Project management method and system based on multi-dimensional data intelligent analysis

Abstract

The invention relates to the technical field of project management, and discloses a multi-dimensional data intelligent analysis-based project management method and system, and the method comprises the steps: constructing a multi-dimensional feature space, constructing a cross-modal association index for project supervision, generating a delay early warning for project management, and generating an adjustment strategy for project management. The system corresponds to the method. According to the application, a system of data acquisition, space-time mapping, knowledge graph recombination, fusion analysis and closed-loop optimization is constructed; the multi-source heterogeneous data is mapped to a multi-dimensional feature space through a space-time rule, so that the data integration efficiency is improved; the dynamic knowledge graph updates tasks, resources and risk association indexes in real time based on event triggering, and recognizes hidden dependency risks in advance; the project management fusion model combines time sequence dependence and feature importance analysis to realize accurate delay duration evaluation; the cross-modal association index is dynamically adjusted based on delay early warning, and the resource utilization rate is increased; and the problems of analysis lag and low adjustment efficiency are effectively solved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of project management, and particularly relates to a project management method and system based on intelligent analysis of multi-dimensional data. BACKGROUND

[0002] The current project management technology has significant defects in dealing with complex projects:

[0003] Firstly, the multi-source data integration capability is weak. Traditional tools rely on manual input of single type data and cannot automatically integrate financial system, IoT device sensor data and unstructured text logs, forming a data island.

[0004] Secondly, the multi-dimensional correlation analysis capability is missing. Existing solutions only present a single progress dimension through static charts such as Gantt charts, and lack real-time correlation modeling of multi-dimensional indicators such as tasks, resources and risks.

[0005] Thirdly, the risk early warning is lagging and relies on manual work. Traditional methods identify risks through simple threshold rules or manual experience, and cannot mine potential risk transmission paths through a knowledge graph.

[0006] Fourthly, the dynamic optimization mechanism is missing. Resource allocation relies on fixed rules such as critical task priority, and when the critical path task is delayed, it cannot automatically trigger cross-modal correlation index adjustment based on real-time data, resulting in low efficiency in delay compensation.

[0007] These defects make it difficult to adapt to the intelligent management needs of multi-source heterogeneous multi-dimensional data.

[0008] A Chinese patent with the authorized publication number CN116127551B discloses a bridge construction progress fine management and control method based on a digital twin multi-dimensional model, but the invention only relies on time and component number for delay management, which is not good for managing different task delays in projects.

[0009] In view of the above, there is an urgent need for a new project management technical solution based on intelligent analysis of multi-dimensional data. SUMMARY

[0010] The application aims to provide a project management method and system based on intelligent analysis of multi-dimensional data to solve the technical problems in the background art.

[0011] To achieve the above-mentioned purpose, the application discloses the following technical solutions:

[0012] In a first aspect, the application discloses a project management method based on intelligent analysis of multi-dimensional data, which comprises:

[0013] S1: collecting multi-source heterogeneous multi-dimensional data in a project implementation process, mapping the multi-dimensional data by using a space-time rule, and constructing a multi-dimensional feature space; wherein the space-time rule comprises a time rule for unifying a time axis and / or a space rule for unifying space encoding, and the multi-dimensional feature space comprises multi-dimensional features corresponding to the multi-dimensional data and a unified space-time axis corresponding to the multi-dimensional features based on the space-time rule;

[0014] S2: reorganizing the multi-dimensional feature space by using a project management knowledge graph, and constructing a cross-modal correlation index for project management, the cross-modal correlation index being tasks, resources, and risks in a project and correlation relationships therebetween; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal correlation index;

[0015] S3: monitoring the cross-modal correlation index by using a project management fusion model, and generating a delay warning for project management; wherein the project management fusion model is used to fuse time features and task features, analyze project duration, and generate the delay warning;

[0016] S4: generating an adjustment strategy for project management based on the delay warning and the cross-modal correlation index, the adjustment strategy being based on the delay warning to adjust the cross-modal correlation index.

[0017] Preferably, the time rule comprises:

[0018] acquiring a planned time sequence at a project planning time, the planned time sequence being used to represent planned start times and planned end times of tasks in the project;

[0019] obtaining an actual time sequence based on a collection time at which the multi-dimensional data is collected, the actual time sequence being used to represent actual start times of tasks, actual progress times of stages of the tasks, and actual end times;

[0020] mapping the planned time sequence and the actual time sequence corresponding to each task to a unified time axis with project planning as a constraint, and querying the planned time sequence and the actual time sequence corresponding to each task on the unified time axis.

[0021] Preferably, the space rule comprises:

[0022] acquiring a planned space coordinate at a project planning time, the planned space coordinate being used to represent implementation positions of tasks in the project;

[0023] obtaining an actual space coordinate based on a collection coordinate at which the multi-dimensional data is collected, the actual space coordinate being used to represent actual implementation positions of tasks in the project;

[0024] mapping the plan space coordinates and the actual space coordinates corresponding to each task to the unified time axis to obtain a unified space-time axis, on which the plan time sequence, the actual time sequence, the plan space coordinates and the actual space coordinates corresponding to each task are queried.

