51 results about "Parametric transformation" patented technology

The invention relates to the technical field of industrial robot control, and particularly discloses an intelligent foreign matter sorting robot control system based on AI recognition, which comprises a dynamic spatial feature extraction module, a manipulator motion state coding module, a collaborative conflict detection module, a dynamic trajectory optimization module and an execution control adjustment module, constructing a three-dimensional dynamic space model through multi-sensor fusion, and extracting spatial topological features by utilizing continuous coherence analysis; manipulator motion parameters are converted into topological space representation, and a track feature coding matrix is established; detecting interaction conflicts among the manipulators in real time by adopting a multi-scale coherence analysis method, and generating graded early warning signals; a collision avoidance track is optimized based on topological constraints and a virtual rejection field technology; precise execution is achieved through inverse kinematics of the Lie group theory and self-adaptive control.

The invention discloses a quantum circuit compiling method and system, a terminal and a computer readable storage medium. The quantum circuit compiling method comprises the steps of obtaining target unitary transformation and hardware constraints; converting a unitary transformation space parameter corresponding to the target unitary transformation into a Riemannian manifold; calculating a geodesic path using a Lagrange action amount based on the Riemannian manifold and the hardware constraint; synthesizing a gate sequence of the geodesic line path; a compiling circuit with authenticated minimum depth and maximum fidelity is output based on the synthesized gate sequence. The compilation circuit obtained by the invention has circuit optimality mathematical guarantee.

The embodiment of the invention provides a DCS man-machine picture conversion method and system, electronic equipment and a medium, and the method comprises the steps: firstly, responding to a conversion request, locking a target DCS system and a required target picture conversion parameter, and then carrying out the locking of the target picture conversion parameter based on a preset general mapping relation and a unique preset extension mapping relation of the target system, and converting the parameters into an intermediate execution file compatible with the target system. Wherein the problem of cross-system common compatibility is solved through general mapping, and it is ensured that icon layout parameters do not fail due to system differences; extended mapping gives consideration to uniqueness of a target system, so that personalized functions of icons are accurately converted. The generated intermediate file can be directly driven to generate a corresponding picture after being imported into a target system, and a design process does not need to be repeated. Therefore, only one conversion request is needed, and the target system and the expansion mapping relation are adjusted, so that the same set of picture requirements can adapt to different systems, the phenomenon that pictures with the same function are redrawn for each system is avoided, and mutual conversion of man-machine pictures among different DCS systems is effectively realized.

The invention discloses an industrial autonomous mobile robot control method, device and equipment and a medium, is applied to a robot control system based on a hierarchical architecture, and relates to the field of automation, and the method comprises the steps: analyzing an input operation text instruction, and generating a navigation task and a grabbing task of a robot based on an obtained structured subtask sequence; calling preset laser equipment according to the navigation task to generate a three-dimensional semantic map of a target end point corresponding to the operation text instruction, and performing path planning and obstacle avoidance through the three-dimensional semantic map to obtain a target path; calling a preset image acquisition device to acquire a material image of a preset material area according to the grabbing task, and processing the material image to generate a material grabbing parameter; and converting the target path and the material grabbing parameter into a robot scheduling instruction corresponding to the robot, and controlling the robot to execute a navigation task and a grabbing task through the robot scheduling instruction. Therefore, autonomous operation of the industrial robot can be realized in combination with semantic understanding.

The application relates to an AI large model access method and system, a storage medium and electronic equipment, and relates to the technical field of artificial intelligence. The method comprises the following steps: if an AI large model selection instruction sent by a terminal is not received, determining the intention complexity corresponding to a target prompt, and selecting a target AI large model adapted to the intention complexity from all to-be-called AI large models; adding a target Token label in the target prompt to obtain a labeled target prompt; based on a preset standard metadata structure, converting the labeled target prompt into a standard parameter, converting the standard parameter into specific manufacturer parameters corresponding to the target AI large model; forwarding the specific manufacturer parameters to the target AI large model, receiving an output result of the target AI large model, processing the output result to obtain a processed result, and sending the processed result to the terminal. The application has the effect of reducing the AI large model access difficulty.

