Multi-Agent Consensus Control Under Saturation Constraints

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing multi-agent systems face challenges in achieving global optimal consensus, especially under input saturation constraints and dynamic communication topologies, where agents struggle to converge to a common optimal state while maintaining stability and optimality.

Innovation Solution

The implementation of a group consensus protocol that divides multi-agent systems into subgroups, with each subgroup minimizing a decentralized objective function, using control protocols that adjust based on the derivative of the objective function, ensuring convergence and stability through Lyapunov analysis and saturation handling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If agents use unbounded control inputs to achieve fast consensus convergence, then convergence speed improves, but control saturation constraints are violated

Engineering Contradiction:

Improveconsensus convergence speedVSAvoidconstraint satisfaction

Core Design Contradiction:

SpeedVSReliability

Solution Approach 1:

The patent transforms the bounded control problem into an unbounded problem by applying a saturation transformation. The actual control input is derived from an unbounded virtual control through a saturation function, allowing the optimization algorithm to operate in an unbounded space while the physical constraints are automatically satisfied in the transformed space.

Inventive Principle:

Principle #35Parameter changes

2Manufacturing precision

If a centralized optimization approach is used to achieve global optimality, then optimality is improved, but system complexity and communication requirements increase

Engineering Contradiction:

Improveoptimality precisionVSAvoidsystem complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent divides the multi-agent system into subgroups, with each subgroup having its own decentralized objective function. This segmentation allows each subgroup to independently optimize its own objective while contributing to the global optimization, reducing communication complexity and enabling distributed computation.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent employs gradient feedback mechanisms where agents use the gradient information of their objective functions to adjust their states. This feedback-driven approach enables distributed optimization without requiring centralized coordination, maintaining global optimality through local gradient descent operations.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If agents communicate frequently to maintain consensus under dynamic topologies, then consensus accuracy improves, but communication overhead and energy consumption increase

Engineering Contradiction:

Improveconsensus accuracyVSAvoidcommunication energy

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent formulates consensus as a continuous optimization problem where agents continuously adjust their states to minimize objective functions. This continuous action approach maintains consensus accuracy without requiring discrete communication events, as the optimization process naturally adapts to topology changes through continuous gradient descent.

Inventive Principle:

Principle #20Continuity of useful action

Data Source

PatentUS11281232B2Systems and methods for multi-agent system control using consensus and saturation constraints

Publication Date: 2022.03.22 KING FAHD UNIVERSITY OF PETROLEUM AND MINERALS

AI summary

Methods, computer readable media, and systems for systems and methods for multi-agent system control using consensus and saturation constraints are described.