基于模糊评估自适应更新的油井动液面软测量建模

doi:10.11949/0438-1157.20190729

摘要/Abstract

摘要：

针对常规动液面软测量方法在面对复杂、多变的工况时缺乏有效的模型更新机制、预测精度不足等问题，提出了一种基于模糊评估的自适应更新建模策略，通过基于模糊推理产液量变化趋势拟合的模型性能评价模块，动态更新模型，实现对原测量模型的反向推理验证。首先离线建立不同工况的动液面多模型预测集，然后根据产液量拟合优度指标对动液面在线输出模型进行实时的输出评估判断，利用相似样本数据进行模型的在线更新，使其能不断适应油井的工况变化，自适应获得更加准确的软测量模型。最后通过辽河油田现场生产数据验证表明，该方法能够有效提高模型的预测精度和泛化能力，可以满足油田现场的生产需求。

关键词: 模糊推理, 多模型, 动态建模, 拟合优度, 预测, 动液面

Abstract:

For the complex and changing operation conditions, the traditional soft sensor methods for the dynamic liquid level lack effective model updating mechanism and have poor prediction accuracy. To solve this problem, an adaptive model updating strategy based on the fuzzy evaluation is proposed in this paper. A model performance evaluation module based on the fuzzy inference for the tendency of the change of the oil liquid is constructed to dynamically update the model, which is used to realize the reverse inference verification for the original model. Firstly, an offline multi-model prediction model for the dynamic liquid level was established; Then, a fitting optimization index of the oil liquid was put forward to do real-time output evaluation for the online output of the dynamic liquid level; Finally, the field production data verification in Liaohe Oilfield shows that the method can effectively improve the prediction accuracy and generalization ability of the model, and can meet the production needs of the oilfield site.

Key words: fuzzy inference, multi-model, dynamic modeling, goodness of fit, prediction, dynamic liquid level

中图分类号:

TQ 063

王通, 段泽文. 基于模糊评估自适应更新的油井动液面软测量建模[J]. 化工学报, 2019, 70(12): 4760-4769.

Tong WANG, Zewen DUAN. Soft sensor modeling for dynamic liquid level of oil well based on fuzzy inference adaptive updating[J]. CIESC Journal, 2019, 70(12): 4760-4769.

图/表 17

表1 产液量变化趋势推理u的模糊专家规则

Table 1 A fuzzy expert rule for trend of liquid change inference u

$Δ x p$	$Δ y ? d$
$Δ x p$	NB	NS	ZO	PS	PB
NB	NB	NB	NB	NS	ZO
NS	NB	NB	NS	ZO	PS
ZO	NS	NS	ZO	PS	PS
PS	ZO	ZO	PS	PB	PB
PB	PS	PS	PB	PB	PB

表1 产液量变化趋势推理u的模糊专家规则

Table 1 A fuzzy expert rule for trend of liquid change inference u

$Δ x p$	$Δ y ? d$
$Δ x p$	NB	NS	ZO	PS	PB
NB	NB	NB	NB	NS	ZO
NS	NB	NB	NS	ZO	PS
ZO	NS	NS	ZO	PS	PS
PS	ZO	ZO	PS	PB	PB
PB	PS	PS	PB	PB	PB

图1 产液量推理u的控制曲面

Fig.1 Control surfaces of liquid fuzzy inference u

图2 自适应更新建模策略

Fig.2 Modeling strategy of adaptive updating

图3 特征参数变化曲线

Fig.3 Characteristic parameters change curve

图4 传统K-means算法聚类结果（h=1）

Fig.4 Traditional K-means algorithm clustering result(h=1)

图5 基于窗口分析的K-means算法聚类结果（h=6）

Fig.5 K-means algorithm clustering result based on window analysis(h=6)

图6 动液面建模预测结果（相同井组）

Fig.6 Predicted result of dynamic liquid levels modeling(same well group)

图7 动液面建模预测结果（不同井组）

Fig.7 Predicted result of dynamic fluid levels modeling(different well group)

表2 不同算法的预测误差

Table 2 Predicted error of different algorithm

井组	算法	MAE	RMSE
相同井组	支持向量机算法	19.0563	24.3677
	AdaBoost算法	16.2779	21.1045
	改进多工况模型算法	12.4856	16.3584
不同井组	支持向量机算法	26.8274	36.0817
	AdaBoost算法	19.9813	28.4624
	改进多工况模型算法	16.9569	25.3182

图8 产液量的波动变化趋势（不同生产措施）

Fig. 8 Fluctuating change trend of flow(different production measures)

图9 动液面输出错误诊断(不同生产措施)

Fig.9 Fault diagnosis for dynamic fluid levels output(different production measures)

