Multi-objective Structural Optimization Of Fibonacci Grid Microphone Array

To address the differing performance requirements of microphone arrays for sound source localization at various frequencies of interest, a multi-objective structural optimization method for Fibonacci-grid microphone arrays—based on an improved Non-dominated Sorting Genetic Algorithm II (NSGA-II)—is proposed. Taking the maximum sidelobe level and mainlobe width as objective functions and the on/off states of array elements as decision variables, the method yields a Pareto-optimal solution set comprising multiple high-performance array configurations via a multi-objective evolutionary process. An individual selection function is then applied to identify an optimized array configuration that exhibits superior performance at the target frequency. Monte Carlo simulations are further conducted to evaluate the robustness of the proposed method against array element position errors, and the feasibility of scaling it up to larger arrays is discussed. Numerical results show that, at 2000 Hz, the optimized array reduces the mainlobe width from 54° to 44° (a reduction of approximately 18.5%) while maintaining the maximum sidelobe level at approximately −10.42 dB—demonstrating that the proposed method enables frequency-specific structural optimization of the array through binary element activation (i.e., switching elements on or off) alone.