37 results about "Indicator vector" patented technology

The invention relates to the technical field of data processing, and discloses a power sample data acquisition and classification method and system. The method comprises the following steps: synchronously acquiring active power, reactive power fluctuation and voltage harmonic data through a multi-point terminal, and constructing six types of characteristic index vector groups; a power load genetic optimization algorithm is used to optimize a classification threshold under power flow constraint; based on the optimal threshold value, clustering analysis is carried out through an EFC-KMeans algorithm in combination with impedance matrix characteristics; and inputting the clustering center into a multi-head power topology attention mechanism modeling node coupling relationship to realize power sample data classification and identification. The physical constraint conditions of the power system are effectively fused in the power sample data acquisition and classification process, so that the physical feasibility and engineering practicability of the classification result are improved.

The invention provides a software quality evaluation method and device for a power grid dispatching automation system. Belongs to the technical field of power dispatching automation. The method comprises the following steps: acquiring running state data of power grid dispatching automation system software, and extracting a third-level index from the running state data; determining subjective and objective fusion weights of the third-level indexes; aggregating the third-level indexes related to the same quality feature according to the subjective and objective fusion weights of the third-level indexes to obtain corresponding second-level indexes; an index vector composed of the second-level indexes is input into the dynamic fuzzy neural network model, and the first-level indexes of the corresponding subsystems are output to serve as comprehensive quality evaluation scores; and determining the software quality grade of the business subsystem according to the comprehensive quality evaluation score and a preset grade threshold value. The method is suitable for power grid dispatching automation system software quality evaluation oriented to complex operation scenes, and comprehensive evaluation and self-adaptive optimization of dispatching software under multi-dimensional and dynamic conditions can be achieved.

The invention provides a reverse design method of a starch hydrogel microstructure, and belongs to the technical field of computer material science. A standardized data set is constructed by collecting different starch hydrogel microstructure images and corresponding target performance index vectors. And constructing a starch hydrogel microstructure reverse generation model, and establishing nonlinear mapping between a macroscopic mechanical property index and a microscopic topological structure in combination with a multi-scale self-attention mechanism and a double-flow physical consistency identification strategy. Meanwhile, a differentiable physical performance prediction agent model and a multi-dimensional physical consistency coupling loss function are designed, adversarial iterative optimization under physical constraints is executed, and a prediction physical performance vector of a generated structure is forced to strictly approach a target performance index vector through a self-supervised feedback closed loop. According to the method, accurate anchoring of the generated structure on the mechanical property is achieved, design scheme recommendation and morphological quantitative analysis based on confidence are provided, and the intelligent level and scientificity of bio-based material design are improved.

The invention discloses a micro-service architecture-oriented fault identification method and system without fault marking, and the method comprises the following steps: constructing a micro-service topological potential energy field based on full-link call data collected in real time, and obtaining the potential well depth of each node depended by a whole network; based on a business semantic type extracted from the call chain; calculating an index vector health consistency coefficient of each node and a direct upstream node under the same service semantic type; performing joint calculation according to the overall dynamic semantic entropy break variable, the potential well depth and the health consistency coefficient to obtain a fault score; when the overall dynamic semantic entropy break variable, the health consistency coefficient and the fault score meet preset triggering conditions at the same time, it is judged that a corresponding node breaks down, and an alarm is output; according to the method, the micro-service calling topological graph with the weight is constructed in the sliding time window, so that the utilization degree of structural information during node state judgment is improved.

The invention relates to the technical field of geological disaster prevention and control, and particularly discloses a freeze-thaw collapse event risk early warning method based on an adaptive weighting algorithm and LSTM, and the method comprises the steps: S1, obtaining and collecting multi-source spatio-temporal data, and obtaining a unified data set; s2, obtaining an index vector of each moment; s3, obtaining a normalized feature vector; s4, obtaining a weighted feature vector at the moment; s5, obtaining prediction risk sequences from a short period to a middle period; s6, obtaining risk early warning output for emergency response and a repair strategy; according to the method, an adaptive weighting algorithm based on contribution degree evaluation is introduced, so that the model can dynamically adjust the relative importance of each input index under different time windows and environment conditions. Compared with a static weight method, the mechanism can automatically amplify physical driving factors directly associated with freeze-thaw collapse, such as the influence of short-term snow melting rate, ground temperature gradient or sudden rainfall accumulation, and meanwhile, inhibits event-independent or noise indexes.

