• Journal of the Korean Society for Precision Engineering
• /
• v.12 no.10
• /
• pp.138-148
• /
• 1995

While Model-based Monitoring systems based on state observer theory have shown much promise in the laboratory, they have not been widely accepted by industry because, inpractice, these systems often have poor performance with respect to accuracy, band-width, reliability(false alarms), and robustness. In this paper, the linitations of the deterministic discrete-time state observer are investigated quantitatively from the machine monitoring viewpoint. The limitations in the transient and steady-state observer performance are quantified as estimation error bounds from which performance indices are selected. Each index represents the conditioning of the corresponding performance. By utilizing matrix norm theory, an unified main index is determined, that dominates all the indices. This index could from the basis for an observer design methodology that should improve the performance of model-based monitoring systems.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.13 no.1
• /
• pp.47-54
• /
• 1993

For the numerical analysis of free oscillation characteristics in a harbor with general boundary and bottom topography, finite element method is applied. The governing Helmholtz equation is transformed into a generalized matrix eigenvalue problem using the standard finite element procedure. A computer code is developed for the numerical evaluation of natural frequencies and free oscillation modes. In the eigensolution process, a shifting strategy is devised for the treatment of numerical singularity. Scaling of coefficient matrix is also found to be effective for the alleviation of numerical ill-conditioning. For the test problems, firstly, analytical and numerical solutions are compared and validity of the code is obtained. Hence the method is successfully applicable for the real-world problems with general geometric boundaries and bottom topography.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.6
• /
• pp.577-582
• /
• 2002

Neural network can be trained to approximate an arbitrary nonlinear function of multivariate data like the mini-mill crown values in Automatic Shape Control. The trained weights of neural network can evaluate or generalize the process data outside the training vectors. Sometimes, the blind modeling of the process data is necessary to compare with the scattered analytical model of mini-mill process in isolated electro-mechanical forms. To come up with a viable model, we propose the blind neural-based range-division domain-clustering piecewise-linear modeling scheme. The basic ideas are: 1) dividing the range of target data, 2) clustering the corresponding input space vectors, 3)training the neural network with clustered prototypes to smooth out the convergence and 4) solving the resulting matrix equations with a pseudo-inverse to alleviate the ill-conditioning problem. The simulation results support the effectiveness of the proposed scheme and it opens a new way to the data analysis technique. By the comparison with the statistical regression, it is evident that the proposed scheme obtains better modeling error uniformity and reduces the magnitudes of errors considerably. Approximatly 10-fold better performance results.

• Smart Structures and Systems
• /
• v.31 no.2
• /
• pp.131-140
• /
• 2023

As well-known, the extended Kalman filter (EKF) is a powerful tool for parameter identification with limited measurements. However, traditional EKF is not applicable when the external excitation is unknown. By using least-squares estimation (LSE) for force identification, an EKF with unknown input (EKF-UI) approach was recently proposed by the authors. In this approach, to ensure the influence matrix be of full column rank, the sensors have to be deployed at all the degrees-of-freedom (DOFs) corresponding to the unknown excitation, saying collocated measurements are required. However, it is not easy to guarantee that the sensors can be installed at all these locations. To circumvent this limitation, based on the idea of first-order-holder discretization (FOHD), an improved EKF with unknown input (IEKF-UI) approach is proposed in this study for the simultaneous identification of structural parameters and unknown excitation. By using projection matrix, an improved observation equation is obtained. Few displacement measurements are fused into the observation equation to avoid the so-called low-frequency drift. To avoid the ill-conditioning problem for force identification without collocated measurements, the idea of FOHD is employed. The recursive solution of the structural states and unknown loads is then analytically derived. The effectiveness of the proposed approach is validated via several numerical examples. Results show that the proposed approach is capable of satisfactorily identifying the parameters of linear and nonlinear structures and the unknown excitation applied to them.