Software development project progress prediction management method based on artificial intelligence

Abstract

The invention discloses a software development project progress prediction management method based on artificial intelligence, and relates to the technical field of project management, and the method comprises the steps: extracting a causal relation from entity relation graph data, and constructing a causal knowledge graph; a graph structure analysis method and a time sequence prediction method are combined, predicted completion time and delay probability are calculated for the causal knowledge graph and the associated time sequence data, and a progress risk assessment result is generated; performing causal chain identification on the progress risk assessment result and the causal knowledge graph by adopting a causal reasoning method to generate delay interpretation information; according to the delay interpretation information, a preset resource priority and a task weight, a rule driving and priority scheduling strategy is adopted to construct a compensation scheduling scheme; and implementing a compensation scheduling scheme, collecting implementation effect data and analyzing a scheduling effect through a closed-loop feedback and dynamic adjustment mechanism, and generating an optimized scheduling management scheme. The intelligent level and the practical value of project progress management are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of project management, and particularly relates to a software development project progress prediction management method based on artificial intelligence. BACKGROUND

[0002] In the current software development life cycle management, accurate prediction and effective control of project progress are of great significance for guaranteeing product delivery on schedule, controlling development cost and optimizing resource allocation. The existing mainstream methods generally use traditional project management tools (such as Gantt chart, WBS structure and PERT chart) to plan and track development tasks, and assist with statistical analysis means to evaluate project execution status. In the aspect of auxiliary tools, there are also some automatic scheduling software based on rule engine, which matches the task priority and resource allocation situation through built-in rules to realize semi-automatic scheduling and adjustment of tasks.

[0003] However, there are obvious limitations in dealing with complex and variable software development environment. Especially in the face of dynamic factors such as multi-project parallel, frequent personnel flow and complex task interdependence, the traditional method is often difficult to identify potential progress delay risks in time due to insufficient data driving degree and inability to automatically perceive and respond to actual progress. Secondly, the current scheduling method lacks in-depth modeling and analysis of the causal relationship between tasks, and it is difficult to realize systematic explanation and scheduling optimization of the causes of progress delay. This leads to that when the project delay is found, the project manager cannot accurately judge the key influencing factors behind the delay, and also lacks a compensation adjustment mechanism driven by global causal reasoning results, which easily causes blindness and inefficiency of scheduling adjustment. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a software development project progress prediction management method based on artificial intelligence to solve the problem that the prior art cannot accurately identify project progress risks and causal relationships.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a software development project progress prediction management method based on artificial intelligence, which comprises: collecting multi-source heterogeneous industrial data and preprocessing to form entity relationship graph data; extracting causal relationships from the entity relationship graph data to construct a causal knowledge graph; combining a graph structure analysis method with a time series prediction method to calculate the expected completion time and delay probability of the causal knowledge graph and associated time series data, and generate progress risk assessment results; using causal reasoning methods to identify causal chains based on the progress risk assessment results and the causal knowledge graph, and generate delay explanation information; using rule-driven and priority scheduling strategies to construct a compensation scheduling scheme based on the delay explanation information and pre-set resource priorities and task weights; implementing the compensation scheduling scheme, collecting implementation effect data and analyzing scheduling effects through a closed-loop feedback and dynamic adjustment mechanism, and generating an optimized scheduling management scheme.

[0008] As a preferred scheme of the software development project progress prediction management method based on artificial intelligence, the entity relationship graph data is formed by the following specific steps,

[0009] Task progress, code submission records, developer attendance records, and task execution logs are collected to obtain multi-source heterogeneous industrial data, which is subjected to missing value filling and format unification to generate structured industrial data;

[0010] Based on the structured industrial data, entities are extracted and logical relationships and temporal dependencies between entities are identified to generate an entity set and an entity relationship set;

[0011] The entity set and the entity relationship set are structured and organized to form the entity relationship graph data.

[0012] As a preferred scheme of the software development project progress prediction management method based on artificial intelligence, the causal knowledge graph is constructed by the following specific steps,

[0013] Causal determination analysis is performed on the entity relationship graph data to extract causal relationships and generate a causal relationship candidate set;

[0014] The causal relationship candidate set is screened and confirmed to form an effective causal relationship set;

[0015] The effective causal relationship set and the entity relationship graph data are structured and fused to construct the causal knowledge graph.

[0016] As a preferred scheme of the software development project progress prediction management method based on artificial intelligence, the progress risk assessment results are generated by the following specific steps,

[0017] A graph structure analysis method is used to analyze the structure of the cause-effect knowledge graph, identify the nodes representing the tasks and the dependent cause-effect path information, and generate a task dependency structure;

[0018] Task execution events with timestamps are extracted from structured industrial data, and the task execution events are classified into each task node in the task dependency structure to generate associated time series data;

[0019] The task dependency structure and the associated time series data are combined, and a time series prediction method is used for calculation to output a predicted completion time series;

[0020] The predicted completion time series is modeled, the delay probability is calculated, and a task delay probability distribution is generated;

[0021] Based on the task dependency structure and the task delay probability distribution, the overall project progress risk is comprehensively evaluated, and a progress risk assessment result is generated.

