added to the same record, and no separate independent record is generated detached from the construction task. The resulting initial construction data table serves as the data verification object for S120, used to unify fields, verify times, and filter records that cannot be packaged into work packages.

[0022] S120. Read the initial construction data table, and perform field unification, time alignment, work surface number verification, responsibility entity mapping, and resource identifier verification on each record to generate a standard construction data table.

[0023] The fields are unified based on a preset field dictionary, which is a pre-built mapping table containing standard construction field names and their corresponding synonyms (specifically stored in key-value pairs, for example, the key is the standard field and the value is a collection of heterogeneous aliases). The system merges fields with the same meaning from different sources into the same field name. For example, "planned start" and "start time" are unified into "planned start time" based on text similarity matching or dictionary lookup. "construction surface" and "work area" are unified into the corresponding work surface. "acceptance conditions" and "release requirements" are mapped into acceptance conditions and release conditions, respectively.

[0024] Time alignment uses the planned start and end times in the construction task package data as the main time axis, and links resource usage data, work surface boundary data, responsibility handover data, and release condition data to the corresponding planned time intervals, and calculates the planned time span for each record:

[0025] in, Let i be the planned time span for the i-th record, where i is the record number in the initial construction data table. Let i be the planned end time of the i-th record. Let i be the planned start time of the i-th record; when When the value is ≤0, the record is marked as a time-reversed record and will not enter the direct encapsulation process.

[0026] The verification of work surface number is based on the work surface boundary data. It verifies whether the work surface exists in the work surface number database and whether adjacent work surface identifiers can form a valid boundary relationship. The mapping of responsible entities is based on the list of responsible entities. It maps the output responsible entity and the receiving responsible entity to a unique entity number. Records that cannot be mapped to a unique entity number are marked as records where the responsible entity cannot be mapped. The verification of resource identifiers is based on the resource ledger. It verifies whether the equipment resource identifier, personnel team identifier, material batch identifier, and transportation channel identifier exist and whether they are repeatedly occupied in the same time period.

[0027] For records lacking process names, planned start times, planned end times, associated work areas, predecessor / successor relationships, or responsibility handover data, they are deemed unable to form the input or output conditions for a work package and are deleted from the initial construction data table. For records with inverted times, conflicting work area numbers, duplicate resource identifiers, or unmapped responsible entities, their fields are deemed identifiable but require review, and an anomaly flag is added. Field completeness is determined using the following formula:

[0028] in, For the field completeness of the i-th record, Let be the number of necessary fields that have passed verification in the i-th record. The total number of necessary fields required to form the job package; when When =1 and no exception flag field exists, the record is entered into the standard construction data table. If the number of missing fields is less than 1 and the missing field is a necessary field, the record is deleted. The standard construction data table retains verified construction task package data, resource usage data, work surface boundary data, time sequence constraint data, and responsibility handover data, and serves as the sole data source for the original work package set encapsulated in S130.

[0029] S130. Read the standard construction data table and encapsulate the process name, planned start time, planned end time, contract section, work surface, predecessor-successor relationship, resource occupation status, work surface boundary status, time sequence constraint status and responsibility handover status corresponding to the same construction task into a work package, and generate the original work package set.

[0030] A work package is the smallest scheduling object used for subsequent interface relationship identification. Each work package carries a work package identifier, input conditions, output conditions, resource occupancy field, boundary occupancy field, timing constraint field, and responsibility handover field. The work package identifier is generated by combining the contract section, work surface, work process name, and planned start time, and remains unique within the standard construction data table.

[0031] in, This is the identifier for the i-th job package. For the segment to which the i-th job package belongs, For the work surface corresponding to the i-th work package, Let i be the process name corresponding to the i-th work package. Let i be the planned start time for the i-th job package. This indicates field concatenation; when two job packages... If the identifiers are the same, continue reading the responsible entity and resource usage fields to generate extended identifiers until there are no duplicate identifiers within the same original job package set.

[0032] The input conditions are formed by the completion status of the preceding work package, the availability status of resources, the availability status of the work surface, the completion status of responsibility handover, and the release status; the output conditions are formed by the work surface boundary released after the completion of this work package, the released equipment resources, the formed acceptance results, the generated release status, and the responsibility handover field transferred to the next work package.

[0033] The resource occupancy field records the equipment resources, personnel resources, material resources, and transportation channel resources occupied by the work package. The boundary occupancy field records the work surface boundaries occupied and released by the work package. The responsibility handover field records the output responsible party, the receiving responsible party, the acceptance conditions, and the release conditions.

[0034] After the original set of work packages is generated, it is input into S210 to identify the candidate interface relationship table based on the relationship between predecessor and successor, the relationship between adjacent work surfaces, the relationship between shared resources, the relationship between shared transportation channels and the relationship between responsibility handover.

[0035] S2 specifically includes the following sub-steps: S210: Read the original set of work packages generated by S130, take any two work packages that have a time sequence relationship or on-site cross relationship as the work package pair to be judged, and verify one by one whether there is a sequential relationship, adjacent relationship of work surface, resource sharing relationship, transportation channel sharing relationship and responsibility handover relationship between the work package pairs to be judged, and generate a candidate interface relationship table.

