基于改进的BP神经网络入侵检测方法研究

针对遗传算法局部搜索能力弱和收敛速度慢,在选择操作之后加上了禁忌搜索算法,并对交叉操作进行改进,最后用禁忌搜索作为变异操作,从而加快算法的收敛速度,并用此改进的遗传算法来优化BP神经网络的权值。实验证明,采用该方法优化BP神经网络权值,能克服BP神经网络收敛速度慢、局部极小问题。     

多振动区水电站短期调峰负荷分配方法

高水头水电站机组存在与水头相关的多个不规则振动区,当参与系统调峰时,随着水头和机组组合的变化,电站要避开动态的多个振动区运行,极大地增加了调峰计划制定的难度和复杂性。针对高水头多振动区水电站调峰负荷分配问题开展研究,提出一种实用的求解方法。首先利用数学组合和集合运算理论确定组合机组振动区,然后采用耦合启发式的约束处理策略与逐次负荷分配方法进行模型求解。南方电网实际应用的算例表明,所提方法能够有效避开振动区,并能满足电站运行其它复杂约束,是一种求解高水头多振动区水电站调峰问题的可行方法。     

适用于不同电价机制的统一机组组合算法

现有电力市场中存在两种结算电价机制：按机组报价结算(一机一价)和按市场出清价格结算(统一电价)。不同市场之间的结算方式也有所不同,例如,双边交易中采用一机一价结算方式,而实时市场中采用边际电价结算方式。不同结算电价机制下,机组组合的目标函数不同,传统机组组合方法必须根据电价机制的不同进行调整。通过研究发现,两种结算方式下机组组合问题的最优条件具有类似的数学表达形式。基于这一统一的最优条件表达形式,提出了一种新的机组组合算法。与传统拉格朗日松弛法相比,新算法能够有效地求解两种电价机制下的机组组合问题。     

纳米钛粉改性环氧底漆对GFEP飞机电磁屏蔽性能的影响

航空电子设备的发展和地面电磁波源的迅速增多对玻璃纤维增强环氧树脂(GFEP)机身结构飞机的运行安全构成了威胁,亟需提升此类机身结构的电磁屏蔽性能。介绍了电磁屏蔽效能的计算方法,提出了在GFEP飞机机身涂覆纳米钛粉改性环氧底漆的应对方案。制备了纳米钛粉质量分数为1.0%的改性环氧底漆,研究了改性底漆对GFEP飞机机身典型结构样品电磁屏蔽性能的影响。结果表明:在300 kHz~1.5 GHz的测试范围内,涂覆改性底漆的测试样品的电磁屏蔽效能良好,达到9.5~29.7 dB。     

Zr-0.75Sn-0.35Fe-0.15Cr-xNb合金在He+辐照下的显微组织演变及其对腐蚀行为的影响

本工作采用2 MeV He⁺在300℃下对Zr-0.75Sn-0.35Fe-0.15Cr-x Nb(x=0,1.0,mass fraction,%)合金进行辐照,然后在360℃/18.6 MPa/0.01 mol/L LiOH水溶液中进行高压釜腐蚀实验,利用扫描电子显微镜和透射电子显微镜研究辐照对合金显微组织及腐蚀行为的影响。结果表明,添加Nb改变了Zr-0.75Sn-0.35Fe-0.15Cr合金中第二相的数量、尺寸和晶体结构。He⁺辐照1 dpa后,合金的辐照损伤峰区域分布有大量的He泡和型位错环,添加Nb对He泡和位错环的数量及第二相的元素扩散等均产生了一定的影响。腐蚀结果表明,He⁺辐照或添加1.0%的Nb均可促进Zr-0.75Sn-0.35Fe-0.15Cr合金的腐蚀,但当二者同时存在时,辐照能减弱Nb促进腐蚀的作用。     

