Torque Estimation Using Neural Network and Acceleration Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing torque estimation methods are inaccurate as they rely on predictions based on information before the torque is generated, lacking real-time data and sensor measurements, which can deviate from actual torque values.
Innovation Solution
A torque estimation device using a trained neural network that processes input data including accelerator operation amount, vehicle acceleration, vehicle speed, brake pressure, road slope, and gear ratio to estimate torque in the power transmission member, allowing for more accurate predictions without the need for torque-sensing sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If torque is predicted based on information before torque is generated (accelerator operation amount), then the estimation can be performed without torque sensors, but the predicted value deviates from actual torque
Solution Approach 1:
The system performs preliminary actions by collecting and storing acceleration data and accelerator operation amount data before torque estimation is needed. The neural network is pre-trained with this historical data so that when estimation is required, the system can quickly provide accurate torque values without needing torque sensors during the estimation process.
Solution Approach 2:
The system introduces feedback by using actual acceleration data (which reflects the real torque effect on vehicle motion) as input to the neural network. This creates a feedback loop where the actual vehicle response to torque is used to improve estimation accuracy, allowing the system to learn from real torque effects and continuously improve its predictions.
2Device complexity
If only accelerator operation amount is used for torque prediction, then the system remains simple without additional sensors, but the prediction accuracy is insufficient
Solution Approach 1:
The system introduces acceleration data as an intermediary element that mediates between the accelerator operation amount and the torque estimation. Acceleration serves as a bridge that reflects the actual torque effect on vehicle motion, allowing the neural network to infer torque more accurately without directly measuring it with sensors.
Solution Approach 2:
The system replaces the mechanical torque measurement system (torque sensors) with an information processing system (neural network). Instead of mechanically measuring torque directly, the system uses computational methods to estimate torque based on acceleration data and accelerator operation amount, substituting mechanical measurement with intelligent estimation.
3Reliability
If prediction is based only on pre-torque information, then real-time torque feedback is unavailable, but implementing sensor-based measurement increases system complexity
Solution Approach 1:
The system performs self-service by using the vehicle's own existing sensors (acceleration sensor and accelerator position sensor) to generate torque estimation. The vehicle's motion response itself provides the feedback needed for accurate estimation, eliminating the need for separate torque measurement systems while maintaining high reliability.
Data Source
AI summary
A control device serving as a torque estimation device includes a storage device and a processing circuit. The storage device stores data of a trained neural network. The trained neural network is trained using training data including data of an actually-measured torque that is measured, data of an accelerator operation amount in a period of a predetermined length up to a time point of measurement of the actually-measured torque, and data of an acceleration of a vehicle from the time point of measurement of the actually-measured torque onward. The processing circuit inputs, to the trained neural network stored in the storage device, input data including the data of the accelerator operation amount and the data of the acceleration of the vehicle, to estimate a torque generated in a power transmission member.