Project strategy optimization method and system and storage medium

Abstract

The embodiment of the invention discloses a project strategy optimization method and system and a storage medium, and the method comprises the steps: building a multi-objective optimization model which is a constraint model of cost, time and residual risk; an improved NSGA-II algorithm is utilized to solve the multi-objective optimization model, a Pareto optimal strategy set is generated, and the improved NSGA-II algorithm comprises an elitism strategy and a Gaussian disturbance variation strategy; a screening threshold value is dynamically adjusted according to the project real-time progress deviation, and the screening threshold value is used for controlling selection of strategies in the Pareto optimal strategy set so as to adapt to the change of the project progress; the method improves the adaptability and flexibility of strategy selection, and is suitable for a dynamic strategy optimization scene under the constraint of complex project resources.

Description

Technical Field

[0001] This invention relates to the field of intelligent project management technology, and in particular to a project strategy optimization method, system, and storage medium. Background Technology

[0002] Risk management and strategy optimization are crucial for project success. In actual project execution, project managers need to balance multiple objectives to achieve optimal resource allocation. However, traditional project management methods often struggle to simultaneously address these objectives, potentially leading to resource waste, schedule delays, or uncontrolled risks during implementation.

[0003] Furthermore, as the project environment constantly changes, project managers need to adjust project strategies based on the latest project schedule and risk profile. However, current project management tools and methods often lack sufficient flexibility and adaptability, making it difficult to meet the needs of such dynamic adjustments.

[0004] Therefore, there is an urgent need for a more adaptable, flexible, and efficient method for optimizing project strategies to achieve a balance between multiple objectives. Summary of the Invention

[0005] The main objective of this invention is to provide a project strategy optimization method, system, and storage medium that can improve the adaptability, flexibility, and efficiency of project management, and achieve a balance between cost, time, and risk.

[0006] To achieve the above objectives, this application provides a project strategy optimization method, comprising:

[0007] A multi-objective optimization model is established, which is a constraint model of cost, time, and residual risk.

[0008] The improved NSGA-II algorithm is used to solve the multi-objective optimization model and generate a Pareto optimal policy set. The improved NSGA-II algorithm includes an elite retention policy and a Gaussian perturbation mutation policy.

[0009] The screening threshold is dynamically adjusted based on the real-time progress deviation of the project. The screening threshold is used to control the selection of strategies in the Pareto optimal strategy set to adapt to changes in project progress.

[0010] On the other hand, this application provides a project strategy optimization system, including:

[0011] A multi-objective optimization modeling unit is used to establish a multi-objective optimization model, wherein the multi-objective optimization model is a constraint model of cost, time and residual risk;

[0012] The solution unit is used to solve the multi-objective optimization model using the improved NSGA-II algorithm to generate a Pareto optimal policy set. The improved NSGA-II algorithm includes an elite retention policy and a Gaussian perturbation mutation policy.

[0013] A dynamic filtering control unit is used to dynamically adjust the filtering threshold according to the real-time progress deviation of the project. The filtering threshold is used to control the selection of strategies in the Pareto optimal strategy set to adapt to changes in project progress.

[0014] In another aspect, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method described in the first aspect.

[0015] This application provides a project strategy optimization method, system, and storage medium. It establishes a multi-objective optimization model, which is a constraint model of cost, time, and residual risk. An improved NSGA-II algorithm is used to solve the multi-objective optimization model to generate a Pareto optimal strategy set. The improved NSGA-II algorithm includes an elite retention strategy and a Gaussian perturbation mutation strategy. A screening threshold is dynamically adjusted based on the real-time project schedule deviation. This screening threshold controls the selection of strategies in the Pareto optimal strategy set to adapt to changes in project schedule. This method, by combining multi-objective optimization modeling and the improved NSGA-II algorithm, achieves real-time adjustment and optimization of strategies in a dynamically changing project environment, improving the adaptability and flexibility of strategy selection. Furthermore, by dynamically adjusting the screening threshold, it ensures that the selected strategies closely match the actual project schedule and requirements. Overall, it helps improve project success rate, optimize resource allocation, reduce risk, and ultimately achieve a balance between cost, time, and risk, providing an innovative and efficient strategy optimization tool for project management. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] in:

[0018] Figure 1 A flowchart illustrating a project strategy optimization method provided in an embodiment of this application;

[0019] Figure 2 This is a schematic diagram of a project strategy optimization algorithm architecture provided in an embodiment of this application;

[0020] Figure 3 This is a schematic diagram of an improved NSGA-II algorithm provided in an embodiment of this application;

[0021] Figure 4 A schematic diagram of a Pareto front solution set provided in an embodiment of this application;

[0022] Figure 5 This is a flowchart illustrating a dynamic filtering rule adjustment method provided in an embodiment of this application. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0024] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0025] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0026] The embodiments of this application are described below with reference to the accompanying drawings.

