12 results about "Minimum cost path" patented technology

The invention provides an evaporation process regulation and control instruction generation method and system, computer equipment and a medium, and belongs to the field of industrial process process index prediction.The method comprises the steps that an aluminum oxide evaporation process historical time sequence working condition sample is collected, and a nonlinear time sequence correlation feature vector is extracted through an encoder; the samples are divided into K subsets by combining a clustering algorithm and expert rules, the subsets are labeled, and the number is counted; generating an adversarial network expansion sample by using structural constraint for the category of which the sample size is less than a threshold value; training an integrated model by using each subset, outputting an initial prediction vector, splicing the initial prediction vector with the feature vector, and obtaining a prediction value through a GBDT element learner; and establishing a working condition switching cost function by using the predicted value and the label, optimizing to obtain a minimum cost path, and generating a working condition regulation and control instruction of the automatic control system. According to the method, joint optimization of the prediction result and the working condition switching path is realized, and the problems of large concentration measurement error, large working condition jump disturbance and unsafe processing process in the aluminum oxide process are solved.

The application discloses an edge computing network service function chain embedding method based on K shortest path, and aims at a distributed service function chain embedding problem under time delay constraint. In the case of meeting the end-to-end time delay constraint, a service function embedding algorithm based on K shortest path algorithm is designed to minimize the deployment cost. The application firstly proposes a service function embedding problem modeling method based on a weighted layered graph, and converts the service function chain embedding problem into a minimum cost path problem under time delay constraint in the weighted layered graph. Secondly, a service function chain embedding mechanism under time delay constraint is designed based on the weighted layered graph and K-linear measurement aggregation algorithm. Compared with the prior art, the application has better high performance and operability, and has practical significance for complex network applications in edge computing networks.

The invention relates to a cross-layer optical network routing optimization method assisted by a large language model. The method comprises the following steps: constructing a framework of a cross-layer optical network, and collecting network physical node composition, link resource parameters and service request characteristics; designing an auxiliary graph model, and converting a cross-layer network routing optimization problem into a minimum cost path calculation problem in a virtual topology; building an auxiliary graph model weight optimization framework endowed by the large language model, and preliminarily generating an edge weight of the auxiliary graph model by the large language model; calculating a minimum cost path based on the edge weight of the initial auxiliary graph, evaluating a network performance index corresponding to the path, inputting an evaluation result as feedback data into the large language model, and iteratively adjusting the edge weight until the performance index is converged to be optimal; and completing routing planning and wavelength allocation of all service requests in the auxiliary graph model according to the optimal edge weight. According to the method, a large language model and an auxiliary graph model framework are combined, so that the flexibility and the multi-target adaptation capability of routing optimization are improved, and global optimization is realized.

The invention relates to the technical field of computer networks, and discloses a hybrid scheduling system fusing flow completion time and delay constraint in a heterogeneous network, which comprises a centralized controller and a plurality of switches, and is characterized in that the centralized controller comprises a self-adaptive throughput sampling module, a hybrid cost target construction module and a minimum cost path distribution module; the invention also discloses a hybrid scheduling method for fusing the flow completion time and the delay constraint in the heterogeneous network, which is used for inputting and sampling the flow to be scheduled and calculating and distributing the hybrid cost. According to the method, the adaptive sampling method is adopted, the prediction of the FCT is much more accurate than that of a traditional method through intelligent identification and skipping of a TCP slow start stage, and a reliable data basis is provided for subsequent intelligent decision making. Two heterogeneous indexes of "throughput / FCT" and "delay / DDL", which seem to conflict, are unified to a computable cost dimension, a traditional heuristic rule is got rid of, and a quantitative optimization decision is made.

The invention discloses an edge computing network service function chain embedding method based on a K shortest circuit, and designs a service function embedding algorithm based on a K shortest circuit algorithm aiming at a time delay constrained distributed service function chain embedding problem under the condition of meeting end-to-end time delay constraints and aiming at minimizing deployment overhead. The method comprises the following steps: firstly, providing a service function embedding problem modeling method based on a weighted hierarchical graph, and converting a service function chain embedding problem into a delay constrained minimum overhead path problem in the weighted hierarchical graph; secondly, designing a time delay constrained service function chain embedding mechanism based on the weighted hierarchical graph and a K-linear metric aggregation algorithm; compared with the prior art, the method has better high performance and operability, and has practical significance for complex network application in the edge computing network.

The application discloses a method, device and system for alignment of heterogeneous modal sequences and a storage medium. The method comprises: extracting frame-level acoustic vectors and fundamental frequencies of an acoustic feature stream to generate a first modal sequence; performing word-level probability coding on a language modal to obtain a word boundary probability vector; extracting a lip movement trajectory from a visual key point matrix to construct a second modal sequence. Further, a mutual information function of the acoustic and language modals is calculated based on a joint probability density, a distance function of the acoustic and visual modals is calculated based on a weighted distance, and a modal weight is dynamically determined based on the distance function to generate a frame-level fusion feature vector. A multi-objective cost matrix containing an inter-frame difference, mutual information loss and key point matching error is constructed by the fusion feature, path searching is performed based on a maximum slope constraint to obtain a minimum cost path, a unified time alignment sequence is output, and accurate timing synchronization of multi-modal information is achieved.

