用分数延迟改进基音预测的CELP编码方案

在ＣＥＬＰ编码器中，通常用延迟为抽样间隔整数倍的长项预测器表征浊音语音的准周期性，然而在低比特率，这种限制降低了编码器的性能。本文在介绍了ＣＥＬＰ编码器原理及激励码本构成后，重点研究了一种新型的基音预测方法；分数延迟基音预测，计算机模拟结果表明，这种方法能对浊音进行更准确的表达，尤其对女性讲话者明显改善了语音质量。     

基于神经网络的河道浅滩演变预测模型

河道浅滩演变是一个复杂的非线性动力学过程, 作者借助神经网络处理非线性问题的优势, 在分析影响河道浅滩演变因素的基础上, 建立了预测河道浅淮演变的BP网络模型, 并对模型中的输入因子和样本的提取进行了探讨. 以闽江竹岐至侯官河段为实例,用“试控法”给出了BP网络模型的建模方案, 用正交设计原理选取相应的训练样本集, 利用该样本集对网络进行学习和训练, 并用训练好的BP网络模型预测浅滩上年内最小水深和年平均淤积厚度. 计算结果表明: 模型预测结果与实际值吻合良好. 这为河道浅滩演变预测研究提供了新方法.     

美国COMSTAC/AST卫星发射市场预测

隶属于美国联邦航空管理委员会（FAA）的美国商业航天运输咨询委员会（COMSTAC）和商业航天运输管理协会（AST）在每年的5-6月份都对未来10年的商业卫星发射市场进行预测。2001年该委员会对商业同步轨道（GEO）和非同步轨道（NGEO）的卫星需求和发射需求都进行了预测。     

应用改进的BP神经网络预测导弹的研制费用

结合导弹武器系统研制的特点，分析并确定了影响导弹武器系统研制费用的主要因素，并在前人研究的基础上，提出了基于改进的BP神经网络的费用预测模型。该模型不仅克服了以往由于专家主观因素而造成的负面影响，而且提高了算法的运行效率和模型的预测精度。通过具体的实例分析比较，充分证明了利用该方法的科学合理性。最后探讨了利用BP神经网络进行数值预测的方法和应该注意的问题，并给出了一些有益的建议。     

Simulation of cup-cone fracture using the cohesive model

The cohesive model is used for the prediction of the crack path during stable crack extension in ductile materials. The problem of crack-path deviation is investigated by means of simulation of crack propagation in a round tensile bar. The respective phenomenon is known as cup-cone fracture. It is shown that the model is able to predict the failure mechanism, which consists of normal fracture in the center and combined normal/shear fracture in the so-called “shear lips” at the specimen’s rim. The damage evolution and crack path predicted by the model are presented. The effect of the normal and shear failure parameters on the crack-deflection point and several aspects of the finite element mesh are discussed.     

Thermomechanical fatigue—damage mechanisms and mechanism-based life prediction methods

An existing extensive database on the isothermal and thermomechanical fatigue behaviour of high-temperature titanium alloy EVII 834 and dispersoid-strengthened aluminum alloy X8019 in SiC particle-reinforced as well as unreinv conditions was used to evaluate both the adaptability of fracture mechanics approaches to TMF and the resulting predictive capabilities of determining material life by crack propagation consideration. Selection of the correct microstructural concepts was emphasised and these concepts were, then adjusted by using data from independent experiments in order to avoid any sort of fitting. It is shown that the cyclic /-integral (δJ_eff concept) is suitable to predict the cyclic lifetime for conditions where the total crack propagation rate is approximately identical to pure fatigue crack growth velocity. In the case that crack propagation is strongly affected by creep, the creep-fatigue damage parameter δ_CF introduced by Riedel can be successfully applied. If environmental effects are very pronounced, the accelerating influence of corrosion on fatigue crack propagation can no longer implicitly be taken into account in the fatigue crack growth law. Instead, a linear combination of the crack growth rate contributions from plain fatigue (determined in vacuum) and from environmental attack is assumed and found to yield a satisfactory prediction, if the relevant corrosion process is taken into account.     

宣汉——达县地区飞仙关组储层预测方法研究

在宣汉-达县地区,以往仅依据地震属性来预测飞仙关组储层分布,钻探结果与实际情况不符,为此本文提出量化分析预测结合主参数分析的新方法和新思路。在量化分析中首先对飞仙关组储层进行高分辨率反演,获得储能系数图;然后结合主振幅剖面和主频率剖面分析,对储层进行预测。在研究区,本方法的预测结果与实钻情况相吻合。     

盐家油田油水井结垢原因探讨

对地层温度65～75℃的盐家油田油水井近井地带、井底、井筒发生严重结垢的原因进行了分析探讨。7口油井采出的地层水均为NaCO3型水．矿化度6．4～10g／L，HCO3浓度高（1．6～3．7g／L），Ca^2＋＋Mg^2＋浓度较低；1口水源井水及1口水井注入水为CaCl2型水，矿化度32～33g／L，Ca^2＋＋Mg^2＋浓度高（3．1～3．8g／L）而HCO3浓度低；采用CaCO3饱和指数法和CaSO4热力学溶解度法预测，在70℃下所有水样均有CaCO3结垢趋势，除3口油井采出的地层水外，均有CaSO4结垢趋势。3口油井采出的地层水与水源水在25℃按不同比例混合后，悬浮物含量大幅度上升。在70℃放置7天后再次大幅度上升，烘干的悬浮物易溶于稀盐酸．其主要成分为碳酸盐．说明注入水、地层水不配伍。1口电泵采油井和1口注水井油管垢样含盐酸可溶物（碳酸盐）超过90％。讨论了碳酸盐、硫酸盐垢的生成条件：温度、压力及pH值。图1表4参3。     

A Novel Channel Predictor Based on Constrained Hidden Markov Model

1　Introduction　Linkadaptationtechniqueisnowwidelyrecognizedasakeysolutiontoincreasethespectralefficiencyofwire lesssystems[1～ 3] .GPRS ,EDGE ,cdma2 0 0 0andWCDMAallincludelinkadaptationasameanstopro videhigherdatarates[4～ 5] .Torealizethepotentialoflinkadaptation ,reliablechannel predictionisneces sary[6 ] .　Therearemanydifferentchannelpredictionmethodsavailablenow ,suchasLongRangePrediction(LRP) [6 ] ,predictionbasedonsubspacesignalprocess ingalgorithms[7] ,predictionbasedonFinit…     

灰色预测法在西安市农村电话预测中的应用

在对西安市农村电话发展现状分析的基础上 ,运用灰色预测法对西安市农村电话未来发展趋势进行预测 ,为制定西安市农村电话发展策略提供参考