Risk quantification and dynamic evaluation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a risk quantification and dynamic evaluation method and device, electronic equipment and a storage medium, and the method comprises the steps: accessing risk related data from a plurality of data sources, and carrying out the preprocessing, so as to obtain risk feature information; calculating a risk weight based on an improved entropy method in combination with the risk diffusion factor and the risk correlation degree; calculating an association degree matrix among the risks by using a time sequence analysis model, and generating a conduction probability through iterative simulation by taking the association degree matrix as an input parameter of conduction simulation; generating a dynamic risk matrix according to the risk weight and the conduction probability, and displaying the dynamic risk matrix in a visual mode; monitoring data change in a data source in real time, triggering recalculation of the risk weight when detecting that a risk state is changed, and dynamically adjusting simulation parameters; quantitative and dynamic assessment of risks can be realized, and the defects of subjective deviation, weak multi-risk association analysis capability, lack of dynamic adaptability and the like of a traditional risk assessment method are overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent project management, and in particular to a risk quantification and dynamic assessment method, device, electronic device and storage medium. Background Art

[0002] In modern project management and business operations, risk assessment is a key step in ensuring the smooth implementation of projects and the stable development of enterprises. Current risk assessment methods mainly include the following:

[0003] Expert experience method: This method relies on the experience and subjective judgment of experts, such as the fuzzy analytic hierarchy process (AHP). Although this method can reflect the expertise of experts to a certain extent, it is easily affected by personal bias, resulting in inconsistent evaluation results.

[0004] Static assessment methods: These methods are typically based on historical data and statistical analysis, but struggle to address the complex relationships between multiple risks, particularly the coupling effects between risks. Furthermore, these methods fail to reflect real-time changes in the project environment and lack dynamic adaptability.

[0005] Traditional quantitative methods: Although some existing quantitative methods can provide certain risk quantification results, they are often unable to process unstructured data (such as text descriptions), which limits their application in complex projects.

[0006] While current methods have made some progress in risk assessment, they still suffer from numerous deficiencies in practical applications, such as significant subjective bias, weak multi-risk correlation analysis capabilities, and a lack of dynamic adaptability. These issues are particularly acute in complex project environments and require urgent improvement. Summary of the Invention

[0007] The main purpose of the present invention is to provide a risk quantification and dynamic assessment method, device, electronic device and storage medium, which can provide a more objective, accurate and dynamic risk assessment method.

[0008] To achieve the above objectives, this application provides a risk quantification and dynamic assessment method, including:

[0009] Access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information;

[0010] Calculating risk weights based on an improved entropy method combined with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on text features in the risk feature information;

[0011] Calculating the correlation matrix between risks using a time series analysis model, and using the correlation matrix as an input parameter for a transmission simulation. Generating transmission probabilities through iterative simulation, wherein the transmission probabilities represent the probability distribution of risk transmission paths.

[0012] Generate a dynamic risk matrix based on the risk weights and the transmission probability, and display it in a visual manner;

[0013] The data changes in the data source are monitored in real time. When a change in the risk status is detected, the risk weight is recalculated and the simulation parameters are dynamically adjusted.

[0014] On the other hand, the present application provides a risk quantification and dynamic assessment device, comprising:

[0015] A data access module is used to access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information;

[0016] A weight calculation module, configured to calculate risk weights based on an improved entropy method in combination with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on text features in the risk feature information;

[0017] A transmission simulation module is used to calculate the correlation matrix between risks using a time series analysis model, and use the correlation matrix as an input parameter for transmission simulation to generate transmission probabilities through iterative simulation, wherein the transmission probability represents the probability distribution of the risk transmission path;

[0018] A matrix generation module, configured to generate a dynamic risk matrix based on the risk weight and the transmission probability, and to display the matrix in a visual manner;

[0019] The real-time monitoring module is used to monitor the data changes in the data source in real time. When a change in the risk status is detected, it triggers the recalculation of the risk weight and dynamically adjusts the simulation parameters.

[0020] On the other hand, the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the first aspect and any possible implementation thereof.

[0021] On the other hand, the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the method described in the first aspect.

[0022] The present application provides a risk quantification and dynamic assessment method, device, electronic device and storage medium, which obtain risk characteristic information by accessing risk-related data from multiple data sources and preprocessing the data; calculating risk weights based on an improved entropy method combined with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on text features in the risk characteristic information; calculating the correlation matrix between risks using a time series analysis model, and using the correlation matrix as an input parameter for conduction simulation, generating a conduction probability through iterative simulation, wherein the conduction probability represents the probability distribution of the risk conduction path; generating a dynamic risk matrix based on the risk weights and the conduction probability, and displaying it in a visual manner; monitoring data changes in the data source in real time, and triggering the recalculation of the risk weights and dynamically adjusting the simulation parameters when a change in the risk status is detected; which can significantly improve the objectivity, relevance, dynamism and visualization capabilities of risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] in:

[0025] Figure 1 A flow chart of a risk quantification and dynamic assessment method provided in an embodiment of the present application;

[0026] Figure 2 A schematic diagram of an LSTM model architecture provided in an embodiment of the present application;

[0027] Figure 3 A flow chart of a dynamic update mechanism provided in an embodiment of the present application;

[0028] Figure 4 A schematic diagram of the entire process of risk quantification and dynamic assessment provided in an embodiment of the present application;

[0029] Figure 5 A schematic diagram of the structure of a risk quantification and dynamic assessment device provided in an embodiment of the present application;

[0030] Figure 6 A schematic diagram of a risk quantification engine architecture provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0032] The terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0033] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0034] In order to illustrate the method steps and effects in this application in combination with application scenarios, this application provides Case 1 and Case 2 for reference, see the subsequent description for details.

[0035] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.

[0036] See also Figure 1 , is a flow chart of a risk quantification and dynamic assessment method provided in an embodiment of the present application, such as Figure 1 As shown in FIG, the risk quantification and dynamic assessment method includes:

[0037] 101. Access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information.

[0038] The embodiment of the present application proposes a risk quantification and dynamic assessment method. The execution subject of the method can be an electronic device, such as a computer, in practical applications.

[0039] First, the multi-source data access and preprocessing in the embodiment of the present application are introduced.

[0040] In an optional implementation, the multiple data sources include:

[0041] Project management tools that provide probability and impact values ​​for risks;

[0042] Document management system, used to provide text information describing risks;

[0043] Manual entry system to provide customized risk indicators;

[0044] The above preprocessing includes:

[0045] Normalize the structured data and map the probability and influence value to the [0,1] interval;

[0046] Use natural language processing technology to vectorize text features.

[0047] Specifically, risk-related data is accessed from the following multiple data sources:

[0048] Project management tools: Provides risk probability and impact values. For example, extracts the frequency and associated impact assessment of change tickets from JIRA (JSON format).

[0049] Document management systems (such as Confluence): Provide textual information describing risks. For example, extract detailed descriptions of requirement changes from Confluence documents (CSV format).

[0050] Manual entry system: Provides customized risk indicators. For example, project managers can enter indicators such as budget fluctuation thresholds and critical path delays through Excel.

