19 results about "Combinatorial explosion" patented technology

The invention discloses a quantum bit mapping optimization method for hardware guidance and fidelity perception and a related device, and belongs to the field of quantum computing, and the method comprises the steps: defining each node in a hardware coupling graph of a target quantum processor as an initial community; evaluating all adjacent community pairs according to a reward function defined based on modularity and fidelity of candidate communities, selecting an optimal community pair for fusion, and taking the community with the largest current scale as an optimal candidate area; iteratively executing a community fusion step until a converged single community is obtained; and selecting an optimal candidate region with the physical quantum bit number greater than or equal to the logic quantum bit number of the quantum circuit to be mapped and the minimum scale, and outputting a mapping sub-graph. According to the method, through a recursive community fusion strategy, the operation range of the solver is restrained in a subspace with the scale greatly reduced and the quality better, the problem of combinatorial explosion caused by global search is fundamentally avoided, and the calculation overhead is greatly reduced.

The invention discloses a mass spectrum data high-throughput alignment and parallel qualitative method based on deep characterization learning, and relates to the field of chromatography-mass spectrometry data processing and calculation mass spectrography. The method comprises the following steps: performing streaming, micro-batch and single traversal analysis on original mass spectrum data files of a plurality of samples to be detected; extracting metadata and main data by taking the fragmented spectrogram as a unit, inputting the metadata and the main data into a pre-trained depth representation learning model to generate a high-dimensional embedded vector, and incrementally writing the metadata and the main data into a column type storage database; constructing a sparse approximate nearest neighbor graph; executing a clustering algorithm on the graph to generate a consensus spectrogram set, and generating a consensus feature vector for each clustering cluster and a single example cluster; determining the nature of the consensus feature vector; and constructing a sample * feature matrix based on the cluster affiliation relationship and the qualitative result. According to the method, the technical problems of I / O bottleneck, memory overflow, combinatorial explosion, computing power waste and the like in large-scale mass spectrum data analysis are effectively solved.

The invention relates to the technical field of purchase demand prediction, and discloses a part purchase demand prediction method and system based on machine learning, and the method comprises the steps: obtaining modular product basic data, and generating a BOM graph structure; calculating prior probability distribution through a Bayesian inference algorithm; executing an adaptive probability pruning algorithm to solve the problem of combinatorial explosion; quantizing uncertainty by using a Bayesian deep learning network; calculating a differentiated safety inventory coefficient based on the value-at-risk model; executing an importance sampling algorithm to carry out Monte Carlo simulation on high-risk low-frequency configuration, and verifying a demand coverage rate in an extreme scene; an incremental learning mechanism is utilized to update probability distribution and a pruning threshold according to the new order data, and a self-adaptive optimization demand prediction result is output; according to the method, the inventory cost is remarkably reduced, the stockout risk is reduced, and the balance between the calculation efficiency and the prediction accuracy is realized.

The invention relates to the technical field of large language models, in particular to a multi-dimensional data attribution analysis method and device based on a large language model, which can reduce the problem of combinatorial explosion in a super-multi-dimensional scene, more accurately identify key influence factors and improve the accuracy of multi-dimensional data attribution analysis. The method comprises the steps of obtaining target data for attribution analysis based on natural language description; determining an initial analysis dimension for attribution analysis through a first large language model, determining an influence probability of the initial analysis dimension through a first prediction model, and screening target analysis dimensions with the influence probability higher than a preset probability threshold; performing service prediction based on the target data and the target analysis dimension through a second prediction model to obtain a prediction index value; and performing attribution analysis based on the actual index value and the predicted index value corresponding to the target analysis dimension through the first large language model to obtain an attribution analysis result.