The invention discloses a model deployment method, system and device, electronic equipment and a medium, and belongs to the technical field of electronics. The method comprises the following steps: determining a target hardware parameter of first hardware equipment based on a performance operation result fed back by the first hardware equipment for N times; wherein the performance operation result fed back by the first hardware equipment for the ith time is obtained by performing a performance operation test after the first hardware equipment deploys a model execution file sent by the electronic equipment for the ith time, N is a positive integer, and i belongs to [1,..., N]; determining target model configuration parameters matched with the target hardware parameters; and sending the neural network model obtained by converting the target model configuration parameters to the first hardware equipment.

The invention relates to the technical field of network security, in particular to a WEB application mimicry judgment system and method.The WEB application mimicry judgment method based on runtime data is characterized in that a parameter collection module is used for implanting instrumentation codes in key functions of WEB application programs and collecting runtime parameters of the WEB application programs in real time; the event conversion module is used for converting the acquired runtime parameters of the plurality of heterogeneous execution bodies into structured runtime events; the sequence alignment module is used for carrying out time sequence alignment on runtime events from different heterogeneous executors according to the business operation labels; and the risk judgment module is used for vectorizing the aligned text data of the runtime events, calculating a plurality of judgment factors and performing weighted fusion to generate a final confidence score, and performing feedback control to judge whether to trigger an execution body rotation operation or not according to the confidence score. According to the method, the WEB application runtime parameters are collected, through event conversion, sequence alignment and multi-factor weighted decision, the WEB application abnormal attack recognition accuracy is greatly improved, and the defect of traditional flow decision is effectively overcome.

The invention provides an assembly pose prediction method and device, electronic equipment, a medium and a slave program product, and relates to the technical field of industrial intelligent assembly. The method comprises the steps of determining a target entity from a plurality of to-be-selected entities based on the matching degrees corresponding to the plurality of to-be-selected entities by determining the matching degrees between a specific entity and the plurality of to-be-selected entities, and determining the target entity from the plurality of to-be-selected entities based on first position information of the specific entity and second position information of the target entity. Determining an optimized pose parameter of the target entity, converting the optimized pose parameter into an affine transformation matrix, and controlling the target entity to perform pose adjustment according to the affine transformation matrix so as to realize corresponding assembly with the specific entity; according to the method, automatic identification and high-precision pose adjustment of the target entity in the entity assembly process are realized through a collaborative process of intelligent matching, optimization solution and precise transformation execution, and the efficiency and accuracy of assembly modeling are improved.

The invention discloses a robot dexterous operation synthetic data generation method and system based on reinforcement learning, and the method comprises the steps: carrying out the preprocessing of motion capture data containing hand joint angles and object poses, and converting the posture parameters of a hand parameterized model into a unified control space; generating an action instruction through an accumulated residual action control strategy; performing strategy training on the action instruction by utilizing a reinforcement learning strategy, wherein an observation state space comprises a hand joint state, relative position information and an object pose; in the strategy execution process, a hand object contact event is detected, and contact force information containing multi-coordinate system representation is generated; operation track data is generated based on the original motion capture data, synthetic operation data containing complete contact force information is generated through a hybrid strategy, and data quality screening is executed; the invention provides a technical scheme capable of efficiently generating robot flexible operation synthetic data containing contact force information with high quality.

The methods and systems for model parameter compression includes selecting a total number of bits in which to store a plurality of parameters wherein the number of bits is divided into a signal bit, an exponent set of bits and a fractional set of bits. Additionally, a maximum parameter and a minimum parameter of the plurality of parameters is received. The method maximizes the number of bits in the fractional set of bits subject to two constraints: 1) the fractional set of bits and the exponent set of bits is less than or equal to one less than the total number of bits, and 2) a relational model of the number bits in the exponent set of bits to a range of parameters. Once the fractional set of bits and the exponent set of bits are thus determined, the system may convert the plurality of parameters into a plurality of converted values utilizing a mapping function.