[0051] The data may be collected periodically, for example, 5 times per second in real time; the data types may include:

[0052] Structured data: risk probability, impact value (numeric type);

[0053] Unstructured data: IRA ticket text description (e.g., "Requirement changes lead to code refactoring delays")

[0054] Optionally, the risk documents and customized risk indicators in the data access module adopt a structured template + dynamic expansion mechanism. The preset template ensures data parsability, and the dynamic expansion adapts to different project needs. For example:

[0055] 1. JIRA risk record (JSON format)

[0056] Default fields:

[0057] {

[0058] "risk_id":"REQ-042",

[0059] "probability": 0.7, / / Automatically count the frequency of work orders

[0060] "impact_level":"High",

[0061] "description":"Requirement changes lead to code refactoring delays"

[0062] }

[0063] Dynamic extension: allows custom label fields (such as critical_path: true) for path weight correction in Monte Carlo simulation.

[0064] 2. Confluence Risk Document (CSV format)

[0065] The preset template may include document ID, associated risk ID, text content, and semantic analysis tags;

[0066] Dynamic update: trigger re-parsing when the document version changes (e.g. interval ≤ 15 minutes)

[0067] 3. Customized risk indicators (manual entry in Excel)

[0068] The preset columns can be found in Table 1:

[0069] Indicator name Numeric Types Influence coefficient Project Phases Budget floating threshold ≤10％ 0.8 Full cycle Critical path delay ≤15 days 1.2 Implementation phase

[0070] Table 1

[0071] Dynamic adjustment: Project managers can add or delete rows at any time, but they must comply with data type constraints (such as impact coefficient ∈ [0, 2]).

[0072] The threshold settings can be:

[0073] Budget fluctuation ≤ 10% (regular projects) vs ≤ 20% (high-risk projects)

[0074] Modifying the budget floating threshold through Excel directly affects the impact value calculation in the entropy method. For example: when the actual budget floats by 12%, if the threshold = 10% → I norm =1.2 (over-limit alarm), please refer to the subsequent description for details.

[0075] Specifically, preprocessing can mainly include:

[0076] Numerical normalization: Map the probability / influence value to the [0,1] interval. The following formula can be used:

[0077]

[0078] Among them, P norm is the normalized probability value, which is mapped to the interval [0,1]; P raw is the probability value in the original data; P min is the minimum probability value in the data set; P max is the maximum probability value in the data set;

[0079] Text feature extraction: Use the BERT model to convert text into high-dimensional vectors (e.g., 768 dimensions) for subsequent analysis. Example: Encode "technical debt accumulation" as [0.23, -0.56, ..., 0.78].

[0080] BERT (Bidirectional Encoder Representations from Transformers) mentioned in the embodiments of this application is a pre-trained language model based on the Transformer architecture, which can convert text into a vector representation of fixed dimension.

[0081] 102. Calculate the risk weight based on the improved entropy method in combination with the risk diffusion factor and the risk correlation. The risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on the text features in the risk feature information.

[0082] The risk weight in the embodiment of the present application can be used as a basis for resource allocation. The higher the weight, the higher the priority of resource investment.

[0083] The basic weight is calculated as follows:

[0084]

[0085] On this basis, an improved entropy method is used in the embodiments of the present application, and a risk diffusion factor is introduced to correct the weight calculation.

[0086] In an optional embodiment, the above improved entropy method formula is:

[0087] W i =(1-E i ) / ∑(1-E j )+λ·Corr(R i , R j )

[0088] Among them, W i is the final weight of the ith risk (i.e. the above risk weight), E i is the information entropy of the ith risk, λ is the risk diffusion factor, Corr(R i , R j) is the risk correlation degree mentioned above, which indicates the strength of the correlation between risks.

[0089] For example, the simulation experiment data table for verifying the theoretical value of the risk diffusion factor λ is as follows:

[0090]

[0091]

[0092] Table 2

[0093] In this application, a long short-term memory (LSTM) network can be used to analyze the extracted text features and calculate the correlation between risks. LSTM is a special type of recurrent neural network (RNN) that can effectively process long-term dependencies in time series data.

[0094] Specifically, the correlation function is calculated using an LSTM neural network. The text feature vectors extracted by BERT serve as the LSTM input (the text description feature vectors of risks i and j). For each risk, the corresponding text feature vector is input. The LSTM model is trained using historical data, enabling it to learn the correlations between risks. Training data can include text descriptions of historical risk events and their correlations. The LSTM model outputs the correlation between each risk and other risks. The correlation is a value between 0 and 1 that indicates the strength of the correlation between two risks. For example, the correlation between requirement changes and technical debt might be 0.65.

[0095] That is, in specific applications, Corr(R i , R j ) function can analyze the text description of the JIRA work order / risk register through the LSTM network and output the correlation coefficient in the range [0,1].

[0096] To further improve the accuracy of correlation calculation, the attention mechanism can be combined. The attention mechanism can automatically assign weights between different risk text features and highlight the impact of important features on correlation.

[0097] Attention weight calculation: The attention mechanism is introduced into the LSTM model to calculate the attention weight of each time step. The attention weight indicates the contribution of the feature of the current time step to the overall relevance.

[0098] Figure 2 A schematic diagram of an LSTM model architecture provided in an embodiment of the present application. Figure 2 As shown in Figure 2, the model structure and its functions specifically include:

[0099] 1. Risk text input: Receive the original text of the JIRA ticket / risk register (Case 1 input data)

[0100] 2. Text segmentation: Use BERT tokenizer for subword segmentation)

[0101] 3. BERT feature extraction layer: load the pre-trained BERT-base model to generate text features

[0102] 4.768-dimensional feature vector: Outputs a context-dependent text representation with a dimension of 768

[0103] 5. LSTM hidden layer: Bidirectional LSTM with 128 hidden units to process time series features

[0104] 6. Context encoding: Generates an encoding vector containing sequence position information

[0105] 7. Attention mechanism module: Calculating the association weights between different risk texts

[0106] 8. Association weight assignment: Output attention weight matrix α_ij∈[0,1] (0.65 association case of case 1)

[0107] 9. Fully connected layer: 128 → 64 → 1 dimension transformation network (model structure details)

[0108] 10. Sigmoid activation: compresses the output to the interval [0,1];

[0109] 11. Normalized output: finally generate Corr(R_i,R_j)∈[0,1] correlation coefficient;

[0110] 12. Correlation coefficient: output to the Monte Carlo module for path probability calculation;

[0111] 13. Monte Carlo module: receives correlation data for risk diffusion simulation;

[0112] 14. Attention visualization: Displays a heat map of attention for "Requirement Change → Technical Debt" (Case 1 effect);

[0113] 15. Debug console: allows manual modification of attention weights (interactive function).

[0114] Furthermore, the risk correlation calculation results are dynamically adjusted based on real-time data. For example, when a new risk event occurs, the correlation is recalculated to reflect the latest risk relationship.

[0115] For example, in an ERP project, the risk weight of requirement changes is calculated as follows:

[0116] The basic weight is 0.38 (calculated based on information entropy);

[0117] The correlation between requirement changes and technical debt is 0.65, calculated using the LSTM network;

[0118] The risk weight of the final demand change is 0.42 (0.38+0.04).

