Aiming at the problem that the output signal of a flexible fiber Bragg grating tactile sensor applied to the electronic skin is the nonlinear and multi-dimensional coupling of the position and size information of the applied load, based on the mechanical finite element simulation of the fiber Bragg grating tactile sensing array, the decoupling methods are proposed, in which error back propagation (BP) neural network and radial basis function (RBF) neural network are applied to the haptic signals in simulation and experiment. The neural network decoupling results of experiment data show that compared with the error BP neural network, the RBF neural network has stronger anti-noise ability and can better approximate the mapping relationship between noisy tactile multi-dimensional nonlinear experiment data. After decoupling with the radial basis function neural network, the spatial resolution of the sensor array is 5 mm, and the minimum relative errors of the pressure position and size perception are 3. 00% and 4. 82% . The research results in this paper have certain practical value for the research and promotion of electronic skin flexible tactile sensors.