[0022] As a preferred scheme of the software development project progress prediction management method based on artificial intelligence provided by the application, the specific steps of generating delay explanation information are as follows,

[0023] According to the progress risk assessment result, the task nodes in the cause-effect knowledge graph are matched and screened, and the task nodes with a delay probability higher than a set delay probability determination threshold are identified to form a set of tasks to be analyzed;

[0024] The cause-effect path information dependent on the set of tasks to be analyzed is extracted from the cause-effect knowledge graph to generate a set of candidate cause-effect chains;

[0025] The set of candidate cause-effect chains is analyzed and screened by a cause-effect reasoning method to form a set of task delay cause-effect paths, and upstream factors triggering the delay are extracted to generate delay explanation information.

[0026] As a preferred scheme of the software development project progress prediction management method based on artificial intelligence provided by the application, the specific steps of constructing a compensation scheduling scheme are as follows,

[0027] Resource factors and constraint conditions affecting progress are extracted from the delay explanation information to generate a set of resources to be scheduled;

[0028] A scheduling priority list is constructed according to the set of resources to be scheduled and the preset resource priority and task weight;

[0029] According to the scheduling priority list, the developers, computing resources, and test environment resources required by the tasks are matched to generate an initial resource configuration scheduling scheme;

[0030] The initial resource configuration and scheduling scheme is subjected to constraint verification and resource conflict resolution to generate a compensation scheduling scheme.

[0031] As a preferred embodiment of the AI-based software development project schedule prediction and management method of the present invention, the preset resource priorities and task weights are defined in the following steps.

[0032] Collect historical project execution records and task completion status to obtain information on resource usage frequency, task completion time, and causal path.

[0033] Based on the frequency of resource usage and the timeliness of task completion, assess the importance of resources and generate an initial resource priority score;

[0034] Based on causal path information, assess the impact of tasks on the overall progress and generate task weight scores.

[0035] The initial resource priority score and task weight score are normalized to form resource priority and task weight.

[0036] As a preferred embodiment of the AI-based software development project schedule prediction and management method of the present invention, the specific steps for generating the optimized scheduling management scheme are as follows:

[0037] The compensation scheduling scheme is optimized to generate a preliminary scheduling execution plan;

[0038] Through a closed-loop feedback mechanism, feedback information is collected during the initial scheduling and execution process, and statistical analysis and performance evaluation are performed to generate scheduling effect evaluation results.

[0039] Based on the evaluation results of the scheduling effect, a dynamic adjustment mechanism is adopted to optimize the compensation scheduling scheme and generate an optimized scheduling management scheme.

[0040] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the artificial intelligence-based software development project schedule prediction and management method as described in the first aspect of the present invention.

[0041] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the artificial intelligence-based software development project schedule prediction and management method as described in the first aspect of the present invention.

[0042] The beneficial effects of this invention are as follows: By introducing causal reasoning into project schedule management, it breaks through the limitations of traditional methods based on statistical regression or simple time series forecasting, and realizes the in-depth revelation and utilization of causal relationships between project tasks. Combined with time series data, it dynamically reflects the actual execution of tasks, making risk assessment more timely and accurate. It can not only accurately identify potential delay risks, but also quantify the probability of delays, providing scientific risk warnings and decision support for project management, and greatly improving the intelligence level and practical value of project schedule management. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Fig. 1 This is a flowchart of the AI-based software development project schedule prediction and management method of the present invention.

[0045] Fig. 2 This is a flowchart of the causal knowledge graph construction in this invention.

[0046] Fig. 3 This is a flowchart illustrating the generation of delay explanation information in this invention.

[0047] Fig. 4 This is a flowchart illustrating the generation of the optimized scheduling management scheme in this invention. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0051] Reference Figs. 1-4This is one embodiment of the present invention, which provides a software development project schedule prediction and management method based on artificial intelligence, including the following steps:

[0052] S1. Collect multi-source heterogeneous industrial data and preprocess it to form entity relationship graph data.

[0053] S1.1 Collect task progress, code submission records, developer attendance records and task execution logs to obtain multi-source heterogeneous industrial data, and fill in missing values ​​and unify the format to generate structured industrial data.

[0054] Specifically, task progress data, including fields such as task status updates and task start and end times, is extracted from the task management platform via data interfaces; code change content, commit time, and submitter information for each commit are extracted from the version control platform as code commit records; daily check-in and check-out times, leave records, and project participation information are exported from the personnel attendance platform as developer attendance records; and information including task identifier, execution time, executor, and related execution status is exported from the project execution platform as task execution logs. Structured data tables are created for the task progress data, code commit records, developer attendance records, and task execution logs, and missing values ​​are filled for any missing fields. For example, forward filling is used to fill in missing attendance records within consecutive time periods, or the default average task duration is used to fill in unregistered task completion times. Subsequently, the time format in the data tables is standardized to "YYYY-MM-DD HH:MM:SS", the task identifier field format is standardized, and the field names, data types, and timestamp formats are standardized to generate structured industrial data.