This application provides a method, apparatus, device, and storage medium for driving the action of a target object, relating to the field of artificial intelligence technology. The method includes: acquiring source speech and a target video, wherein the target video includes a target object; performing facial parameter conversion processing on the speech parameters of the source speech at each moment to obtain source parameters of the source speech at the corresponding moment; extracting parameters from the target video to obtain target parameters; reconstructing the target object in the target video based on a combination parameter obtained by combining the source parameters and the target parameters to obtain a reconstructed image; and generating a synthesized video from the reconstructed image, wherein the synthesized video contains the target object, and the action of the target object corresponds to the source speech. This application enables a smoother and more realistic synthesized video with speech-driven target object action, improving the visual effect of video synthesis.

The invention discloses a kettle-based multi-library-table data automatic filing method, device and equipment and a medium, and relates to the technical field of computers, and the method comprises the following steps: carrying out multi-dimensional verification on configured filing strategy parameters; if the multi-dimensional verification is passed, acquiring a target driver and a target adaptive plug-in from the multi-data source adaptive library by using the archiving strategy parameter, and establishing a connection relationship based on the target driver and the target adaptive plug-in; performing analysis and logic legality verification on the archiving strategy parameter, if the verification is passed, extracting target to-be-archived data from the source database table based on a generated archiving extraction instruction, and performing data conversion on the target to-be-archived data; and determining a target archiving storage mode based on the target archiving library table information and the converted data, generating an archiving task in combination with the archiving strategy parameters, the converted data and the connection relationship, executing the archiving task, and generating a corresponding archiving report. The multi-library table data archiving cost is reduced, and the universality is improved.

This invention discloses an artificial intelligence-based method for recommending buried pipeline layout, belonging to the field of intelligent recommendation technology. The method includes: converting the laying environment, construction purpose, and specific parameters of the buried pipeline into input vectors and inputting them into a pre-trained buried pipeline layout recommendation model to obtain an initial recommended layout scheme; mining historical new problems and solutions for each historical pipeline project during the pipeline laying process, forming several problem-strategy pairs; using a hierarchical clustering algorithm to perform cluster analysis on all problem-strategy pairs under all historical pipeline projects; calculating the similarity between the current buried pipeline laying environment and construction purpose and each historical pipeline project based on an autoencoder similarity algorithm to obtain the required retained problem-strategy pairs; and improving the initial recommended layout scheme according to the layout improvement type. The pipeline layout scheme is optimized and improved to generate a layout scheme that meets actual needs while also considering potential risk prevention.

The application discloses a kind of patent retrieval statistical engine based on Elasticsearch, comprising: standard request parameter acquisition module, for obtaining standard request parameter;Elasticsearch query statement conversion module, for converting standard request parameter into Elasticsearch query statement;Patent retrieval module, for carrying out patent retrieval in Elasticsearch using Elasticsearch query statement;Standard response result obtains module, for converting Elasticsearch retrieval result into standard response result output;The application defines an interface specification of patent retrieval statistics, implements patent retrieval statistical engine based on Elasticsearch, can handle a variety of patent retrieval statistical needs including complex condition retrieval, multidimensional data statistical analysis, patent deduplication document merging and asynchronous patent retrieval.

This application provides a seven-parameter transformation geometric quality inspection method and system for building components, belonging to the field of quality inspection technology. The method includes: constructing a point cloud dataset; performing semantic partitioning of the CAD model using the parsing results; establishing partition identification features of the semantic partitions using the CAD model, and performing global search matching of the point cloud dataset using the partition identification features as matching features, configuring the partition mapping using the global search matching results; performing partition feature matching of the CAD model based on the partition mapping using the point cloud feature set, and establishing initial values ​​for the seven parameters; performing recursive iterative optimization to complete fine registration; calculating local residuals based on the fine registration results, and generating quality inspection results using the local residual calculation results. This application can solve the technical problem of insufficient registration accuracy between point cloud data and CAD models in key functional areas in the prior art, achieving the technical effect of improving the registration accuracy of building component quality inspection.

