Frequency response functions are often utilized to characterize a system's dynamic response. For a wide range of engineering applications, it is desirable to determine frequency response functions for a system under stochastic excitation. In practice, the measurement data is contaminated by noise and some form of averaging is needed in order to obtain a consistent estimator. With Welch's method, the discrete Fourier transform is used and the data is segmented into smaller blocks so that averaging can be performed when estimating the spectrum. However, this segmentation introduces leakage effects. As a result, the estimated frequency response function suffers from both systematic (bias) and random errors due to leakage. In this paper the bias error in the H_1 and H_2-estimate is studied and a new method is proposed to derive an approximate expression for the relative bias error at the resonance frequency with different window functions. The method is based on using a sum of real exponentials to describe the window's deterministic autocorrelation function. Simple expressions are derived for a rectangular window and a Hanning window. The theoretical expressions are verified with numerical simulations and a very good agreement is found between the results from the proposed bias expressions and the empirical results.

When dealing with nonlinear dynamical systems, it is important to have efficient, accurate and reliable tools for estimating both the linear and nonlinear system parameters from measured data. An approach for nonlinear system identification widely studied in recent years is "Reverse Path". This method is based on broad-band excitation and treats the nonlinear terms as feedback forces acting on an underlying linear system. Parameter estimation is performed in the frequency domain using conventional multiple-input-multiple- output or multiple-input-single-output techniques. This paper presents a generalized approach to apply the method of "Reverse Path" on continuous mechanical systems with multiple nonlinearities. The method requires few spectral calculations and is therefore suitable for use in iterative processes to locate and estimate structural nonlinearities. The proposed method is demonstrated in both simulations and experiments on continuous nonlinear mechanical structures. The results show that the method is effective on both simulated as well as experimental data.

With growing demands on product performance and growing complexity of engineering structures, efficient tools for analyzing their dynamic behavior are essential. Linear techniques are well developed and often utilized. However, sometimes the errors due to linearization are too large to be acceptable, making it necessary to take nonlinear effects into account. In many practical applications it is common and reasonable to assume that the nonlinearities are highly local and thus only affect a limited set of spatial coordinates. The purpose of this paper is to present an approach to finding the spatial location of nonlinearities from measurement data, as this may not always be known beforehand. This information can be used to separate the underlying linear system from the nonlinear parts and create mathematical models for efficient parameter estimation and simulation. The presented approach builds on the reverse-path methodology and utilizes the coherence functions to determine the location of nonlinear elements. A systematic search with Multiple Input/Single Output models is conducted in order to find the nonlinear functions that best describe the nonlinear restoring forces. The obtained results indicate that the presented approach works well for identifying the location of local nonlinearities in structures. It is verified by simulation data from a cantilever beam model with two local nonlinearities and experimental data from a T-beam experimental set-up with a single local nonlinearity. A possible drawback is that a relatively large amount of data is needed. Advantages of the approach are that it only needs a single excitation point that response data at varying force amplitudes is not needed and that no prior information about the underlying linear system is needed.