应用萤火虫算法的混沌信号盲源分离

Blind Source Separation of Chaotic Signals by Using Firefly Algorithm

针对现有的独立成分分析法分离混合混沌信号精度不理想的问题,提出了一种新的混沌信号盲分离方法。该方法以求解最优解混矩阵为目标,利用峭度构造目标函数,将混沌信号的盲源分离转化为一个优化问题,并用萤火虫算法求解。同时,通过预白化和正交矩阵的参数化表示降低优化问题的维数,能有效提高分离精度。仿真结果表明,无论是处理混合的混沌映射信号还是混合的混沌流信号,该方法都能快速收敛,并且其分离精度在各项实验中都优于独立成分分析法等现有的盲源分离方法。

Existing independent component analysis（ICA） method is not quite accurate when dealing with chaotic signals. To address this issue, a new blind source separation method based on the firefly algorithm is proposed. Kurtosis is used to design the objective function so that the blind source separation issue is transformed into an optimization problem and the solution is obtained by the firefly algorithm. In addition, pre-whitening process and parameterized representation of orthogonal matrices is employed to reduce the dimension of the optimization process. Therefore, better separation accuracy can be achieved. Simulation results show that the proposed method converges very fast when dealing with linearly mixed chaotic signals. In every simulation test, the proposed method is more accurate than ICA method and other blind source sep- aration methods.