[0025] Preferably, the multi-dimensional feature space comprises:

[0026] acquiring the unified space-time axis, and performing cross-modal feature extraction on the multi-dimensional data on the unified space-time axis to obtain the multi-dimensional features, wherein the cross-modal feature extraction does not change the modal of the multi-dimensional data;

[0027] mapping the multi-dimensional features to the unified space-time axis to obtain the multi-dimensional feature space.

[0028] Preferably, the project management knowledge graph comprises:

[0029] acquiring the knowledge rules about tasks, resources and risks and their corresponding association relationships at the time of project planning, and constructing a project management initial knowledge graph using the knowledge rules;

[0030] acquiring event trigger conditions at the time of project planning, and updating the project management initial knowledge graph in real time based on the event trigger conditions to obtain the project management knowledge graph; wherein the event trigger conditions are used to represent changes in any one or more of tasks, resources and risks, and the project management initial knowledge graph is updated in real time when there is an event trigger condition; wherein a plurality of tasks constitute a project, resources at least include time resources required to complete the task, and risks at least include an estimated value of time resources caused by failure to complete the task.

[0031] Preferably, the cross-modal association index comprises:

[0032] mapping the multi-dimensional features in the multi-dimensional feature space and the unified space-time axis corresponding thereto to the project management knowledge graph based on the project management knowledge graph to obtain the cross-modal association index, which is used at least for real-time task query in project management.

[0033] Preferably, the project management fusion model comprises:

[0034] extracting the time sequence dependency relationship and the feature importance relationship between corresponding tasks based on the cross-modal association index; wherein the time sequence dependency relationship is used to represent the time sequence association between tasks, and the feature importance relationship is used to represent the importance association of different tasks at the same time;

[0035] Based on the process of extracting the temporal dependency relationship and the feature importance relationship, a preset project management fusion learning model is trained based on the temporal dependency relationship and the feature importance relationship; the training process of the project management fusion learning model includes:

[0036] A1: training the project management fusion learning model to extract the corresponding temporal dependency relationship and the feature importance relationship based on the tasks, resources, and risks in the project and the relationships between them;

[0037] A2: The project management fusion learning model is trained to monitor the cross-modal association index based on the temporal dependency and the feature importance relationship, and generate the delay warning. The training comprises reorganizing the cross-modal association index based on the temporal dependency and the feature importance relationship to generate a predicted cross-modal association index, and generating a delay warning based on the cross-modal association index and the predicted cross-modal association index.

[0038] A3: Output the project management fusion learning model that has completed training and define it as the project management fusion model.

[0039] Preferably, the delay warning includes:

[0040] The project management fusion model extracts the corresponding temporal dependency relationship and the characteristic importance relationship based on the tasks, resources and risks in the project and the relationship between them;

[0041] The project management fusion model reorganizes the cross-modal association index based on the temporal dependency and the feature importance relationship to generate the predicted cross-modal association index, and generates a delay warning based on the cross-modal association index and the predicted cross-modal association index;

[0042] The generation of the predicted cross-modal association index includes: judging whether there are time anomalies in the tasks in the cross-modal association index based on the temporal dependency relationship, where the time anomaly includes at least task delay and its corresponding resource change and risk change; reorganizing the tasks, resources and risks in the cross-modal association index and the association relationships therebetween based on the feature importance relationship and the time anomaly to generate the predicted cross-modal association index;

[0043] Among them, the generation of the delay warning includes: counting the number of multidimensional features that have changed between the cross-modal association index and the predicted cross-modal association index, and generating a delay probability based on the number, and the delay probability is used to characterize the possibility of project delay; obtaining the maximum value of resource change and the corresponding maximum value of risk change based on the predicted cross-modal association index, and generating the delay duration based on the delay probability, the maximum value of resource change and the corresponding maximum value of risk change.

[0044] As preferred, the adjustment strategy comprises:

[0045] Based on the delay warning and a preset progress reward, a reward value is generated, and any one or more of the tasks and resources in the cross-modal association index are adjusted based on the reward value, and the adjustment is defined as the adjustment strategy; wherein the progress reward is used to encourage the generation of the adjustment strategy, and the adjustment of any one or more of the tasks and resources in the cross-modal association index is determined based on a preset adjustment record.

[0046] In a second aspect, the application discloses a project management system based on intelligent analysis of multi-dimensional data, which is suitable for the project management method based on intelligent analysis of multi-dimensional data as described above, and the system comprises:

[0047] A space construction module collects multi-source heterogeneous multi-dimensional data in a project implementation process, maps the multi-dimensional data using space-time rules, and constructs a multi-dimensional feature space; wherein the space-time rules include time rules for unifying a time axis and / or space rules for unifying space encoding, and the multi-dimensional feature space includes multi-dimensional features corresponding to the multi-dimensional data and their corresponding unified space-time axis obtained based on the space-time rules;

[0048] An index construction module reorganizes the multi-dimensional feature space using a project management knowledge graph, and constructs a cross-modal association index for project supervision, wherein the cross-modal association index is a task, a resource, and a risk in a project and the association relationship therebetween; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal association index;

[0049] A delay warning module supervises the cross-modal association index using a project management fusion model, and generates a delay warning for project management; wherein the project management fusion model is used to fuse time features and task features, analyze project duration, and generate the delay warning;

[0050] A strategy generation module generates an adjustment strategy for project management based on the delay warning and the cross-modal association index, and the adjustment strategy is based on the delay warning to adjust the cross-modal association index.

