A data-driven artificial intelligence assisted decision making method and system

Abstract

The present application relates to the technical field of building information management, in particular to a data-driven artificial intelligence assisted decision-making method and system, comprising: recording project data with conflicts based on the division mode of building life cycle; defining the conflict feature matrix of each building component based on the date attribute, allocation attribute and site attribute of each building component in each stage in the project data, and the conflict type of each building component; taking the conflict feature matrix of each building component as a node, spatially clustering each node, and constructing the corresponding decision-making workflow of each building component; time-sequentially associating the decision-making workflow, deducing the decision-making interval, and constructing a multi-level decision tree with the project data in each decision-making interval; using the multi-level decision tree for collaborative analysis, determining the decision path output by the decision tree, and obtaining the current construction processing strategy; the efficiency and accuracy of conflict data processing are realized.

Description

Technical Field

[0001] This invention relates to the field of building information management technology, specifically a data-driven artificial intelligence-assisted decision-making method and system. Background Technology

[0002] With the rapid development of industrialization and intelligence in the construction industry, modern construction projects are characterized by high modularity, multiple construction stages, complex participants, and strict resource constraints. Conflicts among building components will occur at each stage of the building life cycle. If these conflicts are not handled properly, they will directly lead to cost overruns, delays, resource waste, and even quality and safety accidents.

[0003] For example, Chinese Patent Publication No. CN116662393A discloses a construction project control method and system based on distributed data query. The method involves splitting construction project operation information in a distributed control center to obtain multiple distributed nodes; sequentially filtering Q pieces of construction project operation information in each distributed node for noise, and calculating the executable value of each node after filtering; when noise filtering of P distributed nodes is completed, the executable value of the P distributed nodes after noise filtering is obtained based on the total executable value before and after filtering; and determining whether preset construction project operation information is detected based on the executable value of the P distributed nodes. If preset construction project operation information is detected, then the preset construction project operation information is used for construction project control based on distributed data query.

[0004] Existing technologies quantify the value of construction operations by breaking down construction projects into distributed nodes; however, quantitative analysis of distributed nodes alone is insufficient to explain conflicting data in construction project analysis, resulting in low efficiency in collaborative processing of conflicting data and poor data mapping effects. Summary of the Invention

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a data-driven artificial intelligence-assisted decision-making method, including: S1, recording project data that conflict in the current stage or the previous stage based on the division of the building life cycle.

[0006] S2, based on the date attributes, allocation attributes, and site attributes of each building component in the project data at each stage, defines the conflict feature matrix of each building component in a single stage according to the conflict type of each building component.

[0007] S3 integrates the conflict states of building components at each stage, uses the conflict feature matrix of each building component as a node, performs spatial clustering of each node, and constructs the decision-making workflow corresponding to each building component.

[0008] S4 establishes a temporal correlation between the decision-making process, derives the decision intervals under the temporal correlation, and constructs a multi-level decision tree using the project data within each decision interval.

[0009] S5 uses a multi-level decision tree for collaborative analysis to determine the decision path output by the decision tree each time, and determines the current construction processing strategy based on the output of the decision path.

[0010] A data-driven AI-assisted decision-making system includes: a conflict identification module, used to record project data with conflicts in the current or previous stage based on the division of the building life cycle.

[0011] The feature extraction module is used to define the conflict feature matrix of each building component in a single stage based on the date attribute, allocation attribute and site attribute of each building component in the project data at each stage, and the conflict type of each building component.

[0012] The conflict component integration module is used to integrate the conflict states of building components at each stage. Using the conflict feature matrix of each building component as a node, the nodes are spatially clustered to construct the decision-making workflow corresponding to each building component.

[0013] The temporal correlation module is used to correlate the decision-making process in a temporal sequence, deduce the decision intervals under the temporal correlation, and construct a multi-level decision tree based on the project data within each decision interval.

[0014] The decision analysis module is used to perform collaborative analysis using a multi-level decision tree, determine the decision path output by the decision tree each time, and determine the current construction processing strategy based on the output results of the decision path.

[0015] The beneficial effects of this invention are as follows: First, this invention sorts the conflict project data of each stage according to the number of conflicts and the number of building components involved, determines the mapping relationship of the project data of each stage under the time progress, and updates it synchronously; based on the mapping relationship, it traces the scope of the impact of the conflict and defines the output project data; it prioritizes the screening of conflict data with greater impact on the project, provides a data basis for the subsequent cross-stage impact scope identification and conflict handling decision setting, and ensures the pertinence and effectiveness of subsequent decision data.

[0016] Second, this invention locates the associated components of a building component by using date, allocation, and site attributes as constraints; it breaks down the associated components into distributed nodes, calculates and merges the probability of conflict data points, and determines the overall conflict probability of the associated components; it divides the associated components into levels according to the conflict probability and selects the target entities to be output first; it constructs a conflict feature matrix in stages according to the output order of the target entities; it clarifies the data boundaries of the current conflict analysis, so that the conflict feature matrix constructed under the conflict analysis has priority guidance and stage differentiation.