[0055] S1.2 Based on structured industrial data, extract entities and identify the logical relationships and temporal dependencies between entities to generate entity sets and entity relationship sets.

[0056] Specifically, field names, values, and timestamps are extracted from structured industrial data. Based on field semantics, core entities containing information such as task identifier, task name, task status, executor, resource number, submission time, and execution start and end times are identified. Fields of different semantic categories are categorized into entity types such as tasks, personnel, resources, and events. The "allocation" relationship between tasks and personnel is identified based on the matching relationship between the task identifier and executor fields; the "occupation" relationship between tasks and resources is identified based on the task start and end times and resource number fields; and the "triggering" sequence relationship between submission actions and task status updates is identified based on the chronological order of code submission records and task status change times. For multiple status nodes within the same task, "precedence-follower" relationships are identified based on the order of status update times. The identified entities, logical relationships, and temporal dependencies are organized into entity sets and entity relationship sets.

[0057] S1.3 Organize the entity set and entity relationship set in a structured manner to form entity relationship graph data.

[0058] Specifically, each entity in the entity set is assigned a unique identifier, and the node type is labeled according to the semantic type of the entity (e.g., task, person, resource, and event). Next, for each relationship in the entity relationship set, a directed edge is established based on the two entity identifiers that connect them, and the relationship type of the edge is labeled (e.g., "allocation", "occupation", "trigger", and "precedence-follower"). A graph data structure is used to construct the correspondence between entity nodes and edges. All entity nodes and their corresponding relationship edges are written into the graph data storage structure according to the data format of the graph structure. An example can be represented using an adjacency list or a graph database format. Finally, the time information of each relationship is appended in timestamp order to complete the structured organization of the entity set and the entity relationship set, forming entity relationship graph data.

[0059] S2. Extract causal relationships from entity relationship graph data and construct a causal knowledge graph.

[0060] S2.1 Perform causal analysis on the entity relationship graph data, extract causal relationships, and generate a candidate set of causal relationships.

[0061] Specifically, for the directed edge relationships between entity nodes in the entity relationship graph data, based on the timestamp information attached to each directed edge, the entity pairs are arranged in chronological order, and entity pairs that meet the chronological order condition are selected. For each pair of entity nodes that meet the chronological order condition, the historical change sequences of the preceding entity attribute values ​​and the subsequent entity attribute values ​​are extracted. The Granger causality test method is used to perform lag correlation analysis on the historical change sequences of the preceding entity attribute values ​​and the subsequent entity attribute values. By comparing the differences in the change trends before and after the introduction of the preceding entity attribute values, it is determined whether the changes in the preceding entity attribute values ​​have a significant predictive power on the changes in the subsequent entity attribute values. The Granger causality test output results are analyzed for significance. Specifically, the statistical probability values ​​of the preceding and subsequent entity attributes in each entity pair are extracted and compared with the confidence level. If the statistical probability value between the preceding and subsequent entity attributes is less than the confidence level (e.g., less than 0.05), it is determined that there is a significant causal relationship between the preceding and subsequent entity attributes, and only the entity pairs consisting of the preceding and subsequent entity attributes are retained as the analysis results. For each retained entity pair, the causal direction, causal strength, and influence delay time are recorded as the contents of the causal relationship candidate set. All entity pairs that meet the conditions are compiled and summarized to generate the causal relationship candidate set.

[0062] It should also be explained that the specific steps for setting the confidence level are as follows: clarify the causal judgment criteria used in the causal relationship identification process, such as screening based on statistical significance or probability confidence; pre-set the confidence level threshold according to the tolerance of the application scenario for the risk of misjudgment, such as setting it to 95% or 99%; then use the confidence level as the criterion for determining whether the causal relationship is valid, and mark all causal relationships that meet the confidence level requirements as valid candidate relationships.

[0063] S2.2 Filter and confirm the candidate set of causal relationships to form a valid set of causal relationships.

[0064] Specifically, each causal link in the candidate set is read one by one. Based on the entity identifiers, entity attributes, and associated task execution order involved in the causal link, the corresponding entity relationship path is searched in the entity relationship graph data. Then, the temporal sequence between the causal entity and the result entity of each causal link is compared, and causal links with unreasonable time sequence or missing entity association paths are eliminated. The retained causal links are checked for logical consistency to determine whether there are duplicate, circular, or conflicting entity connection relationships. Finally, all causal links that meet the temporal rationality and logical consistency are summarized to form a valid causal link set.

[0065] It should also be explained that the specific steps for determining whether there are duplicate, cyclic, or conflicting entity connections are as follows: Read each entity connection path between the causal entity and the result entity in the causal relationship and number all paths; then map each pair of entity connections to a directed edge and construct a directed graph structure; perform path traversal on the directed graph structure to check if there is a path starting from a certain entity and eventually returning to itself; if so, it is determined to be a cyclic connection; simultaneously, during the traversal, check if there are multiple connection paths with the same causal entity and result entity; if so, mark them as duplicate connections; further determine if there are any logical contradictions or entity attribute conflicts between the connection direction between the causal entity and the result entity and other paths; if so, mark them as conflicting connections; finally, output all entity connections marked as duplicate, cyclic, or conflicting for subsequent removal processing.