[0036] The candidate interface relationship table is a data table used to record the construction interfaces that may be formed between two work packages. Each record includes at least the previous work package identifier, the next work package identifier, the relationship source field, the relationship hit item, the interface relationship strength value, the shared resource identifier, the shared transportation channel identifier, the adjacent work surface identifier, and the responsibility handover field.

[0037] Precedence / Precedence Relationship Match: This means that the preceding task field of the subsequent task package contains the identifier of the preceding task package; Adjacent Task Surface Relationship Match: This means that the task surfaces belonging to two task packages share the same boundary or are adjacent to each other; Resource Sharing Relationship Match: This means that two task packages call the same equipment resources, personnel resources, or material resources; Transportation Channel Sharing Relationship Match: This means that two task packages use the same transportation channel within the planned time interval; Responsibility Handover Relationship Match: This means that there is a handover correspondence between the output responsibility entity of the preceding task package and the receiving responsibility entity of the subsequent task package. The interface relationship strength value is determined according to the following formula:

[0038] in, This represents the interface strength value between task package a and task package b, where task package a is the preceding task package being evaluated, and task package b is the following task package being evaluated. The hit value is the value for the preceding and following relationships. The hit value is the adjacent relationship of the working surface. This is the hit value for resource sharing relationships. This is the hit value for the shared transportation corridor relationship. This represents the hit value for the handover of responsibilities. Each hit value is assigned a value of 1 if it is true, and 0 if it is false. When ≥1, add job package a and job package b to the candidate interface relationship table; when When =0, no corresponding candidate interface relationship record is generated. The candidate interface relationship table is input to S220 to further determine whether the output conditions of the previous job packet and the receiving conditions of the next job packet can form an initial set of interface job packets.

[0039] S220. Read the candidate interface relationship table and the original job package set. For each record in the candidate interface relationship table, match the output conditions of the previous job package with the receiving conditions of the next job package item by item. Determine whether the job package pair to be judged has time continuity requirements, space transfer requirements, resource release requirements, responsibility confirmation requirements or release acceptance requirements. Combine the job package pairs to be judged that meet the matching conditions into the initial set of interface job packages.

[0040] The initial set of interface work packages is a collection of construction interface objects obtained by further filtering the candidate interface relationship table. Each interface work package includes at least the previous work package identifier, the next work package identifier, the interface type, the interface triggering condition, the interface acceptance condition, and the responsibility handover field.

[0041] When the completion of one work package is required for the subsequent work package to start, the interface type is determined to be a time-connection interface; when the current work package needs to release the work surface boundary for the subsequent work package to enter or receive, the interface type is determined to be a space-transfer interface; when the equipment resources, personnel resources, material resources, or transportation channel resources released by the current work package are called by the subsequent work package, the interface type is determined to be a resource release interface; when the output responsibility entity of the current work package needs to complete the handover to the receiving responsibility entity of the subsequent work package, the interface type is determined to be a responsibility handover interface; when the acceptance result or release status of the current work package is used as the start condition for the subsequent work package, the interface type is determined to be a release acceptance interface. The condition matching value is determined according to the following formula:

[0042] in, This represents the conditional matching value between task package a and task package b. This represents the number of conditions that have matched the output conditions of job packet a with the receiving conditions of job packet b. This represents the total number of conditions that need to be validated. The conditions to be validated include time continuity conditions, space transfer conditions, resource release conditions, responsibility confirmation conditions, and release acceptance conditions. If any of the following conditions are met: consistent field names, consistent resource identifiers, adjacent work surface boundaries, corresponding responsible entities, or release conditions that can be accepted, this number is counted as part of the number of matched conditions.

[0043] when When >0, the corresponding job package to be judged will be written into the initial set of interface job packages; when When =0, the task package to be judged is not written into the initial set of interface task packages, but its candidate interface relationship record in the candidate interface relationship table is retained. The initial set of interface task packages is input to S230 and used to perform deduplication, completion and priority screening in combination with historical bridge construction scheduling and disposal samples.

[0044] S230: Read the initial set of interface operation packages, candidate interface relationship table, original operation package set and historical bridge construction scheduling and disposal samples, perform imitation learning correction, deduplication, completion and priority screening on the initial set of interface operation packages, and generate the interface operation package set.

[0045] The historical bridge construction scheduling and handling samples are interface handling records of bridge construction scheduling processes that have occurred. They include at least historical interface blockage records, manual scheduling adjustment records, interface responsibility handover records, interface recovery records, and subsequent delay records. The input for imitation learning is the interface feature fields in the historical bridge construction scheduling and handling samples. The interface feature fields include at least the interface type, relationship hit items, historical blockage duration, manual scheduling actions, recovery duration, and whether subsequent delays occurred. The goal of imitation learning is to learn the retention, merging, deletion, and prioritization results of different interface relationships in manual scheduling and handling. The output of imitation learning is an interface discrimination weight record, which includes at least the interface type weight, relationship hit item weight, lag duration weight, recovery duration weight, and subsequent delay weight. The interface discrimination weight record is used to determine the relevant weights in the initial screening value of the interface and the subsequent interface breakage risk value.

