Silencing the Noise: Advanced Active Noise Control for Automotive and Outdoor Environments using FxNLMS algorithm

Active Noise Control (ANC) is a possible technique for reducing noise in the surroundings of appliances, consumer electronics, industrial equipment, mainly emitting in the low frequency range. The ANC technique is based on the cancellation of an noise control filters by generating an anti-noise using electronically controlled speakers. Noise reduction filters are applied to the control signal before being emitted by the speakers, which is how sound waves that are identical to amplitude but in opposition in phase compared to the unwanted noise are produced. Adaptive algorithms are used to update those noise control filters. This thesis discusses the development of ANC algorithms that have been applied to both indoor and outdoor environments. The proposed algorithms are based on the Filtered-x Normalized Least Mean Squares (FxNLMS), typically employed for active noise cancellation problems. In particular, ANC techniques are employed for two specific applications and signals: the first one considers the reduction of a complex, but a priori known, signal (e.g. the siren of an ambulance), while the second one refers to a stationary pure tone noise produced by large industrial machinery (e.g. the emission of an electric reactor). For the ambulance siren’s case the objective was to protect drivers from the noise reaching inside the cabin, as they are exposed to loud and repeated sounds during field operations for extended periods of time. Environmental noise reduction, and particularly the noise generated by emergency vehicles, has been the focus of increased interest in the last decade due to its negative impact on people's physical and mental health. One of the most promising noise-mitigating solutions is the implementation of intelligent systems whose acoustic performance is actively controlled. The algorithms were first implemented in a Single Input Single Output (SISO) configuration, which served as a step towards a multichannel architecture. The adaptive filtering system consists of an error microphone and a control (secondary) loudspeaker, without employing a reference microphone to detect the noise that needs canceling, as the signal to be attenuated is known in advance. 8 Referring to the reduction of stationary noise produced by large industrial machines, the activity has been mainly focused on electric reactors, operating within a local electric substation, implementing active noise control techniques designed for noise reduction of the stationary, pure tone, humming noise they generate during operation. Such noise has been characterized, modeled, and investigated as a relevant case study of interest. Finally, experimental activities were conducted to demonstrate the effectiveness of the developed control logic for both the SISO and Single Input Multiple Outputs (SIMO) configurations. For all the above studies and application, the chosen approach involved actively intervening at the noise sources themselves to reduce their impact, rather than merely attenuating the noise downstream. The results showed that intelligent noise cancellation system could achieve almost complete noise reduction within designated areas of interest, highlighting the successful performance of the systems studied and implemented.