This paper presents the nonlinear analysis and processing of signals based on chaos aspects. Chaotic invariants are described, including Lyapunov exponents and fractal dimensions. The problems of applying Takens' Embedding Theorem to phase space reconstruction are discussed, including those of embedding dimension and time delay. Then RBF neural network used in nonlinear processing of chaotic signals is presented.