9 results about "Cluster algorithm" patented technology

The invention provides a sintering process working condition identification method and system considering a time sequence. The method takes the time sequence data of the process parameters of a sintering process as the input and takes a sintering process working condition as the output, and comprises the steps of firstly, utilizing a Spearman rank correlation analysis method and an information entropy analysis method for parameter selection and combination, and obtaining the combined decision parameters; then, using a fuzzy C-means clustering algorithm based on the dynamic time warping distance for clustering the time sequence data, and obtaining a clustering result of the combined decision parameters; and finally, using a naive Bayes classifier for the working condition recognition, and obtaining the recognized sintering process working condition. According to the working condition recognition method, the effective recognition of the working condition during the sintering process is achieved, and the important economic value and the application value are achieved.

The invention provides an abnormal data detection method and device, and the method is applied to the technical field of data processing and comprises the steps of obtaining the to-be-processed data;clustering the to-be-processed data according to a clustering algorithm to obtain a cluster; pruning the cluster and determining an outlier score of each data point in the pruned cluster; and if the outlier score of a certain data point is higher than a preset threshold value, determining the data point as abnormal data. According to the abnormal data detection method and device provided by the invention, the abnormal value of the to-be-processed data can be accurately detected.

Gating clocks has been a widely adopted technique for reducing dynamic power. The clock gating strategy employed has a huge bearing on the clock tree synthesis quality along with the impact to leakage and dynamic power. This invention is a technique for clock gate optimization to aid the clock tree synthesis. The technique enables cloning and redistribution of the fanout among the existing equivalent clock gates. The technique is placement aware and hence reduces overall clock wire length and area. The technique involves employing the k-means clustering algorithm to geographically partition the design's registers. This invention improves the clock tree synthesis quality on a complex design.

The invention discloses a multi-working-condition frame topological optimization method based on a grey clustering algorithm model, and the method comprises the specific steps: carrying out the initial topological optimization of a structure through employing the grey clustering algorithm model, carrying out the pseudo-density of a unit set A, carrying out the clustering of a sample group in the set A according to the correlation between a unit and the structure, determining a clustering index according to a specific actual engineering condition, and then dividing the set A into a deleted unit set B and a reserved unit set C; and filtering and classifying difficult-to-cluster units in a sample according to a distance selection threshold value from the difficult-to-cluster units to a force transmission path, performing grey clustering analysis, dividing A into a deletion unit set B or a reservation unit set C, and taking a frame as an implementation carrier. According to the method, the grey clustering algorithm is applied to the topological optimization design of the structure, the checkerboard phenomenon is effectively inhibited, the manufacturability of the topological optimization result is improved, the frame structure meeting the dynamic and static characteristics is designed, the design and manufacturing period of a product is shortened, and the research and development efficiency and quality of the product are improved.

The invention discloses a social group determination method, apparatus and device, and a computer storage medium, and belongs to the field of communication social contact. The social group determination method comprises the following steps: obtaining a set of user positions, wherein the set of user positions records the positions of a plurality of users; according to the positions of a plurality of users, through a clustering algorithm, dividing a plurality of users into a plurality of clusters, and establishing a plurality of social groups in one-to-one correspondence with the plurality of clusters, wherein each social group comprises users in the corresponding cluster; and when receiving a current position provided by a target user, determining the distance between the current position of the target user and each cluster according to the current position and the positions of the plurality of users in the user position set, and dividing the target user into a social group corresponding to the target cluster in the plurality of clusters according to the distance between the current position of the target user and each cluster. According to the social group determination method, theproblem of poor flexibility of a social group determination method in related technologies is solved. The effect of improving the flexibility of the social group determination method is achieved.

The invention discloses a non-intrusive household load identification method based on an improved FCM clustering algorithm and an MLP neural network, and the method comprises the steps: carrying out the classification of household loads, and extracting the power variation characteristics of the loads; using an entropy weight method to improve an FCM clustering algorithm, and using the improved clustering algorithm to carry out primary load identification; extracting steady-state difference current harmonic characteristics of the load; an MLP neural network is constructed, and the trained MLP neural network is used to carry out secondary identification on the load; and integrating the primary identification result and the secondary identification result to obtain a complete load identification result. The non-intrusive load identification method covers data features such as active power, reactive power and current harmonics of the load, the accuracy of non-intrusive load identification is improved through selection of multiple features, meanwhile, the workload of load identification only based on a deep learning method is reduced through two identification processes, accurate identification of household electric equipment can be achieved, and the identification efficiency of the household electric equipment is improved. And the method has a good identification effect on the household load with overlapped power characteristics.

The invention relates to a fuel cell service life rapid prediction method based on a clustering algorithm, and belongs to the technical field of fuel cells. Firstly, a collected high-dimensional operation data set of the fuel cell system is projected to a three-dimensional feature vector space through a principal component analysis (PCA) method; and inputting the three-dimensional feature vector after dimension reduction into a clustering algorithm, such as K-means and density-based noise application space clustering DBSCAN (Density-Based Spatial Clustering of Applications with Noise). According to a clustering result, the fuel cell voltage data with similar operation conditions can be classified. Finally, the selected voltage data may be used to accurately represent the true performance degradation of the on-board fuel cell stack. Compared with unprocessed original data, the method can obtain more accurate and more reliable fuel cell performance degradation.

Embodiments of the invention provide a big data mining method. The method comprises the following steps of: obtaining quantity of commodities in user browsing data; classifying users according to thequantity of commodities in each piece of user browsing data so as to determine a target user set; obtaining a single user feature vector of each user in the target user set; and grading the target user set on the basis of a clustering algorithm according to the single user feature vectors so as to determine graded user sets, wherein the quantity of the graded user sets is equal to the quantity ofpredetermined grades. By applying the big data mining method, the data volume of operation can be decreased in a more targeted manner on one hand, and interference, for the clustering effect, of different types of user data can be eliminated on the other hand, so that the user population division is more correct, and accurate and personalized services can be conveniently carried out according to the user population division result.