CAJ | 학술논문

为了解决和提高不稳定流场条件下的超声波流量传感器的测量精度问题，文章将广义回归神经网络引入到流量传感器系数获取当中，并通过建立流量传感器系数模型和样本输入训练，获得了较为准确的流量传感器系数和较高的测量精度，从而验证了本方法对超声波流量传感器在不稳定流场中进行流量测量的可行性，具有总要的实际应用借鉴价值。
위료해결화제고불은정류장조건하적초성파류량전감기적측량정도문제，문장장엄의회귀신경망락인입도류량전감기계수획취당중，병통과건립류량전감기계수모형화양본수입훈련，획득료교위준학적류량전감기계수화교고적측량정도，종이험증료본방법대초성파류량전감기재불은정류장중진행류량측량적가행성，구유총요적실제응용차감개치。
In order to solve and improve the measurement accuracy of ultrasonic flow sensor in unsteady flow condi-tions,the general regression neural network was introduced to obtain the flow sensor coefficient. And through the es-tablishment of the flow sensor coefficient model and the training of input samples,more accurate flow sensor coeffi-cients and higher measurement accuracy were obtained. So the feasibility of this method for the flow measurement in non-ideal flow field conditions by ultrasonic flow sensors had been verified,and the practical reference value of this method also had been reflected.