[0119] The value of the risk diffusion factor λ can be determined through a dynamic optimization mechanism. The value selection rules are as follows:

[0120] 1. Experimental verification of benchmark values

[0121] Conduct multiple groups of experiments (11 groups are used as an example) in the interval λ∈[0,0.5] with a step size of 0.05;

[0122] Input a data set, such as an ERP project risk data set, and calculate the quantitative accuracy under different λ values;

[0123] Select the interval with the highest quantization accuracy and stable variance. For example, in this application, λ∈[0.1,0.3] can be set. Optionally, a default value of λ can be set as needed. For example, in this application, a peak value λ=0.2 can be selected as the default value.

[0124] Figure 3 The following is a graph showing the relationship between the λ value and the quantization accuracy provided in the embodiment of the present application, and the above-mentioned λ value setting basis is explained based on the graph. Figure 3 As shown, where:

[0125] X-axis: λ value (0-0.5); Y-axis: quantization accuracy (%);

[0126] This curve is characterized by a peak at λ=0.2 (88% accuracy) and a subsequent decrease.

[0127] Test environment: Intel Xeon 8-core CPU / 64GB memory, dataset contains 500+ historical risk events.

[0128] For example, the experimental process may include:

[0129] 1. Experimental parameter configuration. Set the test range of λ∈[0,0.5] and the step size of 0.05;

[0130] 2. λ range setting. Generate 11 groups of experimental parameters: 0, 0.05, 0.10, ..., 0.50; (as shown in Table 2 experimental group)

[0131] 3. Load the ERP dataset. Input training data containing 500+ historical risk events (data specifications for Case 1);

[0132] 4. Historical risk event input. Structured data can include probability, impact, and text description fields;

[0133] 5. Quantitative accuracy calculator. Compare the system output with the actual risk occurrence;

[0134] 6. Accuracy comparison analysis. Calculate the accuracy value under each λ value ( Figure 3 Y-axis data generation logic);

[0135] 7. Coefficient of variance check. Verify that β = σ / μ < 0.01 (β value requirements in Table 2);

[0136] 8. Determine the stability of the results. Screen the data set with a qualified coefficient of variance (qualified mark λ = 0.1-0.3 in Table 3);

[0137] 9. Optimal interval selection. Determine λ∈[0.1,0.3] as the optimal interval for accuracy and stability;

[0138] 10. Curve data generation. Construct a set of {(λ, precision)} data points, including 88% peaks at λ = 0.2 ( Figure 3 Curve data);

[0139] 11. Accuracy peak mark, highlighting the highest accuracy point of λ=0.2 in the visual interface ( Figure 3 Peak mark);

[0140] 12. Visual rendering. Generate interactive curve charts and support hovering data points to view details ( Figure 3 illustrate)

[0141] 13. Experimental verification report. Output includes a summary of indicators such as accuracy, variance, and hardware time consumption (Case 1 technical effect);

[0142] 14. Hardware resource configuration: Deploy an Intel Xeon 8-core / 64GB memory computing environment (test environment);

[0143] 15. Computing cluster. Distributed processing of 11 sets of parallel computing tasks for lambda values;

[0144] 16. Parallel computing instructions. Start multiple sets of quantitative precision calculation processes at the same time;

[0145] 17. Parameter adjustment signal. For example, when β ≥ 0.01, the λ value reconfiguration is triggered (dynamic optimization mechanism).

[0146] The core logic closed loop in the above process includes:

[0147] Experimental optimization closed loop: 7→8→17→1, realizing dynamic tuning of λ value (corresponding to experimental verification method);

[0148] Resource scheduling closed loop: 14 → 15 → 16 → 5, ensuring large-scale computing efficiency (based on test environment configuration);

[0149] Data verification closed loop: 5→6→7→8→9, forming a double verification of accuracy and stability.

[0150] Table 2 is a schematic diagram of a dynamic adjustment mechanism for risk diffusion factors provided in an embodiment of the present application.

[0151] Adjust the scene Technical Implementation Industry characteristics adaptation For financial projects, λ = 0.25 (high risk transmission), and for infrastructure projects, λ = 0.15 (low correlation) Real-time data feedback When the Monte Carlo simulation variance coefficient β>0.01, the λ value is automatically lowered by 0.05 User-defined Manually enter the lambda value through the debug console, limited to the range [0.1, 0.3]

[0152] Table 2

[0153] Further, the invention is described below with reference to specific embodiments.

[0154] Take the ERP project (default λ = 0.2) as an example:

[0155] Value selection basis: Data shows that when λ = 0.2, the quantization accuracy is 88% (highest);

[0156] Coefficient of variance β = 0.007 (most stable);

[0157] Calculation formula: W i =(1-E i ) / ∑(1-E j )+0.2·Corr(R i , R j );

[0158] The weight of demand change increased from 0.38 in the traditional entropy method to 0.42 (+10.5%)

[0159] Taking the infrastructure project (dynamic λ = 0.18) as an example:

[0160] Adjustment reason: The transmission speed of geological risk is relatively fast (the transmission time from typhoon suspension to steel price increase is less than 2 weeks), so λ needs to be lowered to suppress overfitting;

[0161] System automatic optimization: λ was reduced from 0.2 to 0.18, and the variance coefficient β was reduced from 0.011 to 0.009.

[0162] In one embodiment, the LSTM network includes an attention mechanism module that can calculate the association weight of risk text.

[0163] 103. Use the time series analysis model to calculate the correlation matrix between risks, and use the above correlation matrix as the input parameter of the transmission simulation. Generate the transmission probability through iterative simulation. The above transmission probability represents the probability distribution of the risk transmission path.

[0164] The above iterative simulation may be a Monte Carlo simulation.

[0165] In an embodiment of the present application, Monte Carlo risk diffusion simulation is a key module in the risk quantification and dynamic assessment device (system), which is used to quantify the coupling effects between multiple risks and generate the probability distribution of risk transmission paths.

[0166] The Monte Carlo risk diffusion simulation in the embodiments of the present application is introduced as follows, wherein the specific numerical values ​​involved are only examples and can be adjusted as needed.

[0167] Parameter settings:

[0168] The input data for Monte Carlo simulation include:

[0169] Improved entropy method weight: The risk weight calculated by the improved entropy method reflects the relative importance of each risk.

[0170] LSTM correlation matrix: The correlation matrix between risks is calculated using the LSTM network combined with the attention mechanism, indicating the strength of the correlation between risks.

[0171] In an alternative embodiment, 10,000 iterations of the Monte Carlo simulation satisfy:

[0172] The variance coefficient of the risk transmission path probability β<0.01;

[0173] The output results include the Pearson correlation coefficient matrix of risk association.

[0174] The number of iterations of the Monte Carlo simulation can be set to a desired value, for example, 10,000 times, to ensure that the variance coefficient β of the simulation results is less than 0.01, thereby ensuring the stability and reliability of the simulation results.

[0175] Path probability calculation:

[0176] Monte Carlo simulation is used to generate the probability distribution of risk transmission paths. The simulation output is a triple: {source_risk:target_risk:probability}, which represents the transmission probability from one risk to another.

