Project risk analysis method, device, equipment, medium and product

Abstract

The invention relates to a project risk analysis method and device, equipment, a medium and a product, and the method comprises the steps: receiving a risk analysis task of a target project, calling a first preset large language model, and generating an analyzable analysis plan based on the risk analysis task and context parameters of the target project; the analysis plan comprises a dynamic risk decomposition structure establishment task, a risk category traversal task and a risk assessment task; executing the dynamic risk decomposition structure establishment task to obtain a dynamic risk decomposition structure of the target project; based on the dynamic risk decomposition structure, executing a risk category traversal task, and executing a risk assessment task for at least one traversed risk category to obtain a risk priority number; and generating a risk analysis result according to the risk priority number. By adopting the method, the reliability and the accuracy of the finally obtained risk analysis result can be improved.

Description

Technical Field

[0001] This application relates to the technical field of project management, and in particular to a project risk analysis method, apparatus, equipment, medium, and product. Background Technology

[0002] Project management encompasses a series of processes from project initiation, planning, execution, monitoring to closure, and risks are omnipresent throughout the entire project lifecycle. To ensure the safety and stability of project execution, it is necessary to accurately identify and assess potential risks throughout the entire project lifecycle.

[0003] In related technologies, the assessment of potential risks in a project often relies on human experience. However, such an assessment method can easily lead to low accuracy in the risk analysis results. Summary of the Invention

[0004] Therefore, it is necessary to provide a project risk analysis method, apparatus, computer equipment, computer-readable storage medium, and computer program product to address the aforementioned technical problems.

[0005] Firstly, this application provides a project risk analysis method, which includes:

[0006] The system receives the risk analysis task for the target project and calls the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes tasks for establishing a dynamic risk decomposition structure, traversing risk categories, and assessing risks.

[0007] Perform the task of establishing a dynamic risk decomposition structure to obtain the dynamic risk decomposition structure of the target project;

[0008] Based on the dynamic risk decomposition structure, a risk category traversal task is performed, and for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number.

[0009] Based on the risk priority number, generate risk analysis results.

[0010] In one embodiment, a dynamic risk decomposition structure establishment task is performed to obtain the dynamic risk decomposition structure of the target project, including:

[0011] Obtain multi-dimensional data for the target project; multi-dimensional data includes structured data and unstructured text data;

[0012] Preprocess the multi-dimensional data to obtain at least one key prompt word;

[0013] Input the key prompts into the first preset large language model to obtain the dynamic risk decomposition structure.

[0014] In one embodiment, for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number, including:

[0015] Based on risk categories, search for historical failure mode subgraphs related to the risk categories from a pre-defined knowledge graph;

[0016] Based on the historical failure mode subgraph and the context parameters of the target project, the second preset large language model is invoked to obtain the risk priority number.

[0017] In one embodiment, based on the historical failure mode subgraph and the context parameters of the target project, a second preset large language model is invoked to obtain a risk priority number, including:

[0018] Candidate information is extracted from the historical failure mode subgraph; the candidate information includes failure mode, failure effect, root cause and countermeasures;

[0019] The candidate information and the context parameters of the target project are input into the second preset large language model to obtain the risk priority number.

[0020] In one embodiment, candidate information and context parameters of the target project are input into a second preset large language model to obtain a risk priority number, including:

[0021] The candidate information and the context parameters of the target project are input into the second preset large language model, so that the second preset large language model can perform inference on failure mode, failure impact, root cause and countermeasures, and obtain the inference result;

[0022] Based on the reasoning results, the severity, occurrence, and detectability are quantitatively scored using a second pre-set large language model, and the risk priority number is determined based on the product of severity, occurrence, and detectability.

[0023] In one embodiment, the method for constructing the preset knowledge graph includes:

[0024] Acquire historical data from historical projects and perform entity identification and relationship extraction on the historical data; the historical data includes historical failure modes, historical failure impacts, historical root causes, and historical countermeasures;

[0025] The processed entities and relationships are stored in a graph database; the node types in the graph database include failure modes, root causes, potential impacts, and countermeasures; the edges between nodes in the graph database are used to represent either causal relationships or mitigation relationships.

[0026] In one embodiment, the method further includes:

[0027] If the project data of the target project changes, the task of establishing a dynamic risk decomposition structure is re-executed based on the risk analysis results and the changed project data to obtain a new dynamic risk decomposition structure.

[0028] Secondly, this application also provides a project risk analysis device, which includes:

[0029] The risk analysis plan generation module receives the risk analysis task of the target project and calls the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes tasks for establishing a dynamic risk decomposition structure, traversing risk categories, and assessing risks.

[0030] The risk decomposition structure determination module is used to perform the task of establishing a dynamic risk decomposition structure and obtain the dynamic risk decomposition structure of the target project.

[0031] The risk priority number determination module is used to perform a risk category traversal task based on a dynamic risk decomposition structure, and perform a risk assessment task for at least one risk category traversed to obtain a risk priority number.

[0032] The risk analysis results determination module is used to generate risk analysis results based on risk priority numbers.