[0066] S2.3. Structure and integrate the effective causal relationship set with the entity relationship graph data to construct a causal knowledge graph.

[0067] Specifically, the process involves reading all causal relationship pairs from the valid causal relationship set and matching them with the entity set in the entity relationship graph data to determine the entity nodes involved in the causal relationships; treating the causal relationships as directed edges to connect the corresponding cause entity node and result entity node; then merging all the directed edges corresponding to the causal relationships with the entity nodes in the entity relationship graph data according to a unified data structure format to generate a fused graph structure containing entity nodes and causal directed edges; validating the fused graph structure to ensure that all causal relationships correspond to valid entity nodes and that no isolated nodes exist; and finally, outputting the fused causal knowledge graph.

[0068] S3. By combining graph structure analysis and time series forecasting methods, the estimated completion time and delay probability are calculated for the causal knowledge graph and associated time series data, and the progress risk assessment results are generated.

[0069] S3.1. Using graph structure analysis methods, perform structural analysis on the causal knowledge graph, identify the nodes representing tasks and the dependent causal path information, and generate the task dependency structure.

[0070] Specifically, the process involves traversing all entity nodes in the causal knowledge graph and selecting a set of task nodes; extracting causal paths connecting task nodes from the directed edges in the causal knowledge graph and constructing a set of causal path information according to the dependency order between nodes; organizing the task node set and the causal path information set according to graph structure rules to form a task dependency structure graph representing the dependencies between tasks; and finally, verifying the integrity of the task dependency structure graph to ensure that all task nodes are included in the dependency structure and that the causal paths are unbroken, thus completing the generation of the task dependency structure.

[0071] S3.2 Extract time-stamped task execution events from structured industrial data, and classify the task execution events into each task node in the task dependency structure to generate associated time series data.

[0072] Specifically, the process involves scanning all records in the structured industrial data and filtering out task execution event data containing timestamp fields. Based on the entity identifiers in the task execution event data and the task node names in the task dependency structure, an entity mapping method is used to match the corresponding relationships. An attribute alignment method is used to compare and verify the relevant attributes of the task execution events with the attributes of the task nodes, confirming the accuracy of the matching. Successfully matched task execution events are categorized into the corresponding task nodes, arranged in timestamp order, and the corresponding associated time series data is generated. The associated time series data undergoes an integrity check to ensure that the timestamps are continuous and without omissions, thus completing the generation of the associated time series data.

[0073] S3.3 Combine the task dependency structure with the associated time series data, use time series forecasting methods to calculate, and output the expected completion time series.

[0074] Specifically, the process involves obtaining the task nodes and their dependencies within the task dependency structure, along with the corresponding associated time series data. Following the chronological order of the task dependency structure, the time series data of the task nodes are synchronized to ensure consistency across the timeline. A suitable time series forecasting method is selected, such as a moving average based on historical execution time or an autoregressive method, to predict the execution time trend of each task node and obtain its estimated completion time. Combining the task dependencies, the estimated start and completion times of subsequent task nodes are calculated sequentially to form an estimated completion time series. Finally, the estimated completion time series of all task nodes are summarized to output a complete estimated completion time series.

[0075] S3.4 Model the expected completion time series, calculate the delay probability, and generate the task delay probability distribution.

[0076] Specifically, obtain sample data for the expected completion time series, and calculate the mean and variance, as expressed by the following expressions:

[0077] ;

[0078] ;

[0079] in, This represents the mean of the expected completion time series. This indicates the expected number of samples in the time series. Indicates the first A sample value of the estimated completion time. This indicates the sample index in the expected completion time series. This represents the variance of the expected completion time series;

[0080] With mean and variance As parameters of the Gaussian distribution, the probability density function is constructed, and its expression is:

[0081] ;

[0082] in, Indicates the estimated completion time The corresponding probability density value, The variable representing the expected completion time of the probability density to be calculated. This represents the natural exponential function. Represents pi;

[0083] Then, for the set task deadline, the delay probability P is calculated using the following formula:

[0084] ;

[0085] in, Indicates the probability of delay. This indicates the deadline set for the task. The cumulative distribution function represents the standard normal distribution. The standard deviation of the expected completion time series , Indicates the delay;

[0086] The calculated delay probability is used to form a task delay probability distribution with the corresponding estimated completion time.

[0087] It should also be noted that the specific steps for setting task deadlines are as follows: Based on the project plan or task requirements, determine the latest completion time for each task; based on the nature, priority, and relevant resource availability of the task, formulate corresponding deadline parameters; record the deadline parameters in a uniform format to form the task deadlines for each task; if adjustments are needed, dynamically update the task deadlines based on task progress and actual conditions to ensure that all task deadlines have clear and traceable time nodes.