The application discloses a high-entropy alloy target performance reverse rapid proportioning method based on active learning, belongs to the technical field of material genetic engineering and artificial intelligence, and comprises the following steps: establishing component normalization, preparation process and phase stability engineering constraints; constructing digital characterization and continuous design space of high-entropy alloy components; constructing and pre-training a differentiable performance predictor based on an attention mechanism; fixing the predictor parameters, constructing a composite loss function driven by the target performance, iteratively optimizing the component vector through gradient back propagation and normalized projection, and outputting the theoretical optimal candidate component; screening high-value samples based on an active learning strategy to verify and incrementally update the model, and forming a closed-loop optimization. The application breaks through the limitations of traditional forward prediction models, realizes rapid and accurate reverse design from target performance to component proportioning, and effectively solves the combination explosion and data scarcity problems in the high-dimensional design space of high-entropy alloys.

The invention discloses a train automatic driving heuristic test case generation method for an intelligent high-speed railway, and the method comprises the steps: firstly analyzing an ATO system driving mode conversion process, extracting fault features of related equipment, and constructing a test input parameter table; then, according to a parameter coverage requirement, determining the dimensionality of a variable force coverage table, and obtaining a small-scale test case set; then, the genetic algorithm is used for controlling evolution and variation of the test case, and expansion of the vertical dimension is achieved; generating a local optimal solution of the test case in combination with a greedy strategy to realize expansion of a horizontal dimension; and finally, after the vertical dimension and the horizontal dimension both meet the coverage requirements, generating a final initial test case set. According to the method, the generation number of invalid or repeated test cases is reduced, and combinatorial explosion is avoided; through test verification, the method has good function test requirement coverage rate and fault detection capability, and provides important reference for verification of function safety, reliability and integrity of the ATO system.

The application provides an OpenAPI multi-step state fuzzy testing method and related device, and relates to the technical field of network security. Through two stages of offline knowledge construction and online state exploration, the problems of too large sequence space, poor request executability and disconnection of exploration are solved, the search space is effectively compressed, semantic guidance is provided, and combination explosion is avoided; through runtime load multiplexing, the executability of deep state paths is ensured; deep-first search scheduling and snapshot rollback are adopted to ensure exploration efficiency and result reproducibility, thereby having the advantages of efficiently detecting multi-step state vulnerabilities.

The invention discloses an electric power spot day-ahead clearing method considering load increment declaration constraints, and relates to the technical field of electric power system operation decisions. The method comprises the following steps: acquiring declaration parameters to construct a boundary parameter set containing physical increase and decrease output variables, and fusing the boundary parameter set with an objective function and mutual exclusion constraints into a system state control matrix; discrete decision variables in the matrix are mapped into an energy physical model, and a lowest energy ground state solution is used as a unit and load combination state reference; and a dynamic evolution system containing a scheduling strategy and disturbance nodes is constructed, the dynamic evolution system is evolved to a stable limit cycle state in a phase space through the constraint of a differential equation set, and a final day-ahead clearing plan and a node marginal electricity price are generated. The method is used for solving the problems that a traditional algorithm is prone to falling into combinatorial explosion when processing massive discrete variables, and a static clearing model lacks dynamic evolution elasticity and a robust safety boundary when coping with high-dimensional continuous disturbance.

This application discloses a multi-vehicle cooperative simulation decision-making method and system integrating vehicle, road, and cloud technologies. It employs a fast filtering algorithm based on intent spectrum and conflict cost matrix to compress the combinatorial explosion problem into a feasible solution space, significantly improving cloud-based solution efficiency. A cooperative weight bidding mechanism is introduced to resolve conflicts while considering system benefits and vehicle preferences, ensuring the fairness and optimality of the scheduling scheme. Vehicles perform local replanning and safety verification based on cloud-based suggestions, adhering to cooperative intent while maintaining local real-time response capabilities. The system's cooperative status is monitored through a consistency index; when a deviation is triggered, only the deviating vehicle needs to provide a new intent spectrum, allowing for rapid cloud-based optimization and simplified scheduling. This avoids large-scale global replanning, enabling low-latency online dynamic adjustment of cooperative strategies and significantly improving the system's long-term stable operation in complex traffic scenarios.

