Linearized Model Predictive Control for Powertrain Torque Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional engine control systems fail to accurately control engine output torque, especially in propulsion systems with continuously variable transmissions, leading to suboptimal balance between drivability, performance, and fuel economy due to the nonlinear relationship between axle torque and transmission ratio.
Innovation Solution
A model predictive control system is implemented to calculate and subtract a disturbance based on the relationship between engine output torque and transmission ratio, allowing for the optimization of axle torque and fuel economy by linearizing requested and measured values, and generating sets of possible command values to determine the lowest cost solution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional engine control systems are used to control engine output torque, then the control system is simple and easy to implement, but the torque control accuracy is insufficient and cannot achieve optimal balance between drivability, performance, and fuel economy
Solution Approach 1:
The patent transforms the nonlinear control problem into a linear one by introducing a disturbance parameter that captures the nonlinear effects. The controller operates in linearized coordinates where the relationship between control inputs and outputs is linear, allowing simple linear control techniques to achieve accurate torque control while maintaining system simplicity
Solution Approach 2:
The patent introduces an intermediary transformation layer that converts between nonlinear physical coordinates and linearized control coordinates. This intermediary framework allows the controller to work with linear relationships while accurately representing the nonlinear plant dynamics, resolving the contradiction between control accuracy and system complexity
2Use of energy by moving object
If model predictive control systems are used to optimize axle torque and fuel economy, then fuel economy and performance are improved, but the systems cannot work well with nonlinear elements such as the relationship between axle torque output and transmission ratio
Solution Approach 1:
The patent changes the parameter space by introducing linearized coordinates that transform the nonlinear relationship between axle torque and transmission ratio into a linear relationship. This allows model predictive control to effectively handle the nonlinear elements while optimizing fuel economy, as the controller operates in a linearized framework where standard MPC techniques are applicable
3Ease of operation
If the nonlinear relationship between axle torque and transmission ratio is directly controlled, then the control approach is straightforward, but the control accuracy deteriorates due to the nonlinear elements
Solution Approach 1:
The patent applies parameter transformation by defining linearized coordinates that simplify the nonlinear control relationships. The controller operates with linearized parameters where the relationship between transmission ratio and axle torque is linear, maintaining ease of operation while achieving high control accuracy through the transformed parameter space
Data Source
AI summary
A propulsion system, control system, and method are provided for optimizing fuel economy, which use model predictive control systems to generate a plurality of sets of possible command values and determine a cost for each set of possible command values based on weighting values, a plurality of predicted values, and a plurality of requested values. The set of possible command values having the lowest cost is determined. A linearized axle torque requested value and a linearized axle torque measured value are each created by subtracting an estimated disturbance. The estimated disturbance is determined based on a model of a relationship between measured engine output torque and measured transmission ratio. The linearized axle torque measured value is used to compute the predicted values, which are used to determine the cost. The linearized axle torque requested value is also used to determine the cost.