The application provides a multi-source heterogeneous data fusion digital talent portrait analysis system, relates to the field of artificial intelligence and big data processing technology, models talent ability and post demand as probability quality distribution, constructs cross-dimension substitution cost compensation impedance matrix based on skill ontology graph, introduces optimal transmission theory to solve the minimum cost path of surplus ability to missing dimension, realizes the post adaptation evaluation under the flexible substitution effect, solves the rigid adaptation problem that the traditional vector similarity algorithm cannot compensate across dimensions due to the dimension orthogonal assumption. At the same time, through the neighborhood co-occurrence frequency statistics and normalized point mutual information purification synergy affinity, the topological distance is adaptively discounted by bottleneck constraint and Gaussian decay gate screening optimal bridge node pair, eliminating the systematic distance overestimation bias caused by the shortest path algorithm ignoring the third party bridge skill, so that the compound talent obtains the actual adaptation score.

A graph-based method comprises generating a configuration file according to partition information, and performing a graph-based algorithm according to the configuration file to find a minimum cost path starting from a starting node, passing an intermediate node and terminating at an ending node. The configuration file comprises vertexes and adjacency information.

The invention relates to the technical field of warehousing and sorting, in particular to a plate warehousing and sorting algorithm which comprises the steps that an information two-dimensional code is generated for each plate, the two-dimensional code information comprises a unique number and plate length and width information, the plates are divided into virtual groups according to requirements, and a plurality of virtual groups of plates in the requirements are virtually combined into a package; the information two-dimensional code is scanned when a single plate is put on a shelf, plate information is recognized, an edge selection rule is dynamically matched according to the plate information, and the edge selection rule comprises an efficiency balance edge selection side and a storage capacity balance edge selection side; the optimal placement position is calculated according to the matched edge selection rule, and then a left side suction cup or a right side suction cup is called for feeding; the method has the advantages that by solving the minimum cost path, the suction cup is guided to select the relatively effective access path, the invalid moving distance of the mechanical arm is effectively reduced, equipment energy consumption and mechanical abrasion are reduced, and the economical efficiency and stability of system operation are improved.

The invention discloses a cross-cloud link bandwidth allocation method and system based on auction in a cloud environment, and relates to the technical field of cloud computing and network resource management. According to the method, when a user arrives at random, a bandwidth demand request and a plurality of selectable deadline-quotation pairs are received, and an optimal transmission path is determined by adopting an improved minimum cost path algorithm on the premise of meeting link capacity constraints; generating a plurality of feasible scheduling schemes based on primal-dual optimization, calculating user utility and selecting a maximum value as a bid winning scheme; the bandwidth unit price is updated through a dynamic pricing function increasing along with the link utilization rate index, and the underestimated value request is prevented from occupying resources in advance. Compared with an existing static or off-line scheduling method, online decision making can be achieved without future information, the resource utilization rate and the social welfare are improved, the trueness, the individual rationality and the calculation efficiency are guaranteed, and the method is suitable for a large-scale distributed data center and wide area network interconnection scene.

The invention discloses a heterogeneous modal sequence-oriented alignment method, device and system and a storage medium. The method comprises the following steps: extracting a frame-level acoustic vector and a fundamental frequency of an acoustic feature flow, and generating a first modal sequence; performing word-level probability coding on the language modality to obtain a word boundary probability vector; and extracting a lip movement track from the visual key point matrix, and constructing a second modal sequence. And further, calculating a mutual information function of the acoustic and language modals based on the joint probability density, calculating a distance function of the acoustic and visual modals based on the weighted distance, dynamically determining a modal weight according to the distance function, and generating a frame-level fusion feature vector. A multi-target cost matrix containing inter-frame differences, mutual information loss and key point matching errors is constructed through fusion features, path search is executed based on maximum slope constraint, a minimum cost path is obtained, a unified time alignment sequence is output, and accurate time sequence synchronization of multi-modal information is achieved.

The invention discloses a method and equipment for generating a main / standby multicast tree, and a medium, and relates to the technical field of communication. The method comprises the following steps: in an initialization stage, defining an original topological graph, a source node, a destination node set, a maximum delay constraint and an iteration threshold value, creating double trees only containing the source node, and binding independent residual graphs respectively to ensure initial resource isolation; a current growth tree is dynamically selected through a random alternate mechanism, a minimum cost path is calculated by using a super node compression technology in combination with a Dijkstra algorithm, a newly added path edge is removed in real time, and a dynamic isolation environment is constructed from another tree remainder graph; in the preferential output stage, the coverage integrity and the total cost are compared through multi-round iteration, the optimal dual-tree combination is reserved, multicast tree optimization which is free of common edges, low in delay and minimum in total cost is achieved, and the network reliability and the calculation efficiency are greatly improved.