[0033] Thirdly, this application also provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0034] The system receives the risk analysis task for the target project and calls the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes tasks for establishing a dynamic risk decomposition structure, traversing risk categories, and assessing risks.

[0035] Perform the task of establishing a dynamic risk decomposition structure to obtain the dynamic risk decomposition structure of the target project;

[0036] Based on the dynamic risk decomposition structure, a risk category traversal task is performed, and for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number.

[0037] Based on the risk priority number, generate risk analysis results.

[0038] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0039] The system receives the risk analysis task for the target project and calls the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes tasks for establishing a dynamic risk decomposition structure, traversing risk categories, and assessing risks.

[0040] Perform the task of establishing a dynamic risk decomposition structure to obtain the dynamic risk decomposition structure of the target project;

[0041] Based on the dynamic risk decomposition structure, a risk category traversal task is performed, and for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number.

[0042] Based on the risk priority number, generate risk analysis results.

[0043] Fifthly, this application also provides a computer program product comprising a computer program that, when executed by a processor, performs the following steps:

[0044] The system receives the risk analysis task for the target project and calls the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes tasks for establishing a dynamic risk decomposition structure, traversing risk categories, and assessing risks.

[0045] Perform the task of establishing a dynamic risk decomposition structure to obtain the dynamic risk decomposition structure of the target project;

[0046] Based on the dynamic risk decomposition structure, a risk category traversal task is performed, and for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number.

[0047] Based on the risk priority number, generate risk analysis results.

[0048] The aforementioned project risk analysis method, apparatus, equipment, medium, and product receive a risk analysis task from a target project and invoke a first preset large language model to generate a parsable analysis plan based on the risk analysis task and the context parameters of the target project. The analysis plan includes a dynamic risk decomposition structure establishment task, a risk category traversal task, and a risk assessment task. The dynamic risk decomposition structure establishment task is executed to obtain the dynamic risk decomposition structure of the target project. Based on the dynamic risk decomposition structure, the risk category traversal task is executed, and for at least one traversed risk category, a risk assessment task is performed to obtain a risk priority number. Based on the risk priority number, a risk analysis result is generated. This application employs the above method, which, through the first preset large language model, can generate a reliable analysis plan based on the risk analysis task and the context parameters of the target project, thereby improving the efficiency and accuracy of subsequent risk analysis. Furthermore, by constructing a dynamic risk decomposition structure, the risk analysis results can be adjusted and optimized in real time according to changes in project data, achieving flexibility in dynamic project management, and further improving the reliability and accuracy of the final risk analysis results. Attached Figure Description

[0049] Figure 1 Flowcharts of project risk analysis methods provided for some embodiments of this application;

[0050] Figure 2 A flowchart illustrating the dynamic risk decomposition structure provided in some embodiments of this application;

[0051] Figure 3 A flowchart for obtaining a risk priority number is provided for some embodiments of this application;

[0052] Figure 4 A flowchart for obtaining a risk priority number is provided for some embodiments of this application;

[0053] Figure 5 A flowchart for determining a risk priority number is provided for some further embodiments of this application;

[0054] Figure 6 A flowchart for constructing a knowledge graph is provided for some embodiments of this application;

[0055] Figure 7 Structural block diagram of a project risk analysis device provided in some embodiments of this application;

[0056] Figure 8 This is an internal structural diagram of a computer device provided in some embodiments of this application. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0058] The project risk analysis method provided in this application can be applied to user terminals loaded with a project management system. These user terminals include, but are not limited to, computers, personal laptops, and other devices. The project management system is based on a modular intelligent agent architecture of ReWOO, including a planning module, an execution module, and a memory module. The planning module is driven by a large language model, whose inputs are the task objective and context from the memory module. The large language model analyzes the steps required to achieve the objective and outputs a clear and parsable action plan (analysis plan). The execution module is also determined by the large language model, responsible for parsing the action plan and executing each step of the plan. It uses predefined tools to pass parameters from the plan and other steps to designated tools and obtains the execution results. The memory module is implemented by a temporary, structured data storage area (such as an in-memory database or a specific data structure) to record the complete state of the current task session, including: original input, generated plan, input and output of each step, execution status (success / failure), and final result. This design enables the agent in the system to perform multi-step reasoning without re-observing all the original data at each step, significantly reducing unnecessary calls to the large language model and data transmission, thereby improving the reliability and efficiency of the project management system.

[0059] When a user terminal needs to analyze the potential risks of a new project, i.e., the target project during its future execution, the project management system's planning module first receives the risk analysis task for the target project and calls the first preset large language model. Based on the risk analysis task and the context parameters of the target project, it generates a parsable analysis plan. The analysis plan includes a dynamic risk decomposition structure establishment task, a risk category traversal task, and a risk assessment task. Then, the system's execution module executes the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project. Based on the dynamic risk decomposition structure, the risk category traversal task is executed, and for at least one risk category traversed, a risk assessment task is performed to obtain a risk priority number. Finally, based on the risk priority number, the risk analysis results are generated. This improves the reliability and accuracy of the final risk analysis results.