图10 动液面建模预测结果(不同生产措施)

Fig. 10 Predicted result of dynamic fluid levels modeling(different production measures)

表3 动态与静态模型预测误差(不同生产措施)

Table 3 Predicted error of different model(different production measures)

算法	MAE	RMSE
静态模型	44.5780	57.5668
动态模型	34.6773	46.8661

图11 产液量的波动变化趋势(不同生产状态)

Fig.11 Fluctuating change trend of flow(different production conditions)

图12 动液面输出错误诊断(不同生产状态)

Fig.12 Fault diagnosis for dynamic fluid levels output(different production conditions)

图13 动液面建模预测结果(不同生产状态)

Fig.13 Predicted result of dynamic fluid levels modeling(different production conditions)

表4 动态与静态模型预测误差(不同生产状态)

Table 4 Predicted error of different model(different production conditions)

算法	MAE	RMSE
静态模型	30.5454	44.5067
动态模型	23.0401	35.8512

参考文献 30

1	李颍川. 采油工程 [M]. 2版. 北京: 石油工业出版社, 2009: 108-114.
	Li Y C. Petroleum Engineering [M]. 2nded. Beijing: Petroleum Industry Press, 2009: 108-114.
2	王通, 高宪文, 刘文芳. 自适应软测量方法在动液面预测中的研究与应用[J]. 化工学报, 2014, 65(12): 4898-4904.
	Wang T, Gao X W, Liu W F. Adaptive soft sensor method and application in determination of dynamic fluid levels[J]. CIESC Journal, 2014, 65(12): 4898-4904.
3	Li X Y, Gao X W, Cui Y B, et al. Dynamic liquid level modeling of sucker-rod pumping systems based on Gaussian process regression[C]//Wang H Y, Yuen S Y, Wang L P, et al. 2013 Ninth International Conference on Natural Computation(ICNC). Shenyang, China: IEEE, 2013: 917-922.
4	李翔宇, 高宪文, 李琨, 等. 基于多源信息特征融合的抽油井动液面集成软测量建模[J]. 化工学报, 2016, 67(6): 2469-2479.
	Li X Y, Gao X W, Li K, et al. Ensemble soft sensor modeling for dynamic liquid level of oil well based on multi-source information feature fusion[J]. CIESC Journal, 2016, 67(6): 2469-2479.
5	李琨, 韩莹, 黄海礁. 基于IBH-LSSVM的混沌时间序列预测及其在抽油井动液面短期预测中的应用[J]. 信息与控制, 2016, 45(2): 241-247+256.
	Li K, Han Y, Huang H J. Chaotic time series prediction based on IBH-LSSVM and its application to short-term prediction of dynamic fluid level in oil wells[J]. Information and Control, 2016, 45(2): 241-247+256.
6	王通, 高宪文, 刘文芳. 基于改进即时学习算法的动液面软测量建模[J]. 东北大学学报(自然科学版), 2015, 36(7): 918-922.
	Wang T, Gao X W, Liu W F. Soft sensor for determination of dynamic fluid levels based on enhanced just-in-time learning algorithm[J]. Journal of Northeastern University( Natural Science), 2015, 36(7): 918-922.
7	汪世杰, 王振雷, 王昕. 基于JIT-MOSVR的软测量方法及应用[J]. 化工学报, 2017, 68(3): 947-955.
	Wang S J, Wang Z L, Wang X. Soft-sensor based on JIT-MOSVR and its application[J]. CIESC Journal, 2017, 68(3): 947-955.
8	王通, 段泽文, 李琨. 基于改进AdaBoost的油井动液面自适应集成建模[J]. 电子测量与仪器学报, 2017, 31(8): 1342-1348.
	Wang T, Duan Z W, Li K. Adaptive ensemble modeling for dynamic liquid level of oil well based on improved AdaBoost method[J]. Journal of Electronic Measurement and Instrumentation, 2017, 31(8): 1342-1348.
9	熊伟丽, 李妍君, 姚乐, 等. 一种动态校正的AGMM-GPR多模型软测量建模方法[J]. 大连理工大学学报, 2016, 56(1): 77-85.
	Xiong W L, Li Y J, Yao L, et al. A dynamically corrected AGMM-GPR multi-model soft sensor modeling method[J]. Journal of Dalian University of Technology, 2016, 56(1): 77-85.
10	郭晨龙, 仇振安, 孙瑞彬. 基于增量学习的合成孔径雷达目标识别算法[J]. 电光与控制, 2019, 26(1): 31-33.
	Guo C L, Qiu Z A, Sun R B. Synthetic aperture radar target recognition based on incremental learning algorithm[J]. Electronics Optics & Control, 2019, 26(1): 31-33.
11	苏志刚, 王培红, 于向军, 等. 中储式制粉系统出力在线监测软测量建模[J]. 中国电机工程学报, 2007, 27(29): 90-95.
	Su Z G, Wang P H, Yu X J, et al. Soft sensor modeling for on-line monitoring the capacity of coal pulverizing system[J]. Proceedings of the CSEE, 2007, 27(29): 90-95.
12	Lee Y H, Kim M, Chu Y H, et al. Adaptive multivariate regression modeling based on model performance assessment[J]. Chemometrics and Intelligent Laboratory Systems, 2005, 78(1/2): 63-67.
13	汤健, 柴天佑, 刘卓, 等. 基于更新样本智能识别算法的自适应集成建模[J].自动化学报, 2016, 42(7): 1040-1052.
