202 results about "Iteration process" patented technology

The invention relates to the technical field of intelligent manufacturing, and discloses a multi-objective optimization method and system for an aviation airborne product, so as to improve the overall performance of the aviation airborne product. The method comprises the steps of obtaining an initial population based on a series of initial model files of a target aviation airborne product, and determining a constraint condition and a target function of the target aviation airborne product; the constraint condition comprises the range of each iteration parameter; in each iteration process of obtaining the global optimal Pareto frontier according to the constraint condition, performing non-dominated sorting and congestion degree calculation on the tth generation of population through an algorithm, selecting individuals with excellent performance from a sorting result, and applying gradient-guided Gaussian disturbance to generate a (t + 1) th generation of new solution; the tth generation of population comprises cross progenies of the (t-1) th generation of population; and when the target iteration termination condition is met, iteration is terminated, and the global optimal Pareto front is obtained.

The invention relates to the technical field of computers, and discloses an information source coding matrix data processing method based on a modified conjugate gradient algorithm. According to the method, sparsity constraint and a fixed-point quantization mechanism are fused in iterative optimization, and a sparse fixed-point coding matrix meeting hardware deployment requirements is directly generated through gradient shielding, correction conjugate parameter calculation and Armijo criterion line search and synchronous execution of hard threshold sparseness and fixed-point quantization after each update. According to the method, the sparsity constraint and the fixed-point quantization mechanism are embedded in the iteration process of the modified conjugate gradient algorithm, so that end-to-end alignment of coding matrix optimization and hardware deployment requirements is realized.

Some aspects utilize an iterative process to synthesis a logic network within an ASIC using lookup table (LUT) optimization techniques. The logic network may be represented by an And-Inverter graph (AIG), as described in more detail below. On each iteration, the current AIG representation of the logic network is mapped to a network of k-LUTs. A k-LUT is a lookup table that can represent any function of k variables. This network may then be improved using LUT optimization techniques. The improved network of k-LUTs is decomposed into a trial AIG representation, which may be further improved for example by applying Boolean-based optimization techniques. The current AIG representation may then be updated in accordance with the quality of the trial AIG representation produced by the current iteration. The final AIG representation from the iterations is synthesized to a netlist of standard cells.

The invention provides a sample generation method and device, a storage medium and a program product. The sample generation method comprises the following steps: inputting a first type of prompts into a sample generation model to obtain a sample set; taking the sample set as a current sample set in a first round of iteration process, and executing the following iteration process: performing clustering processing on the current sample set to obtain a plurality of class clusters; if the iteration ending condition is not met, determining a to-be-enhanced class cluster and a to-be-inhibited class cluster according to the number of samples contained in each class cluster; based on the to-be-enhanced class cluster and / or the to-be-inhibited class cluster, constructing a second class prompt comprising a negative constraint condition and / or a positive constraint condition; the negative constraint condition is used for indicating to reduce generation of samples belonging to the to-be-inhibited class cluster, and the positive constraint condition is used for indicating to increase generation of samples belonging to the to-be-enhanced class cluster; and inputting the second type of prompts into the sample generation model to obtain a new sample set, and taking the new sample set as a current sample set in the next round of iteration process.

The invention relates to a self-adaptive information negotiation method applicable to a continuous variable quantum key distribution system, which comprises the following steps of: estimating an optimal code rate that an LDPC (Low Density Parity Check) code should reach under a current quantum channel, converting an initial code rate into the optimal code rate by a receiver, and obtaining a low-reliability bit number l which needs to be disclosed in conversion calculation, the method comprises the following steps of: decoding an LLR (Local Library Ratio) by using Bob, sending the LLR to Alice through a classic channel in combination with an encryption technology, sorting absolute values of the LLR by Alice in a decoding iteration process, selecting l bit positions with low reliability, sending the bit positions to Bob, determining a bit value by Bob according to the bit positions with low reliability sent by Alice, and feeding back and sending the bit value to Alice, and correcting the LLR value by Alice according to bit information sent by Bob. According to the method disclosed by the invention, the code rate of the LDPC code can be flexibly adjusted along with the change of the state of the time-varying channel, and the negotiation efficiency is improved, so that the key rate of the system is improved. The method is excellent in performance in a time-varying channel environment, negotiation efficiency under a low signal-to-noise ratio condition can be remarkably improved, and an efficient and reliable information negotiation scheme is provided for practical application of a quantum secret communication system.