[0027] Please see Figure 1 This is a flowchart illustrating a project strategy optimization method provided in an embodiment of this application. Figure 1 As shown, the strategy optimization methods for this project include:

[0028] 101. Establish a multi-objective optimization model. The multi-objective optimization model is a constraint model of cost, time and residual risk.

[0029] This application proposes a project strategy optimization method. The execution subject of this method can be a project strategy optimization system, which in practical applications can be an electronic device, such as a computer.

[0030] In this embodiment, a project model can be established based on project cost, project time, and project risk. The objective function of the project model can be to minimize the sum of project cost and project time, and satisfy the constraint that the project risk is less than a preset risk threshold, thereby establishing the above-mentioned multi-objective optimization model.

[0031] Among them, corresponding weighting coefficients can be set for cost and time respectively as needed.

[0032] In one optional implementation, the objective function of the above multi-objective optimization model is:

[0033] min{Σα_i C_i,Σβ_j T_j}, with the constraint R_k≤S_k;

[0034] Where C_i is the cost of the i-th strategy, T_j is the execution time of the j-th strategy, R_k is the residual risk threshold, S_k is the preset risk threshold, α_i is the cost weight coefficient, and β_j is the time weight coefficient.

[0035] The weighting coefficients can be set as needed, for example, cost weight α∈[0.1,0.5], time weight β∈[0.2,0.6].

[0036] The residual risk threshold R_k involved in this application embodiment is an important concept in project management and risk management, used to quantify the level of risk that remains in a project after implementing a specific strategy or action plan. This threshold can be set according to the nature of the project, objectives, resources, and risk tolerance to ensure that the project proceeds within an acceptable level of risk. Optionally, the aforementioned S_k can be set as needed, for example, with the constraint S_k ≥ 0.7.

[0037] In one alternative implementation, a specific method for weighting project characteristics may include:

[0038] 1. Predefined expert experience

[0039] Operating steps:

[0040] Prioritize costs, time, and risks using expert scoring cards during the project initiation phase (e.g., using the Analytic Hierarchy Process, AHP).

[0041] Map the sorting results to the weight coefficient range (α_i∈[0.1,0.5], β_j∈[0.2,0.6]), for example:

[0042] Cost priority: α_i = 0.5, β_j = 0.3;

[0043] Time priority: α_i = 0.2, β_j = 0.6;

[0044] Risk sensitive: α_i = 0.3, β_j = 0.4, and the remaining weights are allocated to the risk constraints.

[0045] 2. Historical Data-Driven Learning

[0046] Operating steps:

[0047] Extract historical project data (such as JIRA task logs and Confluence documents) and construct a feature set (project size, risk type, and resource consumption).

[0048] The weight prediction model is trained using supervised learning (such as random forest or gradient boosting), taking item features as input and outputting recommended values ​​α_i and β_j.

[0049] Adjusting the weights through simulation (e.g., 1000 Monte Carlo iterations) ensures improved Pareto solution set coverage (15% increase in literature citations).

[0050] 3. Dynamic feedback adjustment

[0051] Operating steps:

[0052] Initial stage: Based on the project size N (total number of tasks), set β = 0.1 / N, β_j = 0.6 - 0.1 * N (the larger N is, the lower the time weight).

[0053] Execution phase: Through real-time feedback of ΔP (schedule deviation rate), the cost weight is dynamically adjusted according to the formula α_i(t+1)=α_i(t)×(1+ΔP), and the time weight is adjusted synchronously in the opposite direction.

[0054] Termination condition: When the fitness change rate is <1% for 10 consecutive generations (Monte Carlo parameter), the final weight is locked.

[0055] By employing constraint priority binding, historical data learning, and dynamic feedback mechanisms, adaptive weight allocation is achieved. In practical applications, expert experience and data-driven methods can be combined to maximize the effectiveness of multi-objective optimization.

[0056] 102. The improved NSGA-II algorithm is used to solve the above multi-objective optimization model and generate a Pareto optimal policy set. The improved NSGA-II algorithm includes the elite retention policy and the Gaussian perturbation mutation policy.

[0057] In this embodiment of the application, the elite retention rate can be set to a fixed value of 10%, that is, the above-mentioned elite retention strategy is the optimal strategy of retaining 10% of the elites in each generation and directly entering the next generation population.

[0058] Furthermore, introducing Gaussian perturbation mutation enhances population diversity and avoids local optima. For example, the aforementioned Gaussian perturbation mutation strategy can apply a Gaussian distributed random perturbation to 20% of the offspring, with the standard deviation σ ranging from 0.08 to 0.12.

[0059] In an alternative implementation, the improved NSGA-II algorithm further includes using a fitness evaluation function to evaluate the fitness of the strategy;

[0060] The fitness evaluation function is B / C = (risk reduction rate) / (α·budget + β·time), where α and β are the cost weighting coefficient and the time weighting coefficient, respectively.