	Tang J, Chai T Y, Liu Z, et al. Adaptive ensemble modeling approach based on updating sample intelligent identification[J]. Acta Automatica Sinica, 2016, 42(7): 1040-1052.
14	彭小奇, 孙元, 唐英. 一种软测量模型性能监测评价及其自适应校正方法[J]. 化工学报, 2012, 63(5): 1474-1483.
	Peng X Q, Sun Y, Tang Y. Performance monitoring and assessment of a soft-sensor and its adaptive correction[J]. CIESC Journal, 2012, 63(5): 1474-1483.
15	Hiromasa K, Kimito F. Application of online support vector regression for soft sensors[J]. American Institute of Chemical Engineers Journal, 2014, 60(2): 600-612.
16	Xiong W L, Zhang W, Xu B G, et al. JITL based on MWGPR soft sensor for multi-model process with dual-updating strategy[J]. Computers & Chemical Engineering, 2016, 90(12): 260-267.
17	王振雷, 唐苦, 王昕. 一种基于D-S和ARIMA的多模型软测量方法[J]. 控制与决策, 2014, 29(7): 1160-1166.
	Wang Z L, Tang K, Wang X. A Multi-model soft sensing method based on D-S and ARIMA model[J]. Control and Decision, 2014, 29(7): 1160-1166.
18	刘佳, 邵诚, 朱理. 基于迁移学习工况划分的裂解炉收率PSO-LS-SVM建模[J]. 化工学报, 2016, 67(5): 1982-1988.
	Liu J, Shao C, Zhu L. Modeling of cracking furnace yields with PSO-LS-SVM based on operating condition classification by transfer learning[J]. CIESC Journal, 2016, 67(5): 1982-1988.
19	Alguwaizani A. Degeneracy on K-means clustering[J]. Electronic Notes in Discrete Mathematics, 2012, 39(1): 13-20.
20	张淑美, 王福利, 谭帅, 等. 多模态过程的全自动离线模态识别方法[J]. 自动化学报, 2016, 42(1): 60-80.
	Zhang S M, Wang F L, Tan S, et al. A fully automatic offline model identification method for multi-model process[J]. Acta Automatica Sinica, 2016, 42(1): 60-80.
21	Deng X, Tian X. Nonlinear process fault pattern recognition using statistics kernel PCA similarity factor[J]. Neurocomputing, 2013, 121(18): 298-308.
22	Dobos L, Abonyi J. On-line detection of homogeneous operation ranges by dynamic principal component analysis based time-series segmentation[J]. Chemical Engineering Science, 2012, 75(25): 96-105.
23	刘元元, 杨功流, 李思宜. BP-AdaBoost模型在光纤陀螺零偏温度补偿中的应用[J]. 北京航空航天大学学报, 2014, 40(2): 235-239.
	Liu Y Y, Yang G L, Li S Y. Application of BP-AdaBoost model in temperature compensation for fiber optic gyroscope bias[J]. Journal of Beijing University of Aeronautics and Astronautics, 2014, 40(2): 235-239.
24	Chang L Q, Bian Z F. Research and implementation of pattern recognition based on Adaboost algorithm[J]. Sensor and Transducers, 2013, 159(11): 7-12.
25	孙曙光, 张强, 杜太行, 等. 基于灰色关联度的框架式断路器故障诊断方法[J]. 仪器仪表学报, 2017, 38(10): 2525-2535.
	Sun S G, Zhang Q, Du T X, et al. Fault diagnosis method of frame-type circuit breaker based on grey correlation degree[J]. Chinese Journal of Scientific Instrument, 2017, 38(10): 2525-2535.
26	Jin X, Wang S, Huang B, et al. Multiple model based LPV soft sensor development with irregular/missing process output measurement[J]. Control Engineering Practice, 2012, 20(2): 165-172.
27	王振雷, 陈登乾, 王昕. 基于多模型切换策略的精馏塔最优控制[J]. 清华大学学报(自然科学版), 2016, 56(4): 430-436.
	Wang Z L, Chen D Q, Wang X. Optimal control of distillation columns based on a multiple model switching strategy[J]. Journal of Tsinghua University(Science and Technology), 2016, 56(4): 430-436.
28	牛大鹏, 刘元清. 基于改进即时学习算法的湿法冶金浸出过程建模[J]. 化工学报, 2017, 68(7): 2873-2879.
	Niu D P, Liu Y Q. Modeling hydrometallurgical leaching process based on improved just-in-time learning algorithm [J]. CIESC Journal, 2017, 68(7): 2873-2879.
29	罗英俊, 万仁溥. 采油技术手册[M]. 3版. 北京: 石油工业出版社, 2007: 75-77.
	Luo Y J, Wan R P. Handbook of Petroleum Production Technology[M]. 3rd ed. Beijing: Petroleum Industry Press, 2007: 75-77.
30	李翔宇, 高宪文, 侯延彬. 基于示功图的抽油井动液面软测量机理建模[J]. 控制工程, 2018, 25(3): 464-471.
	Li X Y, Gao X W, Hou Y B. Soft-sensor mechanism modeling for dynamic fluid level of beam pumping systems based on dynamometer card[J]. Control Engineering of China, 2018, 25(3): 464-471.