[0017] Third, this invention incorporates spatial coordinates and installation sequence into a conflict feature matrix, using the feature values ​​of the conflict feature matrix as indicators to cluster building components at each stage; it verifies the clustering results from three dimensions: construction area, conflict type, and batch processing feasibility; based on the verification results, it constructs a decision-making workflow to complete the batch merging and hierarchical association of conflict data; then, it formulates temporal association rules for the decision-making process according to the construction stage sequence and the decision-making process trigger sequence; it divides the decision interval according to the temporal association rules to form a multi-level decision tree; and it binds the decision-making process with the actual construction sequence, making subsequent decisions more consistent with the building stage corresponding to the current conflict data, avoiding conflicts between the execution of the decision-making process and the construction procedures, and ensuring the rationality of the decision settings.

[0018] Fourth, this invention obtains target feature values ​​for decision paths by targeting conflict resolution rate, engineering cost deviation rate, schedule deviation days, and resource utilization rate; based on the target feature values, it selects effective branches of the decision tree to determine the construction handling strategy to be implemented; and ensures that the selected decision path can achieve the project expectations in terms of conflict resolution, cost control, schedule guarantee, and resource utilization, thereby improving the adaptability and efficiency of the construction handling strategy. Attached Figure Description

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] Figure 1 This is a flowchart illustrating a data-driven, AI-assisted decision-making method.

[0021] Figure 2 This is a flowchart illustrating step S1 of a data-driven, AI-assisted decision-making method.

[0022] Figure 3 This is a flowchart illustrating step S2 of a data-driven, AI-assisted decision-making method.

[0023] Figure 4 This is a flowchart illustrating step S3 of a data-driven, AI-assisted decision-making method.

[0024] Figure 5 This is a flowchart illustrating step S4 of a data-driven, AI-assisted decision-making method.

[0025] Figure 6 This is a flowchart illustrating step S5 of a data-driven, AI-assisted decision-making method.

[0026] Figure 7 This is a system framework diagram of a data-driven, artificial intelligence-assisted decision-making system. Detailed Implementation

[0027] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0028] See Figure 1 A data-driven AI-assisted decision-making method includes: S1, recording conflicting project data in the current or previous stage based on the division of the building life cycle.

[0029] S2, based on the date attributes, allocation attributes, and site attributes of each building component in the project data at each stage, defines the conflict feature matrix of each building component in a single stage according to the conflict type of each building component.

[0030] S3 integrates the conflict states of building components at each stage, uses the conflict feature matrix of each building component as a node, performs spatial clustering of each node, and constructs the decision-making workflow corresponding to each building component.

[0031] S4 establishes a temporal correlation between the decision-making process, derives the decision intervals under the temporal correlation, and constructs a multi-level decision tree using the project data within each decision interval.

[0032] S5 uses a multi-level decision tree for collaborative analysis to determine the decision path output by the decision tree each time, and determines the current construction processing strategy based on the output of the decision path.

[0033] The project data currently acquired generally involves multiple data sources, such as design drawings, construction records, and monitoring equipment. Data from different sources may differ, and during the configuration, installation, and construction, changes in plans and construction quality issues may lead to conflicts in the timing and space of the configuration of various building components. These are the parts that require rapid decision-making and processing to complete the decision-making process for the construction project.

[0034] At this point, conflict information for each building component at each stage will be recorded. A data table will be used to record fields such as building component identifier, stage, conflict type, conflict description, discovery time, and resolution status.

[0035] like Figure 2As shown, the implementation of step S1 also includes: S11, based on the stages where project data conflicts occur, sorting the project data for each stage according to the number of conflicts and the number of building components involved. The sorting is done by first sorting the number of conflicts in descending order, and then by sorting the number of building components in descending order when the number of conflicts is the same, in order to process the conflicting data for each stage. This means prioritizing the stages and components with the most severe conflicts, ensuring that the mapping relationship is established from the most critical conflict point, thus improving processing efficiency; for example, if a certain construction stage has the most conflicts, then the mapping relationship is established first for the data of that stage.

[0036] S12, taking the sorted project data as the starting point, determines the mapping relationship between project data at each stage under the time progression, and updates the project data synchronously with the mapping relationship.

[0037] The mapping relationship at this point is the association between data of the same building component in different stages. For example, if the current building component is a ventilation duct, the ventilation duct has BIM model data in the design stage, on-site construction records in the construction stage, and maintenance and monitoring data in the operation and maintenance stage. If a construction conflict occurs in any stage, it is necessary to link the data to other stages of this building component to determine the mapping of conflict information in multiple stages, and whether the conflict can cause other conflicts, so as to determine the layout and handling of each component during construction.