The invention provides a whole-process production scheduling method considering processing and assembling, which comprises the following steps: constructing a two-stage flexible flow shop scheduling model, scheduling processing tasks of all parts of a plurality of products on a parallel machine in the first stage, and scheduling assembling of the product parts and processing tasks of semi-finished products in the second stage; minimizing the maximum completion time and the total energy consumption of the machine is taken as a double-optimization target; a multi-target swarm intelligence algorithm is adopted to solve the model, population individuals represent a scheduling scheme through two-segment coding, the first segment of coding defines a process execution sequence, and the second segment of coding defines a machine selection result; in an iteration process, alternately executing a Thompson sampling strategy and a dedirectional sampling and generating strategy according to a preset probability, adaptively selecting a bottom layer optimization operator to realize global search, and constructing a guide solution set to realize local mining; and outputting a non-dominated solution set after iteration is ended, and obtaining a corresponding whole-process scheduling scheme after decoding.

The invention discloses a dynamic optimization method based on a particle swarm optimization algorithm, and relates to the technical field of intelligent optimization algorithms, and the method comprises the steps: initializing the particle swarm optimization algorithm; calculating a target vector value of each particle in the particle swarm, executing Pareto comparison, screening the first 15% of particles as elite particles, and updating the particle swarm to join in optimization; adopting Alpha-Stable distribution to generate random disturbance, and executing dynamic mutation operation on particle positions; maintaining and updating the external archive storing the non-dominated solution; s4, performing quantitative evaluation on the current iteration state of the algorithm based on a preset convergence judgment criterion, and if the convergence condition is not met, returning to S3 and S4 to perform loop execution until the convergence condition is met, and terminating the iteration process; according to the method, aiming at the performance defects of the particle swarm, the optimization performance of the algorithm is pertinently improved in different stages through a dynamic variation mode, and a feasible solution is provided for the dynamic variation of the particle swarm algorithm.

The invention discloses a flexible job shop energy-saving scheduling optimization method considering a light storage condition and a load characteristic, and relates to the technical field of industrial scheduling, and the method comprises the steps: building a flexible job shop energy-saving scheduling problem model based on mixed integer programming, and setting an optimization target and a constraint condition; an ant colony algorithm is improved, an ant colony is divided into dynamic multi-level search, and a pheromone matrix is optimized; the updating effect of pheromones in the ant colony algorithm in the iteration process is optimized; optimizing the optimal solution of each generation of the ant colony algorithm by adopting a graph neural network off-line learning neighborhood search method; and stopping iteration when a preset termination condition is met, and outputting an optimal scheduling scheme. According to the method, a more efficient and flexible energy-saving scheduling strategy is developed to effectively coordinate the productivity and the energy efficiency, energy optimization and cost reduction in the production process are achieved, actual technical support is provided for energy-saving scheduling of the flexible job shop, and a method system for the workshop scheduling problem is enriched.

The invention discloses an MBSE model version pedigree iteration method and system based on a graph theory and a knowledge graph. The method comprises the following steps: constructing an MBSE model of an aerospace craft and mapping the MBSE model into an attribute graph; a directed acyclic graph is adopted to record a version iteration process, management is carried out through a version number system comprising a main version number, a combined version number and a revised version number, and a differential storage strategy is adopted; and performing change detection and merging based on a baseline, accurately identifying changes through a model comparison algorithm including internal attributes, internal relationships, overall relationships and difference analysis, and processing conflicts in collaborative design. According to the method, atomic-scale version control is realized, the problems of missing change traceability, low verification efficiency and multi-baseline conflict are solved, and the efficiency and quality of aerospace craft full-life-cycle model management are improved.

