Experiment Design Modification for Supersaturated Factor Identification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In complex systems, identifying active factors and estimating their effects is challenging due to the difficulty in visualizing and testing multiple component options, especially with supersaturated designs where the number of factors exceeds test cases, and current model selection techniques are inadequate.

Innovation Solution

A computer-program product generates a design space with user-defined disallowed combinations and additional constraints, suggesting modifications to create a computationally viable model that accounts for these constraints, allowing for the identification of active factors and their effects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the number of factors in a supersaturated design exceeds the number of test cases, then the experiment can cover more factors with limited resources, but it becomes difficult to identify active factors and estimate their effects using current model selection techniques

Engineering Contradiction:

Improvenumber of factors coveredVSAvoiddifficulty of identifying active factors

Core Design Contradiction:

ProductivityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent transforms the original supersaturated design into a new design by applying disallowed combination constraints and generating computer-suggested modifications. This changes the parameter structure of the design space, enabling factor identification in supersaturated conditions where traditional methods fail.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces disallowed combination constraints as an intermediary mechanism between the factors and the model selection process. These constraints guide the computer-generated design modifications to produce a design that facilitates active factor identification while maintaining supersaturated coverage.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If user-defined disallowed combinations are imposed on the design space, then the experiment avoids undesirable factor combinations, but the user-defined model may become computationally infeasible to generate

Engineering Contradiction:

Improveavoidance of undesirable combinationsVSAvoidcomputational complexity of model generation

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent makes the design generation process dynamic by using computer-generated modifications that adapt to the disallowed combination constraints. Instead of a static model generation approach, the system iteratively generates and evaluates design modifications until a computationally viable model satisfying all constraints is produced.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent performs preliminary actions by generating computer-suggested modifications of the user-defined model before final model generation. These preliminary modifications pre-process the design space to ensure computational feasibility while respecting disallowed combination constraints, making the subsequent model generation tractable.

Inventive Principle:

Principle #10Preliminary action

3Ease of operation

If current model selection techniques are used for supersaturated designs, then the process is simple, but the techniques are inadequate for identifying active factors when factors exceed test cases

Engineering Contradiction:

Improvesimplicity of model selectionVSAvoidaccuracy of factor identification

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent replaces traditional mechanical model selection techniques with a computer-generated design modification approach. This substitution enables handling of supersaturated designs by using computational algorithms to generate designs that facilitate accurate factor identification, overcoming the limitations of conventional methods.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11216603B2Transformation and evaluation of disallowed combinations in designed experiments

Publication Date: 2022.01.04 JMP STATISTICAL DISCOVERY LLC

AI summary

A computing system receives a request for a computer-generated design of an experiment. A design space is defined by candidate inputs for each factor of the experiment. The system receives a disallowed combination indication indicating a user-defined disallowed combination for the design space. The disallowed combination constrains a first set of values from a first set of candidate inputs from being assigned to the first factor if the second factor is assigned one of a second set of values from a second set of candidate inputs. The system determines additional constraint(s) on computer generation of the design. The system evaluates if a user-defined model can be generated according to the disallowed combination and the constraint(s). The system generates a computer suggested modification of the user-defined model such that the design can be generated according to the computer suggested modification.