[0060] In one embodiment, such as Figure 1 As shown, the method is applied to the aforementioned terminal as an example for illustration. In this embodiment, the method includes the following steps:

[0061] Step 102: Receive the risk analysis task of the target project and call the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project.

[0062] The target project refers to the specific project requiring risk analysis, including but not limited to software development projects, construction projects, marketing projects, and new product development projects. The risk analysis task involves the specific requirements for identifying and assessing risks to the target project, either comprehensively or within a specific scope. For example, this might involve a comprehensive analysis of the risks of the target project over the next month, or a detailed assessment of the technical risks associated with the target project.

[0063] The first pre-defined large language model is a model with powerful language understanding and generation capabilities. It can generate a parsable analysis plan based on the input risk analysis task and the contextual parameters of the target project. The contextual parameters of the target project are various information related to the target project stored in the system's memory module, including past risk analyses of the project, historical data, previously identified common risk categories, and corresponding handling measures. This information helps the first pre-defined large language model better understand the target project, thereby generating an analysis plan that is more realistic. The analysis plan includes tasks such as establishing a dynamic risk decomposition structure, traversing risk categories, and risk assessment.

[0064] Optionally, risk analysis tasks can be received through the planning module of the intelligent risk analysis system in the project management system. Then, the first preset large language model is called, and the risk analysis task and the context parameters of the target project obtained from the memory module are input into the first preset large language model. The first preset large language model generates a parsable analysis plan.

[0065] Step 104: Execute the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project.

[0066] The dynamic risk decomposition structure of the target project is a hierarchical and machine-resolvable structure that breaks down the project's risks into different categories, such as technical risks, personnel risks, and schedule risks, with each category further subdivided into specific risk sub-items. Unlike traditional static risk decomposition structures, dynamic risk decomposition structures can be adjusted and optimized in real time based on changes in project data and risk feedback information.

[0067] Optionally, the dynamic risk decomposition structure creation task can be executed through the system's execution module. Specifically, structured and unstructured data related to the target project can be collected from multiple channels, including the project management system, market databases, and industry reports. For structured data (such as project schedule data and cost data), direct cleaning and preprocessing are performed. For unstructured data (such as project documents and meeting minutes), natural language processing techniques are used for text extraction, keyword recognition, and sentiment analysis. Then, data mining algorithms (such as association rule mining and sequence pattern mining) are applied to identify potential relationships and patterns between the data. For example, association rule mining can reveal the correlation between project schedule delays and delivery delays from certain specific suppliers. Next, clustering algorithms (such as hierarchical clustering) are used to group similar risk factors into one category; for example, schedule risks and quality risks caused by technical difficulties can be clustered into a technical risk category. Finally, a dynamic risk decomposition structure is constructed based on the clustering results, displaying the clustered risk categories and subcategories in a hierarchical manner and storing them in the project management system's database.

[0068] In addition, multi-dimensional data (including structured and unstructured data) of the target project can be obtained through a pre-set system interface, and integrated into structured prompt words for the generation task of dynamic risk decomposition structure. These prompt words are input into the first preset large language model, and the first preset large language model outputs the dynamic risk decomposition structure of the target project and stores it in the system's memory module.

[0069] Step 106: Based on the dynamic risk decomposition structure, perform a risk category traversal task, and for at least one risk category traversed, perform a risk assessment task to obtain a risk priority number.

[0070] In this dynamic risk decomposition structure, risk categories serve as the basis for classifying risks. Examples include the aforementioned technical risks, personnel risks, and schedule risks. Each risk category contains multiple specific risk sub-items to facilitate systematic risk analysis and management. Risk priority numbers are the result of a quantitative assessment of the specific risks within each risk category during risk assessment tasks, and are related to severity, occurrence, and detectability. Risk priority numbers measure the importance and priority of each risk, helping project managers determine which risks require priority handling, enabling rational resource allocation, effective risk response measures, and improved project success rates.

[0071] Optionally, the probability distribution of each risk category can be determined first; then, Monte Carlo simulation software can be used to perform multiple random samplings for each risk factor. Each time a sample is taken, a random value is generated based on its probability distribution to represent the value of that risk factor in this simulation; the sampled risk factor values ​​are then substituted into a preset risk assessment model to calculate the risk priority number for each simulation; finally, the above sampling and calculation process is repeated multiple times (e.g., 1000 times) to obtain a set of simulation results for risk priority numbers. Statistical analysis is then performed on this set of results, such as calculating the mean, median, and standard deviation, to determine the final risk priority number.

[0072] Alternatively, based on the dynamic risk decomposition structure stored in the memory module, the execution module can begin traversing risk categories. For each traversed risk category, a preset knowledge graph is invoked, and similar cases are retrieved from the knowledge graph. Information on candidate failure modes, root causes, impacts, and countermeasures is extracted. This information, along with the current project context, is then input into a preset large language model, which performs reasoning and quantitative scoring to calculate the risk priority number.

[0073] Step 108: Generate risk analysis results based on the risk priority number.

