33 results about "Correlation clustering" patented technology

Disclosed is a rolling bearing remaining life prediction method based on feature fusion and particle filtering. According to an index calculation process, firstly, original features are extracted from bearing vibration signals, the extracted original features are clustered by the adoption of a relevance clustering method, then, one typical feature is selected from each cluster to form optimal feature sets, and finally the feature sets are fused by the adoption of a weight fusion method into a final recession index. According to a life prediction process, firstly, smoothing and resampling are carried out on the recession index, the time interval is adjusted to be an expected value, state-space model initial parameters are calculated by the adoption of least square fitting, then, model parameters are updated in real time according to new observation data, and finally the remaining life of a bearing can be predicted. According to the rolling bearing remaining life prediction method based on feature fusion and particle filtering, the difference between the life prediction result and a true value is small, and the application effect is good.

The invention relates to the field of excavation, particularly to an gray correlation clustering method based on LDTW distance, comprises the following steps of: processing an original data set to obtain a pre-processed sequence; constructing a reference sequence for the maximum value of each dimension in the pre-processed sequence; calculating the LDTW distance and the bending path length of thepre-processed sequence and the reference sequence; calculating the gray correlation degree between the pre-processed sequence and the reference sequence based on the LDTW distance; dividing the critical value interval into a plurality of critical sections, according to the result of the gray correlation degree, if the gray correlation degree of the two sequences falls within the same critical section, grouping the two sequences into one type. The invention reduces the error of the similarity measure between the two sequences, and can provide help for biologists to study the function of proteins.

A framework for an automated news recommendation system system for financial analysis. The system includes the automated ingestion, relevancy, clustering, and ranking of news events for financial analysts in the capital markets. The framework is adaptable to any form of input news data and can seamlessly integrate with other data used for analysis like financial data.

A framework for an automated news recommendation system for financial analysis. The system includes the automated ingestion, relevancy, clustering, and ranking of news events for financial analysts in the capital markets. The framework is adaptable to any form of input news data and can seamlessly integrate with other data used for analysis like financial data.

The invention discloses a method for automatically discovering synonyms based on a subject model, which at least comprises the following steps: importing data of synonyms to be discovered; word segmentation being performed on the imported data according to the information of the database; constructing theme model and clustering theme model; performing minimum correlation clustering for topic clustering; outputing synonym. The invention does not need prior knowledge and manual labeling, realizes automatic clustering of synonyms, and improves the efficiency of synonym discovery. To a certain extent, the problem of semantic similarity is solved, and manual intervention is not needed in the implementation process except the final screening, which greatly improves the efficiency of synonym automatic discovery.

The invention discloses an entity resolution method suitable for a big data environment and capable of achieving noise immunity. According to the entity resolution method, an improvement is conducted based on a traditional relevancy clustering method, the neighborhood and the core concept are introduced, and the entity resolution method is achieved through two layers of algorithms. The upper-layer algorithm is used for conducting rough pre-partitioning allowing overlapping on data based on the neighborhood. The lower-layer algorithm is used for precisely defining the relevancy degree between a node and a class through introduction of the core concept so that the belonging of the node can be accurately judged, and therefore the accuracy of relevancy clustering is improved.

The invention discloses a pedestrian multi-target tracking method based on correlation clustering and space-time constraint. The method comprises the steps: pedestrian visual feature extraction, correlation clustering based on visual features, pedestrian trajectory association under a single camera, and pedestrian trajectory matching under a cross-camera condition by using a space-time constraintmethod. Aiming at the problem that pedestrian tracking under a single camera is easy to interrupt, a space-time sliding window is introduced to solve the problem; meanwhile, a space-time constraint method is introduced to associate the same pedestrian in a cross-camera scene, so pedestrian multi-target tracking in a cross-camera scene is realized; by utilizing the method provided by the invention,tracking indexes such as MOTA, MOTP and recall rate in pedestrian tracking can be consistently improved.