[0051] Beneficial effects: the project management method and system based on multi-dimensional data intelligent analysis of the application, through the system of data acquisition, space-time mapping, knowledge graph reorganization, fusion analysis and closed-loop optimization constructed, realizes the improvement of multi-dimensional project management efficiency: first, through the space-time rule mapping of multi-source heterogeneous data to multi-dimensional feature space, breaking through the traditional data island bottleneck, improving the data integration efficiency, supporting the cross-modal correlation analysis of tasks, resources and risks; second, the dynamic knowledge graph updates the task, resource and risk correlation index based on event triggering, which can identify implicit dependency risks in advance; third, the project management fusion model combines time sequence dependence and feature importance analysis to improve the delay prediction accuracy and realize accurate delay time evaluation; fourth, based on the delay early warning dynamic adjustment of the strategy generation mechanism of the cross-modal correlation index, the resource utilization rate is improved, the management labor cost is reduced, the end-to-end closed loop from data intelligent analysis to decision execution is formed, and the problems of analysis lag and inefficient adjustment in the prior art are effectively solved. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0053] Figure 1 The flowchart of the project management method based on multi-dimensional data intelligent analysis provided by the embodiments of the present application;

[0054] Figure 2 The structural diagram of the project management system based on multi-dimensional data intelligent analysis provided by the embodiments of the present application. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0056] In this paper, the term "including" is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the elements defined by the statement "including" do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0057] The embodiment provides a project management method and system based on multi-dimensional data intelligent analysis. The method comprises the following steps: collecting multi-dimensional data of a project in a whole cycle, the multi-dimensional data comprising at least text, image and sensor data; constructing a multi-dimensional feature space by using space-time rules; reorganizing a cross-modal correlation index in combination with a project management knowledge graph; generating delay warning and driving adjustment strategy by using a project management fusion model; and realizing intelligent management of the project.

[0058] Specifically, the first aspect of the embodiment discloses a project management method based on multi-dimensional data intelligent analysis, as shown in the method comprises the following steps: Figure 1

[0059] S1: collecting multi-dimensional data of a project in a whole cycle, the multi-dimensional data comprising at least text, image and sensor data; constructing a multi-dimensional feature space by using space-time rules; wherein the space-time rules comprise time rules for unifying a time axis and / or space rules for unifying space coding, and the multi-dimensional feature space comprises multi-dimensional features corresponding to the multi-dimensional data and a unified space-time axis obtained based on the space-time rules;

[0060] S2: reorganizing the multi-dimensional feature space by using a project management knowledge graph, and constructing a cross-modal correlation index for project management, the cross-modal correlation index comprising tasks, resources and risks in the project and the correlation relationship therebetween; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal correlation index;

[0061] S3: monitoring the cross-modal correlation index by using a project management fusion model, and generating delay warning for project management; wherein the project management fusion model is used to analyze project duration and generate delay warning after fusing time features and task features;

[0062] S4: generating adjustment strategy for project management based on the delay warning and the cross-modal correlation index, the adjustment strategy being adjustment of the cross-modal correlation index based on the delay warning.

[0063] It should be noted that the present embodiment uses existing API, crawler, IoT device, etc. to obtain project progress data (such as task completion rate), cost data (such as budget consumption), resource data (such as personnel load), external data (such as weather, supply chain delay), etc. A combination of a time series database and a graph database is used to construct a multi-dimensional feature space composed of structured data (such as a task list) and unstructured data (such as a meeting minutes text, an image log).

[0064] ​Through the above, the data islands are broken through by the spatiotemporal rule mapping, and the integration efficiency is improved; the dynamic knowledge graph is recombined to index the cross-modal association, and the implicit risks are identified in advance; the project management fusion model combines time and task characteristics to accurately predict delays; the resource utilization rate is improved based on the early warning dynamic adjustment strategy; a closed loop of data, analysis and decision is formed to solve the problems of analysis lag and inefficient adjustment.

[0065] For project management, a plan about the completion time of a task is generally set in the planning stage. In this embodiment, the time rule maps the time sequence of the project plan and the actual execution to a unified time axis as a constraint to realize the spatiotemporal alignment query of the plan and the actual progress.

[0066] Specifically, the time rule includes:

[0067] acquiring a planned time sequence at the time of the project plan, the planned time sequence being used to represent the planned start time and the planned end time of each task in the project;

[0068] obtaining an actual time sequence based on the acquisition time when the multi-dimensional data is acquired, the actual time sequence being used to represent the actual start time of each task, the actual progress time of each stage progress of the task, and the actual end time;

[0069] mapping the planned time sequence and the actual time sequence of each task to a unified time axis as a constraint, and performing a query of the planned time sequence and the actual time sequence of each task on the unified time axis.

[0070] In a simple example, the planned start / end time of a development task corresponds to 2023.6.1-2024.5.31, the actual start time is combined with the acquisition time of the actual construction log, which is 2025.6.3, and the task completion time of each stage is mapped to a unified time axis to query the progress deviation in real time.

