Prefilter Design for Improved Spectral Flatness of Physical Pseudorandom Perturbations

This paper documents recent advances towards an efficient computational method for accurately determining the static and dynamic stability and control (SC. In contrast to the “brute force” approach to filling an entire S&C database for an aircraft, the present approach is to reduce the number of simulations required to generate a complete aerodynamic model of a particular configuration at selected flight conditions by using one or a few complex dynamic motions and nonlinear system identification (SID) techniques. The approach is demonstrated by gathering high-fidelity computational fluid dynamics (CFD) data for a rigid F-16 in prescribed motion that approximates dynamic wind-tunnel testing techniques and SID input signals. The motions are optimized to minimize the computational expense and to take full advantage of the tighter control of the CFD environment. They are specified interactively using a newly developed, GUI-based maneuver file generation tool. Global nonlinear parameter modeling and other SID techniques are then used to identify parametric models from the computed aerodynamic force and moment data. These compact models are used to predict the aerodynamic response to maneuvers that were computed for validation purposes and that were not used to derive the models. Partial derivatives of the analytical models can be used to determine the corresponding static and dynamic stability derivatives. The models can also be used to perform real time 6-DOF/aeroelastic simulations of the vehicle in conditions susceptible to spin, tumble, and lateral/longitudinal instabilities. The main benefits of this effort are: 1) early discovery of complex aerodynamic phenomena that are typically only present in dynamic flight maneuvers and therefore not discovered until flight test, and 2) rapid generation of an accurate aerodynamic model to support aircraft and weapon certification by reducing required flight test hours and complementing current stability and control testing.

<div class="section abstract"><div class="htmlview paragraph">High fidelity mathematical vehicle models that can accurately capture the dynamics of car suspension system are critical in vehicle dynamics studies. System identification techniques can be employed to determine model type, order and parameters. Such techniques are well developed and usually used on linear models. Unfortunately, shock absorbers have nonlinear characteristics that are non-negligible, especially with regard the vehicle's vertical dynamics. In order to effectively employ system identification techniques on a vehicle, a nonlinear mathematical shock absorber model must be developed and then coupled to the linear vehicle model. Such an approach addresses the nonlinear nature of the shock absorber for system identification purposes. This paper presents an approach to integrate the nonlinear shock absorber model into the vehicle model for system identification.</div><div class="htmlview paragraph">Three empirical mathematical models are proposed that describe the relationship between shock velocity and shock force. These models are used to predict the force behavior of the shock absorber when the shock absorber velocities ranged from 0 to 1.5m/s, in both compression and extension direction. Empirical shock absorber models can be quickly developed using experimental data and integrated into the larger vehicle model for complete system identification. The shock velocity calculated from the vehicle model is the input to the shock absorber model and the shock force is the feedback to the vehicle model. The influence of nonlinear shock absorber models on system identification for the half car model was analyzed. The result shows that the type of shock absorber model did not have much influence on vehicle system identification using simulated data generated from the model. Furthermore, the system identification method, integrating the nonlinear shock absorber dynamics model into the vehicle model, works well for the simulated data. Future work will focus on the use of experimental data.</div></div>

The representation of flow field via reduced order models (ROM) has proven to be successful in reducing the degree of freedom of the system. ROM can be developed in two ways: (a) system identification techniques and b) Galerkin projection of Proper orthogonal decomposition (POD) modes onto the Governing equations such as Navier-Stokes equations. In this study, we investigate a canonical problem of incompressible flow past a circular cylinder. We first compute the POD modes and then determine the corresponding temporal coefficients. Some of these coefficients are used as training data for system identification techniques. We employ two regressive methods as a system identification strategy for predicting the temporal coefficients of ROM. On the other hand, we also develop a ROM through Galerkin projection technique and compare it with the system identification techniques. Based on this comparison, we have discussed the accuracy, advantages, and limitations of these methods in ROM.

In this paper, a single-ended quality measurement algorithm for noise suppressed speech is described.The proposed algorithm computes fast approximations of Kullback-Leibler distances between Gaussian mixture (GM) reference models of clean, noise corrupted, and noise suppressed speech and a GM model trained online on the test speech signal. The distances, together with a spectral flatness measure, are mapped to an estimated quality score via a support vector regressor. Experimental results show that substantial improvement in performance and complexity can be attained, relative to the current state-of-art single-ended ITU-T P.563 algorithm. Due to its modular architecture, the proposed algorithm can be easily configured to also perform signal distortion and background intrusiveness measurement, a functionality not available with current standard algorithms.

Real Time Suppression of Howling Noise in Public Address System

A psychoacoustics model that approximates the masking threshold is presented. The significance of the tonality index is highlighted by comparing two well-known methods used in its estimation: unpredictability and spectral flatness measure. The first method has a tendency to give a lower threshold in the low frequency (below 3 kHz) region whereas the second one tends to be lower in the high frequency range. A new method is proposed which combines both techniques to achieve better results. This technique moderates the complexity of the two earlier methods and increases the compression ratio considerably without compromising the audio quality.