The invention provides a source network load storage cooperative control optimization method based on an improved selective transmission algorithm, and belongs to the technical field of power system dispatching. An energy storage device equivalent circuit model considering the temperature effect is established, and the state of charge is estimated in real time through Kalman filtering; a power system is modeled into a directed graph network flow model to set power balance constraints, an improved genetic algorithm based on tent chaotic mapping is adopted to solve an optimization model, a network flow algorithm is embedded in an iteration process to check feasibility, and premature convergence is prevented by adaptively adjusting crossover probability and mutation probability. And finally, outputting a Pareto optimal solution set, and selecting a scheduling scheme with optimal comprehensive performance according to an ideal point method. The technical problems of high system operation cost and insufficient new energy consumption rate caused by the fact that multi-objective optimization solution is easy to fall into local optimum in the source network load storage cooperative control optimization process are solved.

The invention discloses a decoupling design method and system for reliability optimization of an angular contact ball bearing. The method comprises the following steps: determining initial design parameters of the angular contact ball bearing; constructing a reliability optimization design model which takes maximization of the rated dynamic load as a target and comprises probability reliability constraint processed by a penalty function; constructing a design variable and failure probability mapping agent model, specifically, identifying and quantifying uncertainty parameters, constructing a quasi-statics model and a limit state function, calculating the failure probability of a sample point through a probability reliability method, and training the agent model taking the design variable as input and the failure probability as output; and based on the proxy model, solving the reliability optimization design model by using an optimization algorithm, and outputting optimal bearing design parameters. According to the method, the reliability analysis process and the optimization iteration process are decoupled, so that the optimization efficiency is greatly improved while the design robustness is ensured.

The invention relates to the technical field of robot motion, in particular to a variable load working condition heavy-load mechanical arm kinematics parameter calibration method and system. The method comprises the steps that joint angle parameters and theoretical Standard D-H model parameters are determined; the theoretical position of the end effector in the ideal state is determined; the actual position is determined, and the position error of the end effector is determined; the deflection under the preset load condition is determined, and corrected D-H model parameters are obtained; determining a load effect based on the parameter structure characteristics; introducing the load effect as a correction item into a parameter updating formula of a preset LM algorithm, and designing an adaptive compensation coefficient to adjust the contribution degree of the load effect; and dynamically adjusting the adaptive compensation coefficient through a convergence index in an iteration process of the LM algorithm. According to the method, the calibration precision of the kinematics parameters of the mechanical arm can be improved under the variable load working condition.

This invention discloses a noise reduction method, a bearing fault diagnosis method, and a system based on dual sparse dictionary adaptive learning. The method performs wavelet decomposition on the signal to be denoised to obtain high-frequency and low-frequency signals, constructing high-frequency and low-frequency matrices. Then, threshold-adaptive DDTF dictionary learning is applied to both the high-frequency and low-frequency matrices. Specifically, an initial dictionary is set and learned to obtain an initial sparse coefficient matrix. An adaptive threshold update is then applied to the initial sparse coefficient matrix, selecting sparse coefficients from the initial sparse coefficient matrix in descending order as dynamic thresholds during the iteration process to update the initial sparse coefficients. After determining the final threshold, the final sparse coefficient matrix is ​​obtained, and the dictionary is updated again to obtain sub-signals. Finally, inverse transformation and matrix rearrangement inverse operations are performed on the obtained sub-signals to obtain the denoised signal. This invention combines wavelet decomposition and DDTF to construct a dual sparse pattern, effectively improving the sparse representation capability of a fixed basis.

The application discloses a parameter correction method and device, electronic equipment and a storage medium. The method comprises the following steps: obtaining the fitness value of any generation in the iteration process of a preset optimization algorithm, and a pre-calibrated fitness correction coefficient, wherein the fitness correction coefficient comprises a positive correction coefficient and a negative correction coefficient; further, determining a target correction coefficient based on the fitness value in the positive correction coefficient and the negative correction coefficient, correcting the fitness value based on the target correction coefficient to determine a fitness correction value, updating the preset optimization algorithm based on each fitness correction value, improving the adaptability of the preset optimization algorithm, thereby improving the optimization accuracy of the preset optimization algorithm; further, performing parameter optimization on the carbon load model based on the optimization algorithm with high optimization accuracy, improving the accuracy of the model parameters, thereby making the error of the target carbon load value estimated by the carbon load model smaller, and improving the estimation accuracy of the carbon load value.

