As movement pattern recognition of powered lower limb devices is poor,a wearable terrain recognition system is designed. The portable laser distance sensor and inertial measurement sensor installed in the waist are used to collect laser distance and terrain height information,and then the collected data is processed by wavelet denoising algorithm;the feature value is extracted,finally,probabilistic neural network is used to complete terrain recognition. The experimental results show that the portable system can effectively identify five terrains of ground,upstairs,downstairs,upslope and downslope. It proves that this wearable terrain recognition system is effective and feasible for powered lower limb devices.