基于预聚类主动半监督的作战体系效能评估

针对作战仿真实验中体系效能通常依靠专家评估、评估代价较大的问题,提出一种基于预聚类主动半监督学习的作战体系效能评估方法。明确了使用该方法进行作战体系效能评估的基本流程,以及自顶向下的评估模式和二值化的评估标准。重点构建了预聚类主动半监督学习算法,首先,结合作战仿真实验数据的特点,对未评估样本进行预聚类,选择最有价值的样本供专家标注;然后,使用已标注的样本训练主动学习算法和半监督学习算法的公用学习器;最后,利用主动学习算法挑选价值较高的样本交由专家评估,并利用新样本对学习器进行不断更新。作战仿真实验数据表明,该方法在达到预期评估准确度的同时降低了评估代价,能有效应用于大规模作战仿真实验的体系效能评估。

基于图约束和预聚类的主动学习算法在威胁情景感知中的研究

针对现有的威胁感知算法对样本标注代价较大,并且在训练分类器时只使用已标注的威胁样本,提出了一种基于图约束和预聚类的主动学习算法。该算法旨在通过降低标注威胁样本的代价,并且充分利用未标注的威胁样本对训练分类器的辅助作用,训练出更好的分类器以有效地感知威胁情景。该算法用已标注的威胁样本集合训练分类器,从未标注的威胁样本集中挑选出最有价值的威胁样本,并对其进行标注,再将标注后的威胁样本加入已标注的样本集中,同时删去原来未标注样本集中的此样本,最后用新的已标注的威胁样本集重新训练分类器,直到满足循环条件终止。仿真实验表明,基于图约束与预聚类的主动学习算法在达到目标准确率的同时降低了标注代价且误报率较低,能够有效地感知威胁情景,具有一定的研究意义。

基于SOM神经网络聚类以及支持向量机的数控机床热误差建模方法的研究

为了减小主轴季节性热误差影响,提高机床的加工精度,提出了基于针对机床热源进行SOM神经网络预聚类后的支持向量回归机的主轴热误差综合模型。针对一台型号为HTM40100h的车铣复合中心,对主轴的关键温度测点进行了内外热源的划分,并在冬夏两个季节对所有测温点温度和热误差数据进行采集,将外部热源温度数据作为SOM网络的输入变量进行季节性聚类,聚类后的外部热源温度数据连同同时刻的内部热源温度数据一起作为不同季节支持向量回归机模型的输入变量,得到热误差拟合值。将通过聚类预处理的方法与未经聚类的方法进行了对比试验,结果表明:该综合预测模型在冬夏两个季节均获得了较高的建模精度和鲁棒性。

Transformer-based correction scheme for short-term bus load prediction in holidays

To tackle the problem of inaccurate short-term bus load prediction,especially during holidays,a Transformer-based scheme with tailored architectural enhancements is proposed.First,the input data are clustered to reduce complexity and capture inherent characteristics more effectively.Gated residual connections are then employed to selectively propagate salient features across layers,while an attention mechanism focuses on identifying prominent patterns in multivariate time-series data.Ultimately,a pre-trained structure is incorporated to reduce computational complexity.Experimental results based on extensive data show that the proposed scheme achieves improved prediction accuracy over comparative algorithms by at least 32.00%consistently across all buses evaluated,and the fitting effect of holiday load curves is outstanding.Meanwhile,the pre-trained structure drastically reduces the training time of the proposed algorithm by more than 65.75%.The proposed scheme can efficiently predict bus load results while enhancing robustness for holiday predictions,making it better adapted to real-world prediction scenarios.

一种高效的K-means聚类改进算法

针对传统K-means算法在初始质心选取的敏感性以及迭代计算的冗余性这两方面的缺陷,提出一种高效的聚类算法(ECA).根据数据对象的空间分布情况,首先采用空间划分预聚类算法(SDPCA)对数据集实现预聚类划分,然后采用基于邻近簇调整的优化聚类算法(OCANC)对预聚类成果进行优化处理,最终获取聚类成果.实验证明,该改进算法能消除对初始输入的敏感性,以更高的运行效率获取较高质量的聚类结果.

