Optimization back-analysis on pipeline defect identification based on constrained penalty function and nonlinear-least-squares method

The optimization back-analysis model of pipeline defect identification is established based on nonlinear-least-squares method and by adding a penalty function to translate the constrained boundary into unconstrained. The measured stresses and strains of pipeline outer wall are used to identify the dimensions (such as inner radius, eccentricity) and locations of defects in the pipeline inner wall through the optimization back-analysis model. An actual pipeline of China West-East natural gas transmission project is calculated as example and the calculating results are compared with the measured values. The comprehensive study results indicate that the error between calculation results by the optimization back-analysis model and the true value is less than 3%, which is a high precise in the actual engineering. The identification accuracy has an inverse proportion to the random errors of the test data. The model presented in the paper is valuable and universal in solution of back-analysis problems.