14 results about "Correlation clustering" patented technology

The present application relates to the field of bioinformatics. In particular, the present application relates to methods and apparatus for speculating variable splicing functionality based on single cell transcriptome data. The method comprises the following steps: determining a variable splicing mode of each gene in a data set in a cell; determining the incidence relation between the variable splicing mode and the gene expression of each gene; a variable splicing mode module is determined according to the incidence relation between the variable splicing modes and the gene expression, and the variable splicing mode module is a variable splicing mode set obtained through clustering according to the correlation between the variable splicing modes and the cell phenotypes; displaying the cell splicing heterogeneity according to the variable splicing mode module; and / or determining a potential regulatory mechanism between the variable splicing mode and the gene expression according to the variable splicing mode module, the potential regulatory mechanism being used for embodying key splicing factors in the gene expression, and a biological approach in which the variable splicing mode affects the cell phenotype.

The invention relates to the technical field of power equipment on-line monitoring and fault diagnosis, and discloses a converter transformer bushing end screen voltage abnormity detection device and early warning method.The method comprises the steps that voltage data are collected through a wireless intelligent sensor, soft synchronization and self-adaptive windowing are achieved through a commutation notch feature locking mechanism, and a preprocessing signal is output; topological correlation clustering is executed to construct a virtual health reference, a dynamic residual sequence is extracted through time domain difference, and system side and equipment side features are decoupled; constructing a variational mode decomposition model, and calculating a characteristic entropy value of the intrinsic mode function; and comparing the entropy value with a threshold value, triggering an active impedance disturbance verification mechanism, controlling a sensor to access a load and comparing an actual voltage drop rate with a theoretical voltage drop rate, and determining an insulation degradation or poor contact fault. High-precision synchronization can be achieved without a hardware clock, and the problems that weak fault features are difficult to extract and virtual connection and insulation faults are difficult to distinguish in a high-impedance loop are effectively solved.

The invention discloses a multi-source operation exception association analysis and grading disposal method and system. The method comprises the steps of telemetry data collection and caching, non-blocking sending and self-adaptive readaptation, sliding time window and matrix construction, exception association clustering and comprehensive scoring, exception type judgment and grading alarm and self-healing disposal. The method comprises the following steps: continuously sensing a program execution state and an abnormal behavior, realizing data peak clipping storage by local cache, introducing a self-adaptive backoff retry mechanism, decentralizing transmission of abnormal data on the premise of not blocking a main service, and identifying and distinguishing anomalies in combination with abnormal propagation characteristics and an influence coupling relationship. And after quantitative judgment, automatically outputting a matched grading treatment strategy and an operation recovery measure. The closed-loop operation treatment mechanism of coverage sensing, transmission, aggregation, judgment and disposal constructed by the invention can realize rapid sensing, accurate attribution and adaptive response of the abnormity in a complex industrial load environment, and the stability and reliability of the operation of an industrial system are remarkably improved.

The application relates to the technical field of warehouse management and intelligent logistics, and discloses a warehouse position allocation strategy optimization method and system based on reinforcement learning, wherein the method comprises the following steps: constructing a warehouse state vector; generating a warehouse position allocation action set of a commodity unit SKU, and screening and migrating the action by adopting a double-layer policy gradient constraint mechanism; collecting real-time operation indexes and outputting real-time reward signals based on a three-dimensional engine simulation environment; constructing an experience four-tuple and performing policy iteration training by using a deep Q network; and performing closed-loop feedback optimization. Compared with the prior art which depends on static rules or heuristic algorithms for SKU warehouse position allocation, especially in the multi-category large warehouse scene, under the conditions of high SKU correlation and frequent combination picking, it is difficult to realize overall picking efficiency optimization. Since the application is driven by a reinforcement learning driven correlation clustering mechanism, dynamic adaptive cluster type warehouse layout optimization is realized, and the operation efficiency of the warehouse system is improved.

