密度峰值聚类的自适应社区发现算法

为减少社区发现算法中参数的选择对社区划分的影响,同时使算法能够自适应地进行社区划分,本文提出一种基于核密度估计的密度峰值聚类的社区发现算法KDED.首先,定义一种基于信任度的距离度量,将社交网络中的用户关系量化为距离矩阵,使用矩阵元素的大小度量用户关系的紧密程度;然后对距离矩阵进行核密度估计,统计各个节点在网络中的影响大小,结合热扩散模型改进计算流程,使其自适应不同规模的数据集以提高计算精度;结合密度峰值聚类原理和社区属性确定社区中心节点后,可根据节点间的距离得到社区内部层次结构和社区外部的自然结构;最后将剩余节点按距离分配到相应的社区当中以完成社区划分.仿真结果表明:通过可视化软件可观察到,通过KDED算法得到的社区划分结果具有清晰的自然结构和内部层次结构;随着社区规模的提升以及划分难度增加,KDED算法具有出色的稳定性;在真实数据集以及LFR基准网络上均得到较为接近真实划分结果的社区划分,自适应性良好,验证算法的可行性与有效性.     

基于变分贝叶斯推断的半盲信道估计

现有MIMO中继通信系统中,基于张量分解的半盲信道估计不能有效地将信道先验信息引入估计过程中,为此提出一种基于变分贝叶斯推断的信道估计算法.该算法首先利用NP(Nested PARAFAC)张量模型,引入有效精度、噪声精度等隐性超参数,建立信道估计概率图模型;由于所求信道参数后验概率分布较为复杂,传统最大似然和最大后验等点估计方法难以实现,算法采用变分贝叶斯推断,推导出信道矩阵、有效精度及噪声精度的递推公式,使具有因子分解形式的q分布逼近所求信道参数的后验分布;并分析了模型证据的下界、模型的初始化及算法复杂度等.该算法能利用信道先验信息以提高信道估计性能,有效精度和噪声精度等参数可自动调节,且计算复杂度与数据的维度呈线性关系.仿真结果表明:在平稳瑞利衰落信道条件下,与基于交替最小二乘(Alternating Least Square,ALS)的半盲估计算法相比,算法的计算复杂度较低,收敛速度较快;与带监督序列的双线性最小二乘(Bilinear Alternating Least Square,BALS)非盲估计算法,基于ALS及非线性最小二乘(Nolinear Least Square,NLS)的半盲估计算法相比,算法具有较高的估计精度.     

基于强跟踪卡尔曼滤波的电池SOC估计

针对锂电池模型不准确和状态突变导致SOC估计精度不佳的问题,提出了引入时变渐消因子的强跟踪卡尔曼滤波算法.以HPPC试验方法辨识了锂电池的等效二阶RC模型,对比分析了现有的扩展卡尔曼滤波原理及提出的强跟踪卡尔曼滤波算法.通过结合强跟踪原理和卡尔曼滤波算法并引入时变渐消因子,提出的方法能够强制估计残差保持正交特性,并保证残差满足高斯白噪声特性.仿真验证表明,与扩展卡尔曼滤波原理相比,在模型不准确和状态突变的情况下,强跟踪卡尔曼滤波算法具有更高的估计精度,估计误差低于2.5%,提高了近45%.     

Development and Verification of the Equilibrium Strategy for Batteries in Electric Vehicles

The inconsistency of the cells in a battery pack can affect its lifespan,safety and reliability in the electric vehicles.The balanced system is an effective technique to reduce its inconsistency and improve the operating performance.A hybrid equilibrium strategy based on decision combing battery state-of-charge (SOC) and voltage has been proposed.The battery SOC is estimated through an improved least squares method.An equalization hardware in loop (HIL) platform has been constructed.Based on this HIL platform,equilibrium strategy has been verified under the constant-current-constant-voltage (CCCV) and dynamic-stress-test (DST) conditions.Experimental results indicate that the proposed hybrid equalization strategy can achieve good balance effect and avoid the overcharge and over-discharge of the battery pack at the same time.     