泡沫在不同渗透率级差填砂管中的调驱特性研究

为了解决胜利油田稠油油藏开采时蒸汽波及效率低的问题,利用泡沫作为蒸汽流度调剖剂,在温度为90℃的条件下通过物理模拟实验测算了泡沫在不同渗透率级差的两并联岩心管中的阻力因子、高低渗管中产液速度变化特征及泡沫驱油时的采出程度与注入压差的变化特征。实验结果表明：起泡剂FCY产生的泡沫阻力因子随渗透率级差的增大而先增大后减小;泡沫能使高渗管的产液速度减小,而使低渗管的产液速度增大;并联岩心管饱和油后采用泡沫驱,注入0.85PV泡沫时,才能形成有效封堵,而且渗透率级差越大,最终注入压差越低;泡沫驱能同时提高高渗管和低渗管的原油采出程度,但随着渗透率级差的增大,高渗管采出程度略有增加,而低渗管采出程度略有降低。     

均匀动态规划方法及其在水电系统优化调度中的应用

动态规划(Dynamic Programming,DP)是求解多维多阶段复杂决策问题的有效方法,然而维数灾问题极大地限制了它的应用。为缓解这一问题,结合均匀试验设计提出均匀动态规划(Uniform Dynamic Programming,UDP)。UDP以DP为基础框架,将各阶段不同维度离散状态的组合视为多因素多水平试验,利用均匀设计表从全部状态变量中优选少数极具代表性、在可行域内均匀散布的状态变量进行计算,大幅降低各阶段状态变量集合基数。分析表明,UDP空间复杂度和时间复杂度由DP指数增长分别降至线性和平方增长,存储量和运算量显著减少,同时澜沧江梯级水电站群仿真测试结果验证了UDP的高效性和实用性。     

玉米须单体成分纯化及类别鉴定

以玉米须浸膏为原料,采用萃取和液相制备色谱相结合的办法分离得到高纯度玉米须黄酮单体,并优化出纯化工艺。     

Surface homogenizing heterostructure coatings induced by Ti3C2Tx MXene for enhanced cycle performance of lithium-rich cathode materials

The layered lithium-rich manganese-based cathode material (Li_1.2Mn_0.54Co_0.13Ni_0.13) has the significant advantage of high specific capacity, but this material also suffers serious defects, including severe capacity attenuation and voltage attenuation during the cycle. At present, most researchers have been working to optimize the cycle performance of lithium-rich materials. In this work, we propose a surface homogenizing heterostructure coating induced by MXene modification to reduce capacity reduction and voltage decay. It can be found that the initial Coulombic efficiency (ICE) increases from 77.2% for the bare electrode Li_1.2Mn_0.54Co_0.13Ni_0.13 (LMO) to 85.5% for 1.4 wt% MXene (Ti₃C₂T_x, T_x represents the surface terminations: OH, O, F) modified lithium-rich (TO2). Furthermore, the discharge specific capacity of the electrode at 5 C rate increased from 160.7 mAh g⁻¹ for LMO to 200.6 mAh g⁻¹ for TO2. More prominently, the outstanding cycle stability with capacity retention rate is 82.1% for TO2 after 200 cycles, while only 64.7% for LMO, and the average discharge voltage dropped from 0.788 to 0.468 V. In addition, the mechanism for improving the electrochemical performance is systematically studied.     

Tibetan Multi-Dialect Speech and Dialect Identity Recognition

Tibetan language has very limited resource for conventional automatic speech recognition so far. It lacks of enough data, sub-word unit, lexicons and word inventories for some dialects. And speech content recognition and dialect classification have been treated as two independent tasks and modeled respectively in most prior works. But the two tasks are highly correlated. In this paper, we present a multi-task WaveNet model to perform simultaneous Tibetan multi-dialect speech recognition and dialect identification. It avoids processing the pronunciation dictionary and word segmentation for new dialects, while, in the meantime, allows training speech recognition and dialect identification in a single model. The experimental results show our method can simultaneously recognize speech content for different Tibetan dialects and identify the dialect with high accuracy using a unified model. The dialect information used in output for training can improve multi-dialect speech recognition accuracy, and the low-resource dialects got higher speech content recognition rate and dialect classification accuracy by multi-dialect and multi-task recognition model than task-specific models.