The invention discloses a plant flower color identification and classification method based on deep learning, and aims to solve the problems of color cast caused by complex illumination and strong reflection, natural background interference and inaccurate counting caused by shielding overlapping. Performing white balance and color correction, mirror suppression and Retinex enhancement, color and internal color double-branch decoupling attention detection and statistics of Lab color distribution, performing cross-view correlation clustering counting in combination with geometric constraint and color similarity, and introducing domain alignment and small target intensified training; the technical effects of stable flower color judgment, accurate positioning and counting and cross-equipment cross-time-period robust output are achieved.

The application provides a seamless steel pipe inner wall ultrasonic detection data real-time analysis and defect identification method, relates to the seamless steel pipe detection technical field, and comprises the following steps: acquiring an ultrasonic echo signal sequence, performing adaptive mode decomposition to extract a characteristic vector set, establishing an adaptive reference model of a normal inner wall state, calculating a deviation metric value and performing correlation clustering to identify a defect region, and further determining a defect type and position. The application realizes high-precision automatic identification of the seamless steel pipe inner wall defect, and improves the detection efficiency and accuracy.

The invention belongs to the technical field of oat germplasm resource genetic diversity analysis, and discloses an oat germplasm resource genetic diversity analysis and feeding value evaluation method which comprises the following steps: S1, planting oats; s2, recording growth periods, and dividing maturity; s3, recording organ characteristics and hay components, and calculating a feeding value; s4, recording quality character indexes; s5, measuring spike characteristics and seed weight; s6, performing genetic diversity index calculation, correlation, clustering, principal component and multiple correspondence analysis, and analyzing and evaluating genetic diversity and feeding value; according to the method, the genetic diversity index of oat phenotypic characters in each planting area is calculated, correlation analysis, principal component analysis, clustering analysis, structural equation model construction and multiple correspondence analysis among quantitative characters are carried out, the genetic diversity of each resource is explored, the resource suitable for the climate of the planting land is screened out, and the method has the advantages of being high in practicability and the like. And a scientific basis is provided for improvement and innovative breeding of oat germplasm resources.

The invention provides a transformer control method and system based on artificial intelligence, and relates to the technical field of transformer control. The method comprises the steps that historical control data are collected, clustering analysis based on control correlation is carried out, and historical control correlation clustering data are formed; performing control prediction analysis on the historical control correlation clustering data, and establishing control prediction model data; according to the method, real-time monitoring data is acquired, control prediction analysis is carried out in combination with control prediction model data, real-time control prediction data is formed, effective and accurate data prediction is realized through reasonable data processing analysis, and a more intelligent transformer control mode is provided.

The invention relates to the technical field of warehouse management and intelligent logistics, and discloses a warehouse space distribution strategy optimization method and system based on reinforcement learning, and the method comprises the steps: constructing a warehouse state vector; generating a shipping space distribution action set of a commodity unit SKU, and screening and migrating actions by adopting a double-layer strategy gradient constraint mechanism; real-time operation indexes are collected based on the three-dimensional engine simulation environment, and real-time reward signals are output; an experience tetrad is constructed, and strategy iteration training is carried out by using a deep Q network; and executing closed-loop feedback optimization. Compared with the prior art that SKU warehouse space distribution is carried out depending on a static rule or a heuristic algorithm, and especially in a multi-category large warehouse scene and under the condition of high SKU relevancy and frequent combination sorting, the technical problem that the overall sorting efficiency is difficult to optimize is solved. Dynamic self-adaptive class cluster type storage space layout optimization is realized, and the operation efficiency of a warehousing system is improved.

