Application of online SVR on the dynamic liquid level soft sensing

There exist different kinds of drawbacks and security risks in the measurement methods of dynamic liquid level in oilfield, consequently , each of them can´t provide reliable protection to the oilfield for its high efficiency production operations and easily to cause significant economic losses . In order to solve the problems above , soft sensing technology is found as a new way to measure the dynamic liquid level of the oilfield instead of original hardware technology. ANN is a kind of soft sensing technology which is earlier to be used as the way to measure the dynamic liquid level , but there are some problems restricted its use of areas, such as easy to fall into local minimization, slow convergence speed and poor generalization ability. According to the oilfield data´s characteristics, Online SVR algorithm is proposed to be a solution to measure the dynamic liquid level in this paper. As a kind of soft sensing technology, Online SVR not only has the same performance of ANN, but also can update the forcast model by online . In addition, it also solve the problems of ANN .Online SVR algorithm is verified by the experiment in this paper and the result of this experiment shows that Online SVR has higher precision than ANN . The experimental results also show the Online SVR can be used in measuring the dynamic liquid level in oilfield.