[0074] The risk analysis results are a comprehensive and structured collection of information derived from a risk analysis of the target project. They comprehensively reflect the various risks present in the project, providing decision-making support for project managers and stakeholders to take appropriate measures to address risks and ensure the smooth progress of the project. Risk analysis results may include risk identification information, risk assessment indicators, and risk response recommendations. These results can be presented in the form of risk analysis reports, visual charts, or risk event cards.

[0075] Optionally, based on the calculated risk priority numbers, corresponding risk item cards can be created in the project management system. These cards include a risk description, risk priority number, potential impact, and suggested countermeasures. This generates a risk analysis result, which is then recorded in the system's memory module for later retrieval. Additionally, data visualization tools can be used to display risk priority numbers and related risk information in intuitive chart formats. For example, a bar chart can be used to rank the risk priorities of different risk factors, and a pie chart can be used to show the proportion of each type of risk in the overall risk. Furthermore, a report generation module can be developed to automatically generate risk analysis reports based on the visualized content and project management needs. Report content can include a risk overview, risk priority ranking, risk trend analysis, and response recommendations. Moreover, users (project managers) can rate or correct the risk analysis results, and the system can incorporate this feedback into the memory module to optimize subsequent reasoning processes, forming a closed-loop learning process.

[0076] The aforementioned project risk analysis method, through a first preset large language model, can generate a reliable analysis plan based on the contextual parameters of the risk analysis task and the target project, thereby improving the efficiency and accuracy of subsequent risk analysis. Furthermore, by constructing a dynamic risk decomposition structure, the risk analysis results can be adjusted and optimized in real time according to changes in project data, achieving flexibility in dynamic project management, and further improving the reliability and accuracy of the final risk analysis results.

[0077] In one embodiment, such as Figure 2 As shown, the dynamic risk decomposition structure establishment task is executed to obtain the dynamic risk decomposition structure of the target project, including:

[0078] Step 202: Obtain multi-dimensional data of the target project.

[0079] Multidimensional data refers to data related to the target project collected from multiple different angles and levels, covering all aspects of the project and comprehensively reflecting its characteristics and status. Multidimensional data includes structured data and unstructured text data.

[0080] Structured data is data with a fixed format and structure, usually stored in tabular form. Each field in the data has a clear definition and meaning, facilitating processing and analysis. Structured data includes basic project information (such as project name, project number, project start and end dates, project budget, etc.), project cost data (such as labor costs, material procurement costs, and equipment rental costs, etc.), and resource data (such as personnel allocation, equipment usage, and material inventory, etc.).

[0081] Unstructured text data is data without a fixed format or structure. It usually exists in text form, and the organization and expression of the data are relatively free, making it difficult to process and analyze directly by computers. Unstructured text data includes project documents (such as project plans, requirements specifications, test reports, etc.), meeting minutes (such as project kickoff meetings, weekly meetings, technical seminars, etc.), and email correspondence (email communication between project team members, and with clients or suppliers).

[0082] Optionally, structured data can be extracted from internal systems such as project management systems, enterprise resource planning systems, and financial management systems, and unstructured text data can be obtained from the company's document management system and email server. External data related to the project, such as industry reports, market dynamics, and policies and regulations, can also be obtained from market research institutions and industry associations. Additionally, for some publicly available data sources, data interfaces (such as APIs) or web crawling technologies can be used to automatically obtain relevant data.

[0083] Step 204: Preprocess the multi-dimensional data to obtain at least one key prompt word.

[0084] Keyword prompts are representative and significant words or phrases extracted from preprocessed, multi-dimensional data, summarizing the core content and key information of the data. By providing accurate and targeted keyword prompts, the large language model can better understand relevant project information, thereby generating a risk decomposition structure that better reflects the project's actual situation. For example, in a software development project, keyword prompts extracted after data preprocessing might include "application of new technologies," "frequent changes in requirements," "insufficient team experience," and "tight deadlines."

[0085] Optionally, the multi-dimensional data can be cleaned and filtered first, and noise, duplicate data, and erroneous data can be removed to ensure the accuracy and completeness of the data. For unstructured text data in the multi-dimensional data, natural language processing techniques can be used for word segmentation, part-of-speech tagging, and named entity recognition to extract meaningful words and phrases. Finally, keyword extraction algorithms (such as TextRank) can be used to extract key prompt words from the processed text.

[0086] Step 206: Input the key prompts into the first preset large language model to obtain the dynamic risk decomposition structure.

[0087] Optionally, based on the project's needs and characteristics, a large language model with good performance and adaptability can be selected, such as the GPT series or BERT. The selected large language model can then be trained and fine-tuned to obtain a first preset large language model that can better adapt to the project's domain and risk analysis tasks. The preprocessed key prompts can then be input into the first preset large language model, and appropriate parameters and instructions can be set to allow the first preset large language model to generate a dynamic risk decomposition structure.

