Aiming at the problem of insufficient BER performance of traditional Turbo channel decoding algorithms, an intelligent Turbo channel decoding method based on dual-driven model and data is adopted; based on the traditional Max-Log-MAP decoding algorithm, the iterative process is expanded in depth and reconfigured as a multilayered cascade neural network architecture, and an adaptive a-priori information-enhanced APETurbo decoding network model is proposed, and an APENet neural subnetwork is designed in the model. Design APENet neural sub-network; the sub-network adopts learnable weights to linearly adjust the external information, and further extracts nonlinear features based on the fully connected layer with nonlinear activation function, and constructs a residual connection structure by using trainable hybrid coefficients combining linear computation results and nonlinear extraction results, to achieve a more accurate estimation of the a priori information; design the probability space mean-square error loss function based on normalization Simulation results show that the proposed algorithm improves the BER performance of Max-Log-MAP algorithm by about 0.4dB over Max-Log-MAP algorithm at a BER of 10 under AWGN channel.