[0061] Specifically, the dynamic Pareto screening rule in this application mainly involves the following core modules:

[0062] Multi-objective optimization solver: Integrates genetic algorithm for Pareto optimal solution search, supporting up to 100 generations of iteration.

[0063] Fitness assessment engine: Score = B / C = (risk reduction rate) / (α·budget + β·time).

[0064] Dynamic filtering controller: Dynamically adjusts the filtering threshold based on project schedule deviations to ensure that the solution set always meets resource constraints.

[0065] 103. The screening threshold is dynamically adjusted based on the real-time progress deviation of the project. The above screening threshold is used to control the selection of the above Pareto optimal strategy set to adapt to changes in project progress.

[0066] This application employs a dynamic Pareto selection rule, allowing the system to adjust the strategy selection criteria based on the actual progress and changing conditions of the project. This dynamic adjustment capability enables flexible responses to various situations that arise during project execution, thereby improving the adaptability and effectiveness of the strategy.

[0067] In an optional implementation, step 103 includes:

[0068] Obtain the actual working hours and planned working hours of the task, and calculate the real-time progress deviation ΔP of the above project. ΔP is the sum of the differences between the actual working hours and the planned working hours divided by the total planned working hours.

[0069] The real-time progress deviation ΔP of the above projects was smoothed by exponential weighted moving average.

[0070] The filtering threshold T_k is adjusted based on the smoothed real-time progress deviation ΔP of the above projects and the learning rate α. The formula for calculating the filtering threshold T_k is T_k=T_base×(1+α·ΔP), where T_base is the baseline threshold, and the formula for calculating the learning rate α is α=0.1 / N, where N is the total number of project tasks.

[0071] Specifically, the first step is to calculate the real-time progress deviation, which is the difference between the actual progress and the planned progress. In this embodiment, ΔP can be understood as the actual / planned work hour deviation rate based on the critical path task, calculated using a sliding window algorithm (e.g., setting a window period of 7 days). The formula is:

[0072]

[0073] Where i = 1 and n are the parameters of the summation symbol (Σ), representing the summation from the first task to the nth task; n is the total number of tasks to be counted within the window period (e.g., 7 days); i is the task index, representing the number of a single task (e.g., i = 1 represents the first task, i = 2 represents the second task, and so on); T actual,i T represents actual working hours. plan,i T represents planned working hours. plan,total This indicates the total planned working hours.

[0074] This deviation rate reflects the current progress status of the project and is the main basis for dynamically adjusting the screening threshold.

[0075] To smooth out fluctuations in progress deviations and improve the stability of screening threshold adjustments, this application employs an Exponentially Weighted Moving Average (EWMA) algorithm to adjust the weights of historical data. Optionally, the decay factor γ is set to 0.9, and the window period is 7 days. This helps reduce the impact of short-term fluctuations on the screening threshold, making strategy selection more stable and reliable.

[0076] In this embodiment, an EWMA of γ = 0.9 reduces noise sensitivity while maintaining adaptability to trend changes. In other scenarios, the γ value can be redefined based on requirements and experimental verification. For example:

[0077] For projects with high-frequency fluctuations (such as agile development): γ can be reduced to 0.85–0.88, accelerating the response to schedule deviations.

[0078] For long-term stable projects (such as infrastructure projects): γ can be increased to 0.92-0.95, strengthening the reference weight of historical data.

[0079] In one alternative implementation, the method further includes:

[0080] After filtering the Pareto optimal solution set using the dynamically adjusted filtering threshold to obtain the final project strategy set, resource allocation is performed through the resource allocation engine based on the final project strategy set.

[0081] Collect the actual implementation results of the project strategy, and use the above actual implementation results to update the above baseline threshold T_base.

[0082] In this embodiment, the baseline threshold for the next cycle can be adjusted based on feedback data to achieve closed-loop optimization and ensure that the strategy selection process can self-adjust and optimize according to the actual execution effect of the project.

[0083] The dynamically adjusted screening threshold T_k controls the selection of strategies within the Pareto optimal strategy set. Only strategies that meet the current screening threshold are selected and applied to project execution. This ensures that the selected strategies can adapt to changes in the project's actual schedule and conditions.

[0084] The above steps can generate a Pareto front solution set, which includes different strategy combinations, and can also output optimized resource allocation schemes and time schedules.

[0085] Based on the description of the foregoing method embodiments, this application also provides a project strategy optimization system.

[0086] In one embodiment, the project strategy optimization system includes:

[0087] A multi-objective optimization modeling unit is used to establish a multi-objective optimization model, wherein the multi-objective optimization model is a constraint model of cost, time and residual risk;

[0088] The solution unit is used to solve the multi-objective optimization model using the improved NSGA-II algorithm to generate a Pareto optimal policy set. The improved NSGA-II algorithm includes an elite retention policy and a Gaussian perturbation mutation policy.