[0177] For example, in Case 1 mentioned later, the probability of the transmission path from requirement change to technical debt is 0.65, and the confidence level is 0.96. The specific output example is as follows:

[0178] {

[0179] "source":"Requirement Change",

[0180] "target":"Technical Debt",

[0181] "probability":0.65,

[0182] "confidence":0.96

[0183] }

[0184] Based on the results of the Monte Carlo simulation, a Bayesian network topology is constructed to generate a directed acyclic graph (DAG). The steps for constructing a Bayesian network are as follows:

[0185] 1. Node definition:

[0186] Each node represents a risk, and its attributes include risk ID and weight. The weight reflects the relative importance of the risk among all risks.

[0187] 2. Edge definition:

[0188] Each edge represents a transmission path between risks. Edge attributes include transmission probability and correlation. The transmission probability is calculated using Monte Carlo simulation, while the correlation is calculated using an LSTM network.

[0189] For example, in Case 1, the transmission path from requirement change to technical debt can be expressed as:

[0190] node:

[0191] Requirement change: The risk ID is "R001" and the weight is 0.42.

[0192] Technical debt: The risk ID is "R002" and the weight is 0.28.

[0193] side:

[0194] The edge from requirement change to technical debt has a transmission probability of 0.65 and a correlation of 0.65.

[0195] In the embodiment of the present application, the accuracy of the correlation can be verified through Monte Carlo simulation. If the deviation between the simulation result and the actual risk transmission path exceeds a certain threshold (such as 5%), the parameters of the LSTM model are adjusted to optimize the correlation calculation.

[0196] Monte Carlo risk diffusion simulations quantify the coupling effects between multiple risks, generate probability distributions of risk transmission paths, and construct Bayesian network topologies. This approach not only considers the correlations between risks but also ensures the stability and reliability of the results through extensive iterative simulations, providing a scientific and comprehensive basis for risk assessment in project management and business operations.

[0197] 104. Generate a dynamic risk matrix based on the above risk weights and the above transmission probability, and display it in a visual manner.

[0198] In the embodiment of the present application, dynamic risk matrix generation is the core link in the risk quantification and dynamic assessment system, which is used to integrate and visualize information such as risk weights and transmission probabilities so that users can quickly understand and make decisions.

[0199] The input data includes the following categories:

[0200] 1. Risk weight vector: This is the risk weight vector calculated using the improved entropy method. For example, in an ERP project, the risk weight vector might be [0.42, 0.28, 0.18, 0.12], where 0.42 represents the risk weight for requirement changes, 0.28 represents the risk weight for technical debt, and so on.

[0201] 2. Correlation matrix: The risk correlation matrix calculated by the LSTM network. For example:

[0202]

[0203] It means that the correlation between requirement changes and technical debt is 0.65, the correlation between requirement changes and personnel turnover is 0.45, etc.

[0204] 3. Monte Carlo path probability distribution: The probability distribution of risk transmission paths generated by Monte Carlo simulation. For example:

[0205] {source:"Requirement Change",target:"Technical Debt",probability:0.65}

[0206] It indicates that the probability of transmitting requirement changes to technical debt is 0.65.

[0207] 4. Real-time risk status data: Risk status data collected in real time from project management tools (such as JIRA), including risk probability and impact value. For example:

[0208] \{\text{risk_id:"Requirement change",probability:0.7,impact:0.8}\}

[0209] In an optional embodiment, the dynamic risk matrix divides risk levels according to the ISO 31000 standard and introduces a conduction velocity dimension, which is obtained by historical data regression or prediction of the time series analysis model.

[0210] The risk value mentioned in the examples of this application is a comprehensive quantitative indicator used to represent the overall impact of a risk. It is usually calculated by combining the probability (Probability, P) and impact (I) of the risk, reflecting the likelihood of the risk occurring and its potential impact on the project or system.

[0211] The data standardization and risk value calculation steps involved may include:

[0212] Normalize probability and impact values: Normalize the probability values ​​output by the Monte Carlo simulation and the impact values ​​in the real-time risk status data to the range [0, 1]. For example, a probability value of 0.7 is normalized to 0.7, and an impact value of 0.8 is normalized to 0.8.

[0213] Risk value calculation: Calculate the risk value of each risk according to ISO 31000 standard.

[0214] Specifically, in the embodiment of the present application, a dynamic risk value formula may be used:

[0215] R score =W i ×(0.6P norm +0.4I norm )

[0216] Among them, W i is the risk weight obtained by the above calculation, P norm is the normalized probability value, I level is the normalized influence value.

[0217] The above normalized probability values ​​are derived from JIRA / Monte Carlo simulations. The normalized impact values ​​are mapped to the interval [0.2, 1.0] using ISO31000 Level 5. The data are derived from expert evaluation / historical databases.

[0218] For example, based on the example of the above correlation matrix, for a requirement change:

[0219] R 需求变更 =0.42×(0.6×0.7+0.4×0.8)=0.42×0.74=0.31

[0220] R 技术债务 =0.28×(0.6×0.5+0.4×0.6)=0.21

[0221] Risk values ​​are used to comprehensively assess risk, helping decision-makers quickly identify high-risk areas. A higher risk value indicates a greater overall risk impact and requires more proactive response measures.

[0222] Table 3 is a schematic table of a five-level risk classification standard provided in an embodiment of the present application.

[0223] Risk Level Score range Color Coding Disposal strategy Low risk [0,0.2) light green Routine monitoring Low to medium risk [0.2,0.4) green Quarterly Review Medium risk [0.4,0.6) yellow Monthly special meetings Medium to high risk [0.6,0.8) orange color Biweekly emergency drills High risk [0.8,1] red Real-time response team intervention

[0224] Table 3

[0225] Based on Table 3, for example, if the risk value of technical debt is 0.1512, it belongs to the low risk range ([0,0.2)) and is therefore marked in light green.

[0226] Furthermore, users can adjust the risk level classification by dragging the threshold. Specifically, dragging the threshold (for example, changing the "medium-high risk" threshold from 0.6 to 0.55) triggers a recalculation of the weight.

[0227] Optionally, the above visualization methods include:

[0228] Use a heat map to display the above risk weights and the above transmission probability;

[0229] Use a topological network to demonstrate the above risk transmission path.

[0230] The generation of dynamic risk matrix is ​​introduced in detail below.

[0231] 1. Node generation:

[0232] Generate a technical debt node, whose attributes include risk ID, risk level, weight, and current probability.

[0233] For example, the technical debt node:

[0234] {id:"R002",level:"Low Risk",weight:0.28,probability:0.5}

[0235] 2. Edge generation:

[0236] Based on the results of the Monte Carlo simulation, the edges of the risk transmission path are generated. The attributes of the edge include the source risk ID, target risk ID, transmission probability, and correlation.

[0237] For example, the requirement changes to the technical debt side:

[0238] {source:"R001",target:"R002",probability:0.65,correlation:0.65}

[0239] 3. Topology network construction:

[0240] The nodes and edges are integrated into a directed acyclic graph (DAG) to visualize the risk transmission path.

[0241] The decision support value of the heat map in the embodiment of this application includes:

[0242] 1. Risk Positioning

[0243] Dark red nodes (such as ERP requirement changes) indicate priority processing targets;

[0244] Edges with a width greater than 8px (probability greater than 0.71) require a conduction blocking mechanism.

