In order to effectively decrease the loss caused by the multipoint fault of wind speed sensors or wind pressure sensors, and to reduce the complexity of computation and the difficulty of the engineering application, a model needs to be proposed to recover the missing data at the same time. As the traditional multi-channel signal diagnosis uses multivariate empirical mode decomposition (MEMD), the multivariable empirical wavelet transform (MEWT) is proposed to restore the multipoint missing data synchronously. In practical application, the multipoint signals are decomposed into a series of modes at the same time, and then the kernel-based extreme learning machine (KELM) is used to predict, and the cuckoo search (CS) algorithm is used to optimize the regularization parameters of the model and the kernel parameters. For multi-step forecasting, the traditional recursive strategy is replaced by the multiple-input multiple-output (MIMO) strategy. The actual measured multipoint wind pressure on the building surface and the measured multipoint data of the downburst are used to verify the feasibility of the model. Compared with the noise assisted multivariate empirical mode decomposition kernel-based extreme learning machine (NA-MEMD-KELM-CS), the result shows that the proposed model can recover signals simultaneously with high accuracy.