[0089] A dynamic filtering control unit is used to dynamically adjust the filtering threshold according to the real-time progress deviation of the project. The filtering threshold is used to control the selection of strategies in the Pareto optimal strategy set to adapt to changes in project progress.

[0090] The steps performed by each module in the aforementioned project strategy optimization system can be referenced. Figure 1 The specific descriptions of the embodiments shown are not repeated here.

[0091] Figure 2 This is a schematic diagram of a project strategy optimization algorithm architecture provided in an embodiment of this application. Figure 2The diagram illustrates a complete process for generating a multi-objective optimization strategy, including key steps such as multi-objective modeling, solving with the improved NSGA-II algorithm, and dynamic screening control, and involves related system units:

[0092] 1. Multi-objective modeling unit: Establish a multi-objective model considering cost, time, and risk.

[0093] 2. Passing constraints: Pass the constraints (such as input parameters such as budget ≤ 500,000, cycle ≤ 6 months) to the algorithm solution unit.

[0094] 3. Improved NSGA-II algorithm solution unit: The improved NSGA-II algorithm is used to solve the multi-objective model.

[0095] 4. Output Pareto solution set: Generate Pareto optimal solution set.

[0096] 5. Dynamic filtering control unit: Adjusts the filtering threshold according to real-time progress deviation.

[0097] 6. Real-time schedule deviation ΔP feedback: The strategy can be adjusted through the feedback mechanism, and the actual / planned work time deviation rate can be obtained through the JIRA interface.

[0098] 7. Weighting coefficients α_i, β_j: Define the weighting coefficients for cost and time.

[0099] 8. Cost-Time-Risk Model: The objective function can be referred to above.

[0100] 9. Generate multi-objective functions: optimization problems that balance cost, time, and residual risk.

[0101] 10. Elite retention rate of 10%: The strategy of retaining the best 10% of each generation directly into the next generation.

[0102] 11. Elite Strategy Library: Stores the current best strategy for population iteration.

[0103] 12. Injecting the next generation of population: Elite strategies and mutation strategies together constitute the new population.

[0104] 13. Gaussian perturbation σ = 0.1: Apply a random perturbation with a standard deviation σ = 0.1 to 20% of the offspring strategies.

[0105] 14. Mutation Strategy Set: Generate diverse strategies through mutation.

[0106] 15. Update offspring parameters: Adjust the cost and time parameters of the strategy.

[0107] 16. Dynamic threshold T_k adjustment: The threshold can be updated according to the aforementioned formula T_k=T_base×(1+α·ΔP).

[0108] 17. Pareto Set Filter: A strategy for filtering out sets that do not meet the current threshold.

[0109] 18. Output the final strategy set: including optimized solutions for resource allocation and time scheduling.

[0110] 19. Resource Allocation Engine: Executes strategies and allocates resources such as developers and budgets.

[0111] 20. Execution Strategy Feedback: Feedback the actual execution results to the dynamic filtering controller.

[0112] Furthermore, its logical order is explained as follows:

[0113] Starting point: From modeling unit (1), input constraints (2) to algorithm unit (3);

[0114] Algorithm flow: Algorithm unit (3) generates a solution set (4) and passes it to the filtering unit (5), while receiving feedback (6);

[0115] Model building: The modeling unit (1) passes the weight coefficients (7) to the model (8) and generates the objective function (9);

[0116] Genetic optimization: Algorithm unit (3) retains elites (10→11), injects them into the population (12), and performs mutation (13→14→15);

[0117] Dynamic adjustment: The filtering unit (5) adjusts the threshold (16), the filter (17) outputs the strategy (18→19), and the execution result is fed back (20→5).

[0118] The following sections further describe the specific processes related to the improved NSGA-II algorithm, Pareto filtering and execution, and dynamic threshold adjustment in the embodiments of this application.

[0119] Figure 3 This is a schematic diagram of an improved NSGA-II algorithm provided for an embodiment of this application. Figure 3 As shown, the specific steps include:

[0120] 1. Initialize the population: Randomly generate an initial policy set.

[0121] 2. Input constraints: Budget ≤ 500,000, cycle ≤ 6 months, etc.

[0122] 3. Fitness evaluation engine: The scoring formula can be found above.

[0123] 4. Calculate the B / C score: quantify the cost-effectiveness of the strategy.

[0124] 5. Non-dominated ranking: Strategies are ranked according to the Pareto front.

[0125] 6. Generate Pareto hierarchy: Divide the policy into multiple non-dominated hierarchy.

[0126] 7. Elite selection: Retain the top 10% of the best strategies.

[0127] 8. Retain the top 10% of elite individuals: directly enter the elite strategy pool.

[0128] 9. Elite Strategy Library: Stores elite strategies for population merging.

[0129] 10. Inject the next generation: Combine the elite strategy with the offspring to generate a new population.

[0130] 11. Population Merging: Merging the elite strategy pool and the mutation strategy set.

[0131] 12. Crossover operation: Genetic algorithms are used to cross offspring.

