In view of the shortcomings of strapdown inertial measurement unit (IMU) such as large noise and error accumulation, poor precision of traditional attitude calculation algorithms and poor adaptability to environment, this paper proposes an attitude calculation algorithm aided by Elman neural network. For multi-sensor information fusion, not every neural network is applicable, but the Elman neural network structure contains a receiving layer, which stores the information of the previous hidden layer, this structural feature enables the Elman neural network to predict in continuous signal. The simulation results show the effectiveness of the algorithm and improve the accuracy and adaptability of the algorithm.