[0088] In the aforementioned project risk analysis method, by acquiring multi-dimensional data, the target project can be analyzed from multiple perspectives, comprehensively identifying potential risks and avoiding risk omissions due to data gaps. Furthermore, preprocessing the multi-dimensional data and extracting key prompts can transform complex data into concise and targeted information, improving the quality of input information. This allows the dynamic risk decomposition structure generated by the first preset large language model to more accurately reflect the actual risk situation of the target project.

[0089] In one embodiment, such as Figure 3 As shown, for at least one risk category encountered, a risk assessment task is performed to obtain a risk priority number, including:

[0090] Step 302: Based on the risk category, search for historical failure mode subgraphs related to the risk category from the preset knowledge graph.

[0091] The pre-defined knowledge graph is a structured semantic network that can display various entities related to project risks (such as risk categories, failure modes, influencing factors, and countermeasures) and their relationships in a graph form. The historical failure mode subgraph is part of the pre-defined knowledge graph and is a subgraph structure related to a specific risk category. It contains various failure modes of that risk category that have occurred in historical projects and the relationships between these failure modes. It may also involve the relationship between failure modes and other relevant factors (such as influencing factors and consequences).

[0092] Optionally, the query language provided by the graph database can be used to search for nodes and edges related to the risk category in a pre-defined knowledge graph, based on the risk category as the query condition, and extract the historical failure mode subgraph. Alternatively, graph traversal algorithms (such as breadth-first search, depth-first search, etc.) can be used to start from the node representing the risk category, traverse along the edges of the graph, find all failure mode nodes related to that risk category and their relationships, and construct the historical failure mode subgraph.

[0093] Step 304: Based on the historical failure mode subgraph and the context parameters of the target project, call the second preset large language model to obtain the risk priority number.

[0094] Among them, the second preset large language model is a language model that has been specifically trained or adjusted. It can receive historical failure mode subgraphs and context parameters of the target project as input, and use its own language understanding and reasoning ability to assess risks and output risk priority numbers.

[0095] Optionally, the historical failure mode subgraph and the context parameters of the target project can be organized and transformed to be acceptable to the second preset large language model. For example, the graph structure data can be converted into text descriptions or vector representations, and the context parameters can be standardized. Then, the processed data can be input into the second preset large language model using a suitable API or interface, and the corresponding input format and parameters can be set. The large model can then be called to perform inference to obtain the risk priority number.

[0096] In this embodiment, by searching historical failure mode subgraphs from a preset knowledge graph, the experience and knowledge from previous projects can be fully utilized to provide specific reference cases and data support for the risk assessment of the target project, making the risk assessment more comprehensive and accurate. Furthermore, because the second preset large language model has powerful language understanding and reasoning capabilities, the resulting risk priority number, after combining the historical failure mode subgraphs and the contextual parameters of the target project, is more reliable and accurate.

[0097] In one embodiment, such as Figure 4 As shown, based on the historical failure mode subgraph and the context parameters of the target project, the second preset large language model is invoked to obtain the risk priority number, including:

[0098] Step 402: Extract candidate information from the historical failure mode subgraph.

[0099] Candidate information is a collection of key information extracted from historical failure mode subgraphs that is relevant to the risk assessment of the target project. Candidate information includes failure modes, failure effects, root causes, and mitigation measures. A failure mode refers to a deviation from the intended function of a system, product, or process. For example, in software development projects, failure modes might include program crashes, slow operation, or inability to perform functions correctly. Failure effects refer to the consequences of a failure mode, including adverse impacts on project schedule, cost, quality, and safety. For instance, software crashes can lead to project delays, decreased customer satisfaction, and even financial losses. Root causes are the most fundamental factors leading to a failure mode. For example, the root cause of a software crash might be logical errors in the code, memory leaks, or incompatibility with other software. Mitigation measures are solutions or preventative measures taken to address the failure mode and its root cause. For example, for software crashes caused by logical errors in the code, mitigation measures might include code review and debugging, and fixing the erroneous code.

[0100] Optionally, if the historical failure mode subgraph is stored in a graph database, a query statement can be written using the graph database's query language to extract node and edge information related to failure modes, failure effects, root causes, and countermeasures based on the labels and attributes of nodes and edges. For historical failure mode subgraphs described in text form, information extraction algorithms from natural language processing, such as named entity recognition and relation extraction, can be used to identify candidate information. For example, a pre-trained named entity recognition model can be used to identify entities such as failure modes and failure effects in the text.

[0101] Step 404: Input the candidate information and the context parameters of the target project into the second preset large language model to obtain the risk priority number.

[0102] Optionally, the candidate information and the context parameters of the target project can be formatted and encoded to make them acceptable to the second preset large language model. For example, the text information can be converted into a vector representation and the numerical context parameters can be normalized. Then, the API interface provided by the second preset large language model can be used to pass the processed data as input to the second preset large language model, and appropriate input formats and parameters can be set. The second preset large language model can then be called to perform reasoning to obtain the risk priority number.

[0103] In this embodiment, by extracting candidate information, targeted historical data is provided for the second preset large language model. Combined with the context parameters of the target project, the second preset large language model can more comprehensively and accurately assess the risk, thereby helping to improve the accuracy of risk priority calculation.

