A Frequency Domain Analysis of Learning Control

1994 ◽  
Vol 116 (4) ◽  
pp. 781-786 ◽  
Author(s):  
C. J. Goh

The convergence of learning control is traditionally analyzed in the time domain. This is because a finite planning horizon is often assumed and the analysis in time domain can be extended to time-varying and nonlinear systems. For linear time-invariant (LTI) systems with infinite planning horizon, however, we show that simple frequency domain techniques can be used to quickly derive several interesting results not amenable to time-domain analysis, such as predicting the rate of convergence or the design of optimum learning control law. We explain a paradox arising from applying the finite time convergence criterion to the infinite time learning control problem, and propose the use of current error feedback for controlling possibly unstable systems.

Author(s):  
Arcandra Tahar ◽  
John Halkyard ◽  
Mehernosh Irani

The Horn Mountain Spar is located in 1,654 m of water about 135 km from Venice, Louisiana in the Gulf of Mexico. The facility was instrumented extensively to measure key spar and riser response parameters (Edwards et. al. 2003). Halkyard et. al. (2004) and Tahar et. al. (2005) have compared measured spar responses such as motion and mooring line tensions with numerical predictions. This paper extends the work done on comparison of the full scale data during hurricane Isidore. All previous numerical simulations were based on a time domain analysis procedure. One concern related to this method is that it is computationally intensive and time consuming. In the initial stages of a project, a frequency domain solution may be an effective tool compared with a fully coupled time domain analysis. The present paper compares results of time domain and frequency domain simulations with field measurements. Particular attention has been placed on the importance of the phase relationship between motion and excitation force. In the time domain analysis, nonlinear drag forces are applied at the instantaneous position. Whereas in the frequency domain analysis, nonlinear drag forces are stochastically linearized and solutions are obtained by an iterative procedure. The time domain analysis has better agreement with the field data compared to the frequency domain. Overall, however, the frequency domain method is still promising for a quick and approximate estimation of relevant statistics. With advantages in terms of CPU time, the frequency domain method can be recommended as a tool in pre-front end engineering design or in a phase where an iterative nature of design of an offshore structure takes place.


Energies ◽  
2021 ◽  
Vol 14 (12) ◽  
pp. 3606
Author(s):  
Jing-Yuan Lin ◽  
Chuan-Ting Chen ◽  
Kuan-Hung Chen ◽  
Yi-Feng Lin

Three-phase wye–delta LLC topology is suitable for voltage step down and high output current, and has been used in the industry for some time, e.g., for server power and EV charger. However, no comprehensive circuit analysis has been performed for three-phase wye–delta LLC. This paper provides complete analysis methods for three-phase wye–delta LLC. The analysis methods include circuit operation, time domain analysis, frequency domain analysis, and state–plane analysis. Circuit operation helps determine the circuit composition and operation sequence. Time domain analysis helps understand the detail operation, equivalent circuit model, and circuit equation. Frequency domain analysis helps obtain the curve of the transfer function and assists in circuit design. State–plane analysis is used for optimal trajectory control (OTC). These analyses not only can calculate the voltage/current stress, but can also help design three-phase wye-delta connected LLC and provide the OTC control reference. In addition, this paper uses PSIM simulation to verify the correctness of analysis. At the end, a 5-kW three-phase wye–delta LLC prototype is realized. The specification of the prototype is a DC input voltage of 380 V and output voltage/current of 48 V/105 A. The peak efficiency is 96.57%.


Author(s):  
Bonjun Koo ◽  
Jang Whan Kim

The Extendable Draft Platform (EDP) is a deep draft, column stabilized platform with a deck box support for topsides and a single, deep draft heave plate that provides suitable motion characteristics to enable the use of dry tree top tensioned risers. The EDP can be fabricated with topsides installed on the deck box and commissioned quayside in a typical construction yard. With the columns in the retracted position, the EDP floats on its deck box and can be towed, in this configuration, to the location of interest. Once the EDP is transported to its final site, the columns and heave plate are lowered to their final operating draft. During the lowering sequence, the deck box and the lower hull become two relatively independent bodies, mechanically connected by chains that control the lowering of the columns and heave plate, and the guides between the deck box and the columns. This multi-body system is hydrodynamically coupled because of radiated and diffracted waves. The global performance analyses of the installation process (lowering of the lower hull) are carried out by three different methods. The first method is frequency-domain analysis by WAMIT and a frequency domain motion solver. In the frequency domain analysis, all the mechanical connections are modeled as linear springs. The second method is time-domain, partially coupled analysis using HARP/WINPOST. In this analysis, the off diagonal 6×6 hydrodynamic interactions are ignored. The last method is a time domain, fully coupled analysis using HARP/WINPOST. In this analysis, full 12×12 hydrodynamic interactions are considered. In the time domain analyses, the mechanical couplings between each column and deck box are modeled with linear springs and the chain connections are modeled with slender rods by using the nonlinear finite element method. This paper presents and compares analysis results based on the three methods for relative motions and loads between the deck box and the lower hull during the lowering of the columns and heave plate.


