Study On The Algorithms Of Active Noise Control Based On The Diagonal Structure

Nowadays,noise pollution is an increasingly serious phenomenon.Traditional passive noise control has many limitations on the size,price and applicability of the equipment.At present,active noise control(ANC)get more attention,more and more knowledge had been used in practical applications.The core of its research is to optimize the algorithms in active noise control systems.In this paper,we have studied on the basic theory and algorithm of nonlinear adaptive active noise control.Based on the basic algorithms in this field,a bilinear filter with diagonal structure used in nonlinear active noise control system has proposed.Using this structure,we derive the diagonal bilinear filter-x least mean square algorithm(DBFXLMS).The simulation experiments given in the paper shows that this algorithm can improve the control result effectively compared with the two previous basic algorithms:volterra filter-x least square algorithm(VFXLMS)and filter-s least square algorithm(FSLMS).In order to optimize the computational complexity of DBFXLMS algorithm,we reduce the number of input signals to derived the simplified diagonal structure bilinear filter-x least mean square algorithm(SDBFXLMS)and the channel reduced diagonal structure bilinear filter-x least mean square algorithm(CRDBFXLMS).From the simulation results,the proposed two algorithms keep the control effects very well and make the computer complexity greatly reduced compared to the DBFXLMS algorithm.To further optimization of the computer complexity,we also use the knowledge of filter-error on the proposed algorithms and derived bilinear filter-error least mean square algorithm(BFELMS),diagonal structure bilinear filter-error least mean square algorithm(DBFELMS)and channel reduced diagonal structure bilinear filter-error least mean square algorithm(CRDBFELMS).According to the simulations,we give the control results of all the improved algorithms mentioned above,and achieve the almost same excellent control result of the proposed DBFXLMS algorithm.After calculating and analyzing the computational complexity of the algorithms,the derived algorithms have a great improvement compared with DBFXLMS.The complexity results can be reduced nearly 80%of the addition per iteration and 81%of the multiplication per iteration compared with the DBFXLMS algorithm.A series of optimization algorithms based on diagonal structure bilinear filter-x algorithm and bilinear filter-error least mean square algorithm in this paper have showed a better performance compared to the traditional volterra filter-x least mean square algorithm and filter-s least mean square algorithms.