[0132] 13. Child generation: Combine the parent strategy to generate a new strategy.

[0133] 14. Gaussian perturbation σ = 0.1: Apply a perturbation with a standard deviation σ = 0.1 to 20% of the offspring.

[0134] 15. Mutation operation: Adjust strategy parameters to enhance diversity.

[0135] 16. Update offspring parameters: optimize cost and time allocation.

[0136] 17. New generation population: A population that has completed one generation of iteration.

[0137] 18. Iterations ≤ 100 generations?: Set the maximum number of iterations to 100 generations.

[0138] 19. Crowding Calculation: Evaluate the distribution density of the solution set.

[0139] 20. Diversity screening: Avoid local optima.

[0140] 21. Select a non-dominated solution: Proceed to the next round of population merging.

[0141] Furthermore, its logical order is explained as follows:

[0142] Starting point: Initialize the population (1), input constraints (2) to the fitness evaluation engine (3);

[0143] Fitness calculation: Engine (3) calculates B / C score (4) and triggers non-dominated sorting (5);

[0144] Elite strategy retention: Select the top 10% elites after non-dominant sorting (5) (7→8→9) and inject them into the population merge (10→11);

[0145] Genetic operation: crossover (12) generates offspring (13), apply Gaussian perturbation (14→15→16) to update parameters;

[0146] Iteration loop: The new generation population (17) checks the iteration count (18), and if it is not satisfied, it re-initializes (1);

[0147] Diversity control: Crowding calculation (19) and screening (20) ensure uniform distribution of solution sets (21→11).

[0148] based on Figure 3 The process shown, including the closed-loop interactions involved, is explained below:

[0149] Feedback loop 18→1: The algorithm closes through iterative conditions (≤100 generations) to continuously optimize the strategy.

[0150] Feedback link 21→11: Diversity screening results participate in population merging to ensure uniform distribution of solution sets.

[0151] Figure 4 This is a schematic diagram of a Pareto front solution set provided for an embodiment of this application. For example... Figure 4 As shown, the specific steps include:

[0152] 1. Multi-objective optimization solver: The set of candidate solutions output after 100 iterations.

[0153] 2. Generate candidate solution set: strategy combinations including cost, time and risk dimensions.

[0154] 3. Pareto Front Screening: Using a non-dominated sorting algorithm to screen for the optimal boundary.

[0155] 4. Output non-dominated solutions: the set of optimal solutions stored in memory.

[0156] 5. Optimal solution set storage: Temporarily store Pareto front solutions.

[0157] 6. Transfer to dynamic filtering: The solution set is transferred to the dynamic control unit for secondary filtering.

[0158] 7. Dynamic screening control unit: Adjusts the threshold according to ΔP.

[0159] 8. Apply the T_k threshold: dynamically adjust according to the formula T_k=T_base×(1+α·ΔP).

[0160] 9. Final strategy set: Executable strategies that satisfy all constraints.

[0161] 10. Visualization output: A Pareto surface representing cost, time, and risk is displayed in a 3D coordinate system.

[0162] 11. Three-dimensional Pareto front plot: Interactive charts support decision-makers' analysis.

[0163] 12. Decision-maker selection: The project manager selects the final strategy based on business needs.

[0164] 13. Execution strategy feedback: The actual execution results are fed back to the optimizer to form a closed loop.

[0165] 14. Effectiveness evaluation: The effectiveness of the strategy was verified through Monte Carlo simulation.

[0166] Furthermore, its logical order is explained as follows:

[0167] Starting point: The optimizer (1) generates a set of candidate solutions (2), which are then filtered by Pareto (3→4) and the optimal solution is stored (5).

[0168] Dynamic adjustment: The dynamic control unit (7) applies thresholds (8→9) to generate the final strategy set (10);

[0169] Visualization and decision-making: 3D graphs (11) assist decision-makers in making choices (12), and feedback on the effects after implementation (13→14);

[0170] Closed-loop feedback: Effect evaluation (14) data is sent back to optimizer (1) to start the next round of optimization.

[0171] The key interactions involved are explained below:

[0172] Feedback Link 14→1: Feedback the simulation verification results (risk exposure decreased by 38.7%) to the solver to optimize subsequent iterations.

[0173] Dynamic threshold control: The threshold T_k is adjusted in real time based on ΔP (ΔP = Σ(actual working hours - planned working hours) / total working hours).

[0174] Figure 5 This is a flowchart illustrating a dynamic filtering rule adjustment method provided in an embodiment of this application. Figure 5 As shown, the specific steps include:

[0175] 1. Real-time progress data input: Obtain actual / planned work hours for the task via the JIRA interface.

[0176] 2. Obtain JIRA / Confluence data: Import risk records, task lists, and historical data.

[0177] 3. Critical path task identification: Screening core tasks that affect the project cycle.

[0178] 4. Calculate the deviation rate ΔP: Formula ΔP = Σ(actual working hours - planned working hours) / total planned working hours.