This invention provides a method, system, storage medium, and electronic device for rating poor driving behavior based on improved extended belief rule reasoning, relating to the field of road traffic safety technology. In this invention, the extended belief rule reasoning (EBRB) method is introduced, utilizing a data-driven decision model to effectively solve the combinatorial explosion problem of the rule base and improve the system's rationality and interpretability. The SC clustering algorithm is introduced, proposing a new activation rule determination method, which effectively reduces the number of rule accesses, improves system operating efficiency, and enhances the consistency among candidate activation rules. Furthermore, using the aforementioned activation rule determination method, a user poor driving behavior rating model based on SC_EBRB is constructed, improving the objectivity and accuracy of poor driving behavior evaluation.

The application discloses a kind of test case generation method for SCADE model, by reading the XML document of SCADE model automatic generation, obtain the information of safety state machine in model, the state and transition in safety state machine model are converted into the node and edge in directed graph;Then depth traversal directed graph, obtain state transition path;To alleviate the combination explosion problem of transition condition, using combination test algorithm AETG, optimize transition condition, obtain transition condition path;And using shielding algorithm carries out the logic value of meeting MC / DC coverage, according to the mapping of transition condition restriction to specific data, generate specific test case for each transition condition, to realize a kind of test case generation for SCADE model.

The present application relates to the technical field of robot optimization design, and more particularly to a high-speed heavy-load robot scale-structure-driving collaborative optimization design method, comprising establishing a coupling optimization model; constructing a trajectory-driven performance quantification interface function, constructing a parameterized performance modeling module, a stiffness proxy model and a binary gating function, and obtaining a compact candidate driving set through type spectrum screening; an outer loop generates a structure-scale candidate scheme and performs rapid feasibility screening; an inner loop iteratively solves the minimum cycle time and demand envelope of the candidate scheme under the trajectory-driven performance quantification interface function on the compact candidate driving set, and performs trajectory-driving closed-loop iteration on the compact candidate driving set to obtain an optimal driving selection; and the outer loop updates the Pareto optimal solution based on a multi-objective optimization function to obtain a multi-objective optimization solution set. The present application solves the problems of low iteration efficiency, combinatorial explosion and performance post-processing of traditional serial design.

A job scheduling technology computationally selects jobs from a set, excluding at least one job by applying at least one job filter, submits the selected jobs to a scheduler mechanism, receives in response a first phase schedule, and produces a second phase schedule updating the first phase schedule to include at least one excluded job. The update avoids computational costs of combinatorial explosion. In some scenarios, a location-agnostic job is excluded from the first phase schedule and then included in the updated schedule. In some scenarios, applying the filters sorts the jobs for submission to the scheduler mechanism.

The invention discloses a whole body motion planning system and method based on contact mode switching, and belongs to the technical field of robots. The method comprises the following steps: acquiring a centroid feasibility map representing environmental geometric information and physical attributes; determining a centroid target trajectory from an initial state to a target state according to the centroid feasibility map; and based on the centroid target trajectory and the centroid feasibility map, according to a preset cost function, planning to obtain a contact mode sequence. The system includes modules for performing the method. According to the method, the guiding track is planned in the low-dimension centroid space firstly, then the track serves as a strong constraint to reversely solve the contact sequence, the problem of combinatorial explosion in traditional contact state space search is fundamentally avoided, and the technical problems that in the prior art, planning calculation complexity is high, and the real-time requirement is difficult to meet are solved.

This document describes kurtosis-based pruning for a sensor fusion system. The kurtosis-based pruning minimizes the total amount of comparisons performed when fusing large data sets together. Multiple candidate radar tracks can be aligned with one of multiple candidate visual tracks. For each candidate visual track, a weight or other match evidence is assigned to each candidate radar track. The inverse of the match error between each candidate visual and each candidate radar track contributes to this evidence, which can be normalized to generate, for each candidate visual track, a distribution associated with all candidate radar tracks. The kurtosis or shape of the distribution is computed. Based on the kurtosis value, some candidate radar tracks are selected for matching and others remaining candidate radar tracks are pruned. The kurtosis helps determine how many candidates to keep and how many to prune. In this way, the kurtosis-based pruning can prevent combinatorial explosion due to large-scale matching.

