BeBeC-2018-S01 ARRAY METHODS: WHICH ONE IS THE BEST? Ennes Sarradj, Gert Herold and Simon Jekosch TU Berlin, FG Technische Akustik Einsteinufer 25, 10587 Berlin, Germany Abstract When it comes to process measured array data, the question which processing method to choose arises. With the large number of methods available, it is impossible to clearly decide which method is the best, because all seem to have their advantages and disadvantages. Many different criteria, such as precision of results and speed of computation, can be of interest. The contribution discusses some of these criteria and makes an attempt to define them in a way that they can be estimated from computed results of given test cases. It is also shown that a decision for one method should not be based on the result of one single test case only. Considering this, a Monte-Carlo-method is introduced to compare different methods on the basis of a large number of test cases (N = 12 600). The choice of the test cases is explained and results for multiple criteria and different methods are discussed. 1 INTRODUCTION If one considers the use of a microphone array for acoustic testing there is a number of things to consider. It has to be decided how to set up the test and which hardware to use. However, since many different signal processing methods are available, thought must be given also to which array data processing method to choose. Among those methods are such that are derived from beamforming, but more interestingly also such that use deconvolution approaches or can be formulated as inverse methods (see [8] for an overview and a slightly different classification). These methods are so numerous and may have very different properties that it is not easy to understand which method(s) would be the best to use in a particular case. However, it is apparent from both theoretical study and from practical application that different methods may behave different and may yield different results. Consequently, the question which one is the best deserves some attention. Ehrenfried and Koop were among the first who addressed this question to a certain extent by comparing different deconvolution approaches [5] using an example test case. Data from practical array wind tunnel measurements was used for comparison of different methods in 1