基于网站结构的网络使用挖掘树化模型

根据网站的树形结构特点,提出了一种统一的树化模型对用户访问路径进行建模,解决了因使用原始访问路径产生的"过度比较问题",并通过采用适合该模型的网页相对位置的概念,大大减轻了原来由于使用网页绝对访问位置导致的误差。结果表明:树化模型能够提高用户访问行为的相似度识别率。

基于动态遍历的分层特征网络视觉定位

采用分层特征网络估计查询图像的相机位姿,会出现检索失败和检索速度慢的问题。对分层特征网络进行分析,提出采用动态遍历与预聚类的视觉定位方法。依据场景地图进行图像预聚类,利用图像全局描述符获得候选帧集合并动态遍历查询图像,利用图像局部特征描述符进行特征点匹配,通过PnP算法估计查询图像的相机位姿,由此构建基于MobileNetV3的分层特征网络,以准确提取全局描述符与局部特征点。在典型数据集上与AS、CSL、DenseVLAD、NetVLAD等主流视觉定位方法的对比结果表明,该方法能够改善光照与季节变化场景下对候选帧的检索效率,提升位姿估计精度和候选帧检索速度。

异常权限配置下的角色挖掘方案

角色挖掘是构建RBAC系统的常用方法,但目前的角色挖掘方案在设计时未考虑原始系统存在异常权限配置问题,导致角色挖掘的结果可能包含错误的角色权限配置,给系统带来极大的安全风险。针对该问题,文章提出一种异常权限配置下的角色挖掘方案。首先在用户聚类部分引入Canopy预聚类,通过预聚类提取子集交叠数据,缩小后续谱聚类计算量;然后结合预聚类结果优化谱聚类的初始值选取,并针对访问控制数据由布尔值表示的特点,采用杰卡德距离和汉明距离相结合的方式对Canopy预聚类和谱聚类的距离进行度量,提高用户聚类效果;最后对异常权限配置检测规则进行细化,利用修正后的用户聚类结果进行角色挖掘。实验结果表明,该方案能够有效发现异常权限配置,提高角色挖掘效率。

Prediction method of highway pavement rutting based on the grey theory

In order to make a scientific pavement maintenance decision, a grey-theory-based prediction methodological framework is proposed to predict pavement performance. Based on the field pavement rutting data,analysis of variance （ANOVA）was first used to study the influence of different factors on pavement rutting. Cluster analysis was then employed to investigate the rutting development trend.Based on the clustering results,the grey theory was applied to build pavement rutting models for each cluster, which can effectively reduce the complexity of the predictive model.The results show that axial load and asphalt binder type play important roles in rutting development.The prediction model is capable of capturing the uncertainty in the pavement performance prediction process and can meet the requirements of highway pavement maintenance,and,therefore,has a wide application prospects.

Study on Pests Forecasting Using the Method of Neural Network Based on Fuzzy Clustering

Aimed to the characters of pests forecast such as fuzziness, correlation, nonlinear and real-time as well as decline of generalization capacity of neural network in prediction with few observations, a method of pests forecasting using the method of neural network based on fuzzy clustering was proposed in this experiment. The simulation results demonstrated that the method was simple and practical and could forecast pests fast and accurately, particularly, the method could obtain good results with few samples and samples correlation.