[0179] 5. ΔP sliding window calculation: based on the task deviation statistics of a 7-day window period.

[0180] 6. Apply the EWMA algorithm γ = 0.9: Exponentially weighted moving average smooths the fluctuation of ΔP.

[0181] 7. Smoothed ΔP value: The progress deviation index after eliminating short-term noise.

[0182] 8. Calculate the learning rate α = 0.1 / N: α is negatively correlated with the project size (N is the total number of tasks).

[0183] 9. Dynamic threshold T_k adjustment: Update the filtering threshold according to T_k = T_base × (1 + α·ΔP).

[0184] 10. Update T_k = T_base × (1 + α·ΔP): The threshold dynamically scales with ΔP.

[0185] 11. Pareto set filter: a strategy for filtering out sets that do not satisfy T_k.

[0186] 12. Output a set of effective strategies: strategies that meet the constraints of cost, time, and risk.

[0187] 13. Resource Allocation Engine: Executes strategies and allocates personnel and budgets.

[0188] 14. Execute and monitor the strategy: Track the effect of strategy execution in real time.

[0189] 15. Feedback on actual results: Collect indicators such as risk exposure and resource utilization rate.

[0190] 16. Update T_base baseline value: Adjust the baseline threshold for the next period based on feedback data.

[0191] Furthermore, the logical order is explained in detail below:

[0192] Data input and processing: Obtain real-time data from JIRA (1→2), identify critical path tasks (3→4);

[0193] ΔP calculation and smoothing: Calculate the deviation rate (5→6), and eliminate noise using the EWMA algorithm (7→8);

[0194] Dynamic threshold adjustment: Update T_k (11→12) in conjunction with the learning rate α (9→10);

[0195] Strategy execution and feedback: Select effective strategies (13→14), and provide feedback on the results after execution (15→16) to form a closed loop.

[0196] The closed-loop interaction involved is explained as follows:

[0197] Feedback Link 16→1: The actual execution effect is used to update T_base, forming a closed loop of "data input → adjustment → execution → feedback".

[0198] Dynamic adaptability: The threshold T_k is dynamically optimized as the project progresses to ensure the real-time nature of the strategy (ΔP is updated every 7 days).

[0199] To more clearly illustrate the methods in the embodiments of this application, the following examples, in conjunction with specific application scenarios, illustrate the methods and their effects.

[0200] Case 1: Optimization of IT Project Risk Response Strategies

[0201] (1) Input data:

[0202] Risk list: Changes in requirements (probability 0.7 / high impact), technical debt (probability 0.5 / moderate impact)

[0203] Constraints: Budget ≤ 500,000, timeframe ≤ 6 months

[0204] (2) System output:

[0205] Strategy set: A set of Pareto front solutions that includes resource allocation and combinations of response measures;

[0206] Optimization results: Based on the PSPLIB-RCPSP standard cases (j30, j60) and a custom risk event model, the probability distribution of risk events adopts the Poisson distribution (λ=0.5), and the impact value is simulated by the Beta distribution (α=2, β=5). After 1000 Monte Carlo simulations, the results show that the risk exposure decreased by 38.7% (95% CI [36.2%, 41.2%]) and the strategy formulation time was shortened by 72%±5%.

[0207] Monte Carlo simulation parameters are disclosed: Random seed range 1-1000, risk event trigger threshold 0.6, and iterative convergence condition is fitness change rate <1% for 10 consecutive generations.

[0208] (3) Technical effect: According to simulation verification, compared with the traditional single-objective optimization method, the resource utilization rate is increased by 41%, and the strategy effectiveness achievement rate is increased from 68% to 92%.

[0209] Case 2: This example describes the strategy optimization process for an e-commerce platform development project, where the methods used can be... Figure 1 The method steps in the illustrated embodiment.

[0210] I. Project Background:

[0211] An internet company is developing a new e-commerce platform with a project cycle of 8 months and a budget of 3 million yuan. The project involves multiple phases, including front-end development, back-end development, database design, and testing.

[0212] II. Input Data:

[0213] (1) Risk List:

[0214] Requirement change (probability 0.7 / high impact)

[0215] Technical debt (probability 0.5 / impact moderate)

[0216] Personnel turnover (probability 0.3 / high impact)

[0217] Budget overrun (probability 0.4 / moderate impact)

[0218] (2) Constraints:

[0219] Budget ≤ 3 million

[0220] Cycle ≤ 8 months

[0221] Implementation steps:

[0222] (1) Data access:

[0223] Import risk records and task lists from JIRA

[0224] Import project documents and historical data from Confluence

[0225] Manually enter specific constraints for the project

[0226] (2) Strategy optimization:

[0227] Establish a multi-objective optimization model: min{Σα_i C_i,Σβ_j T_j};

[0228] Pareto optimality search was performed using the improved NSGA-II algorithm;

[0229] Dynamically adjust the screening threshold T_k = T_base × (1 + α·ΔP);

[0230] in:

[0231] α: Learning rate coefficient, ranging from 0.01 to 0.1, with a default value of 0.05;

[0232] ΔP: Schedule deviation rate, obtained in real time via JIRA interface.

