Firstly, the uncertain factors in drawing process, which cause the changes of the sensor's displacement and the sensor's vibration, are analyzed in terms of mechanics and statistics. Systemic randomictiy and non linearity affect the performance of traditional auto regulation equipment. Using the minimum error variance adaptive controlling technique to solve the uncertainty of controlled system is the ideal method. Secondly, using generalized least square method to achieve the principle and method of the system identifying model is described. Thirdly, the design steps of the minimum error variance adaptive adjustor and the method of the computer simulation are discussed. The result of test running shows that the output of the system follows the input of the system better, the system restrains the random disturbance more efficiently, and the performance of auto regulation is improved apparently.