Research And Implementation Of Detection Algorithm For Crude Oil Moisture Content In Complex Environment

As science and technology develop,it is obvious to see the importance of the oil industry.The water injection method is often applied to increase the production of crude oil in the process of oilfield exploitation,but the gradual increase of the water content in the crude oil in the later stage of exploitation will be caused.In view of it,an important part of crude oil extraction is reflected in the detection of crude oil water content,so the prediction of oilfield life and production can be facilitated,the improvement of the production and quality of crude oil can be promoted.The disadvantages of low detection accuracy and poor data readability due to single-point measurement are embodied for the existing crude oil water content detection algorithms.From this,a crude oil water cut sampling method by using conductivity method was proposed in this thesis,and a conductivity water cut model calculation method based on the idea of multi-sensor information fusion was also put forward.In addition,a surface fitting method for calculating the water content distribution of oil-water fluid in a porous medium in a plane based on the water content matrix was proposed,and a crude oil water content detection device suitable for this method was also developed.Firstly,the basic idea of conductivity method was adopted in this thesis for the problem of the water content detection method of oil-water fluid in porous media.By inserting multiple electrode rods into sampling holes at equal intervals,the measurement on the conductance of oil-water fluid was conducted by grouping measurement method.The detection on the water content was carried out in line with the difference in the conductivity characteristics of oil and water.Based on the idea of multi-sensor fusion,an improved ELM(Extreme Learning Machine)model that took temperature,measured conductance between electrode rods and 50%water cut oil-water measured conductance as inputs was proposed,so the water content of crude oil was calculated.The three kernel function optimized ELM models were established by the algorithm through the inputting of sample data under different oil-water miscibility conditions.The appropriate model was selected through the optimized Maxwell classification method to make the calculation of the water content,so the relationship between the conductance and the water content of crude oil was established.From the experimental results,it indicates that the algorithm is superior to the general BP(Back Propagation)neural network algorithm and the classification ELM algorithm.Secondly,the proposal of an equivalent expansion scheme of the detected conductance matrix was made.The scatter diagram of the water content of crude oil was obtained by the above algorithm after filtering the conductance matrix.Finally,the plane distribution map of crude oil water content in porous media was established by using the multiple nonlinear regression algorithm based on time-frequency domain decomposition.In accordance with the simulation results,it indicates that the algorithm is better than the general surface fitting algorithm.Finally,the development of a crude oil water content detection device based on conductivity method was conducted in line with the above detection scheme and principle.The AC signal and reference resistance voltage were generated by the excitation source in the hardware part.The voltage signal at the measurement point was sent to the single-chip microcomputer through filtering and AD conversion to realize the detection of the sample conductance.Finally,the water content and distribution were calculated based on the sending of the collected data by the single chip microcomputer to the upper computer through the 485 circuit.In accordance with the experiments,it indicates that the accuracy of the device can satisfy the practical requirements of users,so it can be regarded as a successful design case.