[0088] S3.5 Based on the task dependency structure and task delay probability distribution, comprehensively assess the overall project schedule risk and generate schedule risk assessment results.

[0089] Specifically, the process involves: acquiring all task nodes and their predecessor task relationships within the task dependency structure; sequentially traversing each task node and reading the delay probability value from its corresponding task delay probability distribution; calculating the cumulative delay probability of each task node based on its predecessor task relationships using probability propagation rules, specifically, if a task node has multiple predecessor tasks, employing an independent event joint probability calculation method to combine the delay probabilities of all predecessor tasks and multiply them with the delay probability of the current task to obtain the overall delay probability of the task node; accumulating the overall delay probabilities of critical task nodes along the critical path of the task dependency structure to form the overall delay probability of the critical path; and determining the overall project schedule risk using either the maximum value method or the weighted average method based on the overall delay probabilities of all critical paths, thus generating a schedule risk assessment result.

[0090] It should also be noted that the process of forming the overall delay probability of the critical path is as follows: In the task dependency structure, the path that connects the task start node and the end node and has the longest cumulative duration is selected as the critical path; in the critical path, all task nodes with irreplaceable connections and decisive impact on the overall progress are extracted as critical task nodes; finally, the comprehensive delay probability of each critical task node is accumulated along the critical path to form the overall delay probability of the critical path.

[0091] S4. Based on the schedule risk assessment results and causal knowledge graph, identify the causal chain and generate delay explanation information.

[0092] S4.1 Based on the progress risk assessment results, the task nodes in the causal knowledge graph are matched and filtered to identify the task nodes whose delay probability is higher than the set delay probability judgment threshold, thus forming a set of tasks to be analyzed.

[0093] Specifically, based on the progress risk assessment results, all task node information and corresponding task delay probabilities are read from the causal knowledge graph; a delay probability judgment threshold is set, for example, a sample threshold of 0.3; then, the delay probabilities of task nodes are traversed sequentially, and the delay probability of each task node is compared with the delay probability judgment threshold; all task nodes with delay probabilities higher than the delay probability judgment threshold are filtered out to form a set of tasks to be analyzed.

[0094] It should also be noted that when setting the delay probability judgment threshold, the range of the delay probability judgment threshold should be determined according to project management requirements and risk tolerance, such as selecting an appropriate range between 0 and 1; refer to historical project data or industry standards to select a representative delay probability value as the initial threshold; then adjust the initial threshold according to the specific characteristics and importance of the project to ensure that high-risk tasks and low-risk tasks can be effectively distinguished; the adjusted value is determined as the delay probability judgment threshold.

[0095] S4.2 Extract causal path information that depends on the set of tasks to be analyzed from the causal knowledge graph to generate a set of candidate causal chains.

[0096] Specifically, based on the set of tasks to be analyzed, the corresponding task nodes in the causal knowledge graph are located one by one. For each task node, a graph traversal method is used to search backward or forward along the causal connection edges starting from the node to extract all causal path information that is directly or indirectly related to the task node, generating a set of causal paths containing multiple causal connections. Then, the set of causal paths is filtered to retain paths that have a dependency relationship with the set of tasks to be analyzed and whose path length does not exceed the causal path length threshold, thus forming a set of candidate causal chains.

[0097] It should also be noted that the causal path length threshold originates from parameter settings during the causal knowledge graph construction process. It is typically determined based on the complexity of the actual project task and the hierarchical depth of the causal relationships. The causal path length threshold is used to limit the maximum length of causal paths to avoid excessively long paths leading to redundancy in the causal chain or excessive computational complexity, thus ensuring the accuracy and efficiency of causal chain analysis. In the example, the causal path length threshold can be set based on the results of historical project data analysis.

[0098] S4.3. Use causal reasoning methods to analyze and screen the causal strength of the candidate causal chain set, form a set of causal paths for task delay, extract the upstream factors that trigger the delay, and generate delay explanation information.

[0099] Specifically, for each causal chain in the candidate causal chain set, the causal strength of the causal relationship in the chain is calculated using causal reasoning methods. Based on a preset causal strength threshold, causal chains with causal strength reaching or exceeding the threshold are selected to form a set of causal paths for task delay. Then, the starting node of the causal chain in the set of causal paths for task delay is identified as the upstream factor that triggers the delay. Relevant causal relationships and attribute information are extracted to generate delay explanation information.

[0100] It should also be noted that when setting the causal strength threshold, the distribution characteristics of causal strength are statistically analyzed based on historical causal relationship data. Multiple representative causal relationship samples are selected, and the distribution characteristics of causal strength are analyzed to determine the initial numerical range of the causal strength threshold. Subsequently, the causal strength threshold is adjusted in combination with the specific needs of the project to balance the coverage and accuracy of the causal chain. Finally, the causal strength threshold is determined as the standard for screening effective causal chains.

[0101] S5. Based on the delay explanation information and the preset resource priorities and task weights, construct a compensation scheduling scheme using rule-driven and priority scheduling strategies.