2015 ◽  
Vol 31 (5) ◽  
pp. 292-308 ◽  
Author(s):  
Jianhua Wu ◽  
Toyin Ajisafe ◽  
Matthew Beerse

This study used both time and frequency domain analyses to investigate walking patterns with ankle load in children and adults. Twenty-two children aged 7–10 years and 20 young adults participated in this study. Three levels of ankle load were manipulated: no load, low load (2% of body mass on each side), and high load (4% of body mass on each side). An instrumented treadmill was used to register vertical ground reaction force (GRF) and spatiotemporal parameters, and peak vertical GRFs were determined. A frequency domain analysis was conducted on the vertical GRF data. Results demonstrate that, in the time domain, children showed adult-like spatiotemporal parameters and adult-like timing and magnitude of the 2 peak vertical GRFs under each load. In the frequency domain, children produced a lower power from the second harmonic than young adults, although both groups showed the highest power from this harmonic and increased this power with ankle load. It was concluded that children aged 7–10 years may start showing adult-like neuromuscular adaptations to increasing ankle load and display similar spatiotemporal control of foot falls and foot–floor kinetic interaction; however, a frequency domain analysis is effective in revealing different kinetic and neuromuscular characteristics between children and adults.


2020 ◽  
Vol 12 (09) ◽  
pp. 2050103
Author(s):  
Alvin Maningding ◽  
Mojtaba Azadi

The force response of poroelastic materials including poroelastic gels to indentation is known to be time- and space-dependent (i.e., a function of indenter shape and size). Despite the complexity of the poroelastic response and in contrast to viscoelastic mechanics, poroelastic mechanics can be captured in terms of several intrinsic mechanical properties, such as elasticity, permeability, and Poisson ratio. While these intrinsic properties can be found from time-domain or frequency-domain master curves, indentation is usually conducted and analyzed only in the time domain using stress-relaxation or creep experiments. This paper advocates using frequency-domain analysis of poroelastic gels by reviewing and analyzing the relevant works of the literature. The analysis and methods, proposed here, enable researchers to characterize dynamic moduli of poroelastic gels in frequency domain using only a few experimental defining parameters. The authors have intentionally provided extensive details and background, to make this work useful for researchers who consider using frequency-domain analysis for the first time. This work reviews and explains the instantaneous elastic modulus, depicted over normalized time as a unifying and understandable set of master curves for time-domain stress relaxation tests on poroelastic gels for cylindrical, conical, and spherical indenters. The dynamic elastic modulus, depicted over normalized frequency, are derived symbolically and numerically and explained for the first time as master curves with simple transfer function in the frequency domain for presenting poroelastic mechanics of gels.


2012 ◽  
Vol 429 ◽  
pp. 195-199
Author(s):  
Xiao Lei Zhao ◽  
Ming Rong Ren ◽  
Ya Ting Zhang ◽  
Pu Wang

The research and detection of heart disease depends on the analysis of the characteristic of electrocardio signal. Current analysis methods mainly include: (1) time domain analysis is a common used approach. With experience learned by observation and calculation, researchers examine errors and interferences to calculate means and variances directly within time domain. Analysis quality of this method demands higher request for researchers’ experience and skill although it’s a direct and significant result. (2) Frequency domain analysis, such as spectrum estimation, is largely applied to electrocardio signal researches and clinical applications. The analysis reflects abundant electrocardio activities, but failed to show details of the characteristics due to lack of time information. (3) time-frequency domain analysis describes energy density under different time and frequency of electrocardio signal at one time. It clarifies the relationship of signal frequency’s changing along with time such as wavelet transform method. (4) Nonlinear analysis is generally applied to biomedicine signal research in recent years. Correlation dimension, kolmogorov entropy, lyapunov component are major research methods to estimate some nonlinear dynamic parameters to represent the characteristic of electrocardio signal.


Author(s):  
Chan K. Yang ◽  
D. H. Lee ◽  
M. H. Kim ◽  
B. N. Park ◽  
Y. T. Yang ◽  
...  

In this paper, the float-off operation of a land-built crude oil tank (COT) loaded out and towed on launching dual-submersible barges is numerically simulated by a time-domain vessel-mooring-riser coupled dynamic analysis program with multiple floating bodies. The study is particularly focused on the maximum load prediction on connectors and the minimum gap prediction between barges and the COT. In case of simpler modeling, the time-domain simulation results are compared with frequency-domain results. Then, the time-domain model is run for more realistic situation with hawsers and mooring line. In hydrodynamic analysis, the interactions among the 3 floating bodies are fully taken into consideration. In the frequency-domain analysis, the connectors between barges are modeled by equivalent translational and rotational springs, the stiffness of which is estimated using Euler’s beam theory. In order to assess the possible occurrence of contact between COT and barges, the relative motions between barges and the COT at several points of interest were investigated.


Sign in / Sign up

Export Citation Format

Share Document