液/气高压闸阀内漏量化试验及预测模型构建研究OA

To address the threats of internal leakage in high-pressure gate valves to equipment safety and system energy efficiency optimization,research on quantitative detection and condition diagnosis of internal leakage was carried out.An internal leakage test platform for high-pressure gate valves was established,and ultrasonic testing technology was used to conduct internal leakage tests under various combined conditions involving two media(liquid/gas),inlet pressure from 10～70 MPa,defect sizes from 0.1～0.3 mm,and valve diameters from DN65～DN130.The results show that as inlet pressure and defect size increase,both the internal leakage flow rate and the ultrasonic signal of liquid/gate valves show an increasing trend;as valve diameter increases,both decrease,a phenomenon mainly influenced by wall thickness.Under the same conditions,the gas internal leakage flow rate is approximately 10～431 times that of liquid.The gas ultrasonic signal is generally stronger than that of liquid,with signal amplitude approximately 1.4～3.2 times that of liquid;during internal leakage in liquid/gas gate valves,the outlet ultrasonic signal is generally stronger than the inlet signal,with an average difference maintained at 7.2 dBμV RMS.Based on the Gaussian process regression(GPR)method,using the Matern 5/2 kernel function(length scale of 1,signal variance of 0.707 1),an internal leakage flow rate prediction model was constructed,achieving a prediction accuracy of 93.55%.The obtained laws and the model have good universality,providing reliable application basis and theoretical support for the quantitative assessment and operating condition diagnosis of internal leakage in high-pressure gate valves under different working conditions.