Canny算法和中值滤波法的红外全景图像拼接

因传统拼接手段不能获取有效目标向量,导致拼接后图像模糊、不完整,并且存在图像匹配准确率不高、匹配失误率过高和耗时较长的问题,因此,提出Canny算法和中值滤波法的红外全景图像拼接方法。运用Canny算法提取具有显著性特征的图像边缘向量,同时采用中值滤波法,将图像差异过大的数据做对应的过滤处理。为了降低拼接图像结构层次混乱、不齐的情况,根据过滤结果对待拼接的图像做特征定位处理,并实现目标特征匹配。最终在线性融合的基础上提出基于聚类预筛选的多分辨率融合方法,加强图像整体融合效果,从而得出拼接后的完整红外全景图像。分析实验结果可知,所提方法的匹配失误率最低值仅为8%,说明所提出方法可以在根本上降低图像匹配的失误率,增强匹配的精准度,使拼接后的图像更加清晰,并且耗时较短,充分说明该方法应用价值高。

A Multi-model Approach for Soft Sensor Development Based on Feature Extraction Using Weighted Kernel Fisher Criterion

Multi-model approach can significantly improve the prediction performance of soft sensors in the process with multiple operational conditions.However,traditional clustering algorithms may result in overlapping phenomenon in subclasses,so that edge classes and outliers cannot be effectively dealt with and the modeling result is not satisfactory.In order to solve these problems,a new feature extraction method based on weighted kernel Fisher criterion is presented to improve the clustering accuracy,in which feature mapping is adopted to bring the edge classes and outliers closer to other normal subclasses.Furthermore,the classified data are used to develop a multiple model based on support vector machine.The proposed method is applied to a bisphenol A production process for prediction of the quality index.The simulation results demonstrate its ability in improving the data classification and the prediction performance of the soft sensor.

A two-stage short-term traffic flow prediction method based on AVL and AKNN techniques

Short-term traffic flow prediction is one of the essential issues in intelligent transportation systems(ITS). A new two-stage traffic flow prediction method named AKNN-AVL method is presented, which combines an advanced k-nearest neighbor(AKNN)method and balanced binary tree(AVL) data structure to improve the prediction accuracy. The AKNN method uses pattern recognition two times in the searching process, which considers the previous sequences of traffic flow to forecast the future traffic state. Clustering method and balanced binary tree technique are introduced to build case database to reduce the searching time. To illustrate the effects of these developments, the accuracies performance of AKNN-AVL method, k-nearest neighbor(KNN) method and the auto-regressive and moving average(ARMA) method are compared. These methods are calibrated and evaluated by the real-time data from a freeway traffic detector near North 3rd Ring Road in Beijing under both normal and incident traffic conditions.The comparisons show that the AKNN-AVL method with the optimal neighbor and pattern size outperforms both KNN method and ARMA method under both normal and incident traffic conditions. In addition, the combinations of clustering method and balanced binary tree technique to the prediction method can increase the searching speed and respond rapidly to case database fluctuations.

Preventive effect of gelatinizedly-modified chitosan film on peritoneal adhesion of different types

