2015- 07 -01 Application of Elman neural net- work in short-term reservoir inflow forecasting in hydropower station

Elman neural network has the adaptability to time-var ying characteristics, the sensitivity to historical data and the ad vantage of self-learning, it can approximate to any nonlinea mapping by an arbitrary accuracy. Meanwhile, the gradient de scent method can make function has properties of monotone de creasing and gradient converges to O. Compared with the tradi tional Back Propagation neural network, the method which eom bining the gradient descent method and Elman neural networ] has better prediction precision for the short-term reservoir inflo~ forecasting in hydropower station.

Comprehensive analysis of CBM recovery in high rank coal reservoir of Jincheng area

Coalbed methane (CBM) predicting recovery in high rank coal reservoir varies greatly in Jincheng area and it seriously influences efficient and economic exploitation of CBM resource. In order to predict more accurate CBM recovery, we conducted history matching and productivity prediction of vertical well by using COMET 3 reservoir modeling software, innovatively adopted the gas desorption experiment of bulk coal at fixed test pressure, analyzed the recovery extent method of Daning multiple-hole horizontal well and Panzhuang well group, and calculated recovery by sorption isotherm method of 14 vertical CBM wells at the abandonment pressures 1.0, 0.7, 0.5 and 0.3 MPa, respectively. The results show that the reservoir simulation methods (numerical simulation method and the recovery extent method) is more reliable than the theoretical analysis of coal sample (sorption isotherm method and desorption experiment method). Also, desorption experiment method at fixed pressure is superior to sorption isotherm method. Through the comprehensive analysis and linear correction, CBM recovery ratios in high rank coal reservoir of Jincheng area were found to be 38.64%, 49.30%, 59.30%, and 69.20% at the abandonment pressures 1.0, 0.7, 0.5 and 0.3 MPa, respectively. The research results are of significant importance in the CBM exploration and development in Jincheng area.

A Technical Review on Shale Gas Production and Unconventional Reservoirs Modeling

During the past several years, natural gas production from shale gas is increased and has adsorbed much attention worldwide. The reason behind this is because of advances gained in shale gas recovery techniques from this ultra-low permeability/porosity rock. These techniques are considered the horizontal drilling of the length of 3000 to 5000 ft long and conducting multi-stage hydraulic fracturing along the horizontal portion of the wells. The successful application of above has also driven down the gas prices worldwide and also culminated the security of gas supply for the upcoming decades. This paper is a technical literature review of shale gas production and modeling for future performance evaluation that identifies the current challenges in different stages. Several different and complex physics of gas flow in such a low permeability formation is also explained and the state of the art of the challenges encountered in the modeling process is also explained. As such, gas desorption phenomena, non-Darcy Flow, gas Klinkenberg effect are investigated for different shale formations in the US. This technical review also takes a look over the hydraulic fracturing effects on the economics of shale gas wells due to its straight tie to the production from shale and also the overall recovery from such reservoirs.

FORECASTING OF A NONLINEAR CASCADE FOR WATERSHED RUNOFF

Forecasting of a nonlinear cascade was developed for modeling watershed runoff, and was tested by computing the direct runoff hydrograph for two rainfall- runoff events on a small watershed in China . The forecasting model was superior to Ding s variable unit hydrograph method and the method of limited differences for these two events.

Co-Integration Models for Koyna and Warna Reservoirs, India

Koyna region, a seismically active region, has many time series observations such as seismicity, reservoir water levels, and many bore well water levels. One of these series is used to predict others since these parameters are interlinked. If these series were stationary, we used correlation analysis. However, it is seen that maximum of these time series are nonstationary. In this case, co-integration method is used that is extracted from econometrics and forecast is possible. We have applied this methodology to study time series of reservoir water levels of this region and we find them to be co-integrated. Therefore, forecast of water levels for one of the reservoir is done from the other as these will never drift apart too much. The outcomes demonstrate that a joint modelling of both data sets based on underlying physics resolves to be sparingly useful for understanding predictability issues in reservoir induced seismicity.