Influent generator for probabilistic modeling of nutrient removal wastewater treatment plants

The availability of influent wastewater time series is crucial when using models to assess the performance of a wastewater treatment plant (WWTP) under dynamic flow and loading conditions. Given the difficulty of collecting sufficient data, synthetic generation could be the only option. In this paper a hybrid of statistical (a Markov chain-gamma model for stochastic generation of rainfall and two different multivariate autoregressive models for stochastic generation of air temperature and influent time series in dry conditions) and conceptual modeling techniques is proposed for synthetic generation of influent time series. The time series of rainfall and influent in dry weather conditions are generated using two types of statistical models. These two time series serve as inputs to a conceptual sewer model for generation of influent time series. The application of the proposed influent generator to the Eindhoven WWTP shows that it is a powerful tool for realistic generation of influent time series and is well-suited for probabilistic design of WWTPs as it considers both the effect of input variability and total model uncertainty.     

Self-information-based weighted CUSUM charts for monitoring Poisson count data with varying sample sizes

In many applications, the Poisson count data with varying sample sizes are monitored using statistical process control charts. Among these applications, the weighted CUSUM charts are developed to deal with the effect of the varying sample sizes. However, some of them use limited information of the sample size or the count data while assigning the weights. To gain more information of the process, the self-information weight functions are developed based on both the sample size and the observed count data. Then, the weighted CUSUM charts are proposed with the self-information-based weight. Simulation studies show the self-information-based weighted CUSUM charts perform better than the benchmark methods in detecting small shifts. Moreover, the performance of proposed method with estimated parameters is investigated via simulation. Finally, an example is given to illustrate the application of the proposed weighted CUSUM charts.     

Brownian bridge and statistical error estimation in Monte Carlo reactor calculation

Brownian bridge (BB) is an effective vehicle in processing an output series in Monte Carlo (MC) simulation. However, most estimators based on BB cost the capability of on-the-fly monitoring. Here, on-the-fly implies that statistical error can be computed at every generation except some initial generations. In this work, on-the-fly estimation of standard deviation by the way of BB, which maintains a fixed storage size of tallies, has been investigated within a framework of the iterated integration of simulation output (IISO). Numerical tests on the MC power distribution calculation of a pressurized water reactor core reveal that the IISO approach with a relatively few number of integrations performs fairly well on average. The bias of statistical error can be managed to be about 10% or less.     

Speech and noise power estimation using Gamma modeling

In speech enhancement, having an accurate estimation of the power of the speech and noise signals forming the noisy observation is critical, as it can highly affect the performance of the enhancement algorithm. A method is introduced in which the distributions of the power of the speech and noise periodograms are modeled using the Gamma distribution to extract their shape parameters. These shape parameters are later used in the observed noisy speech to estimate the power when forming speech and noise periodograms. This method results in more accurate and faster power estimation with respect to the well‐known minimum statistics power estimation algorithm and together with the maximum a posteriori speech enhancement algorithm exhibits good speech enhancement performance.     

A geometric approach for adaptive estimation of unknown growth kinetics in bioreactors

This paper proposes a new approach for the estimation of unknown and time-varying specific growth rate in fed-batch bioprocess. A novel adaptive estimation technique based on the concept of invariant manifold is proposed as an effective approach to estimate growth kinetic parameters. An asymptotic nonlinear observer is used to provide simultaneous on-line estimation of biomass concentration and growth kinetic. The method is easy to implement and requires only one tuning parameter. The effectiveness of the proposed algorithm is illustrated with representative bioreactor simulation examples.     

Neural,fuzzy and neuro-fuzzy approach for concentration estimation of volatile organic compounds by surface acoustic wave sensor array

Present study evaluates application of adaptive neuro-fuzzy inference system (ANFIS) for concentration estimation of volatile organic compounds (VOCs) by analyzing response matrix of polymer-functionalized surface acoustic wave (SAW) sensor array. The performance of ANFIS is compared with that of subtractive clustering based fuzzy inference system (SC-FIS) and backpropagation artificial neural network (BP-ANN). For analysis, the raw SAW sensor array data is preprocessed by logarithmic scaling followed by dimensional autoscaling and the feature extraction by principal component analysis (PCA). For concentration prediction, the extracted feature vectors were fed as input to the three methods (ANFIS, SC-FIS and BP-ANN) independently. The performance of the three methods were evaluated on the basis of root mean square error (RMSE) and correlation value involving actual and estimated values of concentration. Five sets of SAW sensor array responses are analyzed. The analysis includes both experimental and synthetic (sensor model generated) data sets. It is found that the ANFIS has the least value of RMSE and highest value of correlation compared to SC-FIS and BP-ANN. This signifies the relative superiority of ANFIS method.