[0046] Deduplication refers to retaining only one interface job package if the previous job package identifier, the next job package identifier, and the interface type are all the same; completion refers to filling in the corresponding fields from the original job package set and the candidate interface relationship table when an interface job package is missing interface trigger conditions, interface acceptance conditions, or responsibility handover fields; priority screening refers to calculating the interface initial screening value based on the interface discrimination weight record and retaining interface job packages that reach the preset initial screening threshold. The interface initial screening value is determined according to the following formula:

[0047] in, The interface of job package a and job package b is the initial screening value of the interface. This represents the interface strength value between task package a and task package b. The comprehensive weight value is determined by the interface's weight record. This represents the conditional matching value between task package a and task package b; The value ranges from 0 to 1. The larger the value, the more likely the same interface is to cause a bottleneck or the more priority it needs to be handled in the historical bridge construction scheduling and handling samples.

[0048] when When the preset initial screening threshold is reached, the corresponding interface work package is retained in the interface work package set; interface work packages that do not reach the preset initial screening threshold are only retained as reference records and are not entered into the subsequent risk calculation process. The interface work package set is input into S310 to generate interface baseline condition records and calculate the interface breakage risk value.

[0049] Specifically, the imitation learning employs a behavior cloning algorithm based on deep neural networks to construct an interface discrimination model; the deep neural network includes an input layer, multiple fully connected hidden layers, and a softmax output layer; the training process of the model includes: Discrete variables in the interface feature fields are one-hot encoded, continuous variables are linearly standardized, and concatenated into a state feature vector, which is then input into the input layer. The manual scheduling actions and their respective priorities in the historical bridge construction scheduling and disposal samples are transformed into probability distribution vectors with corresponding weights, which serve as expert action labels. The error between the weight distribution predicted by the output layer and the expert action label is calculated using the cross-entropy loss function, and the network parameters are updated using the backpropagation algorithm and the Adam optimizer. After training converges, the network connection weight matrix between the last fully connected hidden layer and the output layer of the last iteration is extracted. The degree of influence is determined by calculating the absolute value ratio of the connection weights of each input feature node to the prediction expert's action.

[0050] Specifically, the absolute values ​​of the network connection weights corresponding to interface type, relationship hits, lag duration, recovery duration, and subsequent delay features are subjected to Softmax normalization to decode and generate the weight values ​​of each item in the interface discrimination weight record. to (These correspond to the interface type weight, relationship hit item weight, lag duration weight, recovery duration weight, and subsequent delay weight, respectively, and the sum of each weight value is 1) and the comprehensive weight value. .

[0051] In a specific implementation, the deep neural network includes one input layer, three fully connected hidden layers (with the number of neurons configured as 128, 64 and 32 respectively), and one Softmax output layer. The activation function of the hidden layers is the ReLU function. The LSTM network includes two hidden layers, each with 64 hidden units. To cover common risk propagation depths, the sequence time step is configured as 10.

[0052] The training data for imitation learning comes from more than 10,000 valid samples extracted from the historical bridge construction scheduling logs and manual adjustment records over the past 5 years. The data preprocessing methods include removing abnormal records with missing key fields, performing one-hot encoding on discrete variables, and scaling continuous variables to the [0,1] interval using Min-Max standardization.

[0053] "Expert Action Tag" is defined as the specific scheduling action category taken when the corresponding interface was successfully handled manually in the past (e.g., setting no intervention as category 0, setting buffer time as category 1, adding resources as category 2, and changing handover order as category 3), and is converted into one-hot encoded vector format.

[0054] During training, the Cross-Entropy Loss function is used to measure the error between the probability distribution of each scheduled action predicted by the output layer and the true distribution of expert action labels; the Adam optimizer is used to iteratively update the network parameters, with the initial learning rate set to 0.001 and the batch size set to 64.

[0055] S3 specifically includes the following sub-steps: S310: Read the set of interface job packets generated by S230, extract the output conditions of the previous job packet and the receiving conditions of the next job packet in each interface job packet, and generate an interface baseline condition record.

[0056] The interface baseline condition record is a data record used to describe the baseline state that two job packages should meet at the connection point. Its fields are all derived from the original job package set and the interface job package set, and no additional manually judged fields are introduced.

[0057] Specifically, the completion time of the previous work package, the boundary to be released at the previous work site, the resources to be released, the responsible party for output, the acceptance results, and the release status are derived from the output conditions of the previous work package; the start time of the next work package, the receiving boundary at the next work site, the resources to be received, the responsible party for receiving, the start acceptance requirements, and the release status requirements are derived from the receiving conditions of the next work package.

[0058] For time-connection interfaces, the interface baseline conditions record includes the completion time of the previous job package, the start time of the next job package, and the minimum handover time; for space transfer interfaces, it includes the boundary to be released by the previous job surface, the boundary not released by the previous job surface, the boundary to be received by the next job surface, and the boundary release conditions; for resource release interfaces, it includes the resources to be released, the resources to be received, the resource identifier, the resource release time, and the resource availability status. For responsibility handover interfaces, the output responsible entity, receiving responsible entity, acceptance conditions, and handover confirmation record are written; for release acceptance interfaces, the acceptance result, release status, release time, and subsequent work package startup requirements are written. After the interface baseline condition record is generated, it is bound to the interface work package identifier and serves as the sole baseline data for S320 to perform interface integrity verification, preventing subsequent risk calculations from deviating from the aforementioned work package field source.

