.jpg)
SYSTEM FOR SYNTHESISING AND CONTROLLING MULTIVARIATE RANDOM SIGNALS
WO21240040
Disclosed is a system (1.a) for synthesising and controlling multivariate random signals for the generation of random Gaussian and non-Gaussian signals in order to control a multiaxial vibration machine (1.d) or record the signals for subsequent use, the system comprising an inverse matrix transformation module (2.a), a channel estimator (2.b), a spectral control module (2.c), a kurtosis estimator (2.d), a kurtosis control module (2.e), a non-Gaussian noise synthesiser (2.f) and a matrix transformation module (2.g) for transforming the remote control signals.
- Much improved reproduction of real-world environments for acoustic and vibration tests (e.g., for vehicle components testing) respect to univariate or single-axis systems: by synthesising and controlling multiple signals that are transmitted and connected to a set of electrodynamic or electrohydraulic actuators, which in turn, simultaneously transform the different signals into motions in various directions in space. <br> Unlike other systems, this one proposes as a novel solution the adjustment of the individual kurtosis of the control signals by varying the synthesis parameters of a set of random number generators, without this adjustment affecting the SDM. Thus, the system has as many random number generators as command signals. <p> - Simplification of the system and reduction in the computational cost when it comes to synthesizing and controlling the signals. This solution avoids the need to apply a non-linear transformation function without memory and the use of an iterative process to adjust the transformation function, since the kurtosis adjustment method does not affect the SDM of the process. <br> Does not require the use of a memoryless non-linear transform, or an iterative transform fitting method, to synthesise and control non-Gaussian multivariate random signals by manipulating the phase of the signals. <p> The present invention is applicable in the field of acoustic and vibration testing for automotive vehicle components, as it reproduces the vibrational environment of the vehicle in the laboratory. <br> The invention is also applicable in any test that requiring the synthesis and control of non-Gaussian multivariate random signals.
Our system proposes as a novelty the synthesis and control of multivariate non-Gaussian random signals by manipulating the phase of the signals without the need to use a memoryless non-linear transformation or an iterative method of adjusting the transformation. For this purpose, the phase components of the process are randomly generated and adjusted by varying the parameters of a random phase generator. <p>
The synthesized signals can be used to generate vibrations in a multi-axial vibration machine or can be recorded for later use, for example as an input signal in numerical simulations. <p>
By proper control of the generated command signals it is possible to adjust the power spectral density matrix (SDM) and the kurtosis of the control signals measured on the vibration table or on the test specimen. The generated signals are random, i.e. it is not possible to predict deterministically the value of the signal at each instant of time, but it varies based on a set of statistical distributions. These signals can represent different types of vibration processes common in the transport and automotive industry, such as road rolling, rail travel or rocket flight, among others. <p>
The system is intended to synthesise and adjust a set of command signals so that the resulting control signals have user-defined kurtosis values and a user-defined SDM, compensating for possible distortions caused by the different elements of the vibration system.