Optimizer-Agnostic Explanations for Large-Scale Schedules via Resource Links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing large-scale project scheduling systems lack scalability and interpretability, with state-of-the-art methods taking impractically long to solve resource-constrained critical path problems for projects with thousands of tasks, and users struggle to understand and trust decision optimization models due to lack of transparency.
Innovation Solution
An AI-based method that generates a minimal set of resource links by adding dummy tasks to construct transportation problems, allowing for efficient computation of critical paths and providing explanations for optimized schedules, even with multiple resource types, while maintaining computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If state of the art approaches for identifying resource constrained critical tasks are used, then scheduling accuracy is improved, but computational time becomes impractical (hours) for large scale projects
Solution Approach 1:
The patent segments the complex resource-constrained critical path identification into two independent phases: (1) generate an initial schedule using any optimizer, and (2) extract critical tasks using constraint propagation on the schedule output. This segmentation allows the computationally intensive optimization to be done once, followed by rapid critical path extraction, reducing overall computational time while maintaining accuracy.
Solution Approach 2:
The patent introduces an intermediary explanation layer that translates the black-box optimizer output into interpretable critical path information. This intermediary component uses constraint propagation to compute earliest/latest start times and identify critical tasks, serving as a bridge between the opaque optimization engine and the user-needed critical path analysis, thereby enabling fast extraction without re-running the full optimization.
2Productivity
If black-box schedulers are used to minimize makespan, then scheduling optimization is improved, but interpretability and user trust deteriorate
Solution Approach 1:
The patent implements feedback by using the schedule output from the black-box optimizer as input to the critical path extraction phase. The extraction process feeds back interpretable information (critical tasks, floats, sequences) about the optimization results, allowing users to understand and verify the optimizer's decisions without sacrificing the optimization power of the black-box approach.
Solution Approach 2:
The patent introduces an intermediary explanation layer that translates the black-box optimizer output into interpretable critical path information. This intermediary component uses constraint propagation to compute earliest/latest start times and identify critical tasks, serving as a bridge between the opaque optimization engine and the user-needed critical path analysis, thereby enabling fast extraction without re-running the full optimization.
3Reliability
If traditional critical path methods are applied to resource constrained scheduling, then theoretical completeness is improved, but scalability to large projects deteriorates
Solution Approach 1:
The patent segments the traditionally unified critical path computation into independent forward and backward passes that can be efficiently applied to large schedules. By computing earliest start times via forward pass and latest start times via backward pass on the already-generated schedule, the method maintains theoretical completeness of critical path analysis while achieving linear scalability to projects with thousands of tasks.
Solution Approach 2:
The patent performs preliminary scheduling using any optimizer to establish task start and end times before applying critical path analysis. This preliminary action provides the temporal framework needed for efficient constraint propagation, allowing the critical path method to be applied practically to large-scale projects without requiring re-optimization, thus achieving both theoretical completeness and scalability.
Data Source
AI summary
A computer implemented method using an artificial intelligence (A.I.) module to explain large scale scheduling solutions includes receiving an original instance of a resource constrained scheduling problem. The instance includes a set of tasks and a variety of resource requirements and a variety of constraints. An optimizer process determines a schedule for the set of tasks while minimizing a makespan of the schedule. A minimal set of resource links is generated based on resource dependencies between tasks. The resource links are added to the original instance of scheduling problem, as precedence constraints. All the resource constraints are removed from the original instance of the resource constrained scheduling problem. A set of critical tasks is computed using a non-resource constrained critical path. Schedules are provided with an explanation of an optimized order of the set of tasks based on the use of the non-resource constrained critical path.