[0245] 2. Trend Forecasting

[0246] Color gradients reflect real-time risk evolution (e.g., typhoon shutdown nodes change from orange to red);

[0247] Three-dimensional heat map (infrastructure project) reveals the space-time conduction law;

[0248] 3. Resource optimization

[0249] Node radius guides resource allocation ratio (demand changes account for 42% of the budget);

[0250] The edge color intensity indicates the priority of joint prevention and control (red edge > yellow edge);

[0251] This heat map rendering technology realizes the "human-machine collaborative risk decision-making closed loop" in the embodiment of this application through multi-dimensional data fusion + dynamic visual coding + real-time interaction.

[0252] Further optionally, a three-dimensional dynamic matrix may also be used in the embodiment of the present application.

[0253] The method for generating the three-dimensional dynamic matrix in the embodiment of the present application is intended to more accurately reflect the characteristics of dynamic risk transmission in complex projects, and is an innovative extension of the two-dimensional risk matrix (probability × impact) under the traditional ISO 31000 standard.

[0254] Specifically, the expanded dimension: the conduction velocity (V) is added, and the calculation formula is:

[0255]

[0256] Among them, the transmission path length is the number of path steps output by the Monte Carlo simulation (for example, demand change → technical debt → personnel turnover is 2 steps); the transmission time is obtained through historical data regression or LSTM prediction (in Case 2, the transmission time from typhoon suspension to steel price increase = 2 weeks).

[0257] Three-dimensional risk value:

[0258] Among them, V max Set the maximum transmission speed for the project (V max =4 weeks -1 )

[0259] For example, for Case 2 (Infrastructure Project):

[0260] Typhoon suspension → steel price increase transmission:

[0261] Path length = 1 step (direct conduction);

[0262] Conduction time = 2 weeks (LSTM prediction);

[0263] Conduction velocity:

[0264] Rating: Medium-low risk;

[0265] Dynamic correction: When the weekly volatility of steel prices is >5% in real-time monitoring: Vreal=1 / 1=1.0→R 3D =0.3×1.0×1.0=0.3→ISO rating: medium risk;

[0266] Visualization: The color changes from green to orange, marked as the "conduction acceleration zone".

[0267] The advantages of the above method are mainly:

[0268] 1. Dynamic adaptability: Real-time update frequency of transmission speed: every 15 minutes (Case 2 obtains steel futures data through API); 3D matrix refresh delay: <8 seconds (test data).

[0269] 2. Complex risk modeling: The traditional ISO matrix cannot identify: hidden risks that are transmitted quickly (such as public opinion crises) and cumulative risks that are transmitted slowly but in a long chain (such as technical debt → staff loss → project failure).

[0270] 3. Enhanced decision support:

[0271] Risk treatment priority:

[0272] In Case 2, the typhoon shutdown was elevated to the highest priority due to a sudden increase in V, thus avoiding losses of 120 million.

[0273] The present application utilizes a layered design combining an ISO two-dimensional matrix (compliance foundation) and a three-dimensional dynamic matrix (extension) to meet international standards while overcoming the limitations of traditional approaches. The introduction of the transmission velocity dimension enables the system to more accurately reflect the dynamic transmission characteristics of risk in complex projects.

[0274] The reasons for retaining the ISO matrix in the embodiment of this application mainly include:

[0275] Compliance requirements: meeting audit and industry reporting standards;

[0276] Fast decision-making: 2D evaluation is more timely (generation time < 0.2 seconds).

[0277] The visual display in the embodiments of the present application may include but is not limited to heat map rendering, and may also support user interaction functions.

[0278] Specifically, a heat map can be used to display the risk matrix. The color and size of the node reflect the risk level and weight. The node size can be proportional to the weight, for example, radius (node ​​radius) = 10 + 30 * weight (weight), and the color and width of the edge reflect the transmission probability.

[0279] For example, the technical debt node is displayed in light green (low risk) with a radius of 15.4px (weight 0.28); the edge from requirement change to technical debt is displayed in orange-red with a width of 7.55px (probability 0.65).

[0280] Optionally, users can adjust risk classification by dragging thresholds; double-click a node to view details of the associated risk chain; and export risk topology data to third-party tools (such as Gephi) for further analysis. Exported data can be used for scenarios including third-party visualization analysis, risk transmission simulation, compliance report generation, and cross-platform data sharing.

[0281] The associated risk chain mentioned in the embodiments of the present application is a set of risk transmission paths generated through Monte Carlo simulation, which is manifested as a multi-level transmission relationship of source risk → intermediate risk → terminal risk, and each link includes path probability and association strength.

[0282] The output data in the embodiment of the present application may include but is not limited to:

[0283] Risk weight distribution (demand change weight 0.42);

[0284] Correlation matrix (requirement change → technical debt correlation 0.65);

[0285] Diffusion path probability distribution (Monte Carlo simulation confidence >95%).

[0286] The quantitative results output by the system provide a basis for full-chain decision-making for project management, from risk identification to response strategy formulation.

[0287] Further, examples of application scenarios for risk weight distribution are as follows:

[0288] 1. Resource allocation decisions

[0289] Requirements change (weight 0.42) > Technical debt (0.28) > Staff turnover (0.18) → Invest 60% of the risk management budget in requirements change prevention and control (actually reducing response costs by 30% in the example);

[0290] 2. Key points of process optimization

[0291] Implement biweekly review meetings for requirement changes (increased from once a month to twice a month);

[0292] Set a requirement freeze period (major changes are prohibited in the 3rd to 9th months of the project).

[0293] Effect: In practical applications, it can reduce the efficiency of handling demand changes and the probability of delays caused by changes.

[0294] Furthermore, the following example application scenarios for the correlation matrix (requirement change → technical debt correlation 0.65) are as follows:

[0295] 1. Coupling risk prevention

[0296] Establish a demand-technology joint management and control team;

[0297] Develop a requirement change impact assessment tool (automatically detect technical debt increments);

[0298] 2. Critical path warning

[0299] When the frequency of requirement changes is > 3 times per week, a technical debt red alert is triggered;

[0300] For risks with a correlation exceeding 0.6, emergency plans must be formulated simultaneously.

[0301] Effect:

[0302] Through correlation analysis, it was found that for every additional requirement change, the probability of technical debt growth increased by 22% (regression analysis R 2 =0.81); after the implementation of the joint prevention mechanism, the technical debt backlog decreased by 47%.

[0303] Furthermore, examples of application scenarios for diffusion path probability distribution (confidence level > 95%) are as follows:

[0304] 1. Conduction pathway blockage

[0305] Set up a "circuit breaker mechanism" on the path from requirement change to technical debt (freeze new requirements when the technical debt ratio is >15%).

[0306] To address the personnel turnover → budget overrun path, sign an outsourcing service framework agreement in advance;

[0307] 2. Dynamic monitoring strategy

[0308] Implement real-time monitoring of transmission paths with a probability greater than 0.6 (e.g., requirements → technical debt);

[0309] When the path probability rises by 10%, the emergency plan (such as increasing code review manpower) is automatically triggered.

[0310] Effect: After implementation, the probability of overspending is greatly reduced.

