A R & D On Oil Sucker-Rod Pump With Artificial Intelligence

There is no need to go into details how is the position of petroleum in the development of national economy. The reserves of petroleum in our country is not so rich and makes the peoples worried. Some oil field of large scale in our country are faced to be exhausted with different degrees and a lot of oil wells are closed for "income falling short of expenditures". Besides, a lot of oil well are faced to be "retailed" for too high energy consumption and too high cost of recovering. Improving the recovering technology and reducing the corresponding cost must be extremely significant for saving the old oil wells. It will be of great significance especially in our country.Up to now the oil pumps with sucker-rod are still the main facilities for oil recovering in our country. The oil experts in the world still pay their high attention to reducing the energy consumption for such oil pumps.Dry and empty pumping are the primary factors of energy wasting and low efficiency in the older oil wells. In addition, it makes the equipments quick wear and tea, leads breaks of sucker-rod etc. The major objective of this dissertation focuses on the development of control technology with artificial intelligence to realize automatic identification of dry and empty pumping, automatic inspecting of failures, reducing the energy consumption etc.Because of the complicated geological structure of oil fields, there are lots of problems on non-linear, non-definition, time varying and non-complete characteristics. Therefore it is difficult to establish an accurate mathematical model for controlling. This dissertation deals with an expert controller combined with self learning technology to realize automatic acquisition of dry and empty pumping criterions, automatic tracking of stopping time for maximal daily yields. The experiments in oil fields show that those results are effective for solving the problems of energy saving, daily yield and easy operating.