[0038] S13 uses the mapping relationship between project data at each stage to trace conflicts, determines the scope of impact of each conflict in project data, and regards the data corresponding to the scope of impact as the output project data.

[0039] After obtaining the mapping relationship between project data, the data records at each stage are traversed to determine the location where the conflict occurs. Based on the description information recorded at this time, all data related to the building component are found to predict the chain conflicts that will occur in the current or subsequent stages. The scope of influence set at this time is initially set based on the description of the conflict in the project data, providing data support for subsequent detailed division, description analysis and decision-making.

[0040] For example, during the construction phase, the site engineer reported a new on-site problem: the pipes collided severely with the structural beams during actual installation, making it impossible to construct according to the drawings.

[0041] At this point, a new construction conflict record is created, and the specific location and content of the conflict are recorded as its conflict description. After marking its conflict type, the conflict records that are sorted with the content are automatically queried according to the established mapping relationship. If the pipeline correction requires adjustment of other data, such as the support and hanger positions must also be moved, then the data of the supports and hangers and other related building components and equipment will be recorded in the scope of influence and integrated into the current output project data.

[0042] In one embodiment of the present invention, the date attribute includes features related to the usage time of the building component, such as production date, transportation time window, installation time requirement, and environmentally sensitive time point; the allocation attribute includes features representing the actual allocation location of the building component, such as resource priority, manpower allocation requirements, and equipment dependency; the site attribute includes location coordinates, environmental constraints, and space limitation parameters, which describe the actual installation scenario of the building component and are used to distinguish whether each building component is affected by the environment, thereby filtering out the configuration path of each component in case of conflict.

[0043] In step S1, we focus on the lifecycle of conflicts in the project and describe the existing conflict issues from a time perspective. In step S2, we will address these conflict issues by transforming them into specific characteristic states and focusing on the horizontal dimension of conflict issues. Under the constraints of time, complex scenarios, and spatial physical limitations, we will complete the definition of conflict labels for each building component.

[0044] like Figure 3 As shown, the implementation of step S2 includes: S21, using the currently input date attribute, allocation attribute and site attribute as constraints, determining the associated components of each building component when there is a conflict.

[0045] At this point, based on the date attribute, allocation attribute, and site attribute as conditions, we will view the associated components that have conflicting relationships with the current building component, so that each building component is associated with one or more associated components.

[0046] S22, calculate the conflict probability between the current building component and its associated components, and determine the target entity of each building component under the corresponding conflict type based on the conflict probability of the associated components.

[0047] During conflict probability verification, the associated components are quantitatively analyzed to verify the authenticity of the conflict and the target entity is determined based on the severity. The target entity can be understood as the component that dominates the conflict or is more severely affected. The conflict probability here is actually a conflict severity score.

[0048] Preferably, step S22 is further implemented by: S221, splitting the associated components into multiple distributed nodes according to the time window, each distributed node representing an associated component appearing in the time window, and each associated component being bound to the corresponding building component.

[0049] S222, based on the constraints corresponding to the distributed nodes, the distributed nodes are split into multiple conflict data points related to the associated components according to the conflict type, and the conflict probability of each conflict data point is calculated.

[0050] Conflict data points represent a conflict dimension between related components and building components, such as time conflict, allocation conflict, site conflict, etc., and are used to calculate the conflict probability under different dimensions.

[0051] When calculating the probability of conflict, attributes are quantified based on the current conflict type and the corresponding date, allocation, and site attributes. For example, the date attribute includes quantified values ​​for time overlap, time urgency, and environmental time constraints. Time overlap is the overlap ratio between the installation or transportation time windows of the current building component and its associated components, used to explain the overlap ratio between the corresponding configuration time periods under installation and transportation. Time urgency is the ratio of the current building component installation deadline to the planned installation duration, used to describe the urgency of the current building component configuration and explain the time constraints under installation. Environmental time constraints are the overlap ratio between the building component installation time and environmentally sensitive time points (such as the rainy season or environmental control period), used to explain the impact of the environment on the building component on the current date.

[0052] For example, the installation time window for building component A is 11:20-11:30, and for related component B it is 11:25-12:05, with a time overlap of 5 / 10 = 0.5. Building component A has only 3 days left to complete its installation, with a planned installation time of 10 days, resulting in a time urgency of 0.3. If 11:25-12:15 represents the distribution of environmentally sensitive time points, then we can determine that the environmental time constraint of building component A is 0.6, and the environmental time constraint of related component B is 1. When calculating the conflict probability between the building component and the related component, the average of the environmental time constraints of these two components will be chosen to illustrate the conflict probability value under environmental time constraints.