The application provides a kind of fusion time sequence behavior's internet of things terminal exception detection method and system, it is related to internet of things technical field.The method includes: collecting the time sequence communication behavior data of multiple internet of things terminal devices, constructs the time sequence communication behavior matrix of each internet of things terminal device;Multiple internet of things terminal devices are carried out window behavior correlation analysis and construct multiple communication behavior correlation matrix;According to multiple sets of time sequence communication behavior data, construct multiple communication pressure intervals, extract the interval correlation sample set of each communication pressure interval;Communication behavior correlation clustering is carried out in each communication pressure interval to construct multiple behavior class clusters;The behavior class cluster of multiple communication pressure intervals is analyzed to construct class cluster structure migration path, and group behavior anomaly detection is carried out based on class cluster structure migration path to multiple internet of things terminal devices, and internet of things terminal exception detection result is generated.The application realizes the accuracy of improving internet of things terminal device group behavior anomaly detection.

This invention discloses a multimodal named entity recognition method and system for multi-image scenes. The method includes: S1. Performing object detection on the input text sequence and several associated images to obtain a set of candidate objects in each image; S2. Extracting text features from the text sequence and object features from each candidate object; S3. Performing relevance clustering based on the similarity of object features, and selecting cross-image consistent objects related to named entities from the clustering results; S4. Inputting the text features and consistent object features into a dynamic fusion network, dynamically adjusting the modal weights through a cross-modal attention mechanism to generate a multimodal fusion representation; S5. Decoding the multimodal fusion representation to obtain the named entity recognition result. This invention solves the problems of noise interference and static modal weight allocation in traditional methods in multi-image scenes.

The application relates to a self-correlation clustering ectopic heartbeat recognition method, which comprises the following steps: step 1, collecting electrocardio signal data; step 2, data preprocessing; step 3, moving sliding window; step 4, establishing a function; step 5, obtaining atrial premature beat, ventricular premature beat and normal heartbeat correlation data sets; step 6, obtaining the minimum Euclidean distance of vectors in the atrial premature beat, ventricular premature beat and normal heartbeat correlation data sets; step 7, obtaining the maximum Euclidean distance of vectors in the atrial premature beat, ventricular premature beat and normal heartbeat correlation data sets; step 8, obtaining a to-be-detected electrocardio signal segment and performing data preprocessing, constructing a correlation data set phi of the signal segment, and constructing a function of phi and the Euclidean distance in the previous step; and step 9, judging the type of the to-be-detected electrocardio signal according to the function in step 8. The application has the beneficial effect that the position of an R wave and the heartbeat type can be given simultaneously, and the superposition of errors is avoided.

The invention relates to the technical field of system optimization operation and control, and discloses an electric power system decision dependence uncertainty modeling method, system and device and a medium, and the method comprises the steps: recognizing a basic decision unit, so as to obtain the joint representation of the decision dependence uncertainty of all objects; analyzing the influence of the decision on the uncertainty of the supply and demand parties, including the action mechanism of the scheduling operation decision on the uncertainty of the power supply side and the load side, and obtaining the change rule of the overall uncertainty when the decision variable is changed; establishing a representation model of decision dependence uncertainty; and solving the representation model of the decision dependence uncertainty, converting the decision dependence uncertainty problem into a plurality of normal distribution joint probability problems, simplifying the solving process through correlation clustering, and realizing quantitative representation of the uncertainty of the supply and demand parties. According to the method, the uncertainty of decision dependence of the supply and demand parties of the renewable energy dominated power system can be effectively quantified and represented, and the coincidence degree of a calculation result and an actual situation is relatively high.

The invention provides a seamless steel tube inner wall ultrasonic detection data real-time analysis and defect identification method, which relates to the technical field of seamless steel tube detection, and comprises the following steps: acquiring an ultrasonic echo signal sequence, performing adaptive modal decomposition to extract a feature vector set, and establishing an adaptive reference model of a normal inner wall state; and a deviation metric value is calculated and correlation clustering is carried out to identify a defect area so as to determine the defect type and position. High-precision automatic identification of the defects of the inner wall of the seamless steel pipe is realized, and the detection efficiency and accuracy are improved.