CAJ | 학술논문

利用浓度和速度电容式传感器分别测量粉体的浓度和速度，结合阀门开度、喷吹罐压力、温度等参数构造神经网络粉体流量测量模型，考虑BP算法训练神经网络测量模型时收敛速度慢、动态特性不够理想等不足，用遗传算法来优化神经网络测量模型的参数，以提高测量系统的精度。在现场与电子秤比对，最大满量程误差小于4.2%。具有工程应用价值。
이용농도화속도전용식전감기분별측량분체적농도화속도，결합벌문개도、분취관압력、온도등삼수구조신경망락분체류량측량모형，고필BP산법훈련신경망락측량모형시수렴속도만、동태특성불구이상등불족，용유전산법래우화신경망락측량모형적삼수，이제고측량계통적정도。재현장여전자칭비대，최대만량정오차소우4.2%。구유공정응용개치。
Capacitance sensors are used to measure volumetric concentration and velocity of powders, combined with pressure of puffing tank, valve on-off, temperature parameters, a kind of neural network (NN) flowrate measurement model based on genetic Algorithms (GA) is presented. In order to overcome the defects such as slow convergent speed and unsatisfied dynamic characteristic when backpropagation (BP) is used to train the neural network measurement model, the method using GA to train parameters of NN is proposed to improve the performance of measurement model. Spot experiments show this method can reduce the errors of measurement. The whole accuracy is less than 4.2% according to the electronic-weighing device in work site. It has great engineering application value.