Improvement and Evaluation of the Latest Version of WRF-Lake at a Deep Riverine Reservoir

The WRF-lake vertically one-dimensional(1D)water temperature model,as a submodule of the Weather Research and Forecasting(WRF)system,is being widely used to investigate water-atmosphere interactions.But previous applications revealed that it cannot accurately simulate the water temperature in a deep riverine reservoir during a large flow rate period,and whether it can produce sufficiently accurate heat flux through the water surface of deep riverine reservoirs remains uncertain.In this study,the WRF-lake model was improved for applications in large,deep riverine reservoirs by parametric scheme optimization,and the accuracy of heat flux calculation was evaluated compared with the results of a better physically based model,the Delft3D-Flow,which was previously applied to different kinds of reservoirs successfully.The results show:(1)The latest version of WRF-lake can describe the surface water temperature to some extent but performs poorly in the large flow period.We revised WRF-lake by modifying the vertical thermal diffusivity,and then,the water temperature simulation in the large flow period was improved significantly.(2)The latest version of WRF-lake overestimates the reservoir-atmosphere heat exchange throughout the year,mainly because of underestimating the downward energy transfer in the reservoir,resulting in more heat remaining at the surface and returning to the atmosphere.The modification of vertical thermal diffusivity can improve the surface heat flux calculation significantly.(3)The longitudinal temperature variation and the temperature difference between inflow and outflow,which cannot be considered in the 1D WRF-lake,can also affect the water surface heat flux.

小型水库预报预警系统研发与应用

针对小型水库洪水预报预警研究不足问题,以构建的分布式水文模型、缺资料地区参数移植技术,以及小型水库防洪预警方式与指标体系、预警方案编制等为支撑,利用Java语言、Vue技术、ElementUI组件、EChart组件、Spring Boot、JSON数据交互等技术,设计开发了一套基于B/S架构的标准化、通用型小型水库预报预警系统。介绍了系统的设计模式及功能结构,阐述了基础信息展示、洪水预报、防洪调度、洪水预警、数据管理、基础空间地理分析等主要系统模块的设计,以及系统开发关键技术。以在安徽省的应用为例,展示了系统具备的可靠性、灵活性与实用性。

基于集合Kalman滤波的中长期径流预报

为降低中长期径流预报的不确定性,增加水电站水库的发电效益,针对现有方法侧重于提高单一预报模型确定性预报结果的准确性以降低径流预报不确定性的问题,提出一种基于集合Kalman滤波的入库径流确定性预报方法。以旬为预见期的锦西水库实例验证结果表明:相比传统的单一预报模型和传统的信息融合预报模型,基于集合Kalman滤波的中长期径流预报可使RMSE降低4.78 m^(3)/s,合格率可提高0.56%,且更有效地降低了汛期预报的不确定性,得到了更加准确、可靠的确定性径流预报结果,可为开展流域梯级水电站优化调度提供技术支持。

基于Vine Copula的梯级水库短期发电调度风险估计

基于能准确描述高维变量相关关系的Vine Copula,考虑短期径流预报误差的空间相关性,构建了梯级水库短期发电调度风险估计模型,并将模型应用于长江上游溪洛渡、向家坝和三峡水库,分析了径流预报误差带来的单一水库、梯级水库短期发电调度风险。结果表明:基于C-vine Copula构建的联合分布能较好地描述屏山站、朱沱站、寸滩站和武隆站的日径流预报误差特性;随着水库可调节安全区间范围增大,单一水库发电量不足风险率、弃水风险率均越来越小,梯级水库发电量不足、弃水联合风险率和同现风险率越来越小,即水库调节库容越大,其承担的风险也就越小。

基于多模型融合的中长期径流集成预测方法

中长期水文预报是流域水资源规划与合理配置的重要依据。为提高中长期径流预测精度,提出了一种基于多模型融合的水库中长期径流集成预测方法。该方法将ARMA、BP、LSTM、RF和SVR等5个异质预测模型进行融合,同时采用超参数优化方法确定各模型的最优参数。将其用于青海省龙羊峡水库的中长期径流预报中,结果表明,通过Stacking融合算法建立的集成预测模型相较于单一模型,取得了更高的预测精度(R2值由0.71提升至0.82)。此方法可为提升流域中长期径流预测精度提供一定参考。

基于深度学习集合预报的水库闸门防洪优化调度

为了减轻洪水灾害的损失,有必要开展水库防洪预报调度的研究。针对目前水库防洪调度研究较少考虑闸门实际运行及调度方案可执行性的问题,提出以泄放闸门状态与状态持续时长为决策变量的精细化防洪优化调度模型,并且考虑入库流量预报不确定性的影响,将集合了Long Short-Term Memory(LSTM)、Gated Recurrent Unit(GRU)与Transformer深度学习模型的不确定性集合预报作为调度模型输入,在浙江省椒江流域开展实例研究。结果表明:集合预报精度较单一模型提高了4.6%,其不确定性预报结果可用于水库调度;精细化防洪优化调度相比常规调度优势明显,能降低平均最高坝前水位0.43 m,降低下游控制断面平均洪峰流量32.9 m^(3)/s,方案效果受入库流量不确定性的影响更小,方案可执行性高,对水库防洪调度决策有重要的参考价值。