[0053] The subsequent allocation attributes include resource competitiveness, priority difference, and equipment dependency overlap. Resource competitiveness is the proportion of the current building component's demand for the same manpower or equipment compared to related components. It illustrates the competitive situation of the current building component in overall production. This demand proportion can be set based on the ratio of the building component's historical equipment usage frequency to the total equipment usage frequency of the building component and related components, thus clarifying the resource competition between the two components. Priority difference is the difference between the resource priority of the related component and the resource priority of the current component, expressed as a positive or negative value. This priority can be calculated based on the priorities pre-configured in the database. Finally, equipment dependency overlap is the ratio of the number of overlapping dependent equipment to the total number of equipment required. For example, if building component A and related component B both require tower cranes, the equipment dependency overlap = 1 / 2 = 0.5; component A has a priority of 3, and component B has a priority of 5, so the priority difference = 2.

[0054] Site attributes include spatial overlap and environmental constraint similarity. Spatial overlap is the ratio of the overlapping area of ​​the spatial projections of the site coordinates of the building component and related components to the total installation area, explaining the overlap between the building component and related components during installation. Environmental constraint similarity is the ratio of the number of components subject to the same environmental constraints (such as noise control, soft soil foundation, etc.) to the total number of environmental constraints, explaining the similarity in the installation scenario. For example, if the spatial overlap area between the installation sites of building component A and related component B is 20㎡, and the total installation area is 100㎡, the spatial overlap is 0.2; since both are subject to soft soil foundation constraints, the environmental constraint similarity is 1.

[0055] At this point, the probability of conflict under each dimension will be determined by the mapping relationship between each conflict data point and the quantified values ​​of the date attribute, allocation attribute, and site attribute, and the weighted value of the corresponding quantified value will be regarded as the probability of conflict for each conflict data point.

[0056] During weighted calculation, the quantized value corresponding to the current conflicting data point will be obtained, its quantized value will be normalized, and the weight of each quantized value will be set according to the ratio of each quantized value to the total feature value, thereby completing the weighted calculation of each quantized value.

[0057] The final output conflict probability will also be normalized based on the number of conflicting data points to explain the conflict situations that occur in each dimension.

[0058] S223, perform fusion processing on the conflict probabilities of each conflicting data point, and regard the fused conflict probability as the conflict probability corresponding to each associated component.

[0059] When fusing the conflict probabilities of conflicting data points, the conflict probabilities of multiple conflicting data points will be integrated based on conditional probability. Assuming that there are conflicting data points in three dimensions, namely time conflict C1, allocation conflict C2, and site conflict C3, the conflict probability after fusion is P(C1∩C2∩C3), which is the probability value of all three conflicting dimensions occurring simultaneously.

[0060] S224. If there is only one associated component, filter the conflict data points corresponding to the associated component and regard the point with the maximum conflict probability among the conflict data points as the target entity to be output first in the corresponding dimension.

[0061] If there is only one conflicting component in the current building component, it is necessary to determine the core conflict dimension between the building component and the related components. At this time, the conflict probability value will be retrieved, and the conflict data point with the maximum value will be selected. The dimension where the conflict data point with the most obvious conflict is located will be regarded as the main part to be processed.

[0062] S225, if there are multiple associated components, filter according to the conflict probability of each associated component, and retrieve the target entity that is output first under multiple associated entities by dividing the associated component hierarchy and aggregating global conflict data points.

[0063] Preferably, step S225 is implemented by dividing the associated components into levels according to the conflict probability of each associated component.

[0064] When classifying the levels, the conflict probability values ​​of the currently associated components are sorted from largest to smallest. Then, the levels are divided according to the quantiles of the conflict probability (e.g., the top 30% is the high conflict level, the middle 40% is the medium conflict level, and the bottom 30% is the low conflict level). The associated components are divided into three levels: the part with the conflict probability in the top 30% is in a high conflict situation and is the data point that is processed first; the part with the conflict probability in the middle 40% is the secondary processing object and is used to assist in the analysis of conflict data; the part with the conflict probability in the bottom 30% has very little conflict and is only used to monitor the construction progress process.

[0065] The system sequentially counts all conflict data points at each level and determines the cumulative value of the conflict probability for each dimension; the dimension with the largest cumulative value is considered the target entity for the current output.

[0066] At this point, the accumulated value will be used to aggregate the conflict data points across components according to the conflict data points under each level. The accumulated value will be used to determine the most critical dimension under the corresponding level, and the corresponding dimension will be regarded as the target entity output by that level, so as to find the conflict forms that need to be focused on under the three levels.

[0067] S226, perform loop processing on the data of the associated components until all data is output as the target entity, and record the conflict type corresponding to the target entity each time it is output.

[0068] Meanwhile, to prevent missing output data, the judgment of the target entity will be repeatedly executed in a loop until all data at each level has been output, thus completing the output of all data.

[0069] S23. Construct conflict feature matrices corresponding to the target entities in sequence according to the output order of the target entities under the corresponding conflict types, and divide the conflict feature matrices according to the stages of building production.

