8 results about "Statistical cluster" patented technology

The present invention specifically relates to a correction control method for calculating the fuel injection amount at different altitudes of a port injection engine, a pressure correction control method for optimizing the fuel injection amount under different environmental pressures, a correction curve of the intake manifold pressure and the oil rail pressure difference is creatively introduced in the model calculation, and the correction value at different altitudes and different intake manifold pressures is accurately calibrated to ensure that the air-fuel ratio can be controlled at a theoretical optimal level at different altitudes. The present invention is based on a large amount of road test data, including measurements at different ambient temperatures and different altitudes. Through big data statistical cluster analysis, it is found that the key factor affecting the accurate calculation of the fuel amount of the port injection engine is the pressure difference between the oil rail pressure and the intake manifold. Through this method, on the one hand, the fuel economy at high altitudes can be reduced, while ensuring high power output in plateau environments.

The invention discloses a method and system for identifying the same source of voltage sags. The identification method comprises the following steps: obtaining sag monitoring data recorded by a power grid voltage sag monitoring device; based on the propagation law of voltage sag waveforms between multi-stage transformers, using a Hausdorff distance algorithm to measure the waveform similarity between multiple groups of sag monitoring data, and using an effective value absolute difference method to calculate the duration of the sag; deriving a transfer formula for a phase jump value through a transformer, using a linear interpolation method to accurately calculate the phase jump value in the sag process, and using the Euclidean distance to respectively calculate the similarity of the phase jump values ​​of multiple sag waveforms; obtaining characteristic indicators of voltage sag data, including waveform similarity, duration and phase jump three-dimensional features, forming voltage sag data points, and jointly forming a same source identification clustering matrix; counting the number of clustering result clusters based on the color of data points in a point cloud image, and outputting the number of sag same source data groups and the clustering results.

The invention discloses a multi-source data fusion processing method and system. The multi-source data fusion processing method comprises the following steps: data collection: collecting latest data through data collection equipment; data classification marking: matching the data and carrying out classification marking to generate updated data; data fusion and preprocessing: developing an efficient data preprocessing method, including data cleaning, standardization and missing value processing; data visualization and dynamic association analysis: developing a dynamic data visualization tool to visually display multi-dimensional and multi-modal characteristics of data, including visualization of real-time data streams; exporting the fused data: exporting the data and processing the data to generate the fused data. According to the method, fusion of high-dimensional and multi-modal heterogeneous image data is realized, screening and judgment are carried out based on technical means such as semantic recognition and statistical clustering, cross-industry large model modeling including finance, high and new technology industry and population information characteristics is constructed, and data visualization and dynamic association are realized.

The invention provides a method for evaluating a set of input data, the input data comprising at least one of: clinical data of a subject; genomic data of a subject; clinical data of a plurality of subjects; and genomic data of a plurality of subjects, using a deep learning algorithm. The method includes obtaining a set of input data, wherein the set of input data comprises raw data arranged into a plurality of data clusters and tuning the deep learning algorithm based on the plurality of data clusters. The deep learning algorithm comprises: an input layer; an output layer; and a plurality of hidden layers. The method further includes performing statistical clustering on the raw data using the deep learning algorithm, thereby generating statistical clusters and obtaining a marker from each statistical cluster. Finally, the set of input data is evaluated based on the markers to derive data of medical relevance in respect of the subject or subjects.

The invention relates to a novel electric power system generalized load component online detection and state tracking method, and belongs to the technical field of electric power system operation monitoring. Comprising the following steps: performing generalized load unit selection based on an optimized statistical clustering method; establishing and expanding an off-line sub-model set comprising a distributed power supply and a battery energy storage system, and discretizing the expanded off-line sub-model set; on the basis of an interactive multi-model algorithm, generalized load online modeling is carried out by utilizing system online data. And on-line detection of generalized load components of the novel power system, dynamic tracking of total load power and on-line identification of key internal state variables are realized by using the established on-line model. According to the method, the model fitting precision is high, the iteration speed is high, real-time and accurate recognition and early warning of generalized load abnormity, faults or operation state deviation can be achieved, and key data support is provided for safe and stable operation of the system.

The application discloses a kind of automatic driving data processing method, system and equipment. Among them, the processing method of automatic driving data includes: for each automatic driving related data, the corresponding relationship of automatic driving collection data and event is established;Event is synchronized to work order system, to create the work order of the event by the work order system;The event is analyzed by the work order, to carry out label annotation;When the state change of work order is monitored, the structured association relationship of label and the event is established, and the structured association relationship is written in association table, to complete the label structuring of the event.Using the present application, the data processing efficiency is improved, the repetitive operation of manual participation is reduced, the speed of problem discovery to data available is accelerated, the analysis result can be accurately fed back to data through automatic mapping of data and event, so that label can directly participate in query, statistics, clustering and data mining, support large-scale analysis.

The invention relates to the technical field of civil engineering nondestructive testing, and discloses a method for detecting internal defects of a concrete structure. An excitation source and multiple sound pressure sensors are arranged on the surface of concrete, after sound velocity is calibrated in multiple paths in an intact area, a sound pressure sequence of unified sampling is obtained through unit impact excitation, a failure channel is removed, and maximum time delay is extracted from discrete cross correlation through any sensor and converted into an equivalent propagation path difference matrix. The method comprises the following steps: dispersing a volume to be detected into voxels, constructing a propagation contribution function in combination with each voxel and a sensor geometric path, establishing an optimization model with minimum residual error and sparse constraint, carrying out sparse reconstruction and statistical clustering on voxel impedance equivalent path area contribution, automatically identifying and quantifying positions, shapes and sudden change intensity of internal defects, and carrying out quantitative analysis. Errors caused by sound velocity heterogeneity are effectively suppressed, and imaging precision, reliability and automation level are improved.

The application provides a network abnormal flow intelligence construction method and device based on a knowledge graph, which comprises the following steps: constructing an initial knowledge graph based on static information, dividing features based on dynamic information to obtain statistical features and structured features; performing clustering analysis on the statistical features based on density-guided bisection splitting clustering to obtain statistical clusters, and performing clustering analysis on the structured features based on quality density projection clustering to obtain structured clusters and outliers; performing consensus fusion based on the statistical clusters, the structured clusters and the outliers to obtain feature consensus results, mapping entities corresponding to the dynamic information to the initial knowledge graph based on the feature consensus results to perform dynamic completion, and obtaining a completed knowledge graph; and converting context subgraphs retrieved in the completed knowledge graph with abnormal behavior events as the center into context information and inputting the context information into an AI model to obtain abnormal flow intelligence. The application improves the attack tracing efficiency and decision accuracy for advanced persistent threats.