基于声场分解和稀疏正则化的二维空间次级声源布局优化

During the design of transducers configuration for an active noise control system, current optimization methods need to predetermine the error sensors configuration, which significantly increases the workload of later optimization of the secondary sources configuration. In this study, a new method free from specific error sensors configuration information is presented that higher order microphones are used to capture the sound field so as to formulate the cost function in wave domain. In addition, according to sparsity characteristics of the primary sound field, sparsity-inducing regularization is introduced to optimize the secondary sources configuration, including the number and positions, by calculating a sparse approximate solution to underdetermined equations. Effects of the number of candidate secondary sources are discussed, and the comparison with the uniform configuration and the optimized configuration using the genetic algorithm is performed. Results show that the proposed method can optimize the secondary sources configuration effectively independent of the error sensors configuration information. The noise reduction of the proposed method is close to that by the genetic algorithm, while other evaluation metrics for the system are much better, which would benefit the stability of active noise control system.