[0104] In one embodiment, such as Figure 5 As shown, the candidate information and the context parameters of the target project are input into the second preset large language model to obtain the risk priority number, including:

[0105] Step 502: Input the candidate information and the context parameters of the target project into the second preset large language model so that the second preset large language model can perform reasoning on failure modes, failure effects, root causes and countermeasures, and obtain the reasoning results.

[0106] The inference results are the output of the second pre-defined large language model after receiving candidate information and contextual parameters of the target project, inferring failure modes, failure impacts, root causes, and countermeasures. The inference results may provide a more detailed and realistic description and analysis of these aspects of the target project, such as predicting how a certain failure mode might manifest in the context of the target project, the scope and extent of its impact, the possible root causes in specific stages, and correspondingly more applicable countermeasures. The inference results can provide specific and targeted basis for subsequent risk quantification assessments, thereby helping the large language model to understand the risk status of the target project more deeply, making the quantitative scoring of severity, occurrence, and detectability more accurate and reasonable, and ultimately improving the reliability of risk priority calculation.

[0107] Optionally, the candidate information and the context parameters of the target project can be formatted and encoded first, converting the text information into a vector representation that the model can understand, and normalizing the numerical parameters. Then, the API of the large language model or its inference interface can be used to input the processed data into the second preset large language model, and appropriate inference parameters can be set, such as setting the inference temperature and maximum output length. The second preset large language model can then be started to perform inference and obtain the inference result.

[0108] Step 504: Based on the reasoning results, the severity, occurrence and detectability are quantitatively scored using the second preset large language model, and the risk priority number is determined based on the product of severity, occurrence and detectability.

[0109] Severity refers to the degree of impact a failure mode has on project objectives (such as quality, safety, cost, schedule, etc.) once it occurs. For example, in a construction project, if a failure mode in a certain construction stage causes a major safety hazard to the building structure, its severity will be very high; if it only causes some cosmetic defects, the severity will be relatively low.

[0110] Occurrence rate refers to the likelihood of a failure mode occurring during project implementation. Occurrence rate can be represented by probability, such as high, medium, and low levels. For example, in a software development project, the occurrence rate of a common programming error causing system crash may be high, while the occurrence rate of system problems caused by some very rare hardware failures may be low.

[0111] Detectability refers to the ease with which a failure mode can be detected before or during its occurrence. If a failure mode is easily detected using existing detection methods and processes, its detectability is high; conversely, if it is difficult to detect, its detectability is low. For example, on a production line, some surface defects can be easily detected by the naked eye or simple inspection equipment, resulting in a relatively high detectability; while some quality problems hidden inside the product may require complex inspection techniques to detect, resulting in a relatively low detectability.

[0112] Optionally, the risk priority number can be calculated using the following formula:

[0113] RPN = S × O × D;

[0114] In the formula, RPN is the risk priority number; S is the severity; O is the occurrence; and D is the detectability.

[0115] In this embodiment, detailed reasoning results are obtained first, and then quantitative scoring is performed based on these results. This approach fully considers the specific circumstances and historical experience of the target project, making the risk assessment more accurate and comprehensive. Furthermore, by calculating risk priority numbers, project managers can be provided with a clear basis for risk ranking, which helps in the subsequent rational allocation of resources, prioritizing the handling of high-risk failure modes, and improving the efficiency of project risk management.

[0116] In one embodiment, such as Figure 6 As shown, the preset methods for constructing knowledge graphs include:

[0117] Step 602: Obtain historical data of historical projects and perform entity recognition and relationship extraction processing on the historical data.

[0118] Historical projects can be projects that have been completed or are currently underway and have some similarity or connection to the current project. These projects can be projects carried out at different times within the same organization, or similar projects implemented by other organizations in the same industry. Historical data is information related to project risks collected from historical projects. Historical data includes historical failure modes, historical failure impacts, historical root causes, and historical response measures.

[0119] Optionally, you can first collect documents related to historical projects from the project's document library, such as project reports, test records, and troubleshooting logs. These documents may exist in the form of text, tables, and charts. If historical project data is stored in a database, you can also use database query languages ​​(such as SQL) to retrieve the required historical data from the database. Then, use named entity recognition technology from natural language processing to process the historical data text and identify entities such as historical failure modes, historical failure impacts, historical root causes, and historical countermeasures. Finally, use machine learning methods to extract relationships. For example, you can use supervised learning algorithms (such as support vector machines, deep learning models, etc.) to learn the relationship patterns between entities using labeled training data.

[0120] Step 604: Store the processed entities and relationships in the graph database.

[0121] In the graph database, the node types include failure modes, root causes, potential impacts, and countermeasures; the edges between nodes in the graph database are used to represent either causal relationships or mitigation relationships.

