Stochastic Game Control Using Maximum-Width Path Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optimal control techniques struggle with high-speed calculations while ensuring accuracy in stochastic games, particularly in systems like smart grids and autonomous driving, where recalculating power distribution and control strategies are necessary due to stochastic changes.
Innovation Solution
A system optimal control device that converts stochastic games into weighted graphs, selects paths with maximum width, and determines convergence of upper and lower bounds to ensure high-speed and accurate calculations of the probability of arrival.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the bounded value iteration method is used to guarantee accuracy by calculating both upper and lower bounds, then measurement precision is improved, but calculation speed deteriorates due to more complex calculations required to guarantee convergence of the upper bound
Solution Approach 1:
The patent segments the calculation process into two distinct parts: calculating the lower bound of the probability of arrival (which can be done efficiently) and calculating the upper bound of the probability of arrival (which provides the accuracy guarantee). By separating these calculations and using different approaches for each, the system achieves both speed and accuracy without the full complexity of traditional bounded value iteration.
Solution Approach 2:
The patent introduces an intermediary approach where the upper bound calculation uses a simplified method that doesn't require the full complexity of bounded value iteration. This intermediary calculation provides sufficient accuracy guarantee without the computational burden, acting as a mediator between the need for accuracy and the need for speed.
2Productivity
If the ordinary value iteration method is used for high-speed calculation, then calculation speed is improved, but measurement precision deteriorates because it only calculates the lower bound without accuracy guarantee
Solution Approach 1:
The patent applies partial action by calculating only the necessary components for accuracy guarantee. Instead of performing the full bounded value iteration which calculates both bounds with full convergence guarantees, the system performs a partial calculation that provides sufficient accuracy for the upper bound without the complete computational overhead, achieving speed while maintaining necessary precision.
Data Source
AI summary
A system optimal control technique with accuracy guarantee that enables high-speed calculations is provided. One aspect of the present invention is related to a system optimal control device including a graph converting unit configured to convert, based on an upper bound of an probability of arrival from an initial state to a final state of a stochastic game representing system information, the stochastic game into a flow analysis graph, a path selecting unit configured to select a path having a maximum width among paths from each state node to a final state node in the converted flow analysis graph, a width of each of the paths being defined as a minimum weight of edges forming the path, and a convergence determining unit configured to determine convergence of the upper bound and a lower bound of the probability of arrival of the stochastic game based on information about the selected path.