基于脉搏波和小波神经网络的血压预测

为了进一步提高血压的预测精度，实现无创连续血压测量. 本文提出了一种基于脉搏波和小波神经网络 的血压预测算法，通过脉搏波采集系统采集不同人的脉搏波信号，采用椭圆低通滤波器去除高频噪声，并且 有效提取出脉搏波中关于血压的相关特征，建立BP神经网络和小波神经网络预测模型，不断优化学习速率以 及隐含层层数，从而可以准确地预测出血压值. 结果表明，建立的脉搏波采集系统可以采集到信噪比较高的人 体脉搏波信号；建立的BP神经网络和小波神经网络模型，实现了对血压的无创连续预测，并且通过对比血压 预测的误差验证了小波神经网络比BP神经网络的预测效果更佳.
Inordertofurtherimprovethepredictionaccuracyofbloodpressure,andachievenoninvasivecontinuous bloodpressuremeasurement,abloodpressurepredictionalgorithmbasedonpulsewaveandwaveletneuralnetworkis proposedinthispaper.Thepulsewaveacquisitionsystemisusedtocollectthepulsewavesignalofdifferentpeople,and thehighfrequencynoiseisremovedbyellipticlowpassfilter,andtherelatedcharacteristicsofbloodpressureareex? tractedinaneffectiveway.TheBPneuralnetworkandthewaveletneuralnetworkpredictionmodelareestablishedtoop? timizethelearningrateandthenumberofhiddenlayers,sothatthebloodpressurevaluecanbeaccuratelypredicted. Theresultsshowthattheestablishedpulsewaveacquisitionsystemcancollectthehumanpulsewavesignalwithhigh signal-to-noiseratio.TheBPneuralnetworkandthewaveletneuralnetworkmodelareusedtocontinuouslypredictthe bloodpressureinanon-invasivewayandcomparetheerrorofbloodpressureprediction.Itisprovedthatthewavelet neuralnetworkisbetterthanBPneuralnetwork.