Staged Control Application Hot Swapping Without Prediction Rebuild Delays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for updating model predictive controllers require shutting down live applications to upload new control designs, leading to inefficiencies and downtime, as predictions from the new design take time to rebuild and performance degrades until the new design is verified.
Innovation Solution
Implementing a staged application concept where a new processor runs in standby mode alongside the live processor, using historical data to build predictions and allowing seamless hot-swapping, reducing downtime and rebuilding time for predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the live application is stopped to upload and verify a new control design, then the new application can be tested for correctness, but system downtime increases and productivity decreases
Solution Approach 1:
The new control design is uploaded and verified in advance while the live application continues to run. The staged application is prepared and tested before it needs to replace the live application, allowing verification to occur beforehand rather than causing downtime.
Solution Approach 2:
A staged application (copy) of the new control design is created and run in parallel with the live application. This copy can be tested and verified without affecting the operation of the original live application, allowing safe validation before full deployment.
2Adaptability or versatility
If the live application is stopped to upload a new control design, then the new design can be implemented, but the time to rebuild predictions from scratch increases
Solution Approach 1:
The staged application begins building predictions in advance while the live application is still running. By the time the staged application needs to replace the live application, predictions are already built or nearly complete, eliminating the time loss that would otherwise occur.
Solution Approach 2:
The prediction-building process continues uninterrupted in the staged application while the live application operates. This continuous parallel operation ensures that when the swap occurs, the staged application is ready with pre-built predictions, maintaining continuous useful action without downtime.
3Reliability
If the live application is stopped for verification, then new behavior can be tested, but performance degradation occurs during the stoppage period
Solution Approach 1:
A staged application is created as a copy of the new control design and run in parallel with the live application. This allows testing and verification of new behavior in the staged copy while the live application continues to deliver full performance, avoiding any degradation during verification.
Solution Approach 2:
Verification and testing of the new control design occur in advance in the staged application before it replaces the live application. This preliminary verification ensures that when the new design is deployed, it has already been validated, preventing performance issues after deployment.
Data Source
AI summary
A staged application may be transitioned, or hot swapped, to live application, that is, the staged version of an application is transited to the live version of an application. The staged version may nm in estimation mode alongside the live version. Only one version of an application may be active in a given run time workspace. The staged version may be initialized with one or more parameters or variables. The staged version may be tested or simulations may be performed to ensure the staged version is ready to be transitioned to a live version. The staged version may execute at a lower priority to ensure that the efficiency and operations of the live version are not disturbed. Once the staged version has been transitioned, the replaced live version or another version may be transitioned to the staged version. The transitioned staged version and the replaced live version may be archived.