Low Frequency Detector For Measuring Moisture Content Of Green Sand Based On Dielectric Properties

The methods of measuring moisture content of green sand are mainly divided into two categories: direct measurement method and indirect measurement method by measuring the characteristic parameters related to moisture content of green sand.Although the direct measurement method has high measurement accuracy,the measurement cycle is too long,and thus it is unsuitable for industrial production.In the indirect measurement method,the moisture content of green sand is mostly measured by electrical characteristics,which is fast and suitable for on-line measurement.In order to overcome the defects of traditional capacitance and resistance method,the method of measuring moisture content of green sand based on dielectric property was studied,the measuring detector was designed,and the factors affecting the measurement accuracy were analyzed.Based on the transmission line theory and Kirchhoff's law,the dielectric characteristic method was analyzed,and the voltage difference of the transmission line was determined as the dielectric characteristic parameter,and the relationship between the voltage difference and moisture content was got.At the same time,the RLGC equivalent circuit model was established by using the characteristics of coaxial transmission line.The existing form of water and the polarization process in green sand under alternating electric field were analyzed,and the equivalent circuit model of the whole measurement process was designed by combining with the theoretical model of equivalent capacitor.According to the theoretical analysis,a green sand moisture content testing system composed of signal generator,digital oscilloscope,coaxial transmission line and low-frequency detector was constructed,and six detectors with parallel center two probes,center one probe + edge two probes,center one probe + edge three probes,center one probe + edge four probes,edge three probes and edge four probes were designed.The sand sampler with lower and upper cylinders was designed to ensure the same height of the test sample,and the height of cylinders is 160 mm and the inner diameter is 110 mm.The green sand was mixed by roller mixer,and the real moisture content of green sand was measured by infrared moisture meter.A large number of experiments have been carried out on the low-frequency measuring detectors.Firstly,six kinds of detectors with probe length of 100 mm are used for experiments,and the detector structure of one center probe + three edge probes is selected.Then the electric field intensity when the probe length is 20 mm,40 mm,60 mm,80 mm and 100 mm was analyzed respectively by using ANSYS HFSS software,and the approximate length range of the probe was determined.For the probe length of 50-100 mm with interval of 10 mm,the frequency suitable for measuring the moisture content of green sand was decided to be 29-33 MHz by mathematical regression analysis.Finally,according to the maximum and average value of correlation coefficient,the length of center probe is 80 mm,and the lenth of edge probe is 60 mm.The effects of compactability,clay,coal and holding time on the voltage difference of the transmission line were investigated.The results show that the compatibility is positively correlated with the voltage difference,the contents of clay and coal are negatively correlated,and the influence of holding time is small.Under the excitation frequency of 33 MHz,a three-layer BP neural network prediction model for the moisture content of green sand was established.The input layer nodes are the voltage difference of the transmission line,clay content,coal content and compactability,and the output layer node is the water content of green clay sand,and the number of hidden layer nodes is set to 15.The training,verification and test results of the model show that the BP neural network for calculating the water content of green sand by using parameters such as voltage difference of the transmission line is more accurate,and the error between the predicted value and the real value of water content of green sand is less than 5%.