Machine Learning Feedback Control for Changing Target Dynamics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feedback control systems, such as PID control, lack adaptability when the characteristics of the control target or operator change over time, leading to reduced control accuracy.
Innovation Solution
A control device that incorporates a machine learning model with a tree structure to generate predicted outputs and adjust operation amounts based on feedback data, allowing for adaptive control by invalidating operation amounts that exceed predetermined thresholds and storing data for future learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If fixed gain is used in feedback control, then reliability is maintained, but adaptability deteriorates when control target characteristics change over time
Solution Approach 1:
The patent applies dynamics by transitioning from fixed gain to time-varying gain through machine learning. The controller dynamically adjusts the operation amount based on learned patterns from historical data, allowing the system to adapt to changing control target characteristics while maintaining stability through the learned model.
Solution Approach 2:
The system implements self-service through automated machine learning that performs adaptive control without human intervention. The learned model automatically updates and adjusts control parameters based on accumulated operational data, enabling the system to self-optimize its performance over time.
2Adaptability or versatility
If machine learning model is added to achieve adaptive control, then adaptability is improved, but device complexity increases
Solution Approach 1:
The patent uses a learned model as an intermediary between the control target and the controller. This learned model acts as a mediator that processes sensor data and generates appropriate operation amounts, simplifying the overall control architecture while enabling adaptive behavior through the intermediate learning layer.
Solution Approach 2:
The system performs preliminary action by pre-learning control patterns during idle periods or using historical data before actual control is needed. This allows the machine learning model to be trained and ready in advance, reducing the computational burden during real-time operation and simplifying the control system's runtime complexity.
3Measurement precision
If operation amounts exceeding thresholds are invalidated, then control accuracy is improved, but loss of information increases
Solution Approach 1:
The patent extracts and removes only the problematic portions of control data (operation amounts exceeding predetermined thresholds) while retaining the rest of the valuable control information. This selective extraction approach maintains control accuracy by eliminating outliers while preserving the useful data for continued learning and operation.
Solution Approach 2:
The system converts the potentially harmful effect of invalidating control data into a benefit by using the invalidation process as a learning opportunity. The learned model is trained to recognize and avoid threshold-exceeding conditions, transforming data loss into improved predictive capability and more robust control strategies.
Data Source
AI summary
A control device includes a first controller configured to generate a first operation amount for the device on the basis of an output fed back from the device and a target value, a predicted output generator including a learned model which is machine learned so as to generate a predicted output from the device on the basis of the output fed back from the device and the first operation amount, a second controller configured to generate a second operation amount for the device on the basis of the predicted output and the target value, an integrated operation amount generator configured to generate an integrated operation amount which is an operation amount for the device on the basis of the first operation amount and the second operation amount.