The Parameters Identification Procedure For The Exponential Model With Additive Noise Apply In Oil Field

Exponential model with additive noise is a non-liner mode widely used inmodeling of economic system. Document[1] has given a weighted liner leastsquare iterated method to discriminate parameters of exponential mode withadditive noise, and proven its astringency. This method can not be used tocalculate on-line. In real system, data of input and output are usually detectedsequentially. If iterated method is used to discriminate parameters, when agroup of new information is added, calculation must be redone from thebeginning. Especially when many parameters are used, computing calculationwith high-order liner equation must be done repeatedly. So, a lager portion ofcomputer memory is occupied and more calculation is added, in which, manycalculations is repetition.Output is the most important criterion in oil field and the results of allactivities in oil field production are indicated in the output. We all know that theoutput of a single oil well decrease successively. So, corresponding technicalmethods must be taken to guarantee a high and stable output of an oil field. Atpresent, many methods have already been applied in oil field, includingpressure maintaining by injecting water, pumping unit, and electrical pump well.If objective condition permits, drilling new wells in a large scale is also a keymethod to increase output. For a researcher studying oil reservoir, he would liketo build mathematic model of oil output of an oil field to predict future outputprecisely, which has a very important economic significance in scientificallymanaging oil field and preparing economic plan. This paper tries to establish anew mathematic model to oil field output system, which has a simple structureand high precision. This also has practical importance. This paper selects 5 emulation models to check astringency, convergencerate, convergence precision of this weighted liner least square recursionalgorithm. Two ways used to select initial value of parameters and valueparameters in different range. The results show that this method has someflexibility in select initial value, good astringency as well. Using this method,parameters converge near to real value in 20 steps, and the maximum relativeerror to real value less than 10-6 order of magnitude. Emulation results showthat this weighted liner least square recursion algorithm to discriminateexponential model with additive noise has a rapid convergence rate and highconvergence precision. System mode predicting is one of the main aspects of dynamic systemdiscrimination. To an administrative man, every decision he made is relative toconsideration of prediction, especially in oil field exploration and business areas,where economic benefits mainly depend on precision of planning. One keyfactor to determine precision of planning is the precision of prediction, so,people has extensive interests in theory and practice of prediction. This paper establishes exponential model with additive noise for accumulativeoil output and accumulative liquid output and discriminates model parameterswith weighted liner least square algorithm. After conducting five-step dynamicprediction, satisfactory results are acquired. Considering time varying volatility ofparameters and noise statistics in oil field exploration, virtual noise withtime-variable noise statistical characters is introduced, thus, the precision of...