[0071] Through the above, the time rule maps the planned and actual time sequences to a unified time axis to solve the problem of time data fragmentation in traditional management, improve the progress deviation query efficiency, support accurate positioning in the time dimension, and provide accurate data support for delay warning.

[0072] In the project implementation process, the implementation of a task is generally located at a fixed position, and the change of the position will also cause project abnormalities. The space rule of this embodiment maps the spatial coordinates of the project plan and the actual execution to a unified time axis as a constraint to form a unified time-space axis and realize multi-dimensional data linkage query in the time and space dimensions.

[0073] Specifically, the space rule includes:

[0074] acquire a plan space coordinate at a project plan time, the plan space coordinate being used to represent an implementation position of each task in the project;

[0075] acquire an actual space coordinate based on an acquisition coordinate at which the multi-dimensional data is acquired, the actual space coordinate being used to represent an actual implementation position of each task in the project;

[0076] map the plan space coordinate and the actual space coordinate corresponding to each task to a unified time axis to obtain a unified space-time axis, and perform query on the plan time sequence, the actual time sequence, the plan space coordinate and the actual space coordinate corresponding to each task on the unified space-time axis.

[0077] In a simple example, a plan space coordinate of construction is acquired, which can be a GPS coordinate marked on a design drawing, and an actual construction coordinate acquired by an IoT device on site, such as GPS data of a frequency monitoring camera, is mapped to a unified time axis to query a time deviation and / or a position deviation in real time and a space-time influence on a subsequent task.

[0078] According to the above, the space rule integrates the plan space coordinate and the actual space coordinate through the unified space-time axis, solves the problem that space data and time progress are disconnected in traditional engineering management, can discover a correlation risk of position and time in real time, and provides a precise decision basis in a space-time dimension for project management.

[0079] On the basis of the unified space-time axis, multi-dimensional feature space is constructed by performing cross-modal feature extraction on multi-source heterogeneous multi-dimensional data on the unified space-time axis, mapping the extracted multi-dimensional features to the unified space-time axis on the premise of preserving the original modal of the data, and forming the multi-dimensional feature space.

[0080] Specifically, the multi-dimensional feature space includes:

[0081] acquire a unified space-time axis, perform cross-modal feature extraction on multi-dimensional data on the unified space-time axis to obtain multi-dimensional features, and map the multi-dimensional features to the unified space-time axis to obtain the multi-dimensional feature space.

[0082] acquire a unified space-time axis, perform cross-modal feature extraction on multi-dimensional data on the unified space-time axis to obtain multi-dimensional features, and map the multi-dimensional features to the unified space-time axis to obtain the multi-dimensional feature space.

[0083] It should be noted that the embodiment utilizes existing feature extraction to perform cross-modal feature extraction on multi-dimensional data. For example, entity recognition is performed on unstructured text data (such as emails and meeting records) to extract risk events, responsible persons and time nodes, abnormal detection is performed on device sensor data (such as using an isolation forest algorithm to identify abnormal work hour records), and existing natural language technology is used to extract keywords from text data.

[0084] In a simple example, multi-modal data of construction, such as construction log text, detection image, stress sensor time series data, are extracted across modalities on a unified space-time axis, such as mapping keywords in text, distribution features in image, stress fluctuation values of sensors to a unified space-time axis to generate a multi-dimensional feature space.

[0085] Through the above, the construction of the multi-dimensional feature space preserves the original semantic information of the multi-source heterogeneous multi-dimensional data through cross-modal feature extraction without changing the data modalities, avoids the loss of modal information and the complex process and time consumption of data conversion in traditional feature extraction, improves the feature integrity and the data retrieval efficiency, and realizes the space-time alignment of text, image and time series data by mapping to a unified space-time axis, provides multi-dimensional feature support for subsequent knowledge graph reconstruction and delay prediction, and solves the one-sided analysis problem caused by the fragmentation of data modalities.

[0086] Based on the project management method based on intelligent analysis of multi-dimensional data, the project management knowledge graph constructs an initial graph by acquiring task, resource and risk association knowledge rules in the project planning stage, and updates in real time based on task delay, resource overrun and other event trigger conditions to form a dynamic knowledge graph.

[0087] Specifically, the project management knowledge graph comprises:

[0088] Acquiring knowledge rules about tasks, resources and risks and their corresponding association relationships at the time of project planning, and constructing an initial project management knowledge graph using the knowledge rules;

[0089] Acquiring event trigger conditions at the time of project planning, and updating the initial project management knowledge graph in real time based on the event trigger conditions to obtain the project management knowledge graph; wherein the event trigger condition is used to represent the change of any one or more of the tasks, resources and risks, and when the event trigger condition exists, the initial project management knowledge graph is updated in real time; wherein a plurality of tasks constitute a project, the resources at least include time resources required to complete the task, and the risks at least include an estimated value of time resources caused by failure to complete the task.

[0090] It should be noted that the present embodiment utilizes existing knowledge graph technology to implement the construction of the project management knowledge graph.

[0091] In a simple example, construction tasks, resource requirements and construction delay risks and other knowledge rules are acquired according to the design plan to construct an initial graph; the trigger condition of "equipment supply delay for more than 24 hours" is set, and when the event is detected in actual construction, the construction tasks and the dependency relationship between the tasks in the graph are updated in real time.