顺北油气田断溶体油藏产量预测模型研究及应用

Arps递减曲线方程是油气田开发初期产量预测中应用较为普遍的方法,基本是参考类似油藏拟合参数以定压差生产为开发政策指导进行产量预测,对于顺北断溶体油藏地层压力下降明显,且储集层在地饱压差内的非线性流动特征明显,存在唯一的非线性惯性阻力,无法通过描述线性流动特征的采油指数来表征,存在误差较大、产量预测结果单一的缺点。通过分析顺北油气田产量递减特征,基于物质平衡原理,构建动态储量模型,结合顺北油气田通用产能方程和产能评价指标成果,形成了新的断溶体油藏弹性驱产量预测方法,用于指导断溶体油藏弹性驱开发阶段的生产管理,有更高的适应性和准确率,也为断溶体油藏开展产能建设方案经济评价优化、指导油井开发政策奠定了基础、提供多开发政策方案决策支撑。

基于VIC模型的三峡库区流域小时尺度实时洪水预报

VIC模型在流域面积大于3 000 km^(2)的大型流域日或月尺度洪水模拟中表现出良好的适用性,而在大型流域小时尺度实时洪水预报过程中的适用性尚未得到验证。以三峡库区大型流域为例,通过建立小时尺度VIC分布式洪水预报模型,构建三峡库区流域实时滚动洪水预报方案,将VIC分布式洪水预报模型实地化部署并应用于2022及2023年三峡库区实时洪水预报。结果表明:VIC模型应用于2014—2021年三峡库区各子流域历史洪水模拟,率定期和验证期的洪量、洪峰平均合格率均在80%以上,确定性系数均值在0.70以上,构建的VIC分布式洪水预报模型在三峡库区及其子流域洪水模拟中表现出良好的适用性;在2022及2023年三峡库区4场典型洪水的实时洪水预报中,径流深和洪峰的平均相对误差达到16.3%和5.0%,重点产流区产流量平均相对误差为7.8%,能够准确把握库区重点产流区的洪量及来水过程,因此VIC模型在大型流域小时尺度实时洪水预报中具有较大的应用潜力。

基于机器学习的碳酸盐岩油藏地层压力预测

地层压力涉及到开发方式、配产配注等调整,是油田开发过程中极为重要的参数。但获取地层压力需要关井进行压力恢复,操作繁琐,数值模拟法工作量大且计算耗时,现有公式法不太适合采用复杂工作制度的碳酸盐岩油藏。基于相关性计算和主成分分析等数据预处理过程,结合精英策略遗传算法和支持向量回归模型(SEGA-SVR),建立了基于数据驱动的地层压力预测模型。SEGA-SVR模型在训练集决策系数得分为0.97,均方根误差为0.04;测试集决策系数得分0.95,均方根误差为0.05,对邻区验证井也有较好的表现。SEGA-SVR模型的性能与SVR模型相比有了很大提高,与其他机器学习模型相比,总体来说表现最优。研究结果表明,SEGA-SVR模型无需关井即可预测实时地层压力,且通过遗传算法调参省时省力,数据驱动的方式可更好适应复杂情况。同时该模型具有较好的泛化性和稳定性,预测效果较好,为碳酸盐岩油田地层压力预测提供了新方法。

小型水库洪水预报模型研究及应用

针对缺资料地区小型水库入库洪水预报难的问题,构建了一种小型水库入库洪水预报模型,其产流模块采用SCS模型,汇流模块采用地貌瞬时单位线,模型参数基于高精度地形地貌数据确定,不依赖于实测流量资料进行模型参数率定,并在安徽省永堌水库进行了应用。采用1818号台风“温比亚”期间永堌水库强降雨洪水进行了验证分析,应用结果表明,构建的模型对该场次洪水过程两次洪峰模拟值与专家反推的入库洪峰较为一致,相对误差分别为3.2%和-11.8%,均在±20%以内,峰现时差均为0,径流深模拟相对误差为-19.2%,确定性系数为0.949,说明建立的小型水库入库洪水预报模型对本次降雨洪水模拟效果较好,对其他小型水库入库洪水预报具有较好的参考和借鉴意义。