[0059] S320. Read the interface baseline condition record, perform interface integrity verification on each interface job package, and generate an interface deviation record. The interface deviation record stores the results of interface job packages failing to meet baseline conditions in five dimensions: time, space, resources, responsibility, and release. Each deviation corresponds to a specific field in the interface baseline condition record. The original time interval is calculated using the following formula:

[0060] in, This represents the original time interval between task package a and task package b, where task package a is the previous task package and task package b is the next task package. The planned start time for the next work package, The planned end time of the previous work package; when If the handover time is less than the minimum handover time recorded in the interface baseline condition record, it is determined that there is insufficient time connection or time connection deviation. The specific normalization calculation formula is as follows:

[0061] in, Due to time discrepancies, This is the minimum handover time recorded in the interface baseline condition record. The preset maximum allowable time deviation; when hour, =0; when At that time, The truncation value is 1, ensuring that its value strictly falls between 0 and 1.

[0062] Space overlap is calculated using the following formula:

[0063] in, This represents the overlap ratio of space occupied between task package a and task package b. This refers to the boundary area of ​​the work surface that was not released in the previous work package. This refers to the boundary range of the work surface that the next work package needs to receive. The boundary range where the two overlap; when When the value is greater than 0, it is determined that there is overlapping space occupation.

[0064] Resource release deviation is verified based on the resource identifiers, quantities, release times, and availability status of the resources to be released and received. A resource release deviation is generated when the equipment, personnel, materials, or transportation resources to be released in the previous work package do not meet the receiving requirements of the next work package. Its normalized calculation formula is:

[0065] in, For resource release deviation, K is the total number of resource types that the next job packet needs to receive. The nominal quantity of resources that should be received by class k. This represents the actual quantity of the k-th type of resource that has been released in the previous job package and is currently available; in the cumulative calculation, if... Then the deviation item corresponding to this type of resource Take 0.

[0066] Gap in responsibility handover Verification should be conducted based on the output responsible entity, the receiving responsible entity, the acceptance conditions, and the handover confirmation record; a responsibility handover gap is identified if any field is missing, the entity cannot be matched, or the handover confirmation record is not generated. Record it as 1, otherwise record it as 0. Release status gap. Verification is conducted based on the acceptance results, release status, release time, and the requirements for initiating the next work package; if the previous work package has not reached a release status that meets the requirements for initiating the next work package, a release status gap is identified. Record it as 1, otherwise record it as 0.

[0067] After the interface deviation record is generated, it should include at least the following fields: interface job package identifier, time connection deviation, space occupation overlap ratio, resource release deviation, responsibility handover gap, release status gap, and deviation source field. These fields should be input into S330 to calculate the interface breakage risk value.

[0068] S330 reads the interface deviation record and the interface discrimination weight record generated in S230, calculates the interface breakage risk value for each interface job package, and generates an interface integrity record. The interface breakage risk value is a comprehensive numerical value representing the probability that an interface job package will experience task stalling, handover failure, or subsequent startup restrictions at the connection point. Its calculation uses a weighted summation method.

[0069] in, The interface breakage risk value of the interface formed by work package a and work package b. This refers to the time alignment deviation after S320 normalization. The overlap ratio of space occupancy. Due to resource release deviation, To address the gap in the handover of responsibilities, The gap is in the clearance status. Weighting for time coherence deviation. For space occupancy overlap weight, To release the bias weight for resources, Weighting of responsibility handover gaps The weights are determined by the interface's judgment weight record for the gap in the release status.

[0070] Before participating in the calculation, the resource release deviation is normalized according to the preset maximum allowable deviation so that its value falls between 0 and 1; the space occupation overlap ratio is determined according to the calculation result of S320; the responsibility handover gap and the release status gap are represented by 0 or 1.

[0071] When the interface breakage risk value reaches the preset risk threshold, the corresponding interface work package will be marked as a high breakage risk interface; The preset risk threshold was determined by performing K-Means cluster analysis on historical bridge construction scheduling samples; Specifically, a five-dimensional feature vector is constructed using time connection deviation, space occupation overlap ratio, resource release deviation, responsibility handover gap, and release status gap as the feature dimension for clustering; The K value was determined to be 3 using the Elbow Method (corresponding to high, medium, and low fracture risk categories); Euclidean distance was used as the distance metric. The iteration termination condition for clustering is set to the single change of the cluster center coordinates being less than 0.0001 or reaching the maximum number of iterations of 100.

[0072] After clustering is completed, all samples in the "high-risk" cluster are extracted, and the lower boundary (such as the minimum value or 5th percentile) of the comprehensive interface breakage risk value of the samples in the cluster is calculated. This lower boundary is used as the preset risk threshold for dividing the "safe breakage boundary". Interface operation packages that do not reach the preset risk threshold retain their interface breakage risk value and do not delete the record.

[0073] The interface integrity record must include at least the following fields: interface work package identifier, previous work package identifier, next work package identifier, interface type, time connection deviation result, space occupation overlap result, resource release deviation result, responsibility handover gap result, release status gap result, interface breakage risk value, risk level, and deviation source field. The interface integrity record is input into S410 to determine the subsequent construction task packages, subsequent resource occupation periods, and subsequent work face opening periods affected by the interface work package.