The invention provides an automobile millimeter-wave radar plot condensation method and system. The method comprises the steps: preferentially carrying out track and plot correlation clustering according to a target position prediction value, a Doppler speed and target length and width information fed back by the track information of a previous frame when the current frame of plot data is condensed, and then carrying out the plot-level condensation of remaining plots. Therefore, a problem that large target splitting and adjacent target merging are difficult to merge and process is avoided, and the problem of large target splitting of the automobile millimeter-wave radar is effectively solved. In addition, in the track and plot correlation clustering process, the length and width information obtained after clustering is corrected through a target recognition result, and more reliable target size feature information is ensured to be output.

The invention belongs to the technical field of image processing, and particularly relates to an SLAM autonomous navigation identification method in a closed scene, an SLAM data association module in data association of the method tracks common features of images in different frames, the same feature matching is realized through correlation clustering between the frames, and the SLAM autonomous navigation identification method in the closed scene is realized. The moving condition of the autonomous robot is judged; compared with the prior art, the method has the advantages that (1) compared with a data association method in the existing SLAM (Simultaneous Localization and Mapping), the clustering matching has higher practicability, and the clustering matching effect is not influenced by various complex scenes; and (2) the K-means clustering algorithm is simple in calculation, can be embedded into various systems for feature matching, and is very suitable for visual SLAM in a dynamic environment.

The invention relates to the technical field of power distribution network fault analysis, and discloses a power failure range analysis method and system based on a power failure correlation clustering cluster, and the method comprises the steps: carrying out the clustering analysis of the power consumption data of each user through a K-means clustering algorithm, classifying the power consumption data with the highest similarity into a same cluster group, and carrying out the clustering analysis of the power consumption data with the highest similarity; calculating a power failure correlation coefficient between every two power consumption data in the same cluster, generating a correlation coefficient matrix, determining a corresponding characteristic user, and performing polling of power failure and power consumption states on a transformer area, a feeder line layer connected with the transformer area and other users in the cluster in a multi-level manner by adopting a polling technology based on power failure active alarm of the characteristic user. Therefore, the power failure range can be determined, and the positioning precision of the power failure range of the low-voltage transformer area can be effectively improved.

The present invention belongs to the field of integrated circuit design automation, and particularly relates to an integrated circuit simulation data correlation modeling method and apparatus based onthe correlation clustering and covariance contraction technology. The method comprises: obtaining circuit simulation data required for modeling; constructing a primitive multivariate normal distribution according to the data; and modifying the distribution through the correlation clustering and covariance contraction technology to obtain a correlation model expressed by the modified multivariatenormal distribution. According to the technical scheme of the present invention, the reliability and accuracy of the correlation model are improved, so that the correlation model of the circuit simulation data can be applied to circuits of any scale, and accuracy and efficiency the algorithm developed by the model can be ensured.

An aim of the application is to provide a boundary data partitioning method and device. Specifically, a correlation high-density zone of boundary data is obtained via undisputed data of a correlation clustering group in a clustering result; centralized data in the correlation high-density zone is cut out from the undisputed data of the correlation clustering group, similarity between the boundary data and the centralized data in the correlation high-density zone is analyzed, the boundary data is partitioned based on the similarity, and the boundary data can be accurately classified in a lossless manner.

An EFA-BBN-based computer quantitative prediction method and system for personnel error probability relates to the technical field of data analysis and computer prediction, and is characterized in that a father node PSF in a general BBN model is analyzed by using an EFA method in combination with a current situation condition, PSF correlation is clustered into an intermediate factor, the father node and a child node of the BBN model are connected, and the probability of personnel error is predicted. According to the prediction method, the number of the PSFs is not limited, n PSFs can be clustered into a small number of intermediate factors, the integrity of original information of the PSFs cannot be lost, and the generated EFA-BBN model can make up for the defect that an existing HRA method does not consider the relation between the PSFs. In addition, clustering factor nodes and child nodes in the model acquire a conditional probability table based on double truncated normal distribution (TN), then a success likelihood index (SLIM) method is utilized to estimate a personnel error probability value, and compared with an existing HRA method, the method can reduce subjectivity of expert judgment and more accurately estimate the personnel error probability.