[1]	尹刚, 李伊惠, 何飞, 曹文琦, 王民, 颜非亚, 向禹, 卢剑, 罗斌, 卢润廷. 基于KPCA和SVM的铝电解槽漏槽事故预警方法[J]. 化工学报, 2023, 74(8): 3419-3428.
[2]	闫琳琦, 王振雷. 基于STA-BiLSTM-LightGBM组合模型的多步预测软测量建模[J]. 化工学报, 2023, 74(8): 3407-3418.
[3]	郭雨莹, 敬加强, 黄婉妮, 张平, 孙杰, 朱宇, 冯君炫, 陆洪江. 稠油管道水润滑减阻及压降预测模型修正[J]. 化工学报, 2023, 74(7): 2898-2907.
[4]	于源, 陈薇薇, 付俊杰, 刘家祥, 焦志伟. 几何相似涡流空气分级机环形区流场变化规律研究及预测[J]. 化工学报, 2023, 74(6): 2363-2373.
[5]	邵伟明, 韩文学, 宋伟, 杨勇, 陈灿, 赵东亚. 基于分布式贝叶斯隐马尔可夫回归的动态软测量建模方法[J]. 化工学报, 2023, 74(6): 2495-2502.
[6]	李艳辉, 丁邵明, 白周央, 张一楠, 于智红, 邢利梅, 高鹏飞, 王永贞. 非常规服役超临界锅炉的微纳尺度腐蚀动力学模型建立及应用[J]. 化工学报, 2023, 74(6): 2436-2446.
[7]	吴心远, 刘奇磊, 曹博渊, 张磊, 都健. Group2vec：基于无监督机器学习的基团向量表示及其物性预测应用[J]. 化工学报, 2023, 74(3): 1187-1194.
[8]	张中秋, 李宏光, 石逸林. 基于人工预测调控策略的复杂化工过程多任务学习方法[J]. 化工学报, 2023, 74(3): 1195-1204.
[9]	史克年, 郑景元, 钱宇, 杨思宇. 基于马尔可夫链的蒸汽动力系统两阶段随机规划[J]. 化工学报, 2023, 74(2): 807-817.
[10]	陈家辉, 杨鑫泽, 陈顾中, 宋震, 漆志文. 以离子液体密度为例的分子性质预测模型建模方法探讨[J]. 化工学报, 2023, 74(2): 630-641.
[11]	杨静, 林振康, 汤君, 樊铖, 孙克宁. 电池系统的故障特征以及多故障的诊断与识别[J]. 化工学报, 2022, 73(8): 3394-3405.
[12]	周新杰, 王建林, 艾兴聪, 随恩光, 王汝童. 基于IDPC-RVM的多模态间歇过程质量变量在线预测[J]. 化工学报, 2022, 73(7): 3120-3130.
[13]	周乐, 沈程凯, 吴超, 侯北平, 宋执环. 深度融合特征提取网络及其在化工过程软测量中的应用[J]. 化工学报, 2022, 73(7): 3156-3165.
[14]	王琨, 侍洪波, 谭帅, 宋冰, 陶阳. 局部时差约束邻域保持嵌入算法在故障检测中的应用[J]. 化工学报, 2022, 73(7): 3109-3119.
[15]	吴子睿, 孙瑞, 石凌峰, 田华, 王轩, 舒歌群. CO₂混合工质的气液相平衡的混合规则对比与预测研究[J]. 化工学报, 2022, 73(4): 1483-1492.