The application discloses a parameter conversion method and device, equipment, computer storage medium and program product, and relates to the technical field of big data. The method comprises the following steps: receiving a first target parameter to be converted and a configuration modification instruction corresponding to a target conversion code, a target service type of the first target parameter and the target conversion code having a mapping relationship, and the configuration modification instruction being used for indicating that a parameter conversion mode defined by the target conversion code is modified from an initial conversion mode to a target conversion mode; in response to the configuration modification instruction, adjusting a conversion type and / or a conversion parameter set by a first conversion configuration rule in N conversion configuration rules to obtain a second conversion configuration rule; and converting the first target parameter according to the second conversion configuration rule and the remaining conversion configuration rules in the N conversion configuration rules which have not been subjected to configuration modification to obtain a second target parameter. According to the embodiment of the application, the configuration modification of the parameter conversion mode and the parameter conversion can be efficiently realized.

The invention provides an entropy production constrained digital twinning optimization system, method and device, a storage medium and a program method. The entropy yield is directly used as a control target, and errors caused by intermediate parameter conversion are eliminated. Different regulation and control strategy effects are predicted through digital twinborn simulation, and dynamic optimization is achieved. And the response time is shortened to a second level through edge calculation and real-time feedback. And real-time sensing and accurate suppression of the loss of the thermodynamic system are realized.

This application relates to the field of computer technology, and more particularly to a method, apparatus, electronic device, and storage medium for cross-environment object invocation, which reduces the complexity and improves the flexibility of cross-environment object invocation. The method includes: creating a mirror object of a target ontology object in a second runtime environment within a first runtime environment; capturing input parameters from the caller in the first runtime environment for the mirror object, converting the input parameters into a basic data type, and sending them to the second runtime environment; and in the second runtime environment: invoking the target ontology object, executing object execution instructions generated based on the input parameters, converting the obtained ontology execution result into the basic data type, and sending it to the first runtime environment; and based on the ontology execution result, invoking the mirror object to simulate the target ontology object executing the object execution instructions, thereby obtaining a mirror execution result.

This invention discloses an adaptive granularity management system for a large language model KV cache based on access co-occurrence awareness, relating to the field of large language model technology. It includes a pattern feature extraction module, a head group clustering construction module, a flexible boundary generation module, a semantic weight calculation module, a hierarchical compression storage module, and a data merging and scheduling module. The pattern feature extraction module performs sparse structure analysis on the attention weight matrix of the large language model to extract attention pattern fingerprint data. The head group clustering construction module receives the attention pattern fingerprint data, aggregates multiple attention heads with similar pattern features into head tuples, and generates a parameterized co-occurrence template. The flexible boundary generation module obtains the current inference sequence length and, combined with the parameterized co-occurrence template, converts the relative position parameters within the template into a dynamic mapping table through proportional scaling. The semantic weight calculation module calculates the range of co-occurrence groups based on the dynamic mapping table.

A sensory control method includes a step of receiving, a step of converting, and a step of outputting. In the step of receiving, a sensitivity parameter is received. In the step of converting, the received sensitivity parameter is converted into, among a plurality of physical parameters included in physical properties relating to a sensory presentation, a physical parameter correlated with the sensitivity parameter. In the step of outputting, a sensory presentation signal based on the physical parameter obtained as a result of the conversion is output.