[0233] ΔP is calculated based on the actual / planned time deviation rate of the critical path task using a sliding window algorithm (window period of 7 days). The formula can be found in the previous description and will not be repeated here.

[0234] (3) Output of results:

[0235] Generate a Pareto front solution set containing different policy combinations;

[0236] Output the optimized resource allocation plan and schedule.

[0237] III. System Output

[0238] (1) Optimize strategy set:

[0239] Strategy 1: Increase front-end development resources and shorten the time for handling requirement changes.

[0240] Strategy 2: Introduce code review tools to reduce technical debt.

[0241] Strategy 3: Increase critical path buffers to cope with personnel movement.

[0242] Strategy 4: Optimize budget allocation to reduce the risk of budget overruns.

[0243] (2) Resource allocation recommendations:

[0244] Front-end development: Add 2 developers;

[0245] Backend development: Add 1 developer;

[0246] Testing phase: Add 1 test engineer.

[0247] (3) Time arrangement:

[0248] Requirements analysis phase: 1 month;

[0249] Development phase: 5 months;

[0250] Testing phase: 1.5 months;

[0251] Deployment phase: 0.5 months.

[0252] IV. Technical Effects

[0253] Risk exposure decreased: Compared to the traditional approach, risk exposure decreased by 38.7%.

[0254] Strategy formulation time reduced: Strategy formulation time reduced by 72%.

[0255] Resource utilization efficiency improved: resource utilization efficiency improved by 41%.

[0256] V. Conclusion

[0257] By implementing this solution, project teams can more efficiently formulate and optimize risk response strategies, significantly improving project success rates and resource utilization. In this embodiment, the differences in system output across different projects are jointly determined by input conditions (risk type, constraints), optimization rules (multi-objective model, dynamic filtering), and feedback data (ΔP, Monte Carlo results). The generation of output categories (strategy set, resource allocation, schedule) can follow this logic:

[0258] Risk-driven strategy: Predefined risk-strategy mapping table, Monte Carlo simulation to verify effectiveness;

[0259] Resource and time decoupling: resource allocation is based on task decomposition and load balancing, while time scheduling depends on critical paths and dynamic thresholds;

[0260] Dynamic adaptation closed loop: The output granularity is adjusted in real time through ΔP and T_k to ensure that the strategy is synchronized with the project progress. This mechanism enables the system to flexibly adapt to projects in different fields such as IT, e-commerce, and manufacturing, and output highly targeted optimization solutions.

[0261] Current methods for generating risk response strategies may have the following shortcomings:

[0262] Limitations of single-objective optimization: Traditional methods only focus on the single dimension of cost or time, and cannot balance the triangular constraint of cost, time and residual risk (Deloitte 2023 Project Management Survey Report).

[0263] Static strategy issues: Tools such as JIRA rely on fixed rule bases and cannot adapt to dynamic changes in the project environment.

[0264] Low efficiency of the optimization algorithm: The existing NSGA-II algorithm suffers from premature convergence in project management applications.

[0265] The method in this application adopts a three-tiered architecture of "multi-objective modeling - genetic optimization - dynamic screening." By establishing a multi-objective optimization model that considers cost, time, and residual risk simultaneously, it effectively solves the problem of traditional methods that only focus on a single dimension. This model achieves optimal allocation of project resources and effective risk control by minimizing the weighted sum of cost and time while ensuring that the residual risk does not exceed a preset threshold. This method not only balances multiple key objectives in a project but also improves the overall success rate of the project.

[0266] This application employs a dynamic filtering control unit that dynamically adjusts the filtering threshold based on real-time project schedule deviations, thereby adapting to dynamic changes in the project environment. This dynamic adjustment capability makes strategy selection more flexible and enables timely responses to new situations and challenges arising in the project. Compared to tools that rely on fixed rule bases (such as JIRA), it can provide more personalized and adaptable strategy recommendations.

[0267] This application improves the NSGA-II algorithm by increasing the elite retention rate to 10% and introducing a Gaussian perturbation mutation strategy. These improvements enhance the algorithm's global search capability, avoid premature convergence, and thus improve the algorithm's efficiency and solution quality. In this way, this scheme can more effectively explore the solution space, find better strategy combinations, and improve the effectiveness and practicality of the strategy.

[0268] By dynamically adjusting the screening thresholds, the method in this application ensures that the selected strategy always matches the actual progress and conditions of the project. This adaptability enables the project team to flexibly respond to changes in the project and adjust strategies in a timely manner to address new risks and challenges.