AIM： To comparatively study the preventive effect of gelatinizedly-modified chitosan film on peritoneal adhesions induced by four different factors in rats. METHODS： Chitosan was chemically modified by gelatinization, and made into films of 60 μm in thickness, and sterilized. Two hundred Sprague-Dawley rats were randomly divided into five groups, Sham-operation group （group A）, wound-induced adhesion group （group B）, purified talc-induced adhesion group （group C）, vascular ligation-induced adhesion group （group D）, and infection-induced adhesion group （group E）, respectively. In each group, the rats were treated with different adhesion-inducing methods at the cecum of vermiform processes and then were divided into control and experimental subgroups. Serous membrane surface of vermiform processes were covered with the films in the experimental subgroups, and no films were used in the control subgroups. After 2 and 4 wk of treatments, the abdominal cavities were reopened and the adhesive severity was graded blindly according to Bhatia＇s method. The cecum of vermiform processes were resected for hydroxyproline （OHP） measurement and pathological examination. RESULTS： Adhesion severity and OHP level： After 2 and 4 wk of the treatments, in the experimental subgroups, the adhesions were significantly lighter and the OHP levels were significantly lower than those of the control subgroups in group B （2 wk： 0.199 ± 0.026 vs 0.285 ± 0.041 μg/mg pr, P 〈 0.001; 4 wk： 0.183 ± 0.034 vs 0.276 ± 0.03 μg/mg pr, P 〈 0.001）, D （2 wk： 0.216 ± 0.036 vs 0.274 ± 0.040 μg/mg pr, P = 0.004; 4 wk： 0.211 ± 0.044 vs 0.281 ± 0.047 μg/mg pr, P = 0.003） and E （2 wk： 0.259 ± 0.039 vs 0.371 ± 0.040 μg/mg pr, P 〈 0.001; 4 wk： 0.242 ± 0.045 vs 0.355 ± 0.029 μg/mg pr, P 〈 0.001）, but there were no significant differences in groups A （2wk： 0.141 ± 0.028 vs 0.137 =k 0.026 μg/mg pr, P = 0.737; 4 wk： 0.132 ± 0.031 vs 0.150 ± 0.035 μg/mg pr, P = 0.225） and C （2 wk： 0.395 ± 0.044 vs 0.378 ± 0.043 μg/mg pr, P = 0.387; 4 wk： 0.370 ± 0.032 vs 0.367 ± 0.041 μg/mg pr, P = 0.853）; Pathological changes： In group B, the main pathological changes were fibroplasias in the treated serous membrane surface and in group D, the fibroplasia was shown in the whole layer of the vermiform processes. In group E, the main pathological changes were acute and chronic suppurative inflammatory reactions. These changes were lighter in the experimental subgroups than those in the control subgroups in the three groups. In group C, the main changes were foreign body giant cell and granuloma reactions and fibroplasias in different degrees, with no apparent differences between the experimental and control subgroups. CONCLUSION： The gelatinizedly-modified chitosan film is effective on preventing peritoneal adhesions induced by wound, ischemia and infection, but the effect is not apparent in foreign body-induced adhesion.

Vari-gram language model based on word clustering

Category-based statistic language model is an important method to solve the problem of sparse data.But there are two bottlenecks:1) The problem of word clustering.It is hard to find a suitable clustering method with good performance and less computation.2) Class-based method always loses the prediction ability to adapt the text in different domains.In order to solve above problems,a definition of word similarity by utilizing mutual information was presented.Based on word similarity,the definition of word set similarity was given.Experiments show that word clustering algorithm based on similarity is better than conventional greedy clustering method in speed and performance,and the perplexity is reduced from 283 to 218.At the same time,an absolute weighted difference method was presented and was used to construct vari-gram language model which has good prediction ability.The perplexity of vari-gram model is reduced from 234.65 to 219.14 on Chinese corpora,and is reduced from 195.56 to 184.25 on English corpora compared with category-based model.

Short-Term Wind Power Prediction Using Fuzzy Clustering and Support Vector Regression

A sustainable production of electricity is essential for low carbon green growth in South Korea. The generation of wind power as renewable energy has been rapidly growing around the world. Undoubtedly, wind energy is unlimited in potential. However due to its own intermittency and volatility, there are difficulties in the effective harvesting of wind energy and the integration of wind power into the current electric power grid. To cope with this, many works have been done for wind speed and power forecasting. In this paper, an SVR （support vector regression） using FCM （Fuzzy C-Means） is proposed for wind speed forecasting. This paper describes the design of an FCM based SVR to increase the prediction accuracy. Proposed model was compared with ordinary SVR model using balanced and unbalanced test data. Also, multi-step ahead forecasting result was compared. Kernel parameters in SVR are adaptively determined in order to improve forecasting accuracy. An illustrative example is given by using real-world wind farm dataset. According to the experimental results, it is shown that the proposed method provides better forecasts of wind power.

