System State Estimation Using Log-Guided Diagnostic Actions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for identifying the cause of failures in complex systems require executing all possible user actions, which can overload the system and prolong the identification process, especially when complex feedback is involved.
Innovation Solution
A system state estimation apparatus that uses a model generation unit to associate system states with user actions based on accumulated log data, and an action determination unit to determine effective user actions for identifying the system state, reducing the number of actions needed by calculating probabilities and simulating feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all possible user actions are executed to identify the cause of failures, then the identification accuracy is improved, but the system load increases and the identification time is prolonged
Solution Approach 1:
The patent applies partial action by selecting and executing only a subset of user actions that are most likely to identify the failure cause, rather than executing all possible actions. The action selection unit determines which actions to execute based on the current system state and expected diagnostic value, achieving accurate failure identification with fewer actions, thus improving identification speed while maintaining accuracy.
2Measurement precision
If all possible user actions are executed to identify the cause of failures, then the identification accuracy is improved, but the system load increases
Solution Approach 1:
The patent limits the number of executed user actions to only those necessary for accurate failure identification. The action selection unit evaluates the diagnostic value of each potential action and selects only the most effective ones, reducing system load while maintaining identification accuracy.
Solution Approach 2:
The patent uses a simulated environment to pre-evaluate user actions and their expected outcomes before executing them on the real system. This allows the system to determine which actions will be most effective without actually performing them on the production system, thereby reducing the harmful load on the real system while still achieving accurate failure identification.
3Measurement precision
If comprehensive action logs are collected by executing all user actions, then the failure cause identification accuracy is improved, but the time required for log collection increases
Solution Approach 1:
The patent collects action logs only for the subset of user actions that are most likely to identify the failure cause, rather than collecting logs from all possible actions. This significantly reduces log collection time while maintaining the accuracy needed for effective failure diagnosis.
Solution Approach 2:
The patent performs preliminary evaluation in a simulated environment to determine which user actions will be most effective for identifying the specific failure cause. This preliminary action allows the system to pre-determine the necessary log collection scope, avoiding unnecessary log collection and reducing overall time while ensuring accurate identification.
4Measurement precision
If multiple user actions are executed to obtain complex feedback, then the system state identification accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent uses a simulated environment to model system behavior and predict the outcomes of user actions. This simulation copy allows the system to evaluate multiple actions and their complex feedback without actually executing them on the real system, thereby reducing processing complexity while maintaining identification accuracy.
Solution Approach 2:
The patent introduces a simulated environment as an intermediary between the control system and the real system. This intermediary handles the complex processing of evaluating multiple user actions and their feedback, allowing the real system to remain simple while still achieving accurate state identification through the simulation's analysis.
Data Source
AI summary
A system state estimation apparatus to estimate a system state in a system that outputs log data upon executing a user action, the system state estimation apparatus includes a memory; and a processor configured to execute using log data accumulated in advance, to generate a stochastic model in which a system state and a user action are associated with log data when the user action is executed in the system state; and determining a user action that contributes to identification of the system state in the system, based on a probability distribution of the system state and the stochastic model.


