In order to solve the problem of losing codes and pausing codes in the incremental encoder which conventionally used in the stage speed boom system as speed feedback component and prevent the propagation of fault effect, a fault detection and soft close-loop fault-tolerant control method for encoder faults based on the Takagi-Sugeno Fuzzy Neural Network (T-S FNN) model combined with the data-driven technique was proposed. First, the system of T-S FNN prediction model was established by substracting the system normal operation of historical data, and achieved the residual error information by using measured values of actual encoder and predicted values. Next, encoder fault was detected by using improved Sequential Probability Ratio Test (SPRT) algorithm though the residual error real-time data information, in order to overcome the detection delay and ensure the reliability of fault detection. Then, according to the prediction model output which was used as the output of the encoder failure to accommodate the failure when fault was detected, in order to realize the soft fault-tolerant operation by using close-loop mode. At last, the encoder fault tolerant process for the losing codes and pausing codes was proved by simulation experiment effectively. The simulation results show that the method of this article can detect the encoder fault information rapidly and reliability, and switch from the fault-tolerant mechanism timely and safely by using the reconstruction of the prediction information, in order to realize the soft closed-loop fault-tolerant control of encoder failure and improve the safety and reliability of stage speed boom system operation process.