The invention discloses a robot simulation engine based on a graph neural network and a training method thereof, the robot simulation engine comprises a GNN core calculation engine, the GNN core calculation engine is connected with an API interface layer through a state management interface layer, and the API interface layer is an external interface of the GNN core calculation engine; the state updating layer is connected with the GNN core computing engine and used for executing processing of the current state and the prediction increment to obtain state data; the output layer is used for receiving and outputting the state data; the training method comprises the steps of obtaining a training data set; converting the system state parameters of each time step in the training data set into a graph structure data set; the GNN core calculation engine receives the graph structure data set and predicts a next state to obtain a predicted state; calculating the difference between the prediction state and the real state to obtain loss data; using a gradient descent algorithm to reversely propagate loss data to pass through the whole differentiable GNN core calculation engine, and updating all internal weights of the GNN core calculation engine;

The invention belongs to the technical field of circuit design, and discloses a method for impedance matching design based on a large language model, which comprises the following steps of: 1, collecting impedance matching design demand information for preprocessing, and outputting demand data; 2, semantic understanding, field extraction and structured mapping are carried out, a rule verification mechanism is adopted to carry out conversion unit unification, parameter range verification and missing information complementation, and standard structured design task description is generated; 3, judging a matching task type, selecting a corresponding design according to the matching task type, and generating a topological structure and parameters; 4, converting into a netlist file, performing S parameter simulation, analyzing key performance and indexes, judging initial structure performance, and generating a report and storing the report into an RAG knowledge base when a design target is met; 5, if not, parameter optimization is carried out, a report is generated and stored in an RAG knowledge base after optimization succeeds, modification suggestions are generated if the optimization succeeds for many times, and then redesign is carried out. According to the method, the matching structure design efficiency and accuracy of the large language model are improved.

The invention discloses a three-dimensional assembly method, device and equipment based on chemical static equipment and a medium, relates to the technical field of assembly, and is applied to a three-dimensional assembly system.The method comprises the steps that a target positioning mode type, target parts and special positioning parameter values are determined based on an upstream design instruction through a semantic conversion engine, determining a positioning template and a parameter conversion algorithm corresponding to the target positioning mode type from a target configurable positioning mode library; the positioning template comprises a special positioning parameter type; inputting a special positioning parameter value corresponding to the special positioning parameter type into a parameter conversion algorithm through a semantic conversion engine to obtain a software general positioning parameter type and a software general positioning parameter value; packaging the software general positioning parameter type and the software general positioning parameter value into a software executable command through a software adapter, so that the target three-dimensional software assembles a target part in a target positioning mode type based on the software executable command; the assembling efficiency, accuracy and flexibility can be improved.

The invention discloses a flight visual simulation method and system based on a multi-modal large model, and relates to the technical field of artificial intelligence and flight visual simulation. The method comprises the steps of receiving a natural language instruction of a user; acquiring a current state context of flight simulation; understanding and reasoning the instruction and the context by using a multi-mode large model agent subjected to field expert fine adjustment, and outputting a structured scene control instruction; analyzing the instruction and scheduling or generating resources; and the driving engine generates and fuses a multi-modal simulation effect for presentation. The system comprises a user interaction module, a context sensing module, a multi-mode large model agent module, an instruction analysis and distribution module, a simulation resource library, a resource generator, a dynamic scene generation and fusion engine module, a visual rendering engine and the like. The problems that in a traditional method, scene construction is complex, scene diversity is insufficient, and a man-machine interaction mode is single are solved, parametric transformation from parameter driving to semantic driving is achieved, and the intelligent level, interaction efficiency and training effect of flight scene simulation are improved.

The application provides a lightweight target detection method and system based on structure reparameterization, comprising: obtaining a data sample to be detected and preprocessing; inputting the preprocessed data sample into a network inference model to output a target detection result, wherein the network inference model is a single-path network model converted from a multi-branch network training module by structure reparameterization, and the structure reparameterization mode is branch merging and convolution sequence merging. The training and inference stages are decoupled by the structure reparameterization mode, more semantic information and feature representation are obtained by using the multi-branch structure in the training stage, the weight parameters are iteratively updated, and the single-path structure with the training weight information is used in the inference stage, so that the inference speed of the model is accelerated and the model accuracy is ensured.