In this paper, we present an approach to incorporate discriminative weight training into a statistical model-based voice activity detection (VAD) method. In our approach, the VAD decision rule is derived from the optimally weighted likelihood ratios (LRs) using a minimum classification error (MCE) method. An adaptive on-line means of selecting two kinds of weights based on a power spectral flatness measure (PSFM) is devised for performance improvement. The proposed approach is compared to conventional schemes under various noise conditions, and shows better performance.

The Szego-Kolmogorov-Krein theorem is the natural basis for the spectral flatness measure commonly advocated in linear predictive speech processing and parametric spectrum analysis. From this theorem it follows that the logarithm of any normalized spectrum averages to zero. The normalization constant is the minimum prediction error for the underlying process. Jensen's theorem is introduced as a practical method of computing prediction error and spectral flatness for rational spectrum models. The resulting computation algorithm is closely related to Fejer's factorization of rational spectra. Thus our discussion generalizes Makhoul's discussion of the so-called Fejer method for computing prediction error in MA models. One property of any measure of spectral flatness that would seem essential is the following: the introduction of white noise should increase spectral flatness. This property is established by appealing to a classical information-theoretic theorem.

Mooring designs are a critical concern for the offshore renewable industry. The common design processes for mooring systems involve multiple, lengthy cycles of simulation and experimentation to achieve suitable designs. In this paper, we propose, for the first time, a computationally efficient and physically transparent data-driven design optimisation approach using real measurements from only an initial estimated specification of the physical mooring system. This is applied to a complex renewable energy platform, the multiple float wave energy converter M4, by system identification with frequency response driven optimisation, computing the optimised design within an hour on a laptop. More specifically, although mooring force is highly nonlinear in relation to wave input, it is demonstrated by system identification that it is almost linear in relation to surge motion. Further, the low frequency force, separated from the linear wave frequency force by a wavelet filter, is modelled as a resonant mass spring damper. This approach thus effectively models the nonlinear mooring system using two linear models. Finally, the physical parameters, namely mass, damping ratio, and stiffness, are optimised and compared using parameter sweeping and convex optimisation methods to minimise fatigue loads. The results demonstrate that the proposed approach is cost-effective and computationally efficient for mooring system optimisation for complex floating systems. • A data-driven method for mooring design is introduced using system identification and frequency analysis. • Nonlinear mooring forces are modelled accurately using two linear models. • Demonstrating effectiveness on a complex platform - M4 wave energy converter. • It requires one initial design and yields interpretable, cost-effective results. • The optimisation completes within an hour on a standard laptop. • It halves fatigue estimates compared to initial designs.

In this paper, a novel algorithm for partial discharge (PD) pulse waveform analysis and separation based on orthogonal series expansion using Laguerre polynomials is proposed. A system identification technique is also developed to help increase the accuracy rate of the signal classification. Laguerre polynomial series expansions are applied on the PD pulses to effectively extract important information and well-discriminative features. Laguerre expansion coefficients inherently contain information about the amplitude (intensity) and the relaxation time of a dynamic system and this helps direct characterization of PD pulses. The proposed technique is applied to PD pulses acquired from a laboratory set of multiple PD sources. It is shown that this statistical method can classify partial discharge signals with high accuracies, even when the signals are visually indistinguishable. In this research, two methods are developed for system identification. One is based on a deterministic approach in the form of a recursive formula and the other one is a stochastic method based on a group Lasso methodology. The aim of system identification is to improve the classification accuracy by removing the effect of the measurement system from the observed PD waveforms. The accuracy rate of the proposed algorithm is evaluated using the signals that are captured from simultaneous artificial defects in high pressure SF <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</inf> and in low pressure air.

: This report summarizes research into the application of system identification techniques to simulation model abstraction. System identification produces simplified mathematical models that approximate the dynamic behaviors of the underlying stochastic simulations. Four state-space system identification techniques were examined: Canonical State-space, Compartmental Models, Maximum Entropy, and Hidden Markov Models (HMM). Two stochastic simulation models were identified: the a simulation of two opposing forces, each operating with multiple weapon system types; and the Simulation, a simulation of a squadron of aircraft performing battlefield air interdiction. The system identification techniques were evaluated and compared under a variety of scenarios on how well they replicate the distributions of the simulation states and decision outputs. Encouraging results were achieved by the HMM technique applied to Attrition Simulation - and by the Maximum Entropy technique applied to the Mission Simulation. This report also discusses the run-time performance of the algorithms, the development of suitable model structures, and implications for future efforts.