[0070] The final output target entity will be deployed to different production stages according to its corresponding conflict data points, related components, etc., in the form of a feature matrix, so as to complete the setting of the conflict feature matrix under multiple production stages.

[0071] The resulting conflict feature component uses the associated components corresponding to the target entity as rows and the corresponding dimensions as columns to list the main conflict forms of the current building components, thereby obtaining a conflict feature matrix under the building scene. For example, the feature dimensions will be based on the conflict type under time, allocation and site, site coordinates and other features describing the basic attributes of the building components, to obtain a conflict feature matrix containing conflict probability and multiple feature values.

[0072] In one embodiment of the present invention, when implementing step S3, for each conflicting building component, according to the continuous process and batch conflict processing method, the conflict feature matrix of the building component is used to cluster building component nodes with similar spatial location, similar conflict features and matching business attributes into the same cluster. At the same time, the associated components attached to these building components are comprehensively controlled to complete the decision-making work of the building components, reduce the impact of the overall project changes on the building components, and improve the decision-making efficiency of the project advancement process.

[0073] like Figure 4 As shown, the implementation of step S3 includes: S31, introducing the spatial coordinates and installation order of each building component, and adding the spatial coordinates and installation order to the conflict feature matrix; the spatial coordinates will be used to supplement the spatial features of the building components to improve the features of the feature vector in the spatial dimension during subsequent clustering; as for the installation order, it describes the installation dependencies of each building component during project management to supplement the features of each building component in the business dimension.

[0074] S32 uses the eigenvalues ​​of the conflict feature matrix as clustering indices to cluster building components at each production stage. During clustering, the conflict feature matrix is ​​standardized and converted into feature vectors. Clustering is then performed using the index values ​​corresponding to the feature vectors, such as using cosine similarity or Euclidean distance to quantify the similarity between feature vectors. Then, all feature vectors at that stage are clustered. The K-means algorithm can be used to find the K value according to the minimum sum of squares within the cluster, and the maximum silhouette coefficient is used to verify the K value to select the number of clusters to be divided by the current conflict feature matrix.

[0075] S33. Determine whether there is a conflict between building components in the same cluster. If there is no conflict, retrieve the dependencies between building components in each cluster, update the time corresponding to each building component based on the dependencies, and output the time order of the building components as the decision workflow.

[0076] The implementation of step S33 also includes: determining whether building components in the same cluster are located in the same construction area based on the spatial coordinates of each building component; at this time, it is necessary to consider that building components within the same cluster belong to the same partition, that is, at least 90% of the building components belong to the same construction partition, and the remaining building components belong to adjacent partitions, then the corresponding cluster is considered to meet the condition; otherwise, the dimensions of the building components after clustering are too different, making it difficult to uniformly allocate and process them. Adjacent partitions refer to adjacent construction partitions with the same construction technology and management entity. If there are essential differences in the construction requirements of adjacent partitions, re-clustering is required, and the corresponding clusters of the building components need to be obtained again.

[0077] Based on the conflict type corresponding to each building component, it is determined whether the building components in the same cluster are of the same type of conflict. The part that determines the conflict type requires that a certain conflict type is the main conflict type within the cluster, that is, at least 70% of the building components belong to the same main conflict type, which is considered to meet the condition of the same type of conflict. This main conflict type will be set based on the target entity output in step S22. When these data belong to the same main conflict type, it is easier to control and process them. At this time, it can also be set according to the average similarity of the feature values ​​of each building component in the cluster. It is required that the overall calculated average similarity is greater than 0.7, and then it can be considered to meet the condition of the same type of conflict, which can be used to illustrate the consistency under data conflict.

[0078] The 70% mentioned here is for illustrative purposes only. The required consistency ratio must be greater than the threshold under normal settings. Alternatively, the average value determined from historical data can be used for setting.

[0079] Based on the description of each building component, it is determined whether building components in the same cluster can be processed in batches. The determination method for batch processing is based on the existence of at least one strategy that can cover more than 80% of the building components in the cluster. If this value is not met, it is considered not to meet the criteria. In this case, the logic of using a single strategy for group control decision-making is emphasized to complete the comprehensive processing of multiple building components.

[0080] If all of the above conditions are met, it is determined that there is no conflict in the corresponding cluster.

[0081] If the above conditions are not met, the current clustering dimension is adjusted until all clusters meet these three conditions. The mapping relationship under the corresponding cluster is then regarded as the decision-making workflow output at this time.

[0082] The output will construct the decision system according to the overall strategy of the cluster → the processing method between building components within the cluster → the specific processing method of each building component.

[0083] The clustering process used in step S3 is to further clarify the relationships between multiple building components, and to refine the relationships found through the association search to the level of cluster association, so as to supplement the description of the relationships between each building component.

[0084] For example, by associating components, we can obtain direct, indirect, and hierarchical relationships. The resulting clustering relationships between components will make the logic of conflict analysis more complete.