[0122] Optionally, the processed entity and relation data can be organized according to the format required by the graph database (e.g., CSV file), and then imported into the graph database using an import tool. Alternatively, the programming interface provided by the graph database can be used to insert entity and relation data into the graph database by writing code. During data insertion, historical failure modes, historical root causes, historical failure impacts, and historical countermeasures are used as nodes in the graph database (node ​​types correspond to failure mode, root cause, potential impact, and countermeasure), and corresponding edges are created based on the causal or mitigation relationships between entities. Furthermore, in practical use, this knowledge graph can be automatically added periodically, and graph neural networks can be used to optimize graph query efficiency to maintain the timeliness and coverage of the knowledge base.

[0123] In this embodiment, a knowledge graph is constructed based on historical data of historical projects. This can help project managers quickly understand the full picture of project risks. Furthermore, by analyzing the relationships between nodes in the knowledge graph, it is possible to predict the failure modes and their impacts that may occur in the target project in a timely manner, thereby taking corresponding preventive measures in advance to ensure the stable progress of the target project.

[0124] In one embodiment, the method further includes: if the project data of the target project changes, based on the risk analysis results and the changed project data, re-execute the dynamic risk decomposition structure establishment task to obtain a new dynamic risk decomposition structure.

[0125] Understandably, when project data changes, such as changes in requirements or resource adjustments, re-executing the dynamic risk decomposition structure creation task can ensure that the risk decomposition structure reflects the latest status of the project in a timely manner, ensuring that risk analysis always aligns with the actual project situation, thereby providing the project team with accurate risk information.

[0126] Optionally, a data monitoring module can be set up in the project management system to monitor changes in project data in real time. For example, for project progress data, deviations between actual and planned progress can be checked periodically; for resource data, changes in resource usage and availability can be monitored. Furthermore, change thresholds can be set for different types of project data. When data changes exceed these thresholds, it is considered a change requiring attention. For example, when project costs change by more than a certain percentage of the budget, the task of establishing a dynamic risk decomposition structure is triggered to be re-executed.

[0127] When the system confirms that the project data of the target project has changed, it can first collect the changed project data, including but not limited to data on project scope, schedule, cost, quality, resources, etc.; then clean and preprocess this data to ensure its accuracy and completeness; then, based on the changed project data and the previous risk analysis results, re-identify, assess, and analyze risks to identify new risk factors and assess the likelihood and impact of risks; finally, based on the new risk analysis results, reconstruct the dynamic risk decomposition structure, determine the new risk hierarchy, risk categories, and risk relationships, and update the node and edge information in the risk decomposition structure.

[0128] In summary, the project management system of this application utilizes the generalization and reasoning capabilities of a large language model to dynamically generate or adapt a project-specific dynamic risk decomposition structure based on project-specific data. This structure serves as the thinking framework for the system's intelligent agent, guiding the system to systematically traverse all potential risks within the project. Compared to traditional static rule bases, it can more accurately and promptly identify project risks, ensuring the stable progress of projects. Furthermore, the knowledge graph, constructed based on historical data from past projects, integrates into the intelligent agent, supporting analogical reasoning, risk priority calculation, and root cause analysis, thereby contributing to improved accuracy and reliability of the final risk analysis results.

[0129] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0130] Based on the same inventive concept, this application also provides a project risk analysis device for implementing the project risk analysis method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more project risk analysis device embodiments provided below can be found in the limitations of the project risk analysis method described above, and will not be repeated here.

[0131] In one embodiment, such as Figure 7 As shown, a project risk analysis device is provided, including: a risk analysis plan generation module 702, a risk decomposition structure determination module 704, a risk priority number determination module 706, and a risk analysis result determination module 708, wherein:

[0132] The risk analysis plan generation module 702 is used to receive the risk analysis task of the target project and call the first preset large language model to generate a parsable analysis plan based on the context parameters of the risk analysis task and the target project. The analysis plan includes a dynamic risk decomposition structure establishment task, a risk category traversal task, and a risk assessment task.

[0133] The risk decomposition structure determination module 704 is used to perform the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project.

[0134] The risk priority number determination module 706 is used to perform a risk category traversal task based on the dynamic risk decomposition structure, and perform a risk assessment task for at least one risk category traversed to obtain a risk priority number.

[0135] The risk analysis result determination module 708 is used to generate risk analysis results based on the risk priority number.

[0136] In one embodiment, the risk decomposition structure determination module 704 is further configured to: acquire multi-dimensional data of the target project; the multi-dimensional data includes structured data and unstructured text data; preprocess the multi-dimensional data to obtain at least one key prompt word; input the key prompt word into a first preset large language model to obtain a dynamic risk decomposition structure.

[0137] In one embodiment, the risk priority number determination module 706 includes a search submodule and a risk priority number determination submodule. The search submodule is used to search for historical failure mode subgraphs related to risk categories from a preset knowledge graph based on risk categories. The risk priority number determination submodule is used to call a second preset large language model to obtain the risk priority number based on the historical failure mode subgraphs and the context parameters of the target project.

[0138] In one embodiment, the risk priority number determination submodule includes a candidate information extraction unit and a risk priority number determination unit. The candidate information extraction unit is used to extract candidate information from the historical failure mode subgraph. The candidate information includes failure mode, failure impact, root cause and countermeasures. The risk priority number determination unit is used to input the candidate information and the context parameters of the target project into a second preset large language model to obtain the risk priority number.