Alternative diesel fuels have been the subject of extensive investigation. Fatty acid methyl ester (FAME) based Biodiesel manufactured from vegetable oils or animal fats is an excellent candidate to replace common diesel fuel being renewable, non-toxic and often giving rise to reduced exhaust gas emissions. The transesterification process has been commonly and widely used to produce biodiesel from vegetable oil or animal fat. Vegetable oils or animal fats generally have viscosities higher than standard diesel oil. This means that it is necessary to reduce the viscosity by means of reacting vegetable oil with alcohol in the presence of a suitable catalyst. The target product for this reaction is methyl ester, with glycerol and potentially soap produced as by products with the process of transesterification. Methylester (Biodiesel) is produced by converting triglycerides to alkylesters. A batch transesterification process has two significant mechanisms, and exhibits a mass transfer controlled region that precedes a second order kinetically controlled region. In order to control the conversion process it is useful to employ process monitoring. In particular monitoring of the mass transfer processes that limits the initial reaction rates could prove to be beneficial in allowing for process optimization and control.&#13;\n&#13;\nThis thesis proposes the use of a new method of biodiesel process monitoring using low frequency (15kHz) impedance sensing which is able to provide information regarding the progress of mass transfer and chemical reaction during biodiesel production. An interdigitated (ID) sensor has been used to monitoring the biodiesel process The ID sensor is of simple construction and consists of two sets of interleaved electrodes (fingers). The two sets of electrodes are separated by a gap and when an AC excitation voltage is applied across the interleaved electrodes an oscillating electric field is developed. The response of the fluid surrounding the sensor to the applied excitation was then used to determine progress of the chemical reaction by evaluating the real and complex impedance. A significant and unambiguous change in the components of impedance has been shown to occur during mixing (mass transfer) and transesterification.&#13;\n&#13;\nThe impedance measurements gained during transesterification were then used for the development of a system model. A systematic approach was used to select mathematical models and system identification techniques were evaluated. The system identification investigation used real process measurement data in conjunction with the Matlab system identification toolbox.

The dynamic characteristic of a drum boiler is complex and this complication leads to the difficulty in controlling the output of the system, i.e., steam pressure. Therefore, this study attempts to investigate the application of two model predictive methods, artificial neural network (ANN) and system identification, in order to assess the performance of each method. According to the system, the inputs are feed water flow rate and applied heat while the output is the steam pressure. The ANN method used is based on a training algorithm, Levenberg-Marquardt back propagation. On the other hand, the optimal model of system identification method is the output error (OE). The performance measurement is compared by considering the mean squared error (MSE) after fitting the simulated prediction from each model to the observation. The results show that ANN slightly outperforms the system identification technique. Moreover, another finding is that ANN method is capable of identifying the outlier among the observations so it is robust to the disturbances.

Mobile sensing has emerged as an economically viable alternative to spatially dense stationary sensor networks, leveraging crowdsourced data from today's widespread population of smartphones. Recently, field experiments have demonstrated that using asynchronous crowdsourced mobile sensing data, bridge modal frequencies, and absolute mode shapes (the absolute value of mode shapes, i.e., mode shapes without phase information) can be estimated. However, time‐synchronized data and improved system identification techniques are necessary to estimate frequencies, full mode shapes, and damping ratios within the same context. This paper presents a framework that uses only two time‐synchronous mobile sensors to estimate a spatially dense frequency response matrix. Subsequently, this matrix can be integrated into existing system identification methods and structural health monitoring platforms, including the natural excitation technique eigensystem realization algorithm and frequency domain decomposition. The methodology was tested numerically and using a lab‐scale experiment for long‐span bridges. In the lab‐scale experiment, synchronized smartphones atop carts traverse a model bridge. The resulting cross‐spectrum was analyzed with two system identification methods, and the efficacy of the proposed framework was demonstrated, yielding high accuracy (modal assurance criterion values above 0.94) for the first six modes, including both vertical and torsional. This novel framework combines the monitoring scalability of mobile sensing with user familiarity with traditional system identification techniques.

This paper presents the system identification technique by applying PZT actuators and PVDF sensors. An array of finite-length PVDF films is equally spaced and distributed over a cantilever beans acting as sensors. Two finite-length PZT patches are bonded symmetrically and excited 180°out-of-phase for pure bending excitation. The theoretical base of modal analysis for system identification of piezoceramic beams is presented to derive the frequency response functions (FRFs) in conventional modal format. The physical interpretation of P11 and PVDF mode shape functions are characterized respectively. The FRFs between the PZT actuators and PVDF sensors are experimentally measured. By the operation of a series of FRFs and the application of modal parameter extraction method, the modal parameters. including system natural frequencies, mode shapes and modal damping ratios, can be obtained. Both theoretical and experimental approaches agree very well. The system information can be used for control applications and also be applicable to structural failure diagnosis.