A blood pressure prediction model based on recurrent neural network（ RNN） with context connection layer and its implementation process is introduced. Due to the great difference of the hardware acquisition circuit of signal and human physiology

the characteristic points of photoplethysmography can vary greatly

which will affect on robustness of blood pressure model. To solve the problem

some novel work on data augmentation and model modification has been done

including experiments design to enrich photoplethysmography（ PPG） in data

the normalization method is improved

PPG signals with and without filtering

are used for prediction and add the personal physical information to build blood pressure prediction model based on RNN and adding contextual layers

complexity is the lowest among the similar models. The model has precision of（ 4. 87 ± 7. 80） mm Hg for SBP and（ 4. 60 ± 6. 09） mm Hg for DBP which reaches the AAMI standard.