Neural Network Torque Limits for Steering Firewall Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing steering systems require time-consuming and knowledge-intensive tuning of firewalls to protect against erroneous or malicious torque commands, especially in closed-loop feedback control systems, which are complex and prone to human error.
Innovation Solution
A method and system using a trained artificial neural network to calculate functional limits for assist torque in steering systems, reducing the need for manual tuning and providing a universal solution for both open-loop and closed-loop feedback control systems by processing inputs such as handwheel torque, motor velocity, and vehicle speed to generate assist torque limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional firewall tuning methods are used to protect against erroneous or malicious torque commands, then the steering system gains protection functionality, but the system becomes complex and requires time-consuming, knowledge-intensive tuning that is prone to human error
Solution Approach 1:
The neural network automatically determines appropriate firewall limits by learning from training data containing normal and erroneous torque commands. The system self-configures the protection parameters without requiring manual tuning expertise, thereby maintaining reliability while eliminating the complexity and human error associated with manual firewall tuning
Solution Approach 2:
The patent replaces the manual mechanical tuning process with an automated computational approach using neural networks. The neural network processes torque command data and automatically establishes functional limits, substituting the knowledge-intensive manual tuning process with an automated intelligent system that requires no specialized tuning knowledge
2Reliability
If manual firewall tuning is performed for each steering system configuration, then specific protection requirements are met, but the process becomes time-consuming and requires expert knowledge
Solution Approach 1:
The neural network is pre-trained offline with comprehensive training data representing various normal and erroneous operating conditions. This preliminary training action allows the network to be deployed in steering systems without requiring time-consuming on-site tuning, as the protection functionality is already optimized for diverse scenarios before deployment
Solution Approach 2:
The neural network automatically adapts to specific steering system configurations through self-service learning during operation or through automated training processes, eliminating the need for manual tuning time while maintaining reliable protection specific to each system configuration
3Adaptability or versatility
If separate firewall configurations are created for open-loop and closed-loop feedback control systems, then specific control requirements are met, but the device complexity and tuning burden increase
Solution Approach 1:
The neural network is designed as a universal firewall solution that can be applied to both open-loop and closed-loop feedback control systems without requiring separate configurations. The network learns from training data representing various control system types and automatically adapts its behavior, providing multi-functional protection that eliminates the need for multiple specialized firewall configurations
Data Source
AI summary
A method for calculating at least one functional limit for a requested assist torque in a steering system. The method includes receiving at least one input and communicating the at least one input to an artificial neural network, wherein the artificial neural network is configured to calculate an assist torque limit corresponding to the requested assist torque. The method also includes receiving, from the artificial neural network, the assist torque limit corresponding to the requested assist torque and controlling at least one aspect of the steering system using the requested assist torque and the assist torque limit.