【摘要】：海上油田生產(chǎn)優(yōu)化不同于陸上油田生產(chǎn)優(yōu)化,由于海上生產(chǎn)平臺(tái)空間有限,在一個(gè)局限的空間中分布著幾十口井,井口相聚較近,導(dǎo)致井間干擾現(xiàn)象的出現(xiàn)。因而單井的優(yōu)化方法不能滿足生產(chǎn)需要。本論文建立了海上油氣生產(chǎn)實(shí)驗(yàn)?zāi)M裝置,利用該裝置進(jìn)行了平臺(tái)生產(chǎn)效率和產(chǎn)量的優(yōu)化實(shí)驗(yàn);分別進(jìn)行了單井無模型梯度尋優(yōu)實(shí)驗(yàn)、多井無模型梯度尋優(yōu)實(shí)驗(yàn)、水井無模型梯度尋優(yōu)實(shí)驗(yàn)、無模型一體化優(yōu)化實(shí)驗(yàn),得到了收斂的優(yōu)化結(jié)果,實(shí)現(xiàn)了海上油氣生產(chǎn)的無模型梯度智能尋優(yōu)。將實(shí)驗(yàn)所得數(shù)據(jù)運(yùn)用于BP人工神經(jīng)網(wǎng)絡(luò)的訓(xùn)練,建立了海上生產(chǎn)實(shí)驗(yàn)裝置的BP仿真模型,且該模型計(jì)算精度高;同時(shí)對(duì)海上油氣生產(chǎn)系統(tǒng)的優(yōu)化目標(biāo)進(jìn)行篩選,以產(chǎn)量和系統(tǒng)效率為優(yōu)化目標(biāo),建立了適用于海上油氣生產(chǎn)系統(tǒng)的多目標(biāo)優(yōu)化數(shù)學(xué)模型。在對(duì)NSGA-Ⅱ算法與其他三種算法進(jìn)行性能對(duì)比后,確定選用NSGA-Ⅱ作為優(yōu)化算法。將海上生產(chǎn)系統(tǒng)多目標(biāo)優(yōu)化模型、BP仿真模型與NSGA-Ⅱ算法進(jìn)行結(jié)合,對(duì)海上油氣生產(chǎn)體統(tǒng)進(jìn)行了多目標(biāo)優(yōu)化,得到了精確度較高的優(yōu)化結(jié)果,實(shí)現(xiàn)了最大系統(tǒng)效率、高采油量、低注水能耗的優(yōu)化目標(biāo)。通過實(shí)驗(yàn)驗(yàn)證多目標(biāo)優(yōu)化結(jié)果的有效性與精確度,最后對(duì)無模型梯度優(yōu)化結(jié)果和多目標(biāo)優(yōu)化結(jié)果進(jìn)行了對(duì)比評(píng)價(jià)。
[Abstract]:The production optimization of offshore oil field is different from that of onshore oil field production. Because of the limited space of offshore production platform, dozens of wells are distributed in a limited space, and the wellhead is close to each other, which leads to the phenomenon of inter-well interference. Therefore, the single well optimization method can not meet the production needs. In this paper, a simulation device for offshore oil and gas production is established, and the platform production efficiency and output are optimized by using the device. Experiments of single well without model gradient optimization, multi-well without model gradient optimization, wells without model gradient optimization, without model integration optimization experiment were carried out, and the convergence optimization results were obtained. The model-free gradient intelligent optimization of offshore oil and gas production is realized. The experimental data are applied to the training of BP artificial neural network, and the BP simulation model of offshore production experimental device is established, and the accuracy of the model is high. At the same time, the optimization objectives of offshore oil and gas production system are screened, and the multi-objective optimization mathematical model for offshore oil and gas production system is established with the objective of production and system efficiency as the optimization goal. After comparing the performance of NSGA- 鈪，

本文編號(hào)：2370595