This invention discloses an interpretable fault diagnosis method for Software-Defined Packet Transport Networks (SPTN) based on a manifold learning-based confidence rule base (ML-BRB), belonging to the field of network fault diagnosis and interpretable artificial intelligence technology. Addressing the problems of BRB combinatorial explosion caused by high-dimensional coupling of SPTN network monitoring indicators and the loss of physical semantics of low-dimensional features after manifold dimensionality reduction, leading to the inability to initialize BRB parameters, this invention proposes the "SPTN network state manifold space hypothesis." It utilizes the t-SNE algorithm to map high-dimensional monitoring data to a low-dimensional manifold space. Based on manifold geometry and dynamic neighborhood statistics, it adaptively determines the initial parameters of the BRB, including adaptive adjustment of reference values ​​and confidence frequency estimation. Through the IF-THEN rule structure of the BRB model, it outputs fault types and confidence distributions with physical meaning, and uses the one-to-one correspondence between samples before and after dimensionality reduction to backtrack the original high-dimensional data, forming a diagnostic closed loop of "manifold decoupling—rule interpretation—sample backtracking." This invention achieves transparency and interpretability in the fault diagnosis process while ensuring diagnostic accuracy, meeting the stringent real-time requirement of 50ms switching in SPTN networks.

This invention provides a multi-pipeline layout method based on a niche co-evolutionary algorithm, belonging to the technical field of pipeline layout methods. It includes the following steps: A1: Establishing an environmental model of the pipeline layout space; A2: Applying the niche co-evolutionary algorithm to solve the multi-pipeline layout problem; A3: When the niche co-evolutionary algorithm meets the termination condition, outputting the optimal global solution as the final multi-pipeline layout scheme. Based on this, this invention introduces a niche layout pattern, an elite collaborator selection strategy, a multi-dimensional information sharing mechanism, and an adaptive evolutionary mechanism, overcoming the shortcomings of insufficient information interaction and premature convergence in existing co-evolutionary algorithms. It significantly improves the global search capability and solution quality for solving complex combinatorial optimization problems, thereby solving the problems of simple collaborator selection mechanisms, single and delayed shared information dimensions, and combinatorial explosion inherent in existing co-evolutionary algorithms applied to multi-pipeline layout.

The invention discloses a time-sensitive network system multi-rate task chain completion time optimization method, and relates to the field of industrial Internet of Things. The method comprises the following steps of: firstly, quantitatively describing a complex dependency relationship between asynchronous execution tasks by establishing an accurate model of an interaction relationship between multi-rate task instances; secondly, a dependency analysis judgment formula between task instances is deduced, the interaction relation between the instances is efficiently and accurately recognized, and the problem that calculation is not feasible due to combination explosion is solved. And finally, based on a joint resource allocation algorithm of an approximate analytic solution, efficient cooperative scheduling of calculation and network resources is realized, the calculation complexity is remarkably reduced while the optimization precision is ensured, and real-time performance optimization of the multi-rate industrial information physical system is realized in actual engineering.

The present application relates to a spin-torque nano-oscillator-based reservoir computing system, method and medium, belonging to the field of neuromorphic computing and hardware acceleration, which directly generates nonlinear signals from linear signals by using the inherent nonlinear current-voltage characteristics of spin-torque nano-oscillators, realizes the nonlinear part generation of feature vectors in the next generation of reservoir computing, and effectively solves the three major problems of feature dimension combination explosion, high computational overhead caused by a large number of matrix operations, and insufficient multi-scale dynamics capture capability by replacing traditional polynomial algorithms with physical laws. At the hardware level, adaptive frequency domain response is realized without complex reconfiguration, combined with low computational overhead of analog domain processing, which can improve the timing processing efficiency without significantly increasing the hardware cost.