Edge Control Model Distribution for Resource-Limited Facility Controllers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing control systems face challenges in efficiently performing learning processing for facility control due to limited processing resources at the controller level, especially when large amounts of data are required.
Innovation Solution
A control system architecture that utilizes edge computing by distributing learning processing to a more resource-rich upper control apparatus, which generates control models that are then transmitted to local controllers for execution, allowing selection of optimal control data based on certainty and recency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If learning processing is performed at the controller level, then control decisions can be made locally, but the processing load on the controller becomes excessive when large amounts of data are required
Solution Approach 1:
The system divides the control architecture into two segments: an upper control apparatus that performs learning processing and model generation, and local controllers that execute control decisions using pre-generated control models. This segmentation transfers the heavy processing load to the upper apparatus while maintaining local control capability.
Solution Approach 2:
Control models serve as intermediaries between the upper control apparatus and local controllers. The upper apparatus generates these models through learning processing, transmits them to local controllers, and the controllers use them for decision-making without performing the learning processing themselves.
2Adaptability or versatility
If control models are generated and transmitted to multiple controllers, then learning results can be shared across the facility, but the time and resources required for model generation and distribution increase
Solution Approach 1:
The upper control apparatus performs learning processing in advance to generate control models before they are needed for control decisions. These pre-generated models are then transmitted to and stored in multiple controllers, so when control decisions are required, the controllers can immediately use the available models without waiting for learning processing to complete.
Solution Approach 2:
A single control model generated by the upper apparatus can be transmitted to and used by multiple different controllers across the facility. This allows the learning results to serve universal purposes, with one model potentially being applied in multiple control contexts and locations.
3Measurement precision
If the control model is updated frequently to adapt to environmental changes, then control accuracy improves, but the processing burden on the upper control apparatus increases
Solution Approach 1:
The upper control apparatus performs learning processing and model updates at periodic intervals or in response to detected environmental changes, rather than continuously. This allows the system to maintain adequate control accuracy while managing the processing burden through scheduled rather than constant learning operations.
Data Source
AI summary
Provided is a control system comprising a control apparatus including a learning processing unit for generating a control model by learning, the control model being for calculating control data for controlling a facility according to state data detected by at least one sensor for measuring a state of the facility, and a model transmission unit for transmitting the generated control model to a controller for controlling the facility; and a controller including a model receiving unit for receiving the learned control model from the control apparatus, a state receiving unit for receiving the state data from the at least one sensor, a calculation unit for calculating control data for controlling the facility according to the state data, which is a processing target, by using the control model received by the model receiving unit, and a control unit for controlling the facility by using the calculated control data.