[0139] In one embodiment, the risk priority determination unit includes a reasoning result determination subunit and a risk priority determination subunit. The reasoning result determination subunit is used to input candidate information and context parameters of the target project into a second preset large language model, so that the second preset large language model can perform reasoning on failure modes, failure effects, root causes and countermeasures to obtain reasoning results. The risk priority determination subunit is used to quantify and score severity, occurrence and detectability through the second preset large language model based on the reasoning results, and determine the risk priority number based on the product of severity, occurrence and detectability.

[0140] In one embodiment, the risk priority determination module 706 is further configured to: acquire historical data of historical projects and perform entity identification and relationship extraction processing on the historical data; the historical data includes historical failure modes, historical failure effects, historical root causes and historical countermeasures; store the processed entities and relationships in a graph database; the node types of the graph database include failure modes, root causes, possible effects and countermeasures; the edges between nodes in the graph database are used to represent one of causal relationships or mitigation relationships.

[0141] In one embodiment, the device is further configured to: when the project data of the target project changes, based on the risk analysis results and the changed project data, re-execute the dynamic risk decomposition structure establishment task to obtain a new dynamic risk decomposition structure.

[0142] Each module in the aforementioned project risk analysis device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0143] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes 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, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a project risk analysis method. 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.

[0144] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0145] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0146] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0147] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0149] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0150] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A project risk analysis method, characterized in that, The method includes: The system receives a risk analysis task for the target project and invokes a first preset large language model to generate a parsable analysis plan based on the risk analysis task and the context parameters of the target project. The analysis plan includes a dynamic risk decomposition structure establishment task, a risk category traversal task, and a risk assessment task. Execute the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project; Based on the dynamic risk decomposition structure, the risk category traversal task is executed, and for at least one risk category traversed, the risk assessment task is executed to obtain the risk priority number; Based on the risk priority number, generate risk analysis results.

2. The method according to claim 1, characterized in that, The process of executing the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project includes: Obtain multi-dimensional data of the target project; the multi-dimensional data includes structured data and unstructured text data; The multi-dimensional data is preprocessed to obtain at least one key prompt word; The key prompts are input into the first preset large language model to obtain the dynamic risk decomposition structure.

3. The method according to claim 1, characterized in that, The step of performing the risk assessment task for at least one risk category encountered during the traversal to obtain a risk priority number includes: Based on the risk category, search for historical failure mode subgraphs related to the risk category from a preset knowledge graph; Based on the historical failure mode subgraph and the context parameters of the target project, the second preset large language model is invoked to obtain the risk priority number.

4. The method according to claim 3, characterized in that, The step of invoking a second preset large language model to obtain the risk priority number based on the historical failure mode subgraph and the context parameters of the target project includes: Candidate information is extracted from the historical failure mode subgraph; the candidate information includes failure mode, failure impact, root cause, and countermeasures; The candidate information and the context parameters of the target project are input into the second preset large language model to obtain the risk priority number.

5. The method according to claim 4, characterized in that, The step of inputting the candidate information and the context parameters of the target project into the second preset large language model to obtain the risk priority number includes: The candidate information and the context parameters of the target project are input into the second preset large language model so that the second preset large language model can perform inference on failure mode, failure impact, root cause and countermeasures to obtain the inference result; Based on the reasoning results, the severity, occurrence, and detectability are quantitatively scored using the second preset large language model, and the risk priority number is determined based on the product of the severity, occurrence, and detectability.

6. The method according to claim 3, characterized in that, The method for constructing the preset knowledge graph includes: Acquire historical data from historical projects and perform entity recognition and relationship extraction processing on the historical data; the historical data includes historical failure modes, historical failure impacts, historical root causes, and historical countermeasures; The processed entities and relationships are stored in a graph database; the node types of the graph database include failure modes, root causes, potential impacts, and countermeasures; the edges between nodes in the graph database are used to represent either causal relationships or mitigation relationships.

7. The method according to claim 1, characterized in that, The method further includes: If the project data of the target project changes, the dynamic risk decomposition structure establishment task is re-executed based on the risk analysis results and the changed project data to obtain a new dynamic risk decomposition structure.

8. A project risk analysis device, characterized in that, The device includes: The risk analysis plan generation module is used to receive the risk analysis task of the target project, and call the first preset large language model to generate a parsable analysis plan based on the risk analysis task and the context parameters of the target project; the analysis plan includes a dynamic risk decomposition structure establishment task, a risk category traversal task, and a risk assessment task; The risk decomposition structure determination module is used to execute the dynamic risk decomposition structure establishment task to obtain the dynamic risk decomposition structure of the target project. The risk priority number determination module is used to perform the risk category traversal task based on the dynamic risk decomposition structure, and to perform the risk assessment task for at least one risk category traversed to obtain the risk priority number; The risk analysis result determination module is used to generate risk analysis results based on the risk priority number.

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 method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images