Feedback Control Blending AI and Non-AI Outputs by Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing control systems using machine learning models struggle with stable operation, particularly with untrained data, leading to instability in controlling control target devices.
Innovation Solution
A control apparatus that integrates AI and non-AI processing units, with a correction unit to adjust the selection ratio of output values based on reliability, ensuring stable operation by selectively using AI or non-AI outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a machine learning model is used for feedback control, then efficient results can be obtained, but stable operation cannot be ensured for untrained data
Solution Approach 1:
The system dynamically adjusts the selection ratio between AI processing output and non-AI processing output based on the degree of reliability. This dynamic adjustment allows the system to adapt to different operating conditions, using AI processing when reliable and switching to non-AI processing when the AI model confidence is low, thus resolving the contradiction between control efficiency and operational stability.
Solution Approach 2:
The system changes the parameter of selection ratio between AI and non-AI outputs based on the degree of reliability of the AI model. By adjusting this parameter dynamically, the system can optimize the balance between utilizing AI's efficiency and ensuring operational stability through traditional control methods when needed.
2Productivity
If AI processing is introduced for feedback control, then efficient control results are achieved, but difficulty in ensuring stable operation for extrapolated data arises
Solution Approach 1:
The control system is segmented into two parallel processing paths: AI processing and non-AI processing. Each path handles control tasks independently, and their outputs are combined based on reliability assessment. This segmentation allows the system to leverage AI's efficiency while maintaining a traditional control path for stability, managing the overall complexity through modular design.
Solution Approach 2:
The reliability assessment mechanism acts as an intermediary between AI processing and non-AI processing outputs. It evaluates the degree of reliability of AI results and determines the appropriate selection ratio, mediating between the efficient but potentially unreliable AI output and the stable but less efficient non-AI output.
3Productivity
If a machine learning model is used, then efficient control is achieved, but reliability for untrained data becomes problematic
Solution Approach 1:
The system implements feedback through the reliability assessment mechanism that continuously evaluates the degree of reliability of AI processing outputs. This feedback information is then used to adjust the selection ratio dynamically, ensuring that unreliable AI outputs are appropriately weighted or replaced with non-AI outputs, thus maintaining overall system reliability.
Solution Approach 2:
The selection ratio between AI and non-AI outputs is dynamically adjusted based on real-time reliability assessment. This dynamic adaptation allows the system to maximize the use of efficient AI processing when reliable while automatically reducing reliance on it when reliability concerns arise, effectively resolving the contradiction between efficiency and reliability.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A control apparatus (2) performs feedback control on a control target device (3). The control apparatus includes a first processing unit (13) configured to acquire a first output value related to feedback control on the control target device by using AI processing, and a second processing unit (14) configured to acquire a second output value related to the feedback control by using non-AI processing. The control apparatus includes a correction unit (16) configured to correct a selection ratio of the first output value and the second output value based on a degree of reliability of the first output value, and a control unit (12) configured to output a control signal for performing feedback control on the control target device based on the selection ratio.