The application provides an engineering education system and method fusing explainable multi-objective intelligent optimization, and relates to the technical fields of intelligent education and artificial intelligence optimization. The method comprises the following steps: obtaining an initial engineering design scheme of a user, and obtaining a standardized multi-dimensional index vector through physical simulation analysis and logical analysis; configuring weights with structured reasons for each index, identifying critical weight values leading to decision reversal through sensitivity analysis, and then determining optimization objectives and constraint conditions; performing multi-objective optimization search using an explainable multi-objective optimizer, generating a Pareto optimal solution set and outputting explainable data; simultaneously generating a behavior interaction sequence of the user; generating a capability evaluation report of the user according to the behavior interaction sequence; and adjusting a preset teaching logic sequence according to the capability evaluation report to form a closed-loop teaching process. The application is suitable for cultivating and evaluating the decision-making ability of students in balancing technical performance and environmental sustainability in engineering design.

The invention relates to a historical case similarity-based quality evaluation method, which belongs to the field of software development, and comprises the following steps of: screening out a prior index and a posterior index; collecting data of each stage in the research and development process of each software historical version as historical cases; the specific numerical value of each index is counted; obtaining quality problem source data of the historical cases; taking the historical cases with the labels as input samples, performing model training by adopting an SVM classification algorithm, and constructing a quality problem classification model; constructing a prior index vector matrix for each historical case and the newly-added case, and calculating the similarity between the newly-added case and each historical case; automatically classifying the newly added cases and setting labels; calculating a risk value of each similar case; and averaging the risk values of all the similar cases to obtain a risk coefficient value of the newly added case. According to the method, the influence of manual subjective deviation on index interpretation can be avoided, and an objective quality evaluation standard is established through quantitative similarity analysis.

In some aspects, a computing system can train a risk assessment model, using a training process, for determining a risk indicator. The training process can include: accessing a set of features; determining, using a selector network, a set of selected features and an indicator vector; and training the risk assessment model using the set of selected features and the indicator vector. The computing system can determine the risk indicator for a target entity using the trained risk assessment model. The computing system can transmit, to a remote computing device, a responsive message including at least the risk indicator for use in controlling access of the target entity to one or more interactive computing environments.

The invention provides a method for constructing a data quality evaluation model, and belongs to the technical field of quality evaluation, and the method comprises the steps: collecting a to-be-evaluated data set, and obtaining a business demand and an application scene of the to-be-evaluated data set; determining an assessment index vector of the to-be-assessed data set based on the business demand and the application scene of the to-be-assessed data set; based on the to-be-evaluated data set and the evaluation index vector, determining a training quality feature matrix and a test quality feature matrix; and constructing a quality evaluation model based on the training set and the training quality feature matrix, and evaluating and optimizing quality evaluation performance based on the evaluation index vector, the test set and the test quality feature matrix. The method can enhance the capturing capability of the model for the data quality characteristics, achieves the high-precision evaluation of dynamic weighting, improves the intelligence and stability of the quality evaluation model, and improves the adaptability of the model in the quality management and analysis scenes of complex data.

The invention belongs to the technical field of unmanned ship data analysis, and discloses an intelligent weight prediction method for a dynamic scene, and the method comprises the steps: obtaining unmanned ship data; obtaining an index vector of each unmanned ship according to the unmanned ship data; inputting each index vector into a local attention network to obtain a corresponding first attention weight; calculating a corresponding local attention aggregation result according to each first attention weight; inputting each local attention aggregation result into a global attention network to obtain a corresponding second attention weight; calculating a corresponding global attention output result according to each second attention weight; splicing all global attention output results to obtain an output head vector; and inputting the output head vector into a full-connection feed-forward layer to obtain a weight prediction result. According to the method, the relationship among the parameters in the index vector can be captured to match different formation working scenes, and the index weight prediction precision is remarkably improved.