[0102] S5.1 Extract the resource factors and constraints affecting the schedule from the delay explanation information, and generate a set of resource requirements to be scheduled.

[0103] Specifically, the process involves parsing the text and structured data in the delay explanation information to identify descriptive content related to resource usage; extracting resource factor names and attribute information, such as resource type, quantity, and available time, based on entity recognition methods; matching and extracting constraint content related to resource allocation, capacity limits, and time windows according to a preset constraint template; and unifying the extracted resource factors and constraints to form a set of resource demands to be scheduled, including information such as resource name, demand quantity, and constraints.

[0104] It should also be explained that the specific steps for setting up a constraint template are as follows: Based on the commonly used constraint types in project management and resource scheduling, identify key constraints such as resource allocation, capacity limits, and time windows; define various constraints in a structured format, including constraint name, constraint type, constraint parameters, and applicable scope; develop corresponding constraint templates for different resource types based on the actual task execution environment, such as setting maximum available working hours constraints for personnel resources, maximum usable capacity constraints for equipment resources, and allowed scheduling time windows for time resources; and store the constraint templates in a unified format for easy matching and retrieval in subsequent operations.

[0105] S5.2 Construct a scheduling priority list based on the set of resource requirements to be scheduled and the preset resource priorities and task weights.

[0106] Specifically, the process involves: reading the names and corresponding quantities of resources to be scheduled from the set of resource requests; obtaining resource priority settings to determine the priority values ​​of the resources to be scheduled; obtaining task weight information to determine the weight coefficients of task nodes; calculating the weighted priority value for each resource using a weighted scheduling sorting method; sorting all resources to be scheduled according to their scheduling priority values, with resources having higher priority values ​​listed first; and generating a scheduling priority list based on the sorting results, including the resource name, corresponding task node, and scheduling priority value.

[0107] S5.3. Based on the scheduling priority list, match the developer, computing resources and test environment resources required for the task, and generate an initial resource configuration scheduling scheme.

[0108] Specifically, based on the scheduling priority list, tasks are selected sequentially from the task set, and the developer requirements, computing resource requirements, and test environment resource requirements corresponding to each task are read. Based on the developer information, computing resource information, and test environment resource information in the currently available resource pool, tasks are matched one by one according to their order in the scheduling priority list, prioritizing resource items that meet the constraints of resource type, capability level, and available time period. If multiple options exist for a certain type of resource, the resource item with the highest idle rate is selected based on the task scheduling priority value. Successfully matched developer, computing resource, and test environment resources are recorded in the resource configuration record table of the corresponding task, and the resource usage status is updated. These steps are repeated until all tasks have completed resource matching, generating an initial resource configuration scheduling scheme.

[0109] S5.4 Perform constraint verification and resource conflict resolution on the initial resource configuration scheduling scheme to generate a compensation scheduling scheme.

[0110] Specifically, the process iterates through each task resource allocation record in the initial resource configuration scheduling scheme, verifying each task against constraints such as time constraints, resource capability level constraints, and resource availability periods. Tasks that do not meet the constraints or have overlapping resource allocations are marked. Conflicting resource identifiers are extracted from the initial resource configuration scheduling scheme, and the priority allocation object for conflicting resources is determined based on the task priority information in the scheduling priority list. For tasks with conflicting resources that are not priority objects, alternative resources that are not occupied and meet the task requirements are searched in the current resource pool. If multiple alternative resources exist, the resource with the highest idle level or closest to the original resource capability level is selected for replacement. If no alternative resources exist, the task is marked as a task to be processed by subsequent compensation scheduling. After all conflicting resources are resolved, the updated resource allocation records are integrated to generate a compensation scheduling scheme.

[0111] S6. Preset resource priorities and task weights.

[0112] S6.1 Collect historical project execution records and task completion status to obtain information on resource usage frequency, task completion time, and causal path.

[0113] Specifically, project identifiers, task identifiers, task start times, task end times, task statuses, resource names, and resource usage periods are extracted from historical project execution records to construct a raw execution dataset containing timestamps and resource attributes. Then, the frequency of each resource name appearing in all historical projects is counted to obtain the corresponding resource usage frequency. Next, the time difference between the task start time and task end time is calculated based on the task identifier to obtain the actual completion time of each task. This difference is then compared with the expected completion time of the corresponding task in the project plan, and a timeliness comparison analysis method is used to calculate the difference. Finally, based on the preceding and following execution records and associated marker information between task identifiers, causal path information representing the causal dependencies of tasks is constructed according to the project dimension, generating a set of task causal paths.

[0114] S6.2. Based on the frequency of resource usage and the timeliness of task completion, assess the importance of resources and generate an initial resource priority score.

[0115] Specifically, the usage frequency of each resource in the historical project execution records is counted to obtain the resource usage frequency value; then the actual completion time of the tasks participated in by each resource is summarized, and the average completion time value corresponding to the resource is calculated using a weighted scoring method; then the resource usage frequency value and the average completion time value of all resources are normalized respectively, and the normalized resource usage frequency value and the reciprocal of the normalized resource average completion time value are weighted and superimposed to obtain the initial resource priority score.

