An optimal reconstruction method of synergetic order parameters based on orthogonal polynomials

The order parameter is the key factor of synergetic pattern recognition. However, the conventional methods of reconstruction of order parameters have some shortcomings which limit the applications in real world. With these reasons, in this paper, a new approach to reconstruction of order parameters is proposed based on orthogonal polynomial, and thus we can make full use of the self-learning ability of synergetic neural network to obtain a group of linear transformation parameters for order parameters. In addition, a weights-direct-determination method is presented, which could immediately get the reconstruction network weights in the training process. The experiment results show that this novel algorithm can not only search the reconstruction parameters effectively, but also improve the recognition ability of system.