The invention provides a motion axis fault diagnosis method based on OSPCA-Resformer of a current signal, and the method comprises the steps: collecting a steady-state current signal of a drive end of a motion axis of a machine tool under a normal working condition, extracting a peak value, a root-mean-square value and a kurtosis signal feature value of the current signal, carrying out the dimension reduction of data through a PCA principal component analysis method, and selecting a plurality of feature values as feature index vectors; according to the method, a non-intrusive detection method is adopted, unsupervised online detection and supervised offline model diagnosis are combined into a whole, the calculated amount is greatly reduced, and the method has practical possibility for actual industrial production online detection.

Mechanisms are provided for compressing ciphertext data for data transmission. A sparse vector is received, comprising a plurality of vector elements and a tree is built from the sparse vector where each leaf node corresponds to a vector element in the sparse vector, and each subsequent level of the tree is built from a child level below it in the tree. Nodes of a subsequent level have values determined based on values of child nodes connected to them. The mechanisms execute a level-based copy-and-recurse operation on the tree from a root node of the tree to leaf nodes of the leaf node level. The level-based copy-and-recurse operation computes, at each level of the tree, an indicator vector and a selection matrix that identifies which nodes to recurse into. The mechanisms generate the compressed ciphertext data based on the indicator vectors and the sparse vector.

The invention belongs to the technical field of information system resource control, and particularly discloses an intelligent information system resource control method and system based on a large model, and the method comprises the steps: extracting the text data of a request intention, a task type and a task priority label based on a user request log; extracting external event text data based on the external environment data; splicing the index vector and the text vector to obtain an input matrix; predicting the input matrix by using the large model to obtain a task-level resource demand map and a resource control instruction sequence; and taking the task-level resource demand, the resource control instruction sequence, the current resource state of the system, the dependency relationship between the tasks and the historical performance indexes of the system obtained by the large model as input, and obtaining a final task queue priority ranking result and a resource allocation result through a reinforcement learning algorithm. The large model can continuously learn and improve the prediction accuracy and the effectiveness of the control strategy, and self-adaptive optimization is realized.

This invention discloses a dynamic interactive response method and system for e-commerce applications based on sensor data, comprising the following steps: collecting and preprocessing data from the e-commerce application's operation to generate a standardized interactive input data set; extracting continuous state features and discrete event features to construct a dual-path coupling control sequence; performing stage correlation calculations to generate interactive stage indicator vectors; inputting the dual-path coupling control sequence and interactive stage indicator vectors into an improved NCDE model to generate a dynamic interactive hidden state sequence; performing response sensitivity modulation to construct a dynamic interactive response strategy set; dynamically adjusting the interface based on the dynamic interactive response strategy set to generate dynamic interactive response results; collecting subsequent feedback and writing back updates to optimize the dynamic interactive hidden state sequence and the dynamic interactive response strategy set. This invention improves the real-time performance, accuracy, and adaptive optimization capabilities of e-commerce interactive responses.

The invention discloses a digital economy network flow prediction method and system based on artificial intelligence, and relates to the technical field of digital economy monitoring. Comprising the following steps: acquiring historical network traffic time sequence data of a plurality of digital economic entity nodes, corresponding digital economic characteristic data and economic index data aligned with the historical network traffic time sequence data; based on the digital economic characteristic data, the economic index data and the historical network flow time sequence data, constructing a dynamic adjacency graph evolved along with time; the method comprises the following steps: processing historical network traffic time sequence data through a multi-scale time encoder to obtain a node time representation containing a multi-granularity time dependence feature; after node time representation and digital economic feature data are fused, graph volume accumulation combination is carried out in combination with a dynamic adjacency graph, a joint attention mechanism based on an economic index vector generation offset item is introduced, and node representation fused with space-time and economic semantics is generated; and performing initial traffic prediction based on node representation fusing time-space and economic semantics.