[0085] In one embodiment of the present invention, such as Figure 5 As shown, the implementation of step S4 includes: S41, defining the timing association rules of the decision-making workflow according to the construction phase sequence and the triggering sequence of the decision-making workflow.

[0086] The time sequence of decisions for building components is clarified by focusing on the construction phases of a building project (design → construction preparation → foundation → main structure → decoration and finishing → final acceptance) and the life cycle of the decision-making process (dividing the triggered decision-making process into trigger → execution → verification → closed-loop processing). Multiple coded node descriptions are defined for these divided time sequences.

[0087] The defined temporal association rules will record the decision-making workflow within the same construction phase, the decision-making workflow across construction phases, and new decision-making workflows that emerge due to conflict propagation. New decision-making workflows represent situations where new conflicts arise during the execution of the decision-making workflow. These data will be recorded in the temporal association between the original decision-making workflow and the corresponding construction phase. These recorded data will be mapped into temporal association rules according to the corresponding phase and time point to complete the annotation processing of data under each temporal sequence.

[0088] S42, according to the temporal correlation rules of the decision-making process, is divided into multiple decision intervals in chronological order.

[0089] When dividing decision intervals, we will divide them according to the time dimension. For example, a complete stage of building construction is considered as a stage-based decision interval, a decision interval that includes the complete process of the decision workflow, a decision interval for each step of the decision workflow, and decision intervals before and after the triggering of the decision workflow, to illustrate the distribution of the decision workflow at each point in time.

[0090] S43 transforms the data within the decision interval into a tree-like model of hierarchical nodes and decision paths, and obtains a multi-level decision tree in the time order of each decision region.

[0091] When transforming decision intervals into hierarchical nodes and decision paths, the hierarchical nodes will divide the data under each decision interval into nodes according to the order of construction stage, cluster corresponding to the conflict feature matrix, conflict type, and processing strategy. The construction stage corresponding to the decision interval is the root node, the processing strategy is the leaf node, and the cluster and conflict type of the building component corresponding to the current construction stage are the intermediate nodes. The conflict handling method of the building component under a single stage is determined by construction stage → cluster → conflict type → processing strategy.

[0092] The decision path represents the complete path from the root node to the leaf node for each hierarchical node, representing a set of ways to handle a conflict situation. Finally, a decision tree for dealing with conflicts of building components is constructed according to multiple construction stages.

[0093] In one embodiment of the present invention, such as Figure 6 As shown, the implementation of step S5 includes: S51, taking the conflict resolution rate, engineering cost deviation rate, schedule deviation days and resource utilization rate as targets, and obtaining the target feature values ​​of the decision path under each dimension.

[0094] S52, the decision path corresponding to the target feature value is regarded as the valid branch of the decision tree; and the processing strategy corresponding to the valid branch is used as the current construction processing strategy.

[0095] It should be noted that the conflict resolution rate is the ratio of the overall conflict probability before decision execution to the overall conflict probability after decision execution, divided by the overall conflict probability before decision execution. This quantifies the conflict resolution effect for the installation of building components. The higher the conflict resolution rate, the better the resolution effect for multiple sets of building components.

[0096] The engineering cost deviation rate is calculated by dividing the difference between the actual processing cost of the decision execution and the planned processing cost of the decision execution by the planned processing cost of the decision execution. The actual processing cost is the cost expected to be consumed under the current decision path, and the planned cost originally marked in the project data is used as the planned processing cost at this time to determine whether there is a cost overrun after the decision is executed.

[0097] The number of days for project schedule deviation is the same as the project cost deviation rate. It is also set by subtracting the planned project schedule from the actual project schedule. The larger the value, the more serious the deviation.

[0098] The final resource utilization rate is quantified based on equipment resource utilization rate and human resource utilization rate, which is the actual working hours used divided by the planned working hours. The planned hours described here are all values ​​preset in the project data, and the actual hours used are the working hours expected to be generated after the current decision. The value of adding the equipment resource utilization rate and the human resource utilization rate is normalized to obtain the current resource utilization rate.

[0099] In the current scenario, the target feature value represents the characteristic value of each dimension, such as the feature value under the four dimensions of conflict resolution rate, engineering cost deviation rate, schedule deviation days and resource utilization rate. When selecting the target feature value, it is determined according to the needs of the current decision tree, and the decision path belonging to the effective branch is determined.

[0100] If the current goal is to select the optimal combination of decision paths to solve the problem of poor multi-path coordination, the target feature value will be the weighted value corresponding to the conflict resolution rate, engineering cost deviation rate, construction period deviation days, and resource utilization rate. The target feature value is required to be maximized in order to obtain the optimal combination of multiple decision paths.