[0311] 105. Monitor data changes in the above data sources in real time. When a change in risk status is detected, trigger the recalculation of the above risk weights and dynamically adjust the simulation parameters.

[0312] In an optional embodiment, the real-time monitoring includes:

[0313] Collect data from project management tools at least N times per second;

[0314] Data is collected from the document management system every M minutes, where N and M are positive integers.

[0315] In an optional embodiment, the above-mentioned dynamic adjustment of simulation parameters includes:

[0316] When the variance coefficient β of the above Monte Carlo simulation is greater than 0.01, the number of iterations is automatically increased;

[0317] The calculation formula of the above risk weight is dynamically updated based on the above risk diffusion factor adjustment value input by the user.

[0318] The above N and M can be set as needed, for example, N = 5. Correspondingly, the risk level can also be updated.

[0319] The dynamic update mechanism in the embodiment of this application mainly includes two aspects:

[0320] Real-time data triggers updates:

[0321] When real-time data changes (such as a JIRA ticket status change), the system can complete data updates, weight recalculations, Monte Carlo iterations, and matrix refreshes in a short period of time. For example, in Case 1, during the execution of an ERP project, when the frequency of demand changes suddenly increases from 2 to 5 times per week:

[0322] The system completes the following steps within 8 seconds: data update → LSTM recalculation of Corr → Monte Carlo iteration → matrix refresh

[0323] The risk level of the requirement change is automatically upgraded from "medium-high risk" to "extremely high risk" (dynamic matrix threshold adaptation).

[0324] User interaction drives adjustments:

[0325] When the user adjusts the risk level threshold through the interface, the system recalculates the risk level and updates the matrix within 0.5 seconds. For example, if the user adjusts the low risk threshold from 0.2 to 0.25, the system automatically adjusts the risk level of technical debt from "low risk" to "medium-low risk."

[0326] By generating a dynamic risk matrix, the method in the embodiments of this application integrates and visualizes information such as risk weights and transmission probabilities, helping users quickly understand and make decisions. The calculation of the risk value of technical debt and its application in a dynamic risk matrix demonstrates how, by comprehensively assessing the overall impact of risk, a scientific and comprehensive risk assessment basis can be provided for project management and business operations.

[0327] Traditional risk assessment methods often have the following problems:

[0328] There is a large subjective bias and reliance on expert experience (such as fuzzy analytic hierarchy process);

[0329] The ability to analyze multi-risk correlations is weak and cannot handle coupled scenarios;

[0330] Lack of dynamic adaptability makes it difficult to cope with changes in the project environment.

[0331] The method in the embodiment of the present application systematically overcomes the problems existing in the traditional method through three innovations: improved entropy method, Monte Carlo-BERT-LSTM multi-technology fusion, and dynamic matrix interaction mechanism.

[0332] In order to more clearly illustrate the method in the embodiment of the present application, the dynamic update mechanism is further introduced below.

[0333] See also Figure 3 , Figure 3 This is a flow chart of a dynamic update mechanism provided by an embodiment of the present application. The process starts with data update, and goes through data access, weight calculation, parameter refresh, matrix rendering, and finally realizes the rapid update of the interface.

[0334] like Figure 3 As shown, the specific steps include:

[0335] JIRA Data Update: The process begins with a data update in the JIRA project management tool. This can be understood as a trigger condition, such as a JIRA ticket status change (e.g., a requirement change frequency exceeding a threshold). As a data source, JIRA provides real-time project risk information.

[0336] Data Access Module: Updated data is pre-processed and integrated through the Data Access Module. This module is responsible for receiving and cleaning risk-related data from various data sources.

[0337] API push: The preprocessed data is pushed to the improved entropy method module via the API for subsequent risk weight calculation.

[0338] Improved Entropy Method Recalculation: In this step, risk weights are recalculated using an improved entropy method combined with the risk diffusion factor and risk correlation. This step is the core of risk quantification, as it takes into account the mutual influence and correlation between risks.

[0339] Monte Carlo parameter refresh: The calculated risk weights are used to refresh the parameters of the Monte Carlo simulation. Monte Carlo simulation is used to assess the probability and path of risk transmission.

[0340] Matrix Dynamic Rendering: Based on the refreshed parameters, the risk matrix is ​​dynamically generated and visually rendered. This step integrates risk weights and transmission probabilities into the dynamic risk matrix so that users can intuitively understand the risk distribution.

[0341] Interface updates within 0.2 seconds: Finally, the system interface updates within 0.2 seconds to display the latest risk assessment results. This rapid update ensures that users can obtain the latest risk information in a timely manner, allowing them to respond quickly.

[0342] The entire flowchart reflects the dynamic adaptability and real-time nature of the method in the embodiment of the present application, and realizes the rapid assessment and update of project risks through automated data flow and calculation process.

[0343] Figure 4 This is a schematic diagram of the entire process of risk quantification and dynamic assessment provided by the embodiment of this application, which focuses on the closed loop of each sub-process. Figure 4 As shown, the process includes:

[0344] 1. Multi-source data access: real-time collection of JIRA / Asana data (5 times per second) + manual input in Excel

[0345] 2. Data preprocessing:

[0346] Numerical normalization: Probability / influence values ​​are mapped to [0,1]

[0347] Text feature extraction: BERT model generates 768-dimensional vectors

[0348] 3. Improved entropy calculation (no further details here)

[0349] 4. Monte Carlo simulation:

[0350] Input: weight + correlation matrix

[0351] 10,000 iterations generate {source:target:probability}

[0352] 5. Dynamic matrix generation:

[0353] Five risk levels according to ISO 31000

[0354] Node radius = 10 + 30 * weight, edge color gradient from red to yellow

[0355] 6. User interface interaction:

[0356] Double-click a node to expand the risk chain (e.g., requirement change → technical debt → staff turnover)

[0357] Drag threshold triggers weight recalculation (response time < 0.5 seconds)

[0358] 7. Threshold adjustment feedback: After the user modifies the risk level threshold, the entropy method parameters are updated in a closed loop

[0359] 8. Gephi data export: generate topology files (XML / GraphML) containing node weights / edge probabilities

[0360] 9. Text data input: Unstructured text of JIRA tickets / risk registers

[0361] 10. LSTM Correlation Analysis:

[0362] Calculating risk text similarity through attention mechanism

[0363] Output normalized correlation matrix Corr(R i , R j )∈[0,1]

[0364] 11. Correlation Matrix Injection: Using LSTM output as path probability input for Monte Carlo simulation

[0365] 12. Variance detection: monitor the variance coefficient β of Monte Carlo results (β<0.01 is required)

[0366] 13. Self-optimization iteration: When β>0.01, the number of iterations will be automatically increased to 15,000

[0367] Among them, some of the above steps can refer to Figure 1 or Figure 2 The detailed description of the illustrated embodiments will not be repeated here.