The invention discloses a safe and efficient distribution and incremental updating method and system for an AI large model, and belongs to the technical field of artificial intelligence, cloud computing and data security. The method comprises the following steps: firstly, performing quantification or pruning compression on an original large-scale machine learning model to reduce the volume of data to be distributed; and then encrypting the compressed model and storing the compressed model to distributed nodes in a fragmented manner. Through an intelligent scheduling server, on the basis of the hardware capability and network topology of edge nodes, a proper fragment downloading source is distributed to the edge nodes, and parallel high-speed downloading and local recombination loading are achieved. For model version updating, the central server calculates binary system difference between a new version and an old version and generates an increment updating package, only the increment package is distributed to an edge node, the node locally applies updating to reconstruct a new version model, and updating flow and delay are greatly reduced. According to the method, high bandwidth utilization rate, low transmission overhead, full-link security protection and intelligent resource scheduling of the hundred-GB-to-TB-level large model in the cross-region distribution and iteration process are realized, and the efficiency and security of model deployment and updating are remarkably improved.

An RNN operator computing method, device, equipment, computer program product and medium are provided. The method is for an RNN operator. First, sequence data containing a plurality of time step input matrices is acquired, and the input matrices are subjected to single matrix-matrix multiplication operation with a layer input weight matrix, thereby generating, at one time, intermediate result matrices corresponding to all time steps for gated calculation. Subsequently, based on a binary semaphore mechanism, the production completion state of the batch of intermediate result matrices is synchronized among a plurality of parallel computing units of a chip. Once the production is confirmed to be completed, each computing unit performs iterative update calculation of the hidden space state in time step order using the ready intermediate result matrices. In the iteration process, the production state of the latest hidden space state is also synchronized between adjacent time step computing units through the binary semaphore mechanism, to ensure the correctness of the time sequence dependency, and finally the hidden space state output corresponding to the entire input sequence is obtained.

The invention relates to the technical field of textile manufacturing, in particular to a scheduling method based on fusion of a large language model and an improved simulated annealing algorithm, and the method comprises the following steps: obtaining production data; performing problem description, setting constraint conditions, taking minimization of the maximum completion time as a target function, constructing a scheduling problem model, setting an iteration process, and repeatedly executing the iteration process; generating an initial scheme according to the production data and the problem description; in the iteration process, alternately calling the large language model and the improved simulated annealing algorithm for solving, and generating a to-be-verified scheduling scheme; the verification intelligent agent verifies the scheduling scheme to be verified, and after multiple iterations, an optimal solution is finally determined, and an optimal scheduling scheme is output. After the technical scheme is adopted, the maximum completion time of the generated scheduling scheme is relatively short, the generation speed of the scheduling scheme is relatively high, and the scheduling scheme can be ensured to be executable.

The application discloses a power distribution network prediction auxiliary state estimation method and system based on an improved iteration process, and the method comprises the following steps: establishing a nonlinear discrete time state space model; generating a group of Sigma points through an unscented transformation, and substituting the Sigma points into a state transition equation to calculate a state prediction value and a state prediction error covariance matrix at a current time; substituting the Sigma points into a measurement equation to obtain a measurement prediction value, and calculating a Kalman gain; calculating a process noise covariance matrix at a next time; and correcting the state prediction value and the state prediction error covariance matrix at the current time to obtain an optimal state estimation value and an error covariance matrix thereof. The application avoids the risk of losing semi-positive definiteness of the covariance matrix in the iteration process from the algorithm mechanism, significantly enhances the numerical stability and robustness of the UKF algorithm, and ensures reliable application in high-dimensional complex power distribution network state estimation.

The application discloses a quantum linear solving method and device based on quantum adiabatic linear algorithm, and a medium, and relates to the technical field of quantum computing. The method comprises the following steps: obtaining a linear system to be processed, and constructing a Krylov subspace matched with the linear system; determining a subspace equation set of the linear system based on the Krylov subspace according to the linear system and the Krylov subspace; constructing a quantum circuit corresponding to the quantum adiabatic linear algorithm according to the subspace equation set, and solving an approximate solution of the linear system in the Krylov subspace according to the quantum circuit. The application can combine the subspace method with the quantum adiabatic linear algorithm, perform dimension reduction processing on the original high-dimensional linear equation set in the quantum adiabatic linear algorithm through the subspace method, so that the dimension of the linear equation set actually solved in the iteration process is smaller than the dimension of the original linear equation set. Therefore, the demand of the quantum discrete adiabatic linear algorithm for computing resources is reduced, and the quantum discrete adiabatic linear algorithm can be implemented in an NISQ chip and a classical computer.