[0074] S4 specifically includes the following sub-steps: S410: Read the interface integrity record generated by S330 and the original work package set generated by S130. Starting from the interface work package identifier, previous work package identifier, next work package identifier, interface type, interface breakage risk value and deviation source fields in the interface integrity record, retrieve the objects affected by the interface work package on subsequent construction task packages, subsequent resource occupation periods and subsequent work face opening periods, and generate a list of propagation objects.

[0075] The propagation object list is a data list used to record which construction objects may continue to be affected by the risk of interface failure. Each record includes at least the following fields: interface work package identifier, affected subsequent work package identifier, affected resource identifier, affected transportation channel identifier, affected work surface identifier, affected time interval, association type, and association source field.

[0076] The selection of affected subsequent work packages is based on the preceding and following relationships, continuous relationships of work surfaces, continuous occupancy of resources, continuous occupancy of transportation channels, and responsibility handover chain relationships in the original work package set. When a subsequent work package in an interface work package has a preceding and following relationship with a certain subsequent work package, belongs to the subsequent construction section of the same work surface, continuously occupies the same equipment resources or personnel resources, continuously uses the same transportation channel, or is located in the same responsibility handover chain, and the planned start time of the subsequent work package is later than the planned start time of the subsequent work package in the interface work package, the subsequent work package is added to the propagation target list.

[0077] Affected resources and transportation routes are read from the resource occupancy field and transportation route identifier, respectively. Affected work surfaces are read from the work surface boundary field. The affected time interval is determined by superimposing the restricted time corresponding to the deviation source field of the interface work package with the planned time interval of subsequent work packages. After the propagation object list is generated, it is input into S420 to calculate the interface risk propagation results, avoiding the risk of interface failure remaining at the level of a single interface score.

[0078] S420: Read the list of propagation objects and interface integrity records. Based on the interface breakage risk value, the number of affected objects, the affected time range, and the number of affected resources and work surfaces, calculate the propagation impact value of each interface work package and generate the interface risk propagation result.

[0079] Interface risk propagation results are data records used to describe the spread of interface breakage risk to subsequent construction task packages, resource occupation periods, and work surface opening periods. They include at least the risk propagation path, the number of affected objects, the affected time range, the resource freeze period, the work surface delayed opening period, the restricted status of subsequent work package initiation, and the propagation impact value.

[0080] The risk propagation path starts from the interface work package and extends step by step according to the preceding and following relationships, continuous relationships of work surfaces, continuous occupancy of resources, continuous occupancy of transportation channels, and responsibility handover chain relationships in the original work package set; The path extension stops when there are no subsequent related work packages, or when the planned start time of a subsequent work package exceeds the preset propagation time window (in this embodiment, the specific value of the preset propagation time window is set to 7 days or 168 hours), or when the risk attenuation impact value calculated by the current node is lower than the preset lower limit threshold (e.g., 0.1), or when the subsequent work package belongs to the core control milestone node of the project.

[0081] The propagation impact value is calculated according to the following formula:

[0082] in, This represents the propagation impact value of the interface package formed by task package a and task package b, where task package a is the previous task package and task package b is the next task package. This represents the interface breakage risk value for the interface operation package. The normalized value for the number of affected objects in the propagation object list. The normalized value is the value for the affected time range. This represents the normalized value of the number of affected resources and work areas; the normalized value is limited to between 0 and 1, with a larger value indicating a greater corresponding propagation impact.

[0083] in , and The specific normalization formulas are as follows:

[0084]

[0085]

[0086] in, The actual number of searches performed on the affected objects in the list of objects to be affected. Set a preset threshold for the number of objects; This represents the actual total duration of the affected time range. The preset time diffusion cutoff threshold; This represents the actual cumulative number of affected resources and work areas. Set a preset threshold for resource items; when the calculation result is greater than 1, force the value to be 1.

[0087] The resource freeze period is determined based on the original planned release time and the available time after restriction of the affected resources. The delayed opening period of the work face is determined based on the original opening time and the opening time after restriction of the affected work face. The restricted status of subsequent work package initiation is determined based on whether the input conditions of the subsequent work package are blocked by the interface breakage risk. After the interface risk propagation result is generated, it is input into S430 to generate the interface priority scheduling sequence in combination with historical bridge construction scheduling and disposal samples.

[0088] S430: Read the interface risk propagation results, interface integrity records, and historical bridge construction scheduling and handling samples; perform sorting correction on the interface work package set; and generate an interface priority scheduling sequence. Historical bridge construction scheduling and handling samples must include at least historical risk propagation paths, the number of historically affected objects, historical resource freeze periods, historical work face delay opening periods, historical manual scheduling and handling actions, and historical post-handling recovery results.

[0089] The input to imitation learning is the risk propagation path, number of affected objects, affected time range, resource freeze period, delayed opening period of work surface, historical manual scheduling and handling actions, and historical recovery results after handling in the interface risk propagation results; the goal of imitation learning is to learn the results of advance handling, delayed handling, merged handling, or split handling of interfaces with different propagation characteristics in historical manual handling; the output of imitation learning is the scheduling sorting correction record.