This invention belongs to the field of sensors and signal processing, specifically relating to a wavelet kernel-scale sensitivity-guided denoising method for abrasive particle induced voltage signals. The method includes: acquiring abrasive particle induced voltage signals and performing harmonic cancellation to obtain a preprocessed signal; constructing a wavelet kernel function and calculating a kernel-scale guided spectrum using it and the preprocessed signal; constructing a sparse joint denoising model based on the kernel-scale guided spectrum; processing the sparse joint denoising model to obtain a convex optimization objective function; solving the convex optimization objective function using an adaptive step-size gradient descent method and an adaptive iterative shrinking threshold method to obtain a weight vector characterizing the distribution of abrasive particle characteristic signals; binarizing the weight vector characterizing the distribution of abrasive particle characteristic signals to obtain a feature indicator vector; performing a Hadamard product between the feature indicator vector and the preprocessed signal, followed by low-pass filtering to obtain a denoised signal. This invention can adaptively and non-destructively enhance and denoise abrasive particle characteristic signals under strong interference environments.

The application provides a key PVT corner evaluation method and device based on probability integral transformation and equipment. The method comprises the following steps: simulating N design samples randomly sampled to obtain a performance index vector to form an original data set S; modeling the performance index vector of each PVT corner by using a kernel density estimation method to obtain a probability density function and a cumulative distribution function of a circuit performance parameter; mapping the performance index vector into a standard normal distribution variable through a probability integral transformation; constructing a multivariate Gaussian distribution model based on the mapped data and obtaining new samples, combining the new samples with the original data set S to form a complete data set, counting the frequency of each corner as the worst PVT corner in the complete data set, and sorting the frequency to obtain a key PVT corner set, wherein the highest frequency is the first key PVT corner. The embodiment of the application can simplify simulation calculation, significantly reduce the calculation cost, accurately lock the PVT corner with the greatest impact on the circuit performance, and improve the design reliability.

The invention discloses a data classification and grading-oriented preprocessing method and system. The method comprises the following steps: acquiring multi-dimensional original feature data of to-be-evaluated data assets; based on a preset mapping rule, mapping the original feature data of each dimension into a corresponding standardized score to form an index vector, and processing and mapping the original feature data of the data scale dimension by adopting a nonlinear function; based on an entropy weight method, according to the distribution of the original feature data, determining the weight of each dimension to form a weight vector; according to the index vector and the weight vector, calculating a comprehensive evaluation index of the to-be-evaluated data asset; and generating a structured evaluation data set according to the comprehensive evaluation indexes to support data classification and grading decisions based on different industry standards or business rules. The technical implementation of data classification and grading from subjective experience judgment to a quantifiable and computable objective system is realized.

The application belongs to the technical field of unmanned ship data analysis, and discloses an intelligent weight prediction method for dynamic scenes, which comprises the following steps: acquiring unmanned ship data; obtaining index vectors of each unmanned ship according to the unmanned ship data; inputting each index vector into a local attention network to obtain corresponding first attention weights; calculating corresponding local attention aggregation results according to each first attention weight; inputting each local attention aggregation result into a global attention network to obtain corresponding second attention weights; calculating corresponding global attention output results according to each second attention weight; splicing each global attention output result to obtain an output head vector; and inputting the output head vector into a full connection feedforward layer to obtain a weight prediction result. The application can capture the relationship between each parameter in the index vector to match different formation work scenes, and the prediction accuracy of the index weight is significantly improved.

The application relates to a key region identification method based on a four-dimensional characterization index. The method comprises the following steps: based on a sample set, calculating an information surface entropy, a maximum gradient consistency, a local-global similarity divergence and a local-global extreme point ratio to form a four-dimensional characterization index vector of a key region; fusing the four-dimensional characterization index vector based on a Pareto dominance relationship, dividing candidate key regions into different front levels through non-dominated sorting, calculating the crowding distance of each front region, and screening the key region in the test sample space in combination with the front level and the crowding distance. The method can improve the utilization efficiency of a new type of aero-engine digital twin test resources.