The embodiment of the invention discloses a test case generation method, an automatic test method and a computer program product. The method is used for generating a corresponding test case for an iteration function to be applied to an online environment in a function iteration process of a target application program, and comprises the following steps: obtaining an atomic instruction set, and providing a target page comprising an information submission area and an instruction display area; obtaining a test step submitted by a test user in a natural language description mode through the information submission area, calling an AI model, and generating a test instruction corresponding to the test step by the AI model according to the content described by the natural language and the atomic instruction set; obtaining a test instruction generated by the AI model, and displaying the test instruction through an instruction display area so as to verify the integrity and the execution sequence of the test instruction corresponding to a plurality of test steps split by the target iteration function; and performing instruction combination on the plurality of test instructions to obtain a corresponding test case so as to execute the test case to perform automatic test on the target iteration function.

The application discloses a military news event extraction method based on an integrated method and a confidence score, and comprises the following steps: step 1, initializing a model output confidence score distribution; step 2, evaluating the consistency of the model output; step 3, updating the model output confidence score based on the prior and the consistency; and step 4, selecting the optimal model output, in which the model output with the highest model output confidence score is selected as the final output in the iteration process or after the iteration is completed. The application uses a normal distribution to comprehensively represent and estimate the confidence of a single output, estimate the confidence of a single output, and capture the statistical dispersion between the actual confidence and the estimated confidence; important events can be accurately extracted and classified from a large amount of unstructured texts.

The application discloses a kind of based on multi-objective optimization algorithm's metasurface reverse design dataset generation method, this method includes: determining the regular topological structure of metasurface, structure parameter and resistance lumped element value are as optimization variable, and population is constructed;Set multiple electromagnetic performance indexes with competitive relationship as optimization goal;Using NSGA-II algorithm to iterate optimization to population, in each iteration process, based on optimization goal to the non-dominated sorting and crowdedness calculation of current population individual, filter and generate next generation population;Collect the structure parameter, resistance lumped element value and electromagnetic response data corresponding to all or part of population individual generated in iteration process, construct the metasurface reverse design dataset for training metasurface reverse design network.The application generates data for multiple regular topological structure metasurfaces by setting multiple competitive performance goals, using NSGA-II algorithm, to improve the training efficiency of reverse design network.

This invention relates to the field of data processing and discloses a method, apparatus, computer device, and readable storage medium for generating Gantt charts. The method includes: acquiring user stories of the iterative process of a target task; determining the iteration data corresponding to each user story; determining the cycle of each task and the task completion progress of each target task subject, thereby obtaining the individual work status of each target task subject; and finally, generating a Gantt chart based on the iteration tag, cycle, task completion progress, iteration plan start time, iteration plan end time, and each target task subject. Based on this, the present invention enables the Gantt chart to display the work status of different target task subjects under the same target task, and to reflect the overlap of R&D resources / development resources in the development process of the target task, thereby clearly displaying the development progress of the target task.

The embodiment of the invention provides a fully homomorphic encryption parameter optimization method, which comprises the following steps of: firstly, dividing a target network into a plurality of calculation units based on complexity parameters of each layer in the target network, and taking the calculation units as basic units for subsequent iterative optimization to balance a relationship between optimization complexity and unit granularity; and then, carrying out iterative optimization on the plurality of calculation units to determine fully homomorphic encryption parameters of the target network, in an iteration process, determining hierarchical consumption corresponding to each calculation unit according to the reward function and the input vector of the calculation unit, and according to the hierarchical consumption corresponding to each calculation unit, carrying out encryption on the fully homomorphic encryption parameters of the target network. Determining a reward function of the current iteration process and fully homomorphic encryption parameters of the target network; thus, through iteration of the reward function, the hierarchical consumption and the fully homomorphic encryption parameters in the multiple iteration process, balance between delay and errors can be achieved, and finally the fully homomorphic encryption parameters with the delay minimized on the premise that the errors are not remarkably increased are obtained.