光流法雷达外推产品在突发强降水预报中的应用

地形条件复杂的山丘区中小河流洪水突发性强、汇流时间短,高准确率和长时效性降水短时临近预报产品对提高突发洪水预报精度尤为关键。以2021年9月河南省鸭河口水库出现的千年一遇特大洪水为例,利用国家气象信息中心提供的三源融合格点实况降水资料,检验基于改进光流法的雷达外推降水预报产品在本次洪水过程中0~1 h和0~3 h降水预报的TS评分和预报偏差。结果表明:(1)改进光流法在0~1 h的逐小时降水预报上较为精准,累计雨量在50 mm以下时,TS评分在0.45~0.85之间;雨量在50~70 mm之间时,TS评分在0.35~0.70之间;雨量在70 mm以上时,TS评分在0.25~0.35之间。50 mm以上雨量有较高TS评分表现出改进光流法在极端强降水预报中的优势性。(2)改进光流法在0~3 h的降水预报上,累计雨量在50 mm以下时,TS评分在0.55~0.85之间;在50 mm以上时,TS评分在0.35~0.75之间。该降水预报产品不仅对极端性降水预报效果较好,且预报时效长达3 h,可为防洪调度提供更长的决策时间。(3)改进光流法在0~3 h的降水预报产品与融合实况格点降水相比,雨量在20 mm以下的预报结果比较接近,平均绝对误差在10 mm以内;雨量在20 mm以上时,随雨量增大,平均误差、平均绝对误差、均方根误差均逐渐增大。(4)改进光流法在0~3 h的降水预报产品对影响范围小、降水强度大、维持时间长、累计雨量大的极端强降水有较好的预报表现。研究成果可为洪水预报模型提供一种较为可靠的降水输入预报。

基于改进储备池计算的高精度扭秤动力学状态预测方法

在空间引力波探测任务中,超高精度惯性传感器在入轨之前必须经过地面测试与评价,悬丝扭秤是地面测试的首选装置。为了获得悬丝扭秤的精确动力学模型,以减小其固有系统误差,提出了一种基于改进储备池计算(RC)的扭秤动力学预测模型。所提模型利用注意力机制强化时间序列数据的长期依赖特征,并通过贝叶斯优化算法定位模型最优超参数空间,提高了动力学模型的预测精度。在实验室收集的扭秤转角时序数据上验证了所提方法的有效性,相较于传统RC的模型预测误差平均降低40%以上,可以为惯性传感器的地面测试提供可靠的动力学参考。

入库口湿地干湿状态影响下的洪水模拟研究

水利工程的修建直接影响了流域水文响应过程,增加了流域洪水模拟的难度和不确定性。本文以新立城湿地为研究对象,建立特征指标体系将历史洪水按照湿地蓄水状态划分为湿地干旱、湿润状态下洪水,分析湿地干湿状态下入库洪水的水文特征,通过调整新安江水文模型参数以概化入库口湿地的影响,对入库洪水进行模拟。研究结果表明,采用前期雨量和洪水出现次数可较好地将湿地蓄水状态划分为干旱、湿润两种状态,考虑湿地干湿状态影响后新安江模型对湿地干旱、湿润状态下两类洪水的模拟效果均较好,洪水模拟精度可达甲级水平,能较为准确地模拟出入库口湿地干湿状态影响下流域入库洪水,可为同类其他流域水文建模提供思路。

基于梯度下降法的大房郢水库洪水预报模型参数率定及适用性研究

大房郢水库长期承担合肥市的防洪、供水任务,准确可靠地开展洪水预报对当地防汛抗旱工作具有重要意义。分别采用新安江模型和前期影响雨量(antecedent precipitation index,API)模型构建大房郢水库入库洪水预报模型,基于梯度下降法对模型参数进行优选,并以方案合格率作为评价指标,对2种模型在该水库的适用性进行分析。结果表明:新安江模型和API模型预报方案合格率均为90.7%,预报精度达到甲等;2种模型在不同预报项目中各有优势,对于洪峰和峰现时差,API模型预报效果更好,预报合格率分别为88.9%和100%,较新安江模型分别提高6.7%和5.9%;对于径流深,新安江模型预报效果更为理想,预报合格率可达94.4%,相对于API模型其提升幅度约13.3%。此外,API模型更适用于单式洪水预报,而新安江模型则更适用于复式洪水预报。研究成果可为提升大房郢水库入库洪水预报精度提供理论依据。

基于不同因子筛选指标的丹江口入库月径流预报研究

鉴于筛选识别适宜的预报因子对于提升中长期径流预报精度的重要性,以丹江口入库月径流预报为例,选择Pearson、Kendall、Spearman相关系数及随机森林因子重要性作为因子筛选指标,利用多元回归和随机森林模型,开展基于不同因子筛选指标的丹江口入库月径流预报研究。结果表明,大气环流仍是研究流域降水及产汇流的重要影响因素,部分月份径流与前期海温关系较密切;Spearman相关系数筛选下的随机森林模型全年平均预报效果最优,全年平均合格率为72.02%,因子重要性筛选下的随机森林模型在主汛期效果更优,主汛期平均合格率为69.64%;综合预报因子下的随机森林模型精度有一定的提升,全年平均合格率为75.00%,主汛期平均合格率为71.43%,在全年内不同月份的预报效果更稳定,测试期内12个月合格率的标准差下降较显著。