The invention provides a key PVT angle evaluation method, device and equipment based on kernel density estimation. Comprising the following steps: simulating randomly sampled N design samples under each PVT angle to obtain performance index vectors, and forming an original data set S by all the performance index vectors; for each PVT angle, calculating a first difference vector between the performance index vector of the PVT angle and the worst performance index vector of all the other PVT angle sets except the PVT angle; modeling the first difference vector of each PVT angle by adopting a kernel density estimation method to obtain a probability density function and a first cumulative distribution function of the first difference vector; calculating the probability that each PVT angle becomes the worst PVT angle according to the first cumulative distribution function; and selecting the PVT angle with the maximum probability as the first key PVT angle. According to the embodiment of the invention, the simulation calculation can be simplified, the calculation cost is remarkably reduced, the PVT angle which has the maximum influence on the circuit performance is accurately locked, and the design reliability is improved.

The invention provides a multi-field coupling fire blast risk assessment and evolution trend reasoning method, and the method comprises the steps: receiving operation monitoring data, and carrying out the standardization, preprocessing and synchronous storage; classifying, quantifying and fusing the input data according to five types of risk dimensions, and constructing a structured index vector; based on a graph structure modeling and graph neural network inference technology, identifying a coupling cause relationship in a multi-system cross region, and dynamically evaluating a risk level; reasoning based on an accident knowledge graph and a symbolic logic rule, judging a possible development path of the accident, and outputting a visual accident trend chain graph; and displaying regional risk situation and accident evolution trend maps, generating linkage response suggestions, and performing early warning and control. According to the method, the technical problems that the coupling relationship and the incentive linkage mechanism among multiple systems cannot be revealed, the multi-system composite risk is difficult to identify, and the evolution process of an accident in time cannot be reasoned, so that the capability of dealing with the dynamic coupling risk in the complex urban underground environment is insufficient are solved.

The invention relates to the technical field of ship engineering, and discloses a ship-paddle-rudder integrated hydrodynamic optimization design method for a large container ship, which comprises the following steps: generating an initial database containing geometric samples, hydrodynamic performance index vectors and flow field data snapshots; processing the flow field data by adopting an intrinsic orthogonal decomposition method to obtain a low-dimensional flow field characteristic modal coefficient vector; constructing a two-layer agent model of geometry-flow field feature and flow field feature-performance mapping; hierarchical collaborative optimization is executed, optimal target flow field features are searched for upper-layer optimization, and target geometric design is reversely solved for lower-layer optimization; and selecting a new sample for simulation based on a preset point adding criterion and iteratively updating the model. According to the method, the efficiency of ship-paddle-rudder integrated hydrodynamic optimization design of the large container ship is improved by constructing the hierarchical agent model and executing hierarchical collaborative optimization, and the physical reliability of the design scheme is ensured through the self-adaptive point adding criterion.

The invention discloses an RFID target tag category information collection method based on double time slots, under unknown tag interference, target tags are classified into M possible categories according to environment state information collected by the target tags, in order to classify local tags (located in a coverage area of a reader-writer) in the target tags, a system needs to execute multiple rounds, and the classification efficiency is high. Each round comprises two stages of non-target label filtering and local target label category information collection; in a non-target label filtering stage, two test frames are introduced to allocate two time slots for a known label, on one hand, non-target labels can be iteratively eliminated from the time slots through the two test frames; on the other hand, unknown labels are identified through the 0 component in the double indication vectors; in a target label category information collection stage, category information of a target label is collected through two information frames; according to the method, the identification probability of the unknown tag is improved, and the collection efficiency of the category information of the local target tag is also improved.

This invention provides a method, application, and electronic device for constructing an author name disambiguation model. First, a name block and an initial training set consisting of entity pair training samples are constructed. An author name disambiguation baseline model is trained based on the initial training set. The baseline model is then used to calculate the global feature importance vector for each entity pair training sample through feature attribution calculation. The score for each entity pair training sample is calculated based on the global feature importance vector and a non-empty indicator vector, and a training set is selected. The baseline model is then retrained using the selected training set to obtain the author name disambiguation model. The objective loss function of the author name disambiguation model includes classification loss and attribution prior regularization, improving model interpretability, reducing information loss due to metadata sparsity, suppressing model attention drift, and being compatible with various model structures. It has low engineering modification costs and can be widely applied to name disambiguation scenarios in various academic data processing systems.