The invention discloses a robot control method and device based on a diffusion model, a robot and a computer program product. According to the method, a shared query vector is introduced in a reverse denoising iteration process of a diffusion model, and feature information related to a current denoising stage is extracted from observation characterization and historical action characterization by utilizing the shared query vector, so that observation guide information and action guide information are formed; structured collaborative fusion of multi-modal observation information and historical action information in a strategy generation process is realized. Meanwhile, the contribution weight of the action guide information is adjusted through weight adjustment parameters, so that the generative strategy can be combined with historical action contexts to conduct track inference while responding to current environment observation, and time sequence correlation features are explicitly introduced in the action generation process. Therefore, the utilization efficiency of the multi-modal information in the diffusion strategy can be improved, and the action coherence and the overall stability of the robot in the long-time-sequence decision-making task can be enhanced.

The application discloses an intelligent production scheduling method of a foundry processing production line, and particularly relates to the technical field of production line intelligent production scheduling, and through time sequence collection of multiple source production elements and energy consumption-temperature-output correlation mapping, a scheduling feasible space capable of dynamically sensing thermal field changes is constructed; on the basis, a multi-objective weight self-adaptive drift mechanism is introduced, so that the scheduling algorithm can self-adjust the weight according to the actual load and energy consumption gradient, and continuous self-correction of the optimization direction of the beat and energy consumption is realized; further, in combination with optimization direction phase offset evaluation and scheduling shock risk index calculation, active identification and inhibition capacity of the chain instability phenomenon of 'energy consumption lag-weight drift-direction shock' is formed, so that the optimization process still maintains convergence and stability under dynamic conditions, and through embedding the risk index into the scheduling iteration process, a penalty correction is implemented on the optimization direction.

A nonlinear system control device based on dynamic sparse sampling strategy iteration comprises a controlled system modeling and characterization module used for modeling and characterizing a controlled system and setting a system state equation and a performance index function; the parameter setting module is used for setting the number of layers of the BP neural network, the maximum number of iterations, the number of sampling points of each track and a calculation precision threshold value; the initial admissible control law construction module is used for selecting an initial value function, initializing a BP neural network, constructing a proper initial control law and updating dynamic sparse sampling and an iterative value function; and the iteration process module is used for parameter setting and BP neural network initialization, dynamic sparse sampling, iteration value function updating, iteration control law updating, convergence judgment and termination condition judgment. The method aims at solving the problems that in the prior art, calculation burden is too heavy, dependence on state space prior information is achieved, iteration efficiency is low, and engineering feasibility is poor.

The invention discloses an unmanned surface vehicle path planning method based on an improved cuckoo algorithm. According to the method, a self-adaptive step length updating mechanism combining a pheromone factor and a heuristic factor is introduced in the iteration process of the cuckoo algorithm. The pheromone factor refers to a positive feedback mechanism of an ant colony algorithm to guide search to be carried out towards a historical high-quality solution direction; the heuristic factor dynamically adjusts the step length direction according to the distance between the current optimal solution and the candidate solution, and the local development capability is enhanced; and through a dynamic weight adjustment strategy, self-adaptively balancing global exploration and local search in an iteration process, and performing connectivity optimization on discrete path points output by an algorithm by using a Dijkstra algorithm to obtain a continuous and collision-free sailing path. According to the method, the convergence speed and the solving quality of path planning are remarkably improved, and the autonomous navigation capability of the unmanned surface vehicle in a complex environment is enhanced.

The invention provides an electric power system step backspacing simulation method and system considering switching events, and belongs to the technical field of electric power system electromechanical transient simulation, and the method comprises the steps: executing transient simulation step by step, carrying out the alternate iteration of a differential algebraic equation set, scanning the switching events of a DC system in the iteration process, scanning other switching events after the iteration convergence, and carrying out the step backspacing simulation. Whether step backspacing needs to be carried out or not is checked according to a backspacing criterion of the step backspacing strategy; if so, carrying out a rollback operation and adjusting the simulation step length according to a rollback simulation strategy; switching the scanned switching event, and keeping the state variable of the system unchanged after the switching is completed; according to the scheme, whether the next step of simulation continues to be executed or not is checked, if simulation continues to be executed, the step length of the next step needs to be adjusted according to the criterion of step length increase and step length decrease, and the scheme has good technical compatibility and also has excellent expansibility.