Research On The Key Technology In Motion-force Decoupling Of A Wheel Force Transducer

The wheel force of a vehicle represents the interaction between the ground and the wheel. Real-time wheel force signal plays very important roles in the evaluation of active vehicle safety and performance, including road spectrum acquisition and bench simulation, tire dynamic characteristics, suspension dynamic response, traffic road perception and recognition, etc. The wheel force transducer (WFT) provides an effective tool for such real-time detection of the wheel force. So the development of WFT with independent intellectual property rights improves the vehicle testing technology of China. This paper focuses on a multi-axis WFT design, giving a systematic study on the sensor measurement and decoupling technique in motion state.The main contents are as follows:1. Aiming to improve the low measurement accuracy of our existing WFT, we firstly analyzed the error coupling mechanism, particularly the additional mass impact on the sensor itself. By introducing the measuring and decoupling approach of the spoke structure WFT, the inertial mechanics model is built with the Newton's law, followed by the verification using finite element simulation.2. An inertial measurement unit (IMU) was introduced and integrated into the WFT,so we can have both the motion and force detection. This makes up a multi-sensor system including a force-sensing body, an encoder, and an IMU. The sensor system is divided into several modules such as sensing, sampling, processing and transmission,then optimal design was illustrated in detail.3. The initial value problem of the multi-sensor WFT system was studied. Due to the actual installation, the coordinate systems of sensor-body, encoder, inertial sensor and the wheel are not exactly same, which will cause the initial installation errors. This paper calibrates these errors by using the rotation decoupling technique of the spinning wheels.4. The filtering method of wheel noise reduction and angular acceleration prediction was studied. The angular acceleration will have large drift error if directly differentiating the angular velocity from gyros with time interval. A Singer-QKF based filter which can suppress the noise and estimate real-time state is adapted.5. The nonlinear inertial force model was studied. The steady-state calibration result can be improved with continuous acceleration loading. The least squares support vector machine is used to predict the nonlinear inertia coupling errors.6. The performance calibration is studied from two aspects, one is for the force-force cross coupling issue, and the other is the acceleration calibration for the inertial coupling problem. The least square method was used to resolve their calibrations. To verify the motion-force decoupling, both simulation and real road tests are implemented with different conditions.