[0092] By the foregoing, the project management knowledge graph constructs an initial graph through knowledge rules and updates in real time based on events, solves the problem that a traditional static graph cannot dynamically reflect project changes, enhances the implicit dependence of tasks, resources, and risks, supports dynamic correlation analysis of, for example, task delays caused by resource shortages and risk transmission, provides real-time knowledge support for delay warning, and improves the timeliness and accuracy of project management decisions.

[0093] Based on the constructed project management knowledge graph, the cross-modal correlation index maps multi-dimensional features in a multi-dimensional feature space and a unified space-time axis thereof to the knowledge graph, constructs cross-modal correlation relationships of tasks, resources, and risks, and realizes real-time task query in project management.

[0094] Specifically, the cross-modal correlation index comprises:

[0095] Based on the project management knowledge graph, multi-dimensional features in a multi-dimensional feature space and corresponding unified space-time axes thereof are mapped to the project management knowledge graph to obtain a cross-modal correlation index, which is used at least for real-time task query in project management.

[0096] In a simple example, multi-dimensional features of construction, such as equipment distribution features in a monitoring image, a “delayed equipment maintenance” keyword in a log text, stress data of a sensor, and a space-time axis thereof (2023.9.10, coordinate X), are mapped to the knowledge graph to form a cross-modal correlation index of delayed equipment maintenance and construction equipment resource conflict, which supports real-time query of task abnormalities.

[0097] By the foregoing, the cross-modal correlation index breaks through the limitation of data modality fragmentation in traditional management by mapping a multi-dimensional feature space and a knowledge graph, realizes cross-modal correlation of text, image, and time series data, improves task abnormality query efficiency, supports multi-dimensional correlation analysis of spatial position deviation, time progress delay, and resource allocation conflict, provides structured index support for real-time project supervision, and solves the problem of low cross-modal data correlation efficiency.

[0098] Based on the project management method based on multi-dimensional data intelligent analysis, the project management fusion model extracts task time sequence dependency relationships and feature importance relationships from the cross-modal correlation index, and then trains a fusion learning model to realize supervision of the cross-modal correlation index and generation of delay warning.

[0099] Specifically, the project management fusion model comprises:

[0100] Based on the cross-modal correlation index, the corresponding time sequence dependency relationships and feature importance relationships between tasks are extracted; wherein the time sequence dependency relationships are used to represent the time sequence correlation between tasks, and the feature importance relationships are used to represent the importance correlation of different tasks at the same time;

[0101] The project management fusion learning model is trained based on the extracted temporal dependency relationship and feature importance relationship, and the temporal dependency relationship and feature importance relationship. The training process of the project management fusion learning model includes:

[0102] A1: Training the project management fusion learning model to extract the corresponding temporal dependency relationship and feature importance relationship based on the tasks, resources and risks in the project and the association relationship therebetween;

[0103] A2: Training the project management fusion learning model to supervise the cross-modal association index based on the temporal dependency relationship and the feature importance relationship, and generating delay warning. The training is to reorganize the cross-modal association index based on the temporal dependency relationship and the feature importance relationship, generate a predicted cross-modal association index, and generate a delay warning based on the cross-modal association index and the predicted cross-modal association index.

[0104] A3: Outputting the trained project management fusion learning model and defining it as a project management fusion model.

[0105] In a simple example, the temporal dependency relationship of "task A→task B" is extracted from the cross-modal association index, and the feature importance relationship of "condition C affecting task A"; the model is trained based on historical construction data, so that it can reorganize the index and predict the delay risk of "condition C causing task A delay, further causing resource conflict of task B" based on such relationship.

[0106] By the above, the project management fusion model solves the problem of large prediction deviation of traditional single model by extracting and training the temporal dependency and feature importance relationship, and improves the accuracy of delay prediction. Through dynamic reorganization of the cross-modal association index to generate an early warning, a closed loop is realized from data association analysis, risk prediction, and early warning generation, providing accurate model support for project delay control.

[0107] Based on the project management fusion model, the delay warning extracts the task temporal dependency and feature importance relationship through the project management fusion model, reorganizes the cross-modal association index to generate a predicted index, and generates a delay probability and time length based on the difference between the actual and predicted indexes. It should be noted that in practical applications, project delays follow the short board effect, that is, the longest task has the greatest impact on project delay time. Based on this, the short board effect is applied to determine whether there is a time anomaly in the task in the cross-modal association index based on the temporal dependency relationship, thereby improving the accuracy of delay time prediction, rather than relying on a fixed feature for prediction, and realizing the self-adaptation of delay time prediction to key features.