The invention discloses a rapid optimization method for an output hose laying linetype under the action of multiple sea conditions considering storm correlation, and relates to the field of ocean oil and gas resource development, the method comprises the steps: generating a storm parameter sample with correlation based on Nataf inverse transformation, decoupling a design variable through parameter conversion, obtaining a design variable sample in combination with Latin hypercube sampling, and carrying out the optimization of the output hose laying linetype under the action of multiple sea conditions; constructing a sample matrix; establishing a full-coupling time domain model, and obtaining a hose response extreme value data set through time domain simulation; a multivariate input agent model is constructed based on deep learning and Bayesian optimization; and finally, rewriting the objective function, and incorporating the proxy model into an NSGA-II algorithm to complete multi-objective optimization to obtain a Pareto frontier solution. On the premise of considering the wind wave correlation, efficient and accurate optimization of the hose line type under multiple sea conditions is achieved, the safety and reliability of the hose structure are improved, and the method is suitable for optimization design of the oil and gas output hose of the offshore floating type production platform.

The invention relates to the technical field of visualization, and particularly provides a multi-dimensional visual data processing method, system and device and a storage medium, and the method comprises the steps: constructing a data cube model with unified space-time attribute dimensions; identifying a user decision intention and parameters through interactive operation; converting the intention and the parameter into a data query request by utilizing the query template, and dynamically extracting a target multi-dimensional data subset from the cubic model; a visualization strategy is matched through a rule engine, and the data subset is mapped into a fusion rendering instruction; and executing the instruction by the rendering engine to generate a fusion visualization effect. According to the invention, the conversion from passive data display to active intelligent display is realized, and the efficiency and accuracy of low-altitude command decision making are remarkably improved.

A method may include: a classical computer program receiving a combinatorial optimization problem; initializing QAOA parameters for a quantum circuit; setting an initial depth and a maximum depth for the quantum circuit, a choice of basis, and a number of basis coefficients; determining that a current depth of the quantum circuit is less than or equal to the maximum depth; transforming the QAOA parameters to the basis coefficients; determining that a relative performance improvement for the basis coefficients is less than a threshold; transforming the basis coefficients to the QAOA parameters; simulating execution of a quantum circuit with the QAOA parameters, resulting in a bitstring; updating the basis coefficients based on the bitstring; determining that a current depth of the quantum circuit is greater than the maximum depth; executing the quantum circuit with the optimized QAOA; and outputting a final bitstring as a solution to the combinatorial optimization problem.

The invention relates to a computer-implemented method and computer-readable medium for optimizing a machine line setup, particularly for machine line setups in the food and / or beverage industry. The technology disclosed herein comprises providing an input mask for input parameters for the formal description of a machine line model. The input mask is supported by a generative AI model that, for example, prompts the user for parameters in a dialog box. The input parameters can include boundary conditions and specific requirements entered by a user. Furthermore, the method includes generating an initial set of specific optimization goals based on the input parameters. The generative AI model can then transform the input parameters into the specific optimization goals.These optimization goals can then be used to create a solution using a solver. The solver uses a formal knowledge base and specific optimization goals to create at least one possible layout concept for the machine line and to calculate the layout concepts based on the optimization goals and logical rules.

The invention provides a dynamic adaptation method and system for an aircraft training data packet and multiple analog machines, which is based on a multi-analog machine hardware compatibility mechanism of a hardware abstraction layer (HAL), shields a hardware protocol difference through a uniform interface, and realizes cross-device multiplexing of the data packet. A dynamic parameter mapping algorithm is combined with a least square method and real-time error feedback, a data packet-hardware mapping relation is automatically optimized, and hardware characteristic drift is adapted. And according to the adaptation degree evaluation function and the real-time calibration strategy, dynamic adjustment is triggered through quantitative indexes, and the simulation precision (the error is smaller than 2%) is ensured. According to the standardized parameter conversion method, data packets and hardware parameters are unified into a standardized format, and mapping logic is simplified.