Due to the lack of energy

fault and other reasons cause wireless sensor network node can’t collect the follow-up data

in order to ensure the integrity of data transmission

data forecast of subsequent node should be carried out.At present

lack of effective method to realize node data forecast.Aiming at this problem

a data prediction method which combines wavelet analysis and particle swarm optimization least squares support vector machine（PSO-LSSVM） is proposed.In the method

selected sample set data is decomposed using wavelet decomposition technique

and then decomposition quantity is respectively transmitted into the least squares support vector machine（LS-SVM） model to carry on forecast

the LS-SVM prediction model parameters are optimized by particle swarm optimization（PSO）

and finally

all the predicted result got by LS-SVM models are reconstrucedn by wavelet to get complete forecast results.The simulation results show that wireless sensor networks node data prediction method combines of wavelet analysis and PSO-LSSVM can accurately track actual data trend

this method has high prediction precision.