[0101] At this point, the average values ​​of conflict resolution rate, project cost deviation rate, schedule deviation days, and resource utilization rate of multiple decision paths are selected as the basis for judgment. These four values ​​are normalized, and the ratio of the normalized value to the total value is regarded as its weight. The weighted value of the average conflict resolution rate and average resource utilization rate is subtracted from the weighted value of the average project cost deviation rate and average schedule deviation days to obtain a target feature value. When the target feature value is maximized, the decision path of the corresponding combination is taken as the effective branch of the output.

[0102] The multiple decision paths obtained at this point represent the optimal way to allocate time for multiple building components, and the combination of ways to minimize engineering costs and schedule deviations, in order to complete the output of the combined results under the decision paths.

[0103] The output decision path also needs to satisfy the following constraints: only one of the conflicting paths can be selected; the cost of selecting the path is less than the budget of the corresponding stage; and the total schedule deviation of the output decision path is less than the schedule deviation tolerated in the current stage.

[0104] To reflect the temporal relationship between multiple building components in resolving conflicts, the number of days of schedule deviation will be analyzed and processed, and a waiting time related to the number of days of schedule deviation will be introduced to determine the waiting time for each building component to adjust for conflicts.

[0105] The system then performs waiting time statistics for the currently selected decision path combination. With the goal of minimizing the number of days of schedule deviation and minimizing waiting time, it further selects the corresponding decision path combination from the effective branches and regards the corresponding decision path combination as the output result.

[0106] Alternatively, it can reflect the maximum resource utilization rate by processing the engineering cost deviation rate and resource utilization rate, using the normalized resource utilization rate minus the engineering cost deviation rate as the objective, and finding the decision path combination that maximizes resource utilization rate and minimizes cost, ultimately obtaining the effective branch of the output.

[0107] Under these two processing methods that reflect resource utilization and time-series correlation, decision paths with conflict resolution rate, project cost deviation rate, schedule deviation days and resource utilization rate as target outputs can be further filtered to obtain effective branches of the output. Alternatively, they can be judged individually to determine the output effect under the combination of multiple decision paths, thereby completing the auxiliary decision-making under the current project data conflict.

[0108] like Figure 7 As shown, the present invention also provides a data-driven artificial intelligence-assisted decision-making system, including: a conflict identification module, a feature extraction module, a conflict component integration module, a temporal correlation module, and a decision analysis module; wherein, the output end of the conflict identification module is connected to the feature extraction module, the output end of the feature extraction module is connected to the conflict component integration module, the output end of the conflict component integration module is connected to the temporal correlation module, and the output end of the temporal correlation module is connected to the decision analysis module.

[0109] The conflict identification module is used to record project data that conflict in the current or previous stage based on the building life cycle division method.

[0110] The feature extraction module is used to define the conflict feature matrix of each building component in a single stage based on the date attribute, allocation attribute and site attribute of each building component in the project data at each stage, and the conflict type of each building component.

[0111] The conflict component integration module is used to integrate the conflict states of building components at each stage. Using the conflict feature matrix of each building component as a node, the nodes are spatially clustered to construct the decision-making workflow corresponding to each building component.

[0112] The temporal correlation module is used to correlate the decision-making process in a temporal sequence, deduce the decision intervals under the temporal correlation, and construct a multi-level decision tree based on the project data within each decision interval.

[0113] The decision analysis module is used to perform collaborative analysis using a multi-level decision tree, determine the decision path output by the decision tree each time, and determine the current construction processing strategy based on the output results of the decision path.

[0114] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A data-driven artificial intelligence assisted decision making method, characterized in that, The method comprises the following steps: S1, based on the division mode of the building life cycle, record the project data in conflict in the current stage or the previous stage; S2, based on the date attribute, allocation attribute and site attribute of each building component in each stage in the project data, and combined with the conflict type of each building component, define the conflict feature matrix of each building component in a single stage; S3, integrate the conflict state of the building component in each stage, take the conflict feature matrix of each building component as a node, and perform spatial clustering on each node to construct the corresponding decision workflow of each building component; The implementation mode of step S3 comprises the following steps: S31, introduce the spatial coordinates and installation sequence of each building component, and add the spatial coordinates and installation sequence to the conflict feature matrix; S32, take the eigenvalue of the conflict feature matrix as a clustering index, and sequentially cluster the building components in each production stage; S33, determine whether the building components in the same cluster exist conflict, if not, call the front-back dependency relationship between the building components in each cluster, update the corresponding time of each building component, and output the time sequence of the building component as a decision workflow; The implementation mode of step S33 further comprises the following steps:

2. The data-driven artificial intelligence aided decision making method of claim 1, wherein, Based on the spatial coordinates of each building component, determine whether the building components in the same cluster are located in the same construction area; Based on the conflict type corresponding to each building component, determine whether the building components in the same cluster are the same type of conflict; Based on the description of each building component, determine whether the building components in the same cluster can be batch processed; When the above conditions are all yes, it is determined that there is no conflict in the corresponding cluster; 3. The data-driven artificial intelligence aided decision making method of claim 1, wherein, S4, time sequence correlation of the decision workflow is performed, and the decision interval under time sequence correlation is derived, and a multi-level decision tree is constructed by using the project data in each decision interval; The implementation mode of step S4 comprises the following steps: S41, define the time sequence correlation rule of the decision workflow according to the construction stage time sequence and the triggering time sequence of the decision workflow; S42, divide the decision workflow into multiple decision intervals in time sequence according to the time sequence correlation rule of the decision workflow; S43, convert the data in the decision interval into a tree model of hierarchical nodes and decision paths, and obtain a multi-level decision tree according to the time sequence of each decision interval; S5, use the multi-level decision tree for collaborative analysis, determine the decision path output by the decision tree each time, and determine the current construction processing strategy based on the output result of the decision path. The implementation mode of step S1 further comprises the following steps: S11, based on the stage in which the project data appears conflict, sort the project data in each stage according to the number of conflicts and the number of building components involved; S12, take the sorted project data as a starting point, determine the mapping relationship between the project data in each stage under time advancement, and update the mapping relationship to the project data synchronously; S13, trace the conflict based on the mapping relationship between the project data in each stage, determine the influence range of the project data each time when the conflict occurs, and take the corresponding data of the influence range as the output project data. The implementation mode of step S2 comprises the following steps: S21, determine the associated components of each building component in conflict, with the current input date attribute, allocation attribute and site attribute as constraint conditions; S22, calculate the conflict probability of the current building component and the associated component, and determine the target entity of each building component under the corresponding conflict type according to the conflict probability corresponding to the associated component; S23, according to the output order of the target entity under the corresponding conflict type, the conflict feature matrix corresponding to the target entity is constructed in turn, and the conflict feature matrix is divided according to the stage of building production.

4. The data-driven artificial intelligence aided decision making method of claim 3, wherein, The implementation mode of step S22 further comprises: S221, the associated components are split into multiple distributed nodes according to the time window, and each distributed node represents an associated component appearing in the time window, and each associated component is bound to the corresponding building component; S222, according to the constraint condition corresponding to the distributed node, the distributed node is split into multiple conflict data points related to the associated component according to the conflict type, and the conflict probability of each conflict data point is calculated; S223, the conflict probability of each conflict data point is fused, and the fused conflict probability is regarded as the conflict probability corresponding to each associated component; S224, if the associated component is one, the conflict data points are filtered according to the conflict data points corresponding to the associated component, and the point corresponding to the maximum conflict probability in the conflict data points is regarded as the target entity to be output preferentially in the corresponding dimension; S225, if the associated component is multiple, the conflict probability corresponding to each associated component is filtered, and the target entity to be output preferentially under multiple associated entities is called in the form of dividing the associated component level and global conflict data point aggregation; S226, the data of the associated component is processed in a loop until all data are output as target entities, and the conflict type corresponding to the target entity at each output is recorded.

5. The data-driven artificial intelligence aided decision making method of claim 4, wherein, The implementation mode of step S225 comprises: According to the conflict probability corresponding to each associated component, the level corresponding to each associated component is divided; The cumulative value of the conflict probability in each dimension is determined by sequentially counting all conflict data points in each level, and the dimension with the maximum cumulative value is regarded as the target entity to be output.

6. The data-driven artificial intelligence aided decision making method of claim 1, wherein, The implementation mode of step S5 comprises: S51, taking the conflict resolution rate, the engineering cost deviation rate, the time deviation days and the resource utilization rate as the target, and taking the combination result of any target, the target feature value of the decision path in each dimension is obtained; S52, the decision path corresponding to the target feature value is regarded as the effective branch of the decision tree, and the processing strategy corresponding to the effective branch is taken as the current construction processing strategy.

7. A data-driven artificial intelligence aided decision system for performing the steps of a data-driven artificial intelligence aided decision method according to any one of claims 1-6, characterized in that, Comprise: The conflict identification module is used for recording the project data with conflict in the current stage or the previous stage based on the division mode of building life cycle; The feature extraction module is used for defining the conflict feature matrix of each building component in single stage based on the conflict type of each building component in each stage, the date attribute, the allocation attribute and the site attribute in the project data; The conflict component integration module is configured to integrate conflict states of building components at different stages, take a conflict feature matrix of each building component as a node, perform spatial clustering on the nodes, and construct a decision workflow corresponding to each building component. The time sequence correlation module is configured to perform time sequence correlation on the decision workflow, deduce a decision interval under time sequence correlation, and construct a multi-level decision tree using project data in each decision interval. The decision analysis module is configured to perform collaborative analysis using the multi-level decision tree, determine a decision path output by the decision tree each time, and determine a construction processing strategy currently executed based on an output result of the decision path.

Images