[0108] Specifically, the delay warning includes:

[0109] The project management fusion model extracts corresponding time sequence dependency relationships and characteristic important relationships based on tasks, resources and risks in the project and the correlations therebetween;

[0110] The project management fusion model reorganizes a cross-modal correlation index based on the time sequence dependency relationships and the characteristic important relationships, generates a predicted cross-modal correlation index, and generates a delay warning based on the cross-modal correlation index and the predicted cross-modal correlation index;

[0111] The generation of the predicted cross-modal correlation index comprises: determining whether there is a time anomaly in the tasks in the cross-modal correlation index based on the time sequence dependency relationships, the time anomaly at least including a task delay and corresponding resource changes and risk changes; reorganizing the correlations between the tasks, the resources and the risks in the cross-modal correlation index based on the characteristic important relationships and the time anomaly, and generating the predicted cross-modal correlation index;

[0112] The generation of the delay warning comprises: counting the number of multi-dimensional features that have changed between the cross-modal correlation index and the predicted cross-modal correlation index, generating a delay probability based on the number, the delay probability being used to represent the possibility of the delay of the project; obtaining a maximum value of the resource changes and a maximum value of the corresponding risk changes based on the predicted cross-modal correlation index, and generating a delay duration based on the delay probability, the maximum value of the resource changes and the maximum value of the corresponding risk changes.

[0113] In this embodiment, the project management fusion model is fused based on the existing LSTM (time sequence prediction) and XGBoost (feature importance analysis), the time sequence gradient of the LSTM is back propagated to the feature selection layer of the XGBoost, and the prediction weights of the LSTM (long-term trend) and the XGBoost (short-term feature) are dynamically adjusted according to the task stage. Further, the delay probability is a function of the number of multi-dimensional features that have changed between the cross-modal correlation index and the predicted cross-modal correlation index, and the determination method of the function can be but is not limited to feedback and fitting determination through historical delay probability. Based on this, the delay duration in this embodiment is determined by summing the maximum value of the resource changes and the maximum value of the corrected risk changes, and the correction is to multiply the original risk by the reciprocal of the delay probability.

[0114] In a simple example:

[0115] The number of multi-dimensional feature changes is 3, the initial delay probability is determined to be 70% based on the fitting function of the historical delay data; the maximum value of the resource changes is 2 days of delay (i.e. the number of days of the corresponding resource changes in all the changed multi-dimensional features is 2 days), and the quantitative impact of the corresponding resource; the maximum value of the original risk changes is 2 days of rain season construction delay risk, which is corrected to about 2.8 days through the reciprocal of the delay probability 70%. The sum is determined to be 4.8 days, and the delay duration is obtained by rounding up to 5 days.

[0116] By the above, the delay warning realizes quantitative evaluation of delay probability and time length by fusing time-dependent, feature important relationship reorganization index, solves the hysteresis problem of traditional warning relying on artificial experience, improves delay prediction accuracy, and can dynamically associate resource and risk changes to provide accurate digital warning support for project progress control.

[0117] On the basis of the project management method based on multi-dimensional data intelligent analysis, the adjustment strategy generates a reward value by combining the delay warning and the preset progress reward, and adjusts the tasks and resources in the cross-modal association index based on the reward value. The adjustment is determined based on a preset adjustment record.

[0118] Specifically, the adjustment strategy includes:

[0119] Based on the delay warning and the preset progress reward, a reward value is generated, and any one or more of the tasks and resources in the cross-modal association index is adjusted based on the reward value. The adjustment is defined as an adjustment strategy. The progress reward is used to encourage the generation of the adjustment strategy, and the adjustment of any one or more of the tasks and resources in the cross-modal association index is determined based on a preset adjustment record.

[0120] In this embodiment, the progress reward is based on an existing hierarchical reward function, which decomposes the resource optimization target into sub-targets such as "progress guarantee", "cost control" and "risk avoidance", and generates a Pareto optimal strategy through Monte Carlo tree search. Specifically, the cross-modal association index is adjusted using a defined composite reward value, which is obtained by weighting and summing the preset progress reward, cost penalty and risk penalty, wherein the weight values are dynamically adjusted based on the cross-modal association index. Further, based on the knowledge graph, invalid actions are excluded to adjust any one or more of the tasks and resources in the cross-modal association index.

[0121] In a simple example, after receiving the warning "Task B is delayed for 5 days", the system generates a reward value according to the preset progress reward rule, triggers the strategy "transfer 20 workers from non-critical tasks to Task B" in the adjustment record, and adjusts the resource allocation relationship in the cross-modal association index.

[0122] By the above, the adjustment strategy realizes dynamic optimization of project resources and tasks by fusing the delay warning and the progress reward generation mechanism, solves the problem of manual adjustment hysteresis in traditional management, improves resource utilization, improves delay compensation efficiency, and ensures the feasibility of the strategy through the preset adjustment record, forming a closed-loop management from warning to execution.

[0123] The second aspect of the present embodiment discloses a method for Figure 2An embodiment of the application provides a project management system based on intelligent analysis of multi-dimensional data, which is applicable to the project management method based on intelligent analysis of multi-dimensional data as described above, and comprises:

[0124] a space construction module configured to collect multi-source and multi-dimensional data in a project implementation process, map the multi-dimensional data by using space-time rules, and construct a multi-dimensional feature space, wherein the space-time rules comprise time rules for unifying a time axis and / or space rules for unifying space encoding, and the multi-dimensional feature space comprises multi-dimensional features corresponding to the multi-dimensional data and corresponding unified space-time axes obtained based on the space-time rules;

[0125] an index construction module configured to reorganize the multi-dimensional feature space by using a project management knowledge graph, and construct a cross-modal correlation index for project management, the cross-modal correlation index being a task, a resource, a risk and a correlation relationship therebetween in a project; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal correlation index;

[0126] a delay warning module configured to supervise the cross-modal correlation index by using a project management fusion model, and generate a delay warning for project management; wherein the project management fusion model is used to fuse time features and task features, analyze a project duration, and generate the delay warning;

[0127] a strategy generation module configured to generate an adjustment strategy for project management based on the delay warning and the cross-modal correlation index, the adjustment strategy being based on the delay warning to adjust the cross-modal correlation index.