[0116] S6.3. Based on the causal path information, assess the impact of the task on the overall progress and generate a task weight score.

[0117] Specifically, based on the causal path information, the frequency of each task node in all causal paths is counted using a node frequency statistics method to obtain the path coverage of the task node; then, a key node identification method is used to calculate the frequency of each task node as a key node in the causal path to obtain the key node weight; subsequently, a normalization processing method is used to normalize the path coverage and key node weight respectively; based on a weighted normalization scoring method, the normalized path coverage and key node weight are weighted and summed according to a preset weight ratio to generate the task weight score of the task node.

[0118] It should be noted that the preset weight ratios are determined based on project management literature and historical data. For example, the weight of path coverage is 0.6 and the weight of key nodes is 0.4. These can be adjusted according to the project to ensure that the importance of task nodes is reflected. The weights are derived from project risk management standards and causal analysis studies.

[0119] S6.4 Normalize the initial resource priority score and task weight score to form resource priority and task weight.

[0120] Specifically, based on the initial resource priority score and task weight score, the min-max normalization method is used to map the initial resource priority score and task weight score to a unified numerical range. Specifically, the min-max normalization method is used to calculate the maximum and minimum values ​​for the initial resource priority score set, and the initial score of each resource is linearly converted into a normalized resource priority based on the maximum and minimum values. For the task weight score set, the maximum and minimum values ​​are calculated, and the weight score of each task is linearly converted into a normalized task weight based on the maximum and minimum values, thus obtaining the resource priority and task weight.

[0121] S7. Implement the compensation scheduling scheme, collect implementation effect data and analyze scheduling effect through closed-loop feedback and dynamic adjustment mechanism, and generate an optimized scheduling management scheme.

[0122] S7.1 Optimize the compensation scheduling scheme and generate a preliminary scheduling execution plan.

[0123] Specifically, the compensation scheduling scheme is encoded into a chromosome representation of the genetic algorithm, specifically mapping resource allocation and scheduling order to chromosome gene sequences. The scheduling scheme for each chromosome is evaluated based on a fitness function, which comprehensively considers scheduling completion time, resource utilization efficiency, and constraint satisfaction to calculate a comprehensive score. A higher fitness score reflects the scheduling effectiveness, indicating shorter completion time, higher resource utilization, and better constraint compliance. Next, a roulette wheel selection method is used to select several chromosomes from the current chromosome population as parents. A new generation of chromosomes is generated through single-point crossover and mutation operations, with crossover and mutation probabilities of, for example, 0.8 and 0.05. The fitness of the new generation of chromosomes is then calculated, and an elite retention strategy is used to retain several high-fitness chromosomes. This selection, crossover, mutation, and fitness calculation process is repeated until a preset number of iterations is reached. Finally, the chromosome with the highest fitness is selected as the initial scheduling execution plan.

[0124] S7.2. Through a closed-loop feedback mechanism, collect feedback information during the initial scheduling and execution process, perform statistical analysis and performance evaluation, and generate scheduling effect evaluation results.

[0125] Specifically, through a closed-loop feedback mechanism, feedback information is collected during the initial scheduling and execution plan implementation process, including actual task completion time, resource usage, and abnormal event records. Subsequently, the feedback information is statistically analyzed, using descriptive statistical methods to calculate the task completion rate as the ratio of completed tasks to the total number of tasks. For resource usage, the resource utilization rate is calculated as the ratio of actual resource usage to the total resource amount. For delays, the delay rate is calculated as the ratio of delayed tasks to the total number of tasks. Then, based on the statistical results, a multi-indicator performance evaluation method is applied to comprehensively evaluate the scheduling effect. Evaluation indicators include, but are not limited to, timely task completion rate, resource allocation rationality, and scheduling stability. Finally, a scheduling effect evaluation result is generated based on the evaluation results.

[0126] S7.3. Based on the evaluation results of the scheduling effect, adopt a dynamic adjustment mechanism to optimize the compensation scheduling scheme and generate an optimized scheduling management scheme.

[0127] Specifically, based on the scheduling effect evaluation results, the deficiencies and bottlenecks in the compensation scheduling scheme are identified, and the deviations of key performance indicators are extracted. Then, based on the dynamic adjustment mechanism, a feedback control method is used to adjust parameters such as resource allocation, task priority, and scheduling order. The adjustment magnitude and direction are determined according to the specific deviations of key performance indicators in the evaluation results. Next, the adjusted compensation scheduling scheme is simulated and verified to ensure that the adjusted scheme meets the constraints and that performance is improved. Finally, an optimized scheduling management scheme is generated, which includes the adjusted resource configuration scheme, task scheduling order, and priority settings.

[0128] This embodiment also provides a computer device applicable to the software development project schedule prediction and management method based on artificial intelligence, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the software development project schedule prediction and management method based on artificial intelligence as proposed in the above embodiment.

