Probabilistic neural networks for seismic damage mechanisms prediction

The procedure commonly employed to assess the seismic vulnerability of buildings uses simplified qualitative and quantitative observations obtained from the measured data entered into report forms. In Italy, the data sheets adopted by the National Defence Group against Earthquakes (Gruppo Nazionale per la Difesa dai Terremoti—GNDT) play a unifying and reference role. This paper proposes a method for the processing of the data contained in such report forms which is based on probabilistic neural networks producing a Bayesian classification. The final goal is to exploit the fundamental learning and generalization capabilities of neural networks to obtain an estimate of the vulnerability of structural systems. In particular, the aim is to be able to predict the damage mechanisms which may be triggered in the macro-elements of public worship buildings. Copyright © 1999 John Wiley & Sons Ltd.     

中国东南地区复杂地形下降水概率预报的订正研究

使用TIGGE (the THORPEX interactive grand global ensemble)资料集下欧洲中期天气预报中心(the European Centre for Medium-Range Weather Forecasts, ECMWF)逐日起报的预报时效为24～168 h的日降水量集合预报资料,集合预报共包括51个成员,利用左删失的非齐次Logistic回归方法(left-Censored Non-homogeneous Logistic Regression, CNLR)和标准化的模式后处理方法(Standardized Anomaly Model Output Statistics, SAMOS)对具有复杂地形的中国东南部地区降水预报进行统计后处理。结果表明：采用CNLR方法能够有效改进原始集合预报的平均绝对误差(Mean Absolute Error, MAE)和连续分级概率评分(Continuous Ranked Probability Score, CRPS),提升了降水的定量预报和概率预报的预报技巧。而使用SAMOS方法对数据进行预处理,考虑地形...     

基于Bayes网络的多粒度时空对象地理过程演化建模——以新安江模型为例

多粒度时空对象具有多粒度、多类型、多形态、多参照系、多元关联、多维动态、多能自主特点,可用于直接描述从微观到宏观的现实世界.基于时空对象建模理论构建多尺度地理对象耦合演化的集成表达是多粒度时空对象模型支撑地理分析与建模的关键.本文基于多粒度时空对象建模理论,在概率图和条件概率表的基础上发展了一种基于Bayes网络的地理...     

基于模糊影响图的海洋结构风险分析方法研究

基于模糊影响图(FPID)和失效模式与效果分析(FMEA)建立了一种海洋结构风险分析方法。鉴于风险分析中某些事件发生概率和关系概率两个重要参数确定时需借助专家主观判断,本文将模糊概率的概念引入海洋结构影响图评估方法中。该方法成功地应用到海洋平台人员伤害分析中。虽然这里仅给出了该方法在海洋平台风险评估中的应用,但其具有更广泛应用价值。     

Rocking amplification and strong‐motion duration

This paper characterizes the ability of natural ground motions to induce rocking demands on rigid structures. In particular, focusing on rocking blocks of different size and slenderness subjected to a large number of historic earthquake records, the study unveils the predominant importance of the strong‐motion duration to rocking amplification (ie, peak rocking response without overturning). It proposes original dimensionless intensity measures (IMs), which capture the total duration (or total impulse accordingly) of the time intervals during which the ground motion is capable of triggering rocking motion. The results show that the proposed duration‐based IMs outperform all other examined (intensity, frequency, duration, and/or energy‐based) scalar IMs in terms of both “efficiency” and “sufficiency.” Further, the pertinent probabilistic seismic demand models offer a prediction of the peak rocking demand, which is adequately “universal” and of satisfactory accuracy. Lastly, the analysis shows that an IM that “efficiently” captures rocking amplification is not necessarily an “efficient” IM for predicting rocking overturning, which is dominated by the velocity characteristics (eg, peak velocity) of the ground motion.     

An application of the LTP＿DSEF model to heavy precipitation forecasts of landfalling tropical cyclones over China in 2018

Recently, a track-similarity-based Dynamical-Statistical Ensemble Forecast(LTP_DSEF) model has been developed in an attempt to predict heavy rainfall from Landfalling Tropical cyclones(LTCs). In this study, the LTP_DSEF model is applied to predicting heavy precipitation associated with 10 LTCs occurring over China in 2018. The best forecast scheme of the model with optimized parameters is obtained after testing 3452 different schemes for the 10 LTCs. Then, its performance is compared to that of three operational dynamical models. Results show that the LTP_DSEF model has advantages over the three dynamical models in predicting heavy precipitation accumulated after landfall, especially for rainfall amounts greater than 250 mm. The model also provides superior or slightly inferior heavy rainfall forecast performance for individual LTCs compared to the three dynamical models. In particular, the LTP_DSEF model can predict heavy rainfall with valuable threat scores associated with certain LTCs, which is not possible with the three dynamical models. Moreover, the model can reasonably capture the distribution of heavier accumulated rainfall, albeit with widespread coverage compared to observations. The preliminary results suggest that the LTP_DSEF model can provide useful forecast guidance for heavy accumulated rainfall of LTCs despite its limited variables included in the model.     