Preliminary evaluation of five elephant grass cultivars harvested at different time for sugar production

Five elephant grass cultivars, Pennisetum purpureum, cv. Huanan （Huanan）, P. purpureum, cv. N51 （N51）, P. purpureum, cv. Sumu No. 2 （Sumu-2）, （ Penniseturn americanum x P. purpureum ） x P. purpureum cv. Guimu No. 1 （Guimu-1） and P. americanum cv. Tift23A x P. purpureum cv. Tilt NS1 （Hybrid Pennisetum）, at three harvest stages were studied. With dilute sulfuric acid pretreatment followed by enzymatic hydrolysis, it is found that cel- lulose conversion of the five elephant grass cultivars harvested in August and September is higher than that har- vested in October. The cellulose conversion for elephant grass cultivars harvested in August and September follows an order of Hybrid Pennisetum 〉 Sumu-2 〉 Huanan 〉 Guimu-1 〉 N51. This may be explained by the fact that lignification is gradually strengthened with time, inhibiting degradation of cellulose and hemicellulose. Moreover, cellulose conversions of Hybrid Pennisetum, Sumu-2 and Huanan harvested in August and September are higher based on hierarchical clustering results.

Cluster analysis of the domain of microseismic event attributes for fl oor water inrush warning in the working face

Differences are found in the attributes of microseismic events caused by coal seam rupture,underground structure activation,and groundwater movement in coal mine production.Based on these differences,accurate classification and analysis of microseismic events are important for the water inrush warning of the coal mine working facefloor.Cluster analysis,which classifies samples according to data similarity,has remarkable advantages in nonlinear classification.A water inrush early warning method for coal minefloors is proposed in this paper.First,the short time average over long time average(STA/LTA)method is used to identify effective events from continuous microseismic records to realize the identification of microseismic events in coal mines.Then,ten attributes of microseismic events are extracted,and cluster analysis is conducted in the attribute domain to realize unsupervised classification of microseismic events.Clustering results of synthetic andfield data demonstrate the effectiveness of the proposed method.The analysis offield data clustering results shows that thefirst kind of events with time change rules is of considerable importance to the early warning of water inrush from the coal mine working facefloor.

Solar flare forecasting using learning vector quantity and unsupervised clustering techniques

In this paper, a combined method of unsupervised clustering and learning vector quantity (LVQ) is presented to forecast the occurrence of solar flare. Three magnetic parameters including the maximum horizontal gradient, the length of the neutral line, and the number of singular points are extracted from SOHO/MDI longitudinal magnetograms as measures. Based on these pa- rameters, the sliding-window method is used to form the sequential data by adding three days evolutionary information. Con- sidering the imbalanced problem in dataset, the K-means clustering, as an unsupervised clustering algorithm, is used to convert imbalanced data to balanced ones. Finally, the learning vector quantity is employed to predict the flares level within 48 hours. Experimental results indicate that the performance of the proposed flare forecasting model with sequential data is improved.

Predicting crest settlement in concrete face rockfill dams using adaptive neuro-fuzzy inference system and gene expression programming intelligent methods

This paper deals with the estimation of crest settlement in a concrete face rockfill dam （CFRD）, utilizing intelligent methods. Following completion of dam construction, considerable movements of the crest and the body of the dam can develop during the first impoundment of the reservoir. Although there is vast experience worldwide in CFRD design and construction, few accurate experimental relationships are available to predict the settlement in CFRD. The goal is to advance the development of intelligent methods to estimate the subsidence of dams at the design stage. Due to dam zonifieation and uncertainties in material properties, these methods appear to be the appropriate choice. In this study, the crest settlement behavior of CFRDs is analyzed based on compiled data of 24 CFRDs constructed during recent years around the world, along with the utilization of gene ex- pression programming （GEP） and adaptive neuro-fuzzy inference system （ANFIS） methods. In addition, dam height （H）, shape factor （St）, and time （t, time after first operation） are also assessed, being considered major factors in predicting the settlement behavior. From the relationships proposed, the values ofR2 for both equations of GEP （with and without constant） were 0.9603 and 0.9734, and for the three approaches of ANFIS （grid partitioning （GP）, subtractive clustering method （SCM）, and fuzzy c-means clustering （FCM）） were 0.9693, 0.8657, and 0.8848, respectively. The obtained results indicate that the overall behavior evaluated by this approach is consistent with the measured data of other CFRDs.