[0368] Furthermore, the key closed-loop verification involved in the above process may include:

[0369] Main closed loop: 3→4→5→6→7→3 (user interaction drives algorithm optimization)

[0370] Data closed loop: 5→8→4 (export data to feed back simulation parameters)

[0371] Text analysis closed loop: 1→9→10→11→4 (complete transformation from unstructured data to quantitative computing)

[0372] The following example is verified (ERP project):

[0373] Process Stage Input Data Output Timeliness Data Access JIRA requirement change record Normalized P = 0.7, I = 0.8 real time Entropy calculation λ=0.2,Corr=0.65 Weight 0.42 <1 second Monte Carlo simulation 10,000 iterations Path probability 0.65 (96% confidence) 3 minutes Matrix Generation R_score=0.56 Medium risk (yellow) 0.2 seconds to render User Interaction Drag the threshold to 0.55 Upgraded to medium-high risk (orange) 0.5 second response

[0374] Table 4

[0375] Table 4 verifies the feasibility and effectiveness of the method of the embodiment of the present application in the ERP project through specific examples, demonstrating its significant advantages in risk quantification and dynamic assessment.

[0376] In order to better illustrate the method in the embodiments of the present application, Case 1 and Case 2 are provided as examples.

[0377] Case 1: Large-scale Enterprise Resource Planning (ERP) system development project

[0378] 1. Project Background

[0379] A software development company undertook a large-scale enterprise resource planning (ERP) system development project with a 12-month project cycle and a budget of 5 million yuan. The project involved the development of multiple modules, including financial management, supply chain management, and human resources management. Due to the project's high complexity and the collaboration between multiple departments, risk management was crucial.

[0380] 2. Input data

[0381] (1) Risk List:

[0382] Requirements change (probability 0.7 / high impact)

[0383] Technical debt (probability 0.5 / medium impact)

[0384] Personnel turnover (probability 0.3 / high impact)

[0385] Budget overrun (probability 0.4 / impact)

[0386] (2) Constraints:

[0387] Budget ≤ 5 million

[0388] Cycle ≤ 12 months

[0389] 3. Implementation steps

[0390] (1) Data access:

[0391] Import risk records from JIRA (JSON format)

[0392] Import risk documents from Confluence (CSV format)

[0393] Manual entry of customized risk indicators (Excel format)

[0394] (2) Risk quantification:

[0395] Use the improved entropy method to calculate the weight of each risk

[0396] Predicting the risk diffusion path through Monte Carlo simulation (10,000 iterations)

[0397] Among them, the specific implementation of the correlation function Corr() includes:

[0398] a) Use the pre-trained BERT model to extract risk text features (such as the description of the requirement change work order);

[0399] b) Calculate the attention weights between risk pairs through the LSTM network;

[0400] c) The output layer uses the Sigmoid function to normalize the value to [0, 1].

[0401] (3) Risk matrix generation:

[0402] Generate a five-level risk matrix based on ISO 31000 standards

[0403] Visual risk matrix, support Gephi format export

[0404] 4. System output

[0405] (1) Risk weight distribution:

[0406] Requirement change: weight 0.42

[0407] Technical debt: weight 0.28

[0408] Personnel turnover: weight 0.18

[0409] Budget overrun: Weight 0.12

[0410] (2) Probability distribution of risk diffusion paths:

[0411] Requirements change → Technical debt: Probability 0.65

[0412] Technical Debt → Turnover: Probability 0.45

[0413] Staff turnover → budget overrun: probability 0.55

[0414] (3) Dynamic risk matrix:

[0415] Requirements change: high risk

[0416] Technical Debt: Medium to High Risk

[0417] Movement of people: medium risk

[0418] Budget overrun: Medium risk

[0419] 5. Technical Effect

[0420] Improved risk quantification accuracy: Compared with traditional methods (traditional entropy method), risk quantification accuracy is improved by 42%;

[0421] Reduced strategy formulation time: Reduced strategy formulation time by 72% (based on dynamic matrix optimization);

[0422] Reduced risk response costs: Expected risk response costs are reduced by 30%;

[0423] Calculating the textual correlation between requirement changes and technical debt takes 23ms (NVIDIA T4 GPU).

[0424] In combination with the method in the embodiment of the present application, the input data of Case 1 are: risk list (demand changes, technical debt), constraints (budget ≤ 500,000, cycle ≤ 6 months);

[0425] System outputs include:

[0426] Risk weight distribution (demand change weight 0.42);

[0427] Correlation matrix (requirement change → technical debt correlation 0.65);

[0428] Diffusion path probability distribution (Monte Carlo simulation confidence >95%).

[0429] Conclusion: Through the implementation of this system, project teams can more accurately identify and quantify risks, develop more effective risk response strategies, and significantly improve project success rates.

[0430] Case 2: Risk Assessment of Infrastructure Projects

[0431] 1. Project Background

[0432] A cross-sea bridge construction project with a total investment of 2 billion yuan and a construction period of 36 months involves three core risks: geological risks, material price fluctuations, and construction safety.

[0433] 2. Input data

[0434] (1) Risk List:

[0435] Geological exploration error (probability 0.6 / high impact)

[0436] Steel price fluctuations (probability 0.4 / moderate impact)

[0437] Typhoon suspension (probability 0.3 / extremely high impact)

[0438] (2) Constraints:

[0439] Budget fluctuation ≤ 10%

[0440] Critical path delay ≤ 15 days

[0441] 3. Implementation steps:

[0442] (1) Data access:

[0443] Import geological exploration data (IFC format) from BIM system;

[0444] Get steel futures prices through crawlers (real-time API access).

[0445] (2) Risk quantification:

[0446] Improved entropy method to calculate weights (geological risk weight 0.51);

[0447] The Monte Carlo simulation output shows a transmission probability of 0.72: typhoon shutdown → steel price increase.

[0448] (3) Dynamic matrix generation:

[0449] Generate a three-dimensional risk matrix (probability × impact × transmission velocity);

[0450] The visualization shows the typhoon shutdown as a "deep red" very high risk area.

[0451] 4. Technical effects:

[0452] The indicators compared with the traditional PERT method are shown in Table 5.

[0453] index This system PERT Improvement Risk warning accuracy 89％ 62％ +43.5％ Budget overspending forecast bias ±3.2％ ±8.7％ -63.2％

[0454] Table 5

[0455] Based on the description of the aforementioned method embodiment, the present application also provides a risk quantification and dynamic assessment device.

[0456] Figure 5 This is a schematic diagram of the structure of a risk quantification and dynamic assessment device provided in an embodiment of the present application. Figure 5 As shown, the risk quantification and dynamic assessment system 500 includes:

[0457] The data access module 510 is used to access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information;

[0458] A weight calculation module 520 is configured to calculate risk weights based on an improved entropy method in combination with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to a preset rule, and the risk correlation is calculated based on text features in the risk feature information;

[0459] The transmission simulation module 530 is used to calculate the correlation matrix between risks using the time series analysis model, and use the correlation matrix as an input parameter for the transmission simulation to generate the transmission probability through iterative simulation. The transmission probability represents the probability distribution of the risk transmission path;

[0460] A matrix generation module 540 is configured to generate a dynamic risk matrix based on the risk weight and the transmission probability, and to display the matrix in a visual manner;

[0461] The real-time monitoring module 550 is used to monitor the data changes in the data source in real time. When a change in the risk status is detected, it triggers the recalculation of the risk weights and dynamically adjusts the simulation parameters.

[0462] Among them, the steps performed by each module in the above risk quantification and dynamic assessment system have been Figure 1 、 Figure 2 and Figure 3 The detailed description is given in , which will not be repeated here.