The invention provides a prediction method of a multi-response mixed effect model and a related system. The prediction method comprises the following steps: determining a fixed effect coefficient and a covariance matrix of the model; determining predicted original input data and training data; extracting a preset number of sample data from the training data, and determining predicted to-be-verified input data according to a relationship between the sample data and the original input data; constructing a precision matrix and an indication vector according to the to-be-verified input data, and obtaining a random effect coefficient; and obtaining the prediction result according to the fixed effect coefficient, the covariance matrix and the random effect coefficient. Therefore, the random effect coefficient is estimated through conditional expectation, and the calculation challenge of direct inversion of a large matrix is avoided.

The embodiment of the invention relates to the technical field of evaluation expert recommendation, and discloses an evaluation expert intelligent recommendation method, device and equipment and a storage medium, and the method comprises the steps: carrying out the feature quantification processing of evaluation task demand information, and obtaining a task feature vector; on the basis of the expert information base and the task evaluation scene, constructing an expert portrait of each evaluation expert; obtaining a domain cross adaptation coefficient and a task complexity matching coefficient corresponding to each evaluation expert; generating a time decay factor representing the capability timeliness based on the recent moment of participation in evaluation; performing adaptation degree calculation on the multi-dimensional capability index vector, the domain cross adaptation coefficient, the time attenuation factor, the task complexity matching coefficient and the task feature vector by using an adaptation degree formula to obtain a comprehensive adaptation score of the assessment expert; and generating an evaluation expert recommendation sequence according to each comprehensive adaptation score. A scientific expert recommendation sequence can be automatically generated, and the reliability of recommended assessment experts is improved.

The application provides a multi-party longitudinal GBDT secure bucket aggregation method based on secret sharing and masking, comprising: obtaining a plurality of gradient vector segment values of a plurality of nodes and an indicator vector of a feature owner node; and obtaining sample gradient and segment value of any node by using the plurality of gradient vector segment values and the indicator vector; wherein the plurality of nodes comprise a plurality of passive party nodes, one active party node and one auxiliary party node, and any node in the plurality of passive party nodes is the feature owner node. The application effectively avoids the expensive homomorphic encryption overhead by proposing the multi-party longitudinal GBDT secure bucket aggregation calculation method, and can easily expand to the multi-participation scene by using the mask to protect the original data, thereby significantly improving the performance of the privacy protection GBDT training on the distributed large-scale data set.

The invention belongs to the technical field of financial analysis, and particularly relates to an adaptive multi-criterion decision model, method and device based on monotonic transformation constraint, and the method comprises the steps: S1, obtaining a group of normalized multi-dimensional financial index vectors; s2, dynamically generating an index weight for each sample in the multi-dimensional financial index vector; s3, monotonically increasing nonlinear transformation is carried out on index dimensions in the multi-dimensional financial index vector, and each index dimension is mapped into a new index dimension; and S4, performing element-by-element multiplication on the index weight obtained in the step S2 and the index dimension obtained in the step S3, performing summation on a plurality of index dimensions corresponding to each sample, and performing normalization operation to obtain a comprehensive score corresponding to the sample. According to the method, the strong nonlinear fitting capability is guaranteed, and meanwhile, absolute monotonicity guarantee is provided through the structural design of the monotonous variation network, so that the decision logic of the model is clear and explainable, and accords with field intuition.

The invention relates to the technical field of artificial intelligence machine learning, and discloses a settlement water adaptability intelligent evaluation method and system adopting artificial intelligence machine learning. The method comprises the following steps: constructing a time sequence dynamic heterogeneous graph sequence representing a settlement system; performing space-time alignment and attribute occurrence on the multi-source heterogeneous data; performing spatial neighborhood aggregation and time state evolution by using the space-time diagram neural network model, and extracting high-dimensional space-time features; a multi-dimensional water adaptability evaluation index vector is generated through an evaluation decoder, and water adaptability trend prediction in a future scene is supported. The system comprises a dynamic heterogeneous graph construction module, a multi-source data processing module, a spatial-temporal feature coding module, a water adaptability evaluation decoding module and an evaluation and simulation application module. According to the method, dynamic, fine and multi-dimensional evaluation and prospective prediction of the settlement water adaptability are realized, and the model precision and the decision support capability are remarkably improved.