[0128] It should be noted that the project management system based on intelligent analysis of multi-dimensional data in the embodiment corresponds to the project management method based on intelligent analysis of multi-dimensional data described above. Therefore, the contents not specifically described in the project management system based on intelligent analysis of multi-dimensional data in the embodiment can be, but are not limited to, functional definitions, working principles and technical effects, and can all be referred to the descriptions in the project management method based on intelligent analysis of multi-dimensional data described above. Therefore, the contents will not be described here again.

[0129] In summary, the project management method and system based on multi-dimensional data intelligent analysis of this embodiment achieves multi-dimensional project management efficiency improvement through the constructed system of data collection, spatiotemporal mapping, knowledge graph reorganization, fusion analysis and closed-loop optimization: First, by mapping multi-source heterogeneous data to multi-dimensional feature space through spatiotemporal rules, the bottleneck of traditional data islands is broken through, the efficiency of data integration is improved, and cross-modal correlation analysis of tasks, resources and risks is supported; second, the dynamic knowledge graph updates the task, resource and risk correlation index in real time based on event triggering, which can identify implicit dependency risks in advance; third, the project management fusion model combines temporal dependency and feature importance analysis to improve the accuracy of delay prediction and realize accurate delay duration assessment; fourth, based on the delay warning, the strategy generation mechanism of the cross-modal correlation index is dynamically adjusted to improve resource utilization, reduce management labor costs, and form an end-to-end closed loop from data intelligent analysis to decision execution, effectively solving the problems of analysis lag and adjustment inefficiency in the existing technology.

[0130] In the embodiments provided herein, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a processor, a controller, a microcontroller, a microprocessor, other electronic units designed to implement the functions described herein, or a combination thereof. For software implementation, part or all of the processes of the embodiments can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein the communication media include any medium that facilitates the transmission of a computer program from one place to another. The storage medium can be any available medium that a computer can access. The computer-readable storage medium can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0131] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, modifications or equivalent replacements of the technical solutions described in the foregoing embodiments can still be made by those skilled in the art, or some technical features can be replaced by equivalent features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A project management method based on multidimensional data intelligent analysis, characterized in that: The method includes: S1: Collect multi-source heterogeneous multidimensional data during project implementation, map the multidimensional data using spatiotemporal rules, and construct a multidimensional feature space; wherein the spatiotemporal rules include time rules for unifying the time axis and / or space rules for unifying spatial encoding, and the multidimensional feature space includes multidimensional features corresponding to the multidimensional data and their corresponding unified spatiotemporal axis obtained based on the spatiotemporal rules; S2: Reorganizing the multidimensional feature space using a project management knowledge graph to construct a cross-modal association index for project supervision, wherein the cross-modal association index represents tasks, resources, and risks in a project and the relationships therebetween; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal association index; S3: Using a project management fusion model to monitor the cross-modal association index and generate a delay warning for project management; wherein the project management fusion model is used to fuse time features and task features to analyze the project duration and generate the delay warning; S4: Generate an adjustment strategy for project management based on the delay warning and the cross-modal association index, where the adjustment strategy is to adjust the cross-modal association index based on the delay warning.

2. The project management method based on multidimensional data intelligent analysis according to claim 1, characterized in that: The time rules include: Obtain the planned time series when planning the project. This time series is used to represent the planned start and end time of each task in the project. Based on the collection time when the multidimensional data is collected, an actual time series is obtained, where the actual time series is used to represent the actual start time of each task in the project, the actual progress time of each stage of the task, and the actual end time; The planned time series and the actual time series corresponding to each task are mapped onto a unified time axis with the project plan as a constraint, and the planned time series and the actual time series corresponding to each task are queried on the unified time axis.

3. The project management method based on multidimensional data intelligent analysis according to claim 2, characterized in that: The space rules include: Obtain the planning space coordinates during project planning. The planning space coordinates are used to represent the implementation locations of each task in the project. Based on the acquisition coordinates when the multidimensional data is acquired, actual space coordinates are obtained, where the actual space coordinates are used to represent the actual implementation location of each task in the project; The planned space coordinates and the actual space coordinates corresponding to each task are mapped to the unified time axis with the project plan as a constraint to obtain the unified space-time axis, and the planned time series, the actual time series, the planned space coordinates and the actual space coordinates corresponding to each task are queried on the unified space-time axis.

4. The project management method based on multidimensional data intelligent analysis according to claim 3 is characterized in that: The multidimensional feature space includes: Obtaining the unified spatiotemporal axis, performing cross-modal feature extraction on the multidimensional data on the unified spatiotemporal axis to obtain the multidimensional features; wherein the cross-modal feature extraction does not change the modality of the multidimensional data; The multidimensional features are mapped to the unified space-time axis to obtain the multidimensional feature space.