[0090] The scheduling and sorting correction record includes at least the propagation path weight, the weight of the number of affected objects, the resource freeze weight, the job surface delay opening weight, the historical handling priority weight, and the sorting correction coefficient. The scheduling priority value for each interface job package is calculated using the following formula:

[0091] in, The scheduling priority value for the interface job package formed by job package a and job package b. This represents the interface breakage risk value for the interface operation package. This represents the propagation impact value of the interface operation packet. The sorting correction coefficient is determined by the scheduling sorting correction record; the sorting correction coefficient is used to characterize the degree to which the same type of propagation feature interface is given priority in the historical bridge construction scheduling and disposal samples.

[0092] Specifically, the imitation learning for generating scheduling order correction records employs a sequence prediction model based on a long short-term memory network (LSTM). The node sequence and node characteristics on the risk propagation path are used as the time step input of the LSTM network, and the network parameters are optimized by using the historical recovery results after treatment as the feedback reward signal. After the network training is completed, the activation values ​​of the fully connected layers are extracted as sorting correction coefficients for the corresponding interface types. .

[0093] When sorting multiple interface job packages, an interface priority scheduling sequence is generated from high to low according to the scheduling priority value. When two interface job packages have the same scheduling priority value, the interface job package with a longer affected time range is given priority. If the affected time range is the same, the interface job package involving the responsibility handover gap or the release status gap is given priority.

[0094] The interface priority scheduling sequence includes at least the interface work package identifier, interface breakage risk value, propagation impact value, sorting correction coefficient, scheduling priority value, sorting position, suggested disposal type and disposal triggering reason, and is input into S510 to determine the risk disposal type and generate bridge construction intelligent scheduling risk assessment results.

[0095] S5 specifically includes the following sub-steps: S510 reads the interface priority scheduling sequence generated by S430, the interface integrity record generated by S330, and the interface risk propagation result generated by S420, determines the risk handling type corresponding to each interface work package, and generates the bridge construction intelligent scheduling risk assessment result.

[0096] The risk assessment results of intelligent scheduling for bridge construction are data records used to describe the source, scope, and direction of risk in interface work packages. Each record includes at least the interface work package identifier, the identifier of the previous work package, the identifier of the next work package, the interface type, the interface breakage risk value, the propagation impact value, the scheduling priority value, the deviation source field, the affected objects, and the risk handling type. The risk handling type is determined based on the deviation source field and the interface risk propagation results. When the deviation source field shows that the time connection deviation is the main deviation, the risk handling type is set to interface buffer period setting; when the resource release deviation is the main deviation, the risk handling type is set to resource early release or resource release order adjustment; when the responsibility handover gap is the main deviation, the risk handling type is set to responsibility handover completion; when the release status gap is the main deviation, the risk handling type is set to release status verification; when the interface risk propagation result shows that the subsequent work package startup is restricted, the risk handling type is set to subsequent work package startup condition lock. When the interface risk propagation results indicate a delayed opening period for the work surface, the risk handling type is determined as "Work Surface Opening Period Adjustment." For the same interface work package with multiple deviation sources, the primary handling type is determined by the deviation item in the deviation source field that contributes the most to the interface breakage risk value, and the remaining handling types are written into the auxiliary handling field. The risk handling intensity value is determined according to the following formula:

[0097] in, The risk handling intensity value is the interface package formed by task package a and task package b, where a is the previous task package and b is the next task package. This is the scheduling priority value for the interface job packet. This represents the interface breakage risk value for the interface operation package. This represents the propagation impact value of the interface work package; the higher the risk handling intensity value, the more priority the interface work package needs to generate scheduling adjustment instructions, and the more affected objects the scheduling adjustment instructions need to act upon. The risk assessment results of intelligent scheduling for bridge construction are input into S520 to generate scheduling adjustment instructions.

[0098] S520: Read the risk assessment results of intelligent scheduling for bridge construction, generate scheduling adjustment instructions for each interface work package that needs to be handled, and perform conflict verification on the generated scheduling adjustment instructions.

[0099] Scheduling adjustment instructions are executable data records used to modify the planning, resource, work surface, responsibility handover, or release status in the original work package set. Each record includes at least the instruction identifier, interface work package identifier, risk handling type, target work package identifier, target resource identifier, target work surface identifier, responsible entity, execution start time, execution end time, triggering reason, execution conditions, and write-back field.

[0100] For the interface buffer period setting type, an interface buffer period setting instruction is generated, and the execution condition is that there is a time connection deviation before the start of the next work package; for the resource early release or resource release order adjustment type, a resource release order adjustment instruction is generated, and the execution condition is that the resources required by the next work package have not become available before the receiving time; for the responsibility handover completion type, a responsibility handover node completion instruction is generated, and the execution condition is that there are gaps in the output responsibility subject, receiving responsibility subject, or handover confirmation record; for the release status verification type, a release status verification instruction is generated, and the execution condition is that the acceptance result, release status, or release time does not meet the start requirements of the next work package. For subsequent job package start-up condition lock types, a subsequent job package start-up condition lock instruction is generated, with the execution condition being that the interface risk propagation result shows that the subsequent job package start-up is restricted; for job surface open time period adjustment types, a job surface open time period adjustment instruction is generated, with the execution condition being that the job surface delayed open time period is greater than 0. After the scheduling adjustment instruction is generated, multiple scheduling adjustment instructions with overlapping execution time intervals and acting on the same resource, the same job surface, or the same responsible entity are subject to conflict verification. The instruction conflict value is determined according to the following formula:

[0101] in, Let be the instruction conflict value between the u-th scheduling adjustment instruction and the v-th scheduling adjustment instruction, where u and v are the sequence numbers of the different scheduling adjustment instructions. This is a value used to determine whether the execution time intervals of two scheduling adjustment instructions overlap; it is set to 1 if the times overlap and 0 if they do not. This is a value used to determine whether two scheduling adjustment instructions apply to the same resource, the same work surface, or the same responsible entity; it is set to 1 if the affected objects are the same, and 0 if they are different. When the value is 1, it is determined that there is a conflict between the two scheduling adjustment instructions. According to the interface priority scheduling sequence, the scheduling adjustment instruction with the higher priority value is retained, and the other scheduling adjustment instruction is written into the instruction record to be adjusted. The scheduling adjustment instruction after conflict verification is input to S530 for writing back the original job packet set.

[0102] S530 reads the scheduling adjustment instructions after conflict verification and the original work package set generated by S130, updates the corresponding work package according to the write-back field of each scheduling adjustment instruction, generates an updated work package set, and generates an updated intelligent scheduling scheme for bridge construction based on the updated work package set.

[0103] The interface buffer period setting command is used to update the planned start time or interface buffer period field of the next job package; the resource release order adjustment command is used to update the resource occupancy field, resource release time, and resource availability status of the relevant job package; the responsibility handover node completion command is used to update the responsibility handover field, handover confirmation status, and responsibility subject confirmation time; the release status verification command is used to update the release status field, release verification time, and release result field; the subsequent job package start condition lock command is used to update the start condition field of the next job package or the affected subsequent job package; the work surface open period adjustment command is used to update the work surface boundary field and work surface open period.

[0104] The updated intelligent scheduling scheme for bridge construction includes at least the updated work package set, the interface work package set, the interface integrity record, the interface risk propagation result, the interface priority scheduling sequence, the scheduling adjustment instruction, the written-back fields, the instruction record to be adjusted, and the triggering conditions for the next round of evaluation; among them, the updated work package set is a data set formed by writing scheduling adjustment instructions on the original work package set.

[0105] After the write-back is completed, if the planned start and end time, resource usage field, work surface boundary field, responsibility handover field, or release status field changes, the next round of intelligent scheduling risk assessment for bridge construction will be triggered. Among them, if only the field values ​​change and the interface relationship does not change, the interface integrity record will be regenerated from S310. If the predecessor and successor relationship, work surface boundary relationship, resource sharing relationship, transportation channel sharing relationship, or responsibility handover relationship changes, the interface work package set will be re-identified from S210, thus forming a closed loop of interface identification, risk assessment, scheduling and handling, and scheme update.

[0106] Example 2: Figure 2 As shown, this embodiment provides a bridge construction intelligent scheduling risk assessment system, including: The construction data parsing and work package construction module is used to read construction task package data, resource occupancy data, work surface boundary data, time sequence constraint data and responsibility handover data from the bridge construction site. After field unification, time alignment, work surface number verification, responsibility entity mapping and resource identifier verification, the module generates the original work package set. The construction interface relationship matching module is used to generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships between work packages in the original work package set. It matches the output conditions of the previous work package with the receiving conditions of the next work package and generates an interface work package set through imitation learning correction. The interface baseline condition verification module is used to generate interface baseline condition records based on the interface job package set, perform interface integrity verification on each interface job package, determine time connection deviation, space occupation overlap, resource release deviation, responsibility handover gap and release status gap, and generate an interface integrity record containing interface breakage risk value. The risk propagation chain analysis and priority calculation module is used to generate a list of propagation objects based on the interface integrity record and the original work package set, calculate the propagation impact value of the interface breakage risk, and generate an interface priority scheduling sequence in combination with historical bridge construction scheduling and disposal samples. The scheduling instruction generation and scheme update module is used to generate intelligent scheduling risk assessment results for bridge construction based on the interface priority scheduling sequence, interface integrity record and interface risk propagation results, output and conflict-check scheduling adjustment instructions, write back the conflict-checked scheduling adjustment instructions to the original work package set, and generate an updated work package set and an updated intelligent scheduling scheme for bridge construction.

[0107] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0108] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0109] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0110] In conclusion, the above description is only 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 bridge construction intelligent scheduling risk assessment method, characterized in that, Includes the following steps: S1. Read the construction task package data, resource occupation data, work surface boundary data, time sequence constraint data and responsibility handover data of the bridge construction site. After field unification, time alignment, work surface number verification, responsibility subject mapping and resource identifier verification, generate the original work package set. S2. Generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships among work packages in the original work package set. Match the output conditions of the previous work package with the receiving conditions of the next work package, and generate an interface work package set through imitation learning correction. S3. Generate interface baseline condition records based on the interface job package set, perform interface integrity verification, and generate interface integrity records containing interface breakage risk values. S4. Generate a list of propagation objects based on the interface integrity record and the original job package set, calculate the propagation impact value of the interface breakage risk, and generate an interface priority scheduling sequence.

2. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, Also includes: S5. Generate the bridge construction intelligent scheduling risk assessment result based on the interface priority scheduling sequence, interface integrity record and interface risk propagation result, output and conflict-check the scheduling adjustment instruction, write it back to the original work package set, and generate the updated bridge construction intelligent scheduling scheme.

3. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, S1 includes: Read the construction task package data, resource usage data, work surface boundary data, time sequence constraint data, and responsibility handover data from the bridge construction site, and generate an initial construction data table according to the rule that a single construction task forms one record; The initial construction data table is standardized in terms of fields, time alignment, work area number verification, responsibility mapping, and resource identification verification. Records lacking necessary fields are deleted and abnormal records are marked to generate a standard construction data table. The work packages are encapsulated based on the standard construction data table to generate an original set of work packages. Each work package carries a work package identifier, input conditions, output conditions, resource usage field, boundary usage field, timing constraint field, and responsibility handover field.

4. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, S2 include: Read the original set of work packages, and select work packages with temporal or on-site overlapping relationships as work package pairs to be judged. Verify the relationships of precedence and precedence, adjacent work surfaces, resource sharing, transportation channel sharing, and responsibility handover, and generate a candidate interface relationship table. Read the candidate interface relationship table and the original job package set, match the output conditions of the previous job package with the receiving conditions of the next job package, determine the interface type, interface triggering conditions and interface acceptance conditions, and generate the initial set of interface job packages.

5. The intelligent scheduling risk assessment method for bridge construction according to claim 4, characterized in that, Also includes: Based on historical bridge construction scheduling and handling samples, we perform imitation learning to generate interface discrimination weight records, and perform deduplication, completion and priority screening on the initial set of interface operation packages to generate a set of interface operation packages.

6. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, S3 include: Read the set of interface job packages, extract the output conditions of the previous job package and the receiving conditions of the next job package in each interface job package, and generate interface baseline condition records according to time connection, space occupation, resource release, responsibility handover and release status. Read the interface baseline condition record, perform interface integrity verification on each interface job package, determine time connection deviation, space occupation overlap, resource release deviation, responsibility handover gap and release status gap, and generate interface deviation record; Read the interface deviation record and interface discrimination weight record, calculate the interface breakage risk value, generate the interface integrity record, and use the interface integrity record for subsequent interface risk propagation analysis.

7. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, S4 include: Read the interface integrity record and the original work package set, starting from the interface work package identifier, interface breakage risk value and deviation source field, retrieve the affected subsequent work packages, resources, transportation channels, work surfaces and time intervals, and generate a list of propagation objects; Based on the list of objects to be propagated and the integrity record of the interfaces, the risk propagation results of the interfaces are generated along the relationships of preceding and subsequent relationships, continuous relationships of work surfaces, continuous relationships of resource occupation, relationships of transportation channel occupation, and relationships of responsibility handover, and the propagation impact value is calculated; Read the interface risk propagation results and historical scheduling and handling samples, generate scheduling ranking correction records through imitation learning, and generate interface priority scheduling sequence based on interface breakage risk value, propagation impact value and ranking correction coefficient.

8. The intelligent scheduling risk assessment method for bridge construction according to claim 1, characterized in that, S5 include: Read the interface priority scheduling sequence, interface integrity record and interface risk propagation result, determine the risk handling type based on the deviation source field, interface breakage risk value, propagation impact value and scheduling priority value, and generate intelligent scheduling risk assessment results for bridge construction. Based on the risk assessment results of intelligent scheduling for bridge construction, scheduling adjustment instructions are generated, and conflict verification is performed on scheduling adjustment instructions that overlap in execution time intervals and act on the same resource, work surface, or responsible entity.

9. The intelligent scheduling risk assessment method for bridge construction according to claim 8, characterized in that, The scheduling adjustment instructions after conflict verification are written back to the original work package set, generating an updated work package set and an updated intelligent scheduling scheme for bridge construction, and triggering the next round of risk assessment based on field changes.

10. A bridge construction intelligent scheduling risk assessment system, employing the bridge construction intelligent scheduling risk assessment method according to any one of claims 1 to 9, characterized in that, include: The construction data parsing and work package construction module is used to read construction task package data, resource usage data, work surface boundary data, time sequence constraint data, and responsibility handover data at the bridge construction site. The construction interface relationship matching module is used to generate a candidate interface relationship table based on the preceding and following relationships, adjacent relationships of work surfaces, resource sharing relationships, transportation channel sharing relationships, and responsibility handover relationships between work packages in the original work package set; The interface baseline condition verification module is used to generate interface baseline condition records based on the interface job package set and to perform interface integrity verification on each interface job package. The risk propagation chain analysis and priority calculation module is used to generate a list of propagation objects based on the interface integrity record and the original work package set, and to calculate the propagation impact value of the interface breakage risk. The scheduling instruction generation and scheme update module is used to generate intelligent scheduling risk assessment results for bridge construction based on the interface priority scheduling sequence, interface integrity records, and interface risk propagation results.

Images