Ultrasonic feature signal extraction algorithm based on convolutional neural network model
The flight time difference is a key parameter of the time difference ultrasonic flowmeter,which determines the measurement accuracy of the meter.In this paper,convolutional neural network is used to extract features of ultrasonic echo signal,and the extracted features are used to regression predict the flight time difference.The ultrasonic echo signal is used as the input layer of the model.In order to improve the model performance and speed up the training,the middle layer uses five layers of convolution layer,five layers of pooling layer and RELU Activation function to extract the signal characteristics,and the output layer regression predicts the flight time difference,which improves the accuracy of time difference estimation.Simulation studies have shown that the accuracy of the model predict is higher than 99%and has good generalization ability.Building an experimental platform and conducting experimental research,the results show that the convolutional neural network models are used to predict flight time differences in ultrasonic echo signals which has high measurement accuracy,measurement error is better than±1%,and repeatability is better than 0.2%.
ultrasonic flow meterflight time difference detectionconvolutional neural network modelecho signal processing
万方数据
维普