Probabilistic development of shear strength model for reinforced concrete squat walls

In order to reconcile the larger scatter and avoid the biased estimate from deterministic predictions for the shear strength of reinforced concrete (RC) squat structural walls, a probabilistic shear strength model is developed in this paper based on the strut‐and‐tie model and the generalized likelihood uncertainty estimation (GLUE) method. The strut‐and‐tie model is used to derive an appropriate function form for the probabilistic shear strength model, where four unknown model parameters (e.g. k₁, k₂, k₃ and k₄) are defined carefully to guarantee them having a clear physical‐based meaning so that the corresponding prior distribution ranges can be specified reasonably. Then, the GLUE method is adopted to estimate the posterior cumulative distribution of k₁, k₂, k₃ and k₄ with an available experimental database. Furthermore, to demonstrate the stability of the estimated posterior cumulative distribution, the sensitivity of three major aspects in GLUE method is investigated. Finally, based on the estimated cumulative distribution of k₁, k₂, k₃ and k₄, the developed probabilistic shear strength model is simplified as a mean prediction model and a standard deviation prediction model for facilitate using in engineering practice. Therefore, with the developed probabilistic shear strength model, not only can the squat structural walls be designed in confidence, but the accuracy of those deterministic predictions can be evaluated in a probabilistic manner. Copyright © 2016 John Wiley & Sons, Ltd.     

大渡河猴子岩水库入库洪水过程预报-实时校正-概率预报集成

与传统确定性预报相比,洪水概率预报能够为防洪调度决策提供更为丰富的信息。以大渡河猴子岩水库以上流域为研究区,建立新安江次洪模型,并采用动态系统响应曲线进行实时洪水预报校正。在确定性预报校正基础上,建立基于水文不确定性处理器（HUP）的次洪概率预报模型,定量分析预报不确定性,实现入库洪水概率预报。结果表明：(1)利用猴子岩流域2009 2019年水文气象资料,建立的新安江次洪模型整体精度较高,率定期和验证期的洪量和洪峰相对误差均在±20%以内,平均确定性系数分别为0.69和0.72;经动态系统响应曲线校正后,洪峰和洪量误差均有降低,率定期和验证期的确定性系数分别提高0.13和0.09。(2)以2020年3场洪水未来48 h预报降雨为输入,新安江模型预报精度不高,且随着预见期增长而降低,但经动态系统响应曲线校正后,整体预报精度有所提高,洪量相对误差减小幅度超50%,确定性系数提高幅度超60%。(3)HUP次洪概率预报模型提供的分布函数中位数Q50的预报精度在一定程度上优于校正后的确定性预报;提供的90%置信区间覆盖率均在90%左右,离散度均在0.40以下,能以相对较窄的区间覆盖大部分实测值...     

On the prediction of persistent processes using the output of deterministic models

A problem frequently met in engineering hydrology is the forecasting of hydrological variables conditional on their historical observations and the hindcasts and forecasts of a deterministic model. On the contrary, it is a common practice for climatologists to use the output of general circulation models (GCMs) for the prediction of climatic variables despite their inability to quantify the uncertainty of the predictions. Here we apply the well-established Bayesian processor of forecasts (BPF) for forecasting hydroclimatic variables using stochastic models through coupling them with GCMs. We extend the BPF to cases where long-term persistence appears, using the Hurst-Kolmogorov process (HKp, also known as fractional Gaussian noise) and we investigate its properties analytically. We apply the framework to calculate the distributions of the mean annual temperature and precipitation stochastic processes for the time period 2016–2100 in the United States of America conditional on historical observations and the respective output of GCMs.     

一种基于蓝牙室内指纹定位的贝叶斯改进算法

贝叶斯估计是重要的位置指纹定位算法,但传统的等值贝叶斯先验概率在动态定位中不适用。针对该问题,本文提出了一种基于贝叶斯指纹定位的改进算法。首先,借助陀螺仪获取的航向信息和高斯核函数模型建立概率投票算法,计算先验概率;然后,结合先验概率和信号强度计算待测点位于参考点上的后验概率;最后,选取概率最高的参考点,以概率为权重计算待测点的最或然值。以智能手机为试验对象,在规则路径试验中,改进算法的平均定位误差为1.15 m,定位误差小于2 m的概率为96.1%,不规则路径试验中,平均定位误差为0.50 m,定位误差在1 m的可信度为94.8%;并且改进算法对定位中位置跳变的现象有明显改善,具有较好的稳健性。