5. The project management method based on multidimensional data intelligent analysis according to claim 1, characterized in that: The project management knowledge graph includes: Obtaining the knowledge rules regarding tasks, resources, and risks and their corresponding relationships during project planning, and using the knowledge rules to construct an initial knowledge graph for project management; Acquire event trigger conditions during project planning, and update the project management initial knowledge graph in real time based on the event trigger conditions to obtain the project management knowledge graph; wherein, the event trigger conditions are used to characterize changes in any one or more of tasks, resources, and risks, and when event trigger conditions exist, update the project management initial knowledge graph in real time; wherein, multiple tasks constitute a project, resources include at least the time resources required to complete the task, and risks include at least the estimated value of time resources caused by failure to complete the task.

6. The project management method based on multidimensional data intelligent analysis according to claim 5, characterized in that: The cross-modal association index includes: Based on the project management knowledge graph, the multidimensional features in the multidimensional feature space and their corresponding unified space-time axis are mapped to the project management knowledge graph to obtain the cross-modal association index, which is at least used for real-time task queries in project management.

7. The project management method based on multidimensional data intelligent analysis according to claim 1, characterized in that: The project management integration model includes: Based on the cross-modal association index, the temporal dependency and feature importance relationship between the corresponding tasks are extracted; wherein the temporal dependency is used to characterize the temporal association between tasks, and the feature importance relationship is used to characterize the association between the importance of different tasks at the same time; Based on the process of extracting the temporal dependency relationship and the feature importance relationship, a preset project management fusion learning model is trained based on the temporal dependency relationship and the feature importance relationship; the training process of the project management fusion learning model includes: A1: training the project management fusion learning model to extract the corresponding temporal dependency relationship and the feature importance relationship based on the tasks, resources, and risks in the project and the relationships between them; A2: The project management fusion learning model is trained to monitor the cross-modal association index based on the temporal dependency and the feature importance relationship, and generate the delay warning. The training comprises reorganizing the cross-modal association index based on the temporal dependency and the feature importance relationship to generate a predicted cross-modal association index, and generating a delay warning based on the cross-modal association index and the predicted cross-modal association index. A3: Output the project management fusion learning model that has completed training and define it as the project management fusion model.

8. The project management method based on multidimensional data intelligent analysis according to claim 7, characterized in that: The delay warning includes: The project management fusion model extracts the corresponding temporal dependency relationship and the characteristic importance relationship based on the tasks, resources and risks in the project and the relationship between them; The project management fusion model reorganizes the cross-modal association index based on the temporal dependency and the feature importance relationship to generate the predicted cross-modal association index, and generates a delay warning based on the cross-modal association index and the predicted cross-modal association index; The generation of the predicted cross-modal association index includes: judging whether there are time anomalies in the tasks in the cross-modal association index based on the temporal dependency relationship, where the time anomaly includes at least task delay and its corresponding resource change and risk change; reorganizing the tasks, resources and risks in the cross-modal association index and the association relationships therebetween based on the feature importance relationship and the time anomaly to generate the predicted cross-modal association index; Among them, the generation of the delay warning includes: counting the number of multidimensional features that have changed between the cross-modal association index and the predicted cross-modal association index, and generating a delay probability based on the number, and the delay probability is used to characterize the possibility of project delay; obtaining the maximum value of resource change and the corresponding maximum value of risk change based on the predicted cross-modal association index, and generating the delay duration based on the delay probability, the maximum value of resource change and the corresponding maximum value of risk change.

9. The project management method based on multidimensional data intelligent analysis according to claim 1, characterized in that: The adjustment strategy includes: Based on the delay warning and the preset progress reward, a reward value is generated, and based on the reward value, any one or more of the tasks and resources in the cross-modal association index are adjusted, and the adjustment is defined as the adjustment strategy; wherein the progress reward is used to incentivize the generation of the adjustment strategy, and the adjustment of any one or more of the tasks and resources in the cross-modal association index is determined based on the preset adjustment record.

10. A project management system based on multidimensional data intelligent analysis, the system being applicable to the project management method based on multidimensional data intelligent analysis according to any one of claims 1 to 9, characterized in that: The system includes: A spatial construction module collects multi-source heterogeneous multidimensional data from the project implementation process, maps the multidimensional data using spatiotemporal rules, and constructs a multidimensional feature space; wherein the spatiotemporal rules include time rules for unifying the time axis and / or space rules for unifying spatial encoding, and the multidimensional feature space includes multidimensional features corresponding to the multidimensional data and their corresponding unified spatiotemporal axis obtained based on the spatiotemporal rules; An index construction module, which uses a project management knowledge graph to reorganize the multidimensional feature space and construct a cross-modal association index for project supervision, wherein the cross-modal association index is tasks, resources, and risks in the project and the relationships between them; wherein the project management knowledge graph stores knowledge rules for constructing the cross-modal association index; A delay warning module, which uses a project management fusion model to monitor the cross-modal association index and generate delay warnings for project management; wherein the project management fusion model is used to analyze the project duration and generate the delay warning after fusing time features and task features; A strategy generation module generates an adjustment strategy for project management based on the delay warning and the cross-modal association index, where the adjustment strategy adjusts the cross-modal association index based on the delay warning.

Images