[0463] In one embodiment, the system core module in the embodiment of the present application includes:

[0464] Risk weight calculation module: Integrates API data from platforms such as JIRA and Asana, supporting real-time access;

[0465] Risk Diffusion Simulation Module: Combining Monte Carlo simulation with LSTM network to predict risk transmission paths;

[0466] Matrix visualization module: outputs interactive risk matrix and supports Gephi format export.

[0467] The above Gephi format file can be interactively visualized through the following fields:

[0468] node:<risk_id,level,weight>

[0469] side:<source_id,target_id,correlation,probability> The probability field comes from the path probability distribution output by the Monte Carlo simulation.

[0470] based on Figure 5 The structure shown, Figure 6This is a schematic diagram of a risk quantification engine architecture provided in an embodiment of the present application. Figure 6 As shown in the figure, the risk quantification engine architecture includes:

[0471] 1. Multi-source data access module: Integrates JIRA / Asana API and Excel data input to achieve real-time data collection 5 times per second;

[0472] 2. Data preprocessing: Extract features from unstructured work order text and normalize numerical data to the range [0, 1];

[0473] 3. Original risk dataset: stores standardized risk indicators. The fields include: risk ID, probability, impact value, and text description.

[0474] 4. Structured data stream: Use Avro format to transmit data to ensure data type consistency;

[0475] 5. Improve the entropy calculation module (no further details here);

[0476] 6. λ correction parameter: Dynamically inject the experimentally verified diffusion factor, with the default peak value of 0.2;

[0477] 7. LSTM correlation calculation submodule: A neural network with an attention mechanism analyzes the similarity between texts such as "requirement changes" and "technical debt";

[0478] 8. Correlation matrix: Outputs n×n dimensional correlation coefficient matrix, which is used for conduction probability calculation in Monte Carlo simulation (case 1 with correlation of 0.65);

[0479] 9. Monte Carlo simulation module: performs 10,000 iterations to generate confidence intervals for the risk diffusion path;

[0480] 10. Path probability distribution: Output is a set of triples in the form of {source_risk:target_risk:probability} (output example of Case 1);

[0481] 11. Bayesian Network Builder: Convert path probabilities into a directed acyclic graph with nodes containing prior / posterior probability tables;

[0482] 12. Network topology data: including Gephi-compatible node tables (risk_id, level, weight) and edge tables (source, target, probability);

[0483] 13. Dynamic risk matrix generation module: Constructs a 5×5 matrix based on the ISO 31000 standard, supports heat map rendering and threshold adjustment;

[0484] 14. Gephi export command: triggers the export of topology data, the file formats include graphml and gexf;

[0485] 15. Visual interactive interface: supports double-clicking to view the risk chain and dragging to adjust the risk level threshold;

[0486] 16. Threshold adjustment feedback: After the user modifies the threshold, the weight recalculation is triggered within 0.5 seconds (the basis for the 72% efficiency improvement in Case 1);

[0487] 17. Risk status push: Reversely notify JIRA to update risk labels through webhooks, with a response time of <200ms (Case 1 real-time update mechanism);

[0488] 18. Real-time weight update: When λ or correlation changes, the Monte Carlo simulation parameters are dynamically updated;

[0489] The closed-loop interactive verification involved in the above architecture includes:

[0490] Main closed loop (2→3→...→15→16→5): User interface adjustments trigger core algorithm recalculation, forming a human-computer interaction closed loop;

[0491] Data synchronization closed loop (13→17→1): Matrix status changes are fed back to the data source system in real time (cross-platform linkage mechanism);

[0492] Algorithm collaborative closed loop (5→18→9): Weight updates automatically trigger Monte Carlo simulation refreshes (dynamic adaptability effect of Case 1).

[0493] In one embodiment of the present application, an electronic device is also provided. The electronic device includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the computer program will execute the following Figure 1 、 Figure 2 or Figure 3 Any step in the method embodiment shown. The electronic device may also include an input / output device, etc. In a specific embodiment, the electronic device may be a terminal device, etc.

[0494] In one embodiment, a computer-readable storage medium is further provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor executes the method steps executed by the above system.

[0495] Those skilled in the art will appreciate that all or part of the processes in the method executed by the above-mentioned system can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0496] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, 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.

[0497] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A risk quantification and dynamic assessment method, characterized in that: include: Access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information; Calculating risk weights based on an improved entropy method combined with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on text features in the risk feature information; Calculating the correlation matrix between risks using a time series analysis model, and using the correlation matrix as an input parameter for a transmission simulation. Generating transmission probabilities through iterative simulation, wherein the transmission probabilities represent the probability distribution of risk transmission paths. Generate a dynamic risk matrix based on the risk weights and the transmission probability, and display it in a visual manner; The data changes in the data source are monitored in real time. When a change in the risk status is detected, the risk weight is recalculated and the simulation parameters are dynamically adjusted.

2. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The multiple data sources include: Project management tools that provide probability and impact values ​​for risks; Document management system, used to provide text information describing risks; Manual entry system to provide customized risk indicators; The pretreatment includes: Normalize the structured data and map the probability and influence value to the [0,1] interval; Use natural language processing technology to vectorize text features.

3. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The improved entropy method formula is: W i =(1-E i ) / ∑(1-E j )+λ·Corr(R i ,R j ) Among them, E i is the information entropy of the ith risk, λ is the risk diffusion factor, Corr(R i , R j ) is the risk correlation degree.

4. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The iterative simulation is a Monte Carlo simulation; the dynamic adjustment of simulation parameters includes: When the variance coefficient β of the Monte Carlo simulation is greater than 0.01, the number of iterations is automatically increased; The calculation formula of the risk weight is dynamically updated according to the risk diffusion factor adjustment value input by the user.

5. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The dynamic risk matrix divides risk levels according to the ISO 31000 standard and introduces a conduction velocity dimension, which is obtained by historical data regression or prediction of the time series analysis model.

6. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The visualization methods include: Using a heat map to display the risk weight and the transmission probability; The risk transmission path is displayed using a topological network.

7. The risk quantification and dynamic assessment method according to claim 1, characterized in that: The real-time monitoring includes: Collect data from project management tools at least N times per second; Data is collected from the document management system every M minutes, where N and M are positive integers.

8. A risk quantification and dynamic assessment device, characterized in that: include: A data access module is used to access risk-related data from multiple data sources and pre-process the data to obtain risk characteristic information; A weight calculation module, configured to calculate risk weights based on an improved entropy method in combination with a risk diffusion factor and a risk correlation, wherein the risk diffusion factor is dynamically adjusted according to preset rules, and the risk correlation is calculated based on text features in the risk feature information; A transmission simulation module is used to calculate the correlation matrix between risks using a time series analysis model, and use the correlation matrix as an input parameter for transmission simulation to generate transmission probabilities through iterative simulation, wherein the transmission probability represents the probability distribution of the risk transmission path; A matrix generation module, configured to generate a dynamic risk matrix based on the risk weight and the transmission probability, and to display the matrix in a visual manner; The real-time monitoring module is used to monitor the data changes in the data source in real time. When a change in the risk status is detected, it triggers the recalculation of the risk weight and dynamically adjusts the simulation parameters.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.

Images