Time-series behavior

Forecasting of time series behavior using lead time data (data obtained during current experiment) for prediction of the material response to the similar actions and loads in future or of testing results for twin material specimens during lead time . 

I. INTRODUCTIQN TO QSAR D. Time-series behavior and C. Library design, compound... 

The idea of using mathematical modeling for describing materials behavior under loads is well known. Some physical phenomena, which can be observed in materials during testing, have time dependent quantitative characteristics. It gives a possibility to consider them as time series and use well developed models for their analysis [1, 2]. Usually applied... 

Among its many useful features is the ability to simulate both discrete and continuous CA, run in autorandoinize and screensaver modes, display ID CAs as color spacetime diagrams or as changing graphs, display 2D CAs either as flat color displays or as 3D surfaces in a virtual reality interface, file I/O, interactive seeding, a graph-view mode in which the user can select a sample point in a 1-D CA and track the point as a time-series, and automated evolution of CA behaviors. 

The application of time series techniques to electrochemical data is promising. It is possible to use the ARIMA analysis to study the behavior of a single coating system. It is also possible to use time series analysis to rank coatings with respect to the properties under study. 

In order to analyze both systems, some techniques from nonlinear science are burrowed. Firstly, a phase portrait is constructed from delay coordinates, a Poincare map is also computed, FFT is exploited to derive a Power Spectrum Density (PSD) Maximum Lyapunov Exponents (MLE) are also calculated from time series. Although we cannot claim chaos, the evidence in this chapter shows the possible chaotic behavior but, mostly important, it exhibits that the oscillatory behavior is intrinsically linked to the controlled systems. The procedures are briefly described before discuss each study case. 

Fig. 3. Filtered signal (solid line) and measured time series (dotted line) for the first experiment E.l (see text for details). Temperature oscillation were induced by recycle between heat exchanger and biological reactor. The filtered signal remains the oscillatory behavior of the system.

Figure 6 displays the Poincare maps for all experiments. Note that even the projections in canonical planes (see Figure 5) seem ordered in layers. That is, a toroidal structure can be seen form the Poincare surface. That is, small amplitude oscillations were detected in time series (see Figures 3 and 4) for all experiments. The t3rpical behavior of aperiodic (possibly chaotic) oscillations can be confirmed is one takes a look at the corresponding Poincare section... 

Fig. 7. Power spectrum density. The measured time series comprise several fundamental frequencies. Since frequencies have low-order (< 0.03 Hz) noise effect can be neglected. Note that if the values reference outlet substrate and the control gains decrease (experiment E.ld), then the number of fundamental frequencies in PSD decreases. This leads us to belief that there is a suitable values such that system displays hmit cycle. However, this behavior was not experimentally found.

A set of experiments on gas-liquid motion in a vertical column has been carried out to study its d3mamical behavior. Fluctuations volume fraction of the fluid were indirectly measured as time series. Similar techniques that previous section were used to study the system. Time-delay coordinates were used to reconstruct the underl3ung attractor. The characterization of such attractor was carried out via Lyapunov exponents, Poincare map and spectral analysis. The d3mamical behavior of gas-liquid bubbling flow was interpreted in terms of the interactions between bubbles. An important difference between this study case and former is that gas-liquid column is controlled in open-loop by manipulating the superficial velocity. The gas-liquid has been traditionally studied in the chaos (turbulence) context [24]. 

Essentially, MLE is a measure on time-evolution of the distance between orbits in an attractor. When the dynamics are chaotic, a positive MLE occurs which quantifies the rate of separation of neighboring (initial) states and give the period of time where predictions are possible. Due to the uncertain nature of experimental data, positive MLE is not sufficient to conclude the existence of chaotic behavior in experimental systems. However, it can be seen as a good evidence. In [50] an algorithm to compute the MLE form time series was proposed. Many authors have made improvements to the Wolf et al. s algorithm (see for instance [38]). However, in this work we use the original algorithm to compute the MLE values. 

A series of such mass balance differential equations representing all of the interlinked compartments are formulated to express a mathematical representation, or model, of the biological system. This model can then be used for computer simulation to predict the time course behavior of any given parameter in the model. 

The die-away behavior of the autocorrelation function (the long declining shape of the ACF) hints at a trend, because the die-away of an autocorrelation function in a stationary time series would show an exponential function. 

Figure 35 shows a typical example for a time series in which slight variation of one of the external parameters caused a transition from regular harmonic behavior to period doubling (147). Besides even more complex time series, irregular oscillations suggesting chaotic behavior were also observed, but have not yet, however, been fully analyzed as has Pt(l 10). 

A key factor in modeling is parameter estimation. One usually needs to fit the established model to experimental data in order to estimate the parameters of the model both for simulation and control. However, a task so common in a classical system is quite difficult in a chaotic one. The sensitivity of the system s behavior to the initial conditions and the control parameters makes it very hard to assess the parameters using tools such as least squares fitting. However, efforts have been made to deal with this problem [38]. For nonlinear data analysis, a combination of statistical and mathematical tests on the data to discern inner relationships among the data points (determinism vs. randomness), periodicity, quasiperiodicity, and chaos are used. These tests are in fact nonparametric indices. They do not reveal functional relationships, but rather directly calculate process features from time-series records. For example, the calculation of the dimensionality of a time series, which results from the phase space reconstruction procedure, as well as the Lyapunov exponent are such nonparametric indices. Some others are also commonly used ... 

Although the detailed features of the interactions involved in cortisol secretion are still unknown, some observations indicate that the irregular behavior of cortisol levels originates from the underlying dynamics of the hypothalamic-pituitary-adrenal process. Indeed, Ilias et al. [514], using time series analysis, have shown that the reconstructed phase space of cortisol concentrations of healthy individuals has an attractor of fractal dimension dj = 2.65 0.03. This value indicates that at least three state variables control cortisol secretion [515]. A nonlinear model of cortisol secretion with three state variables that takes into account the simultaneous changes of adrenocorticotropic hormone and corticotropin-releasing hormone has been proposed [516]. 

---