Research On Temperature Prediction Based On The Elman Network Applied In Air Tightness Test

With the more and more use of sealing industries, leak detection technology is concerned by enterprises and research institutes both at home and abroad. Differential pressure method because of its high precision and easy to use, develop a lot. But with the deepening research, people find gas as the medium of detection is vulnerable to the effect of environment, especially the temperature of air, affect the test results. On this background, and according to the actual industrial situation, a differential leak detection method based on the prediction of temperature is proposed in this paper, the theoretical and experimental research of this method has been conducted. The article’s main research contents include:By the comparison of detection methods commonly used, the factors affect the results and solutions, identify the research ideal of this paper. Through the theoretical derivation, prove the influence of temperature to the results, then get the leakage calculation equation with temperature compensation. The influence of detected pressure, leak, container volume and other factors to the temperature of air are proved by the simulation analysis. The results showed that the greater the container volume, leak detection and pressure are, the greater the temperature difference between two cavities and the impact on test results are.According to detect the program requirements, complete the design of detection system, includes the design of hardware and software. The hardware system includes the test process control, data acquisition control and the selection of pneumatic components. Complete the design of software by Lab VIEW, realize the function of parameter configuration, data acquisition, performance test, data processing and management.After the assessment of repeatability and accuracy of the system, then start to test to obtain experimental data. Consider the factors to the temperature of contain, complete the design, training and validation of temperature prediction model based on Elman network.Through the comparison of results predicted by the Elman network and the BP network, and the results predicted in different conditions, prove the precision of prediction of dynamic by Elman network is much better than the other networks. Compare the leakage rate and the standard rate, prove the prediction of temperature is useful to improve the precision of leak detection. A large number of tests are carried by this method,after temperature compensation,the relative error is reduced to 2.52%.