Prediction For Seamless Steel Tube Quality And Optimization Control For Key Parameter

The tube hollow is widely used in automobile manufacture, aviation industry, petrochemical industry, construction industry, boiler manufacture, military industry and some other departments, which is very important in national economy, so it’s called "the blood of the industry". As the economy is developing rapidly, the applying field is enlarging and requirement of the product quality is becoming much higher. In most of the iron and steel company, the detection for the product quality is used in production tube, but it’s unreliable to use the final detection for the final product quality. Because of the errors in inspection technique and some defects or defective products appeared in the process of the production, it’s unreliable to examine the quality of the tube hollow in the end of the production. If the defects or the defective production in the previous process can’t be inspected and corrected in time, some defections will be kept in the next process of the production until the finished product is done, thereby seriously affecting the final quality of the production. For this reason, the production quality in the process of tube hollow production must be controlled in the first stage, and it’s necessary to reinforce the inspection and control of the production quality in order to make sure the final quality which is eligible for the latter process of production.This paper is based on Baosteel company’s tube hollow continuous rolling production line, using quality prediction and control to each link of the tube hollow production process. The work for this paper is shown as follows.Due to the large time delay in the tube blank heating process, which causes the problem of tube blank heating difficulty for quality prediction and control, a soft sensor method, based on TLPCR, is proposed to realize the accurate quality prediction in the tube blank heating process and make the tube blank final temperature kept in the range of the production requirement. This method is better used to solve the control problem of tube blank heating quality prediction in the first process of the tube hollow production, which provides the reliable guarantee of the materials for the latter processes.For the characteristic of multi-period batch process in piercing production and the particularity of the non-Gaussian distribution in the production data, a step-by-step staged MICR scheme, which is suitable for the non-Gaussian distribution data, is developed to establish the accurate model for quality prediction of tube hollow. Using the result and the iterative learning control algorithm, the error control of the wall thickness is well realized, improving the production quality of the tube hollow. Based on the simulation with production data and the experimental effect to the field, the effectiveness of this method is proved.According to the data trapezium distribution characteristics such as typical multi-period and dynamic multivariate, it puts forward a method of step mean value staged MPLS to establish predictive model for shell quality. By means of the result of the predictive model, it improves the shell quality by using the iterative learning control method into the application of the control system of the wall thickness deviation in the continuous rolling process. From the simulation and the experimentation, it shows the effectiveness of this method.Aiming at the shell production of the wall reduction and diameter reduction which involves many product specifications and some batch process characteristics such as typical multi-period and dynamic multivariate in the production of the wall reduction and diameter reduction, this paper presents a method of multi-model mean value staged RMPLS to establish the quality prediction model for the diameter reduction shell. By using the result of the prediction model and applying the iterative learning control method into the deviation control system of the process, it improves the production quality of the reducing diameter shell. From the simulation and the experimentation, it shows the effectiveness of this method.To deal with the difficulty in on-line measurement and control for the guide disc revs, an indirect measurement method based on speed observer is proposed, establishing the model of the guide disc revs. By utilizing the ICR method to establish the soft sensor model, it can realize the real-time prediction for the observer which has the difficulty in measuring the guide disc load torque and meanwhile is also the main disturbance in the control system. After the accurate and useful guide disc revs is gotten, it uses the compound control algorithm which combines the feedforward with inferential controller of the guide disc revs control system to realize the accurate control for the guide disc revs.In the tube hollow production, the quality is not the only index the company cares about. The company also cares about the production efficiency and the cost. In view of this problem, mean value staged MICR method is proposed to build the piercing efficiency and energy consumption prediction model. According to the production constraints and the need of the market, the method is developed to optimize the solution and get the optimal piercing production process parameters to guide the production and make sure the profits maximized.