Real-Time Control Allocation Using Parallel Optimizer Voting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optimization techniques for controlling over-actuated systems in vehicles, such as electric vehicles, often produce erroneous outputs due to convergence problems, instabilities, and malfunctions, which can compromise safety, energy efficiency, and drivability, and are not suitable for real-time control applications.
Innovation Solution
A method involving multiple independent optimization processes that solve an optimization problem, followed by a voting scheme to select the most reliable solution based on differences between solution vectors, with optional quality indicators and dimensionality reduction to enhance reliability and reduce failure rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single optimization process is used to solve the control allocation problem in real-time, then the processing speed meets real-time requirements, but the reliability is compromised due to convergence problems, instabilities, and erroneous outputs
Solution Approach 1:
The patent divides a single optimization process into multiple independent optimization processes (first, second, third, etc.). Each process independently solves the control allocation problem and generates a solution vector. This segmentation increases reliability by distributing the computational task across multiple processes, reducing the impact of convergence failures or instabilities in any single process.
Solution Approach 2:
The patent combines the results from multiple independent optimization processes through a voting scheme. Solution vectors from different processes are compared, and the most frequent solution (or a consensus solution) is selected as the final control signal. This merging approach enhances reliability by leveraging the collective output of multiple processes while maintaining real-time performance.
2Reliability
If multiple optimization processes are executed independently and their results are compared through a voting scheme, then the reliability and failure rate are improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent executes multiple optimization processes in parallel rather than sequentially, performing all computations simultaneously within the real-time deadline. This preliminary action approach ensures that the computationally intensive task of running multiple processes is completed before the control signal is due, maintaining real-time productivity while achieving improved reliability through multiple independent solutions.
Solution Approach 2:
The patent creates multiple copies of the optimization process (first, second, third optimization processes) that each independently solve the same control allocation problem. These copies operate in parallel and generate identical or similar solution vectors, allowing the system to vote on the best solution without significantly increasing overall processing time, thus maintaining productivity while improving reliability.
3Ease of operation
If iterative optimization methods are used in critical control strategies, then optimal control solutions can be obtained, but safety concerns arise when the optimizer fails to compute a good solution within the scheduled time
Solution Approach 1:
The patent prepares multiple independent optimization processes in advance, each capable of independently solving the control allocation problem. By having multiple pre-configured processes ready to execute in parallel, the system creates a safety cushion against failures. If one process fails to converge or produces an erroneous output, the other processes provide backup solutions, ensuring safety in critical control strategies.
Solution Approach 2:
The patent implements a voting scheme that provides feedback on the quality and consistency of solutions from multiple optimization processes. By comparing solution vectors and identifying the most frequent or consensus solution, the system can detect and eliminate erroneous outputs. This feedback mechanism ensures that only reliable control signals are generated, addressing safety concerns while maintaining ease of operation for optimal control.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling a technical system (no) in real time, comprising: sensing a state of the technical system; with the sensed state, initiating independent executions of a plurality of optimization processes (Ρ 1, P 2, P 3,...) configured to solve a predefined optimization problem related to optimal control of the technical system; after a predetermined delay, extracting a current solution vector (u 1, u 2, u 3,...) from each optimization process; computing differences (dij~||ui-uj ||) for pairs of the solution vectors; and on the basis of the differences, selecting at least one of the solution vectors for use in controlling the technical system.