The Methods Of Stream Dryness Measurement In Oil Field

The stream which is poured into the oil wells during heavy oil production is wet stream. The dryness of wet stream is a key point in the production, in order to insure the security of the stream-injection boiler and the effect of oil production, it is necessary to measure dryness quickly and exactly, and it is also a hot topic in two phrases Current of Vapour-Liquid research domain. This thesis proposes a on-line method to forecast the stream dryness, which uses several related variables to constitute the neural network model. The emulational result is presented in the last part of the thesis.First the thesis introduces the basic knowledge of stream dryness measurement, for the prevalent problems which exist in most domestic dryness measurement methods; a new method based on the neural network is brought forward.Then, grounded on analyzed basic knowledge of neural network and the stream driver oil recovery, the thesis proposes the dryness measurement project based on neural network. By analyzing the variables which are collected in oil locale, it can get that the stream dryness has a relationship with the stream pressure, the stream differential pressure and the water flux, those variables can be measured directly. Next the stream dryness model is established by those related variables use BP neural network. By doing a series of experiments, contrast and the result analysis, it is found that the pretreatment of the input and output data, the disturbance added to the model and the circular training are vary important for the model. At last, it is concluded that the model which has two connotative layers and disturbed signal is good for dryness measurement. When contrasted with conventional model, it can be seen this model has advantages both in precision and time.Last, based on the above scheme, a stream dryness measurement system is designed and developed. This system has two parts: Training system and Simulation system. The function of the Training system is to establish network, train network and save network; the function of Simulation system is to forecast stream dryness by the network which has been saved, it also can present the stream dryness and the relative error immediately. This system can largely reduce the workload of worker and increase the intelligence to the stream dryness measurement.