Traditional approaches of random error modeling like GM model and Allan variance method work unsatisfactorily selleck chemicals Idelalisib for MEMS sensors [14]. Moreover, the whole process of modeling the static and dynamic Inhibitors,Modulators,Libraries biases are extremely complex and sometimes do not provide the reliable estimates, thereby affecting the navigation accuracy of the system. Alternatively, artificial intelligence approaches utilizing Neural Network (NN) have been utilized in modeling the MEMS error and are found to perform better than other conventional techniques [14,15]. However in this particular case, NN suffers from poor generalization capability due to the presence of an elevated level of noises in the input-output data to be modeled. Hence, the NN model prediction accuracy is poor and deteriorates after a short time.
Also, the model development process takes longer time, which limits their real-time Inhibitors,Modulators,Libraries implementation [16]. Recently, Support Vector Machines (SVMs) based techniques have been applied to model the MEMS error [17,18]. Support Vector Machines (SVMs) based on the structural risk minimization principle can avoid local minimization and over-fitting problems as encountered in NN, thus improving the prediction accuracy. As opposed to neural networks, it requires less training time, and hence is suitable for real-time implementation. This paper thus proposes the implementation of an enhanced Nu-Support Vector Regression (Nu-SVR) technique for modeling these random and substantial MEMS sensor errors [19]. The proposed approach is different from those presented in [17,18], as it automatically selects the model parameter (i.
e., error margin), and the priori knowledge of the noise model is not mandatory [20]. Like the NN approach, Inhibitors,Modulators,Libraries the Nu-SVR model utilizes the same set of input-output sample pairs to model the errors. Once the Nu-SVR model is trained, it is utilized to predict the desirable output over an independent set of sample pairs. To test the efficacy of the proposed model, a low-cost MEMS Inertial Measurement Unit (IMU) manufactured by Cloud Cap Technology known as Crista Inhibitors,Modulators,Libraries IMU is employed [9].The paper has been divided into five sections. Section 2 covers the conventional approaches of modeling the MEMS sensor errors. Section 3 explains the working of support vector regression. Experimental setup and the calibration results obtained using conventional and the proposed approaches are detailed in Section 4, along with their impact on the navigation solution accuracy.
Finally, Section 5 concludes the paper.2.?Conventional Error Modeling ApproachesThere are numbers of errors like bias, scale factor, cross-axis sensitivity or misalignment, GSK-3 noise and temperature drifts that affect the performance of inertial sensors protein inhibitor [5]. Bias is the output observed when no input is applied. It can be divided into two parts, namely, static bias and dynamic bias.