论文信息 - Data Fusion based on RBF Neural Network for Error Compensation in Resistance Strain Gauge Force Transducers

Data Fusion based on RBF Neural Network for Error Compensation in Resistance Strain Gauge Force Transducers

Many factors, such as environmental temperature and material elasticity, can affect the output of resistance strain gauge force transducers used in vehicle traction force measurements. A data fusion method based on radial basis function (RBF) neural network is proposed to reduce the negative effects and compensate the measurement error. A multiquadrics kernel is utilized as the kernel function for the RBF neural networks. It fuses the environmental temperature in the force measurement while realizing an accurate compensation of errors. Tests have been carried out within temperature ranging from -10deg C to 60degC and the results show that the maximum error with load 80000N is below 0.5 % after compensation while it is greater than 6 % before compensation.

[1] Kevin Swingler,et al. Applying neural networks - a practical guide , 1996 .

[2] Ganapati Panda,et al. Artificial neural network-based nonlinearity estimation of pressure sensors , 1994 .

[3] Zhao Wang. A New Method to Improve Pressure Sensor Precision Based on RBF Neural Network , 2004 .

[5] Weifeng Tian,et al. Application of multisensor data fusion based on RBF neural networks for fault diagnosis of SAMS , 2002, 7th International Conference on Control, Automation, Robotics and Vision, 2002. ICARCV 2002..

[6] Shien Hui,et al. A high-accuracy, high-speed interface circuit for differential-capacitance transducer , 2006 .

[7] Simon Haykin,et al. Neural Networks: A Comprehensive Foundation , 1998 .

[8] James Llinas,et al. An introduction to multisensor data fusion , 1997, Proc. IEEE.