[0129] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0130] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the software development project schedule prediction and management method based on artificial intelligence as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0131] In summary, this invention, by introducing causal reasoning into project schedule management, breaks through the limitations of traditional methods based on statistical regression or simple time series forecasting. It enables in-depth revelation and utilization of causal relationships between project tasks, dynamically reflects the actual execution of tasks by combining time series data, and makes risk assessment more timely and accurate. It can not only accurately identify potential delay risks, but also quantify the probability of delays, providing scientific risk warnings and decision support for project management, and significantly improving the intelligence level and practical value of project schedule management.

[0132] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A software development project schedule prediction and management method based on artificial intelligence, characterized in that: include, Collect multi-source heterogeneous industrial data and preprocess it to form entity relationship graph data; Extract causal relationships from entity relationship graph data and construct a causal knowledge graph; By combining graph structure analysis and time series forecasting methods, the estimated completion time and delay probability are calculated for causal knowledge graphs and associated time series data, and the schedule risk assessment results are generated. Using causal reasoning methods, causal chains are identified from the schedule risk assessment results and causal knowledge graph to generate delay explanation information; Based on the delay explanation information and the preset resource priorities and task weights, a compensation scheduling scheme is constructed using rule-driven and priority scheduling strategies. Implement a compensation scheduling scheme, collect and analyze the implementation effect data through a closed-loop feedback and dynamic adjustment mechanism, and generate an optimized scheduling management scheme.

2. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for forming the entity relationship graph data are as follows: Collect task progress, code submission records, developer attendance records, and task execution logs to obtain multi-source heterogeneous industrial data, and perform missing value filling and format unification to generate structured industrial data; Based on structured industrial data, entities are extracted and logical relationships and temporal dependencies between entities are identified, generating entity sets and entity relationship sets; The entity set and entity relationship set are organized in a structured manner to form entity relationship graph data.

3. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for constructing the causal knowledge graph are as follows. Perform causal analysis on entity relationship graph data, extract causal relationships, and generate a candidate set of causal relationships; The candidate set of causal relationships is screened and confirmed to form a valid set of causal relationships; By structurally integrating the effective causal relationship set with entity relationship graph data, a causal knowledge graph is constructed.

4. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for generating the schedule risk assessment results are as follows: Using graph structure analysis methods, we perform structural analysis on causal knowledge graphs, identify nodes representing tasks and dependent causal path information, and generate task dependency structures. Extract time-stamped task execution events from structured industrial data, categorize these events into task nodes within the task dependency structure, and generate associated time-series data. By combining the task dependency structure with associated time series data, a time series forecasting method is used to calculate and output the expected completion time series. Model the expected completion time series, calculate the delay probability, and generate the task delay probability distribution; Based on the task dependency structure and task delay probability distribution, the overall project schedule risk is comprehensively assessed, and a schedule risk assessment result is generated.

5. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for generating delay explanation information are as follows: Based on the progress risk assessment results, the task nodes in the causal knowledge graph are matched and filtered to identify task nodes whose delay probability is higher than the set delay probability judgment threshold, thus forming a set of tasks to be analyzed. Extract causal path information that depends on the set of tasks to be analyzed from the causal knowledge graph to generate a set of candidate causal chains; The causal reasoning method is used to analyze and screen the causal strength of the candidate causal chain set, forming a set of causal paths for task delays, and extracting the upstream factors that trigger the delays to generate delay explanation information.

6. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for constructing the compensation scheduling scheme are as follows: Extract the resource factors and constraints affecting the schedule from the delay explanation information to generate a set of resource requirements to be scheduled; Based on the set of resource demands to be scheduled and the preset resource priorities and task weights, construct a scheduling priority list; Based on the scheduling priority list, match the developer, computing resources and test environment resources required by the task, and generate an initial resource configuration scheduling scheme. The initial resource configuration and scheduling scheme is subjected to constraint verification and resource conflict resolution to generate a compensation scheduling scheme.

7. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The preset resource priorities and task weights are defined in the following steps. Collect historical project execution records and task completion status to obtain information on resource usage frequency, task completion time, and causal path. Based on the frequency of resource usage and the timeliness of task completion, assess the importance of resources and generate an initial resource priority score; Based on causal path information, assess the impact of tasks on the overall progress and generate task weight scores. The initial resource priority score and task weight score are normalized to form resource priority and task weight.

8. The method for predicting and managing the progress of software development projects based on artificial intelligence as described in claim 1, characterized in that: The specific steps for generating the optimized scheduling management scheme are as follows: The compensation scheduling scheme is optimized to generate a preliminary scheduling execution plan; Through a closed-loop feedback mechanism, feedback information is collected during the initial scheduling and execution process, and statistical analysis and performance evaluation are performed to generate scheduling effect evaluation results. Based on the evaluation results of the scheduling effect, a dynamic adjustment mechanism is adopted to optimize the compensation scheduling scheme and generate an optimized scheduling management scheme.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the artificial intelligence-based software development project schedule prediction and management method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the artificial intelligence-based software development project schedule prediction and management method as described in any one of claims 1 to 8.

Images