[0269] This application improves resource utilization efficiency and reduces waste by optimizing the strategy selection process. Simultaneously, it increases project success rates by more effectively managing project risks. According to simulation results, this solution can reduce risk exposure by 38.7%, shorten strategy development time by 72%, and increase resource utilization by 41%.

[0270] In one embodiment, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, causes the processor to perform the method steps described in the foregoing embodiments.

[0271] Those skilled in the art will understand that all or part of the processes in the methods implemented by the above-described system can be accomplished by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, 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 various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0272] 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.

[0273] 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 patent application should be determined by the appended claims.

Claims

1. A project strategy optimization method, characterized in that, include: A multi-objective optimization model is established, which is a constraint model of cost, time, and residual risk. The improved NSGA-II algorithm is used to solve the multi-objective optimization model and generate a Pareto optimal policy set. The improved NSGA-II algorithm includes an elite retention policy and a Gaussian perturbation mutation policy. The screening threshold is dynamically adjusted based on the real-time progress deviation of the project. The screening threshold is used to control the selection of strategies in the Pareto optimal strategy set to adapt to changes in project progress.

2. The project strategy optimization method according to claim 1, characterized in that, The objective function of the multi-objective optimization model is: min{Σα_i C_i,Σβ_j T_j}, with the constraint R_k≤S_k; Where C_i is the cost of the i-th strategy, T_j is the execution time of the j-th strategy, R_k is the residual risk threshold, S_k is the preset risk threshold, α_i is the cost weight coefficient, and β_j is the time weight coefficient.

3. The project strategy optimization method according to claim 2, characterized in that, The improved NSGA-II algorithm also includes using a fitness evaluation function to evaluate the fitness of the strategy; The fitness evaluation function is B / C = (risk reduction rate) / (α·budget + β·time), where α and β are the cost weight coefficient and the time weight coefficient, respectively.

4. The project strategy optimization method according to claim 1, characterized in that, The dynamic adjustment of the screening threshold based on the real-time progress deviation of the project includes: Obtain the actual working hours and planned working hours of the task, and calculate the real-time progress deviation ΔP of the project, wherein ΔP is the sum of the differences between the actual working hours and the planned working hours divided by the total planned working hours; The real-time progress deviation ΔP of the project is smoothed by an exponentially weighted moving average. The filtering threshold T_k is adjusted based on the smoothed real-time progress deviation ΔP of the project and the learning rate α. The formula for calculating the filtering threshold T_k is T_k=T_base×(1+α·ΔP), where T_base is the baseline threshold, and the formula for calculating the learning rate α is α=0.1 / N, where N is the total number of project tasks.

5. The project strategy optimization method according to claim 4, characterized in that, The method further includes: After filtering the Pareto optimal solution set using the dynamically adjusted filtering threshold to obtain the final project strategy set, resource allocation is performed through the resource allocation engine based on the final project strategy set. Collect the actual execution results of the project strategy and feed them back to update the baseline threshold T_base.

6. A project strategy optimization system, characterized in that, include: A multi-objective optimization modeling unit is used to establish a multi-objective optimization model, wherein the multi-objective optimization model is a constraint model of cost, time and residual risk; The solution unit is used to solve the multi-objective optimization model using the improved NSGA-II algorithm to generate a Pareto optimal policy set. The improved NSGA-II algorithm includes an elite retention policy and a Gaussian perturbation mutation policy. A dynamic filtering control unit is used to dynamically adjust the filtering threshold according to the real-time progress deviation of the project. The filtering threshold is used to control the selection of strategies in the Pareto optimal strategy set to adapt to changes in project progress.

7. The project strategy optimization system according to claim 6, characterized in that, The solution unit is further configured to evaluate the fitness of the strategy using a fitness evaluation function; The fitness evaluation function is B / C = (risk reduction rate) / (α·budget + β·time), where α and β are the cost weight coefficient and the time weight coefficient, respectively.

8. The project strategy optimization system according to claim 6, characterized in that, The dynamic filtering control unit is specifically used for: Obtain the actual working hours and planned working hours of the task, and calculate the real-time progress deviation ΔP of the project, wherein ΔP is the sum of the differences between the actual working hours and the planned working hours divided by the total planned working hours; The real-time progress deviation ΔP of the project is smoothed by an exponentially weighted moving average. The filtering threshold T_k is adjusted based on the smoothed real-time progress deviation ΔP of the project and the learning rate α. The formula for calculating the filtering threshold T_k is T_k=T_base×(1+α·ΔP), where T_base is the baseline threshold, and the formula for calculating the learning rate α is α=0.1 / N, where N is the total number of project tasks.

9. The project strategy optimization system according to claim 8, characterized in that, The dynamic screening control unit is also used for: After filtering the Pareto optimal solution set using the dynamically adjusted filtering threshold to obtain the final project strategy set, resource allocation is performed through the resource allocation engine based on the final project strategy set. Collect the actual execution results of the project strategy and feed them back to update the baseline threshold T_base.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to perform the steps of the method as described in any one of claims 1-5.

Images