319 results about "Iterative search" patented technology

The invention discloses a game reinforcement learning-based reactive power dispatching method for an electric vehicle to participate in a power distribution network, and the method comprises the steps: collecting node voltage, line current, electric vehicle charging and discharging states and topological information, and constructing graph attention embedding; generating an initial reactive power regulation action in the multi-agent game reinforcement learning framework, and shaping and updating a strategy gradient through double rewards; the improved particle swarm is initialized according to the updating strategy gradient, and a self-adaptive inertia coefficient is set; iteratively searching according to the node voltage sensitivity, the line reactive margin and the expected convergence step number to obtain an optimized particle swarm; the particle speed is mapped to a main strategy network through online cooperative training, the inertia coefficient is synchronously adjusted, a cooperative optimization strategy is output, a reactive power dispatching instruction is generated, and the algorithm is updated in a closed-loop mode according to real-time feedback. According to the invention, rapid and cooperative reactive dynamic scheduling of an electric vehicle group is realized, and the voltage stability and the electric energy quality are remarkably improved.

The invention discloses a free-form surface three-axis ball-end cutter equal approximation error finish machining tool path generation method which comprises the following steps: firstly, importing a free-form surface model to be machined, and setting data such as the radius, the line spacing, the approximation error maximum allowable value and the precision of a ball-end cutter; secondly, a group of section planes are planned according to the row spacing to intersect with the curved surface, the intersecting line serves as a cutter contact curve, and an equal-bow-height-error cutter contact iterative search method for driving cutter contact adjustment through the geometric distance is provided for the cutter contact curve on the section planes; the maximum distance between a cutter cutting envelope surface and a cutter contact track line is used as an approximation error, and an adaptive discrete method is adopted to carry out approximation error calculation; and finally, the cutter location points of the equal-bow-height-error cutter contact points serve as initial values, an equal-error cutter location point calculation method of step length self-adaptive adjustment iteration is provided, approximation errors between the cutter location points are all within an allowable range, and therefore the free-form surface three-axis ball-end cutter equal approximation error finish machining cutter path is obtained.

The present disclosure describes methods, systems, apparatus, and media for object identification and classification, utilizing multi-feature and multi-modal data. This includes shape, material, brand, price, odor, taste, tactility, and sound. The system integrates a server space for data processing, a querying device for iterative searches, and a data interface module for refining results. It features AI-driven image optimization, feature extraction, and pattern recognition, employing novel techniques for fusing multi-feature and multi-modal embeddings utilizing multi-head attention. Additionally, a linker module powered by two active learning with feedback loops AI models consolidates scattered data into a unified object information database. The system also employs novel AI algorithms for isolating the object of interest through a saliency map and semantic analysis, as well as for enhancing raw images with a GAN-autoencoder.

The invention discloses a traffic scheduling method and an electronic device, and relates to the technical field of traffic scheduling, and the method comprises the steps: determining the priority weight of a micro-service, and predicting a target traffic according to the historical traffic information of a network device; constructing a graph model according to the topological information of the network equipment and the dependency relationship of the micro-service, and performing embedded learning on nodes in the graph model to generate a state vector representing a network state; the priority weight, the state vector and the target traffic of the micro-service serve as input of a reinforcement learning model, and a traffic scheduling strategy of the network equipment is obtained; performing iterative search according to iterative particles formed by encoding the strategy network parameters of the reinforcement learning model and the feature learning network parameters of the graph model to determine reinforcement learning model parameters; and issuing the traffic scheduling strategy to the network equipment and executing the traffic scheduling strategy so as to solve the technical problem that a traffic scheduling method in related technologies is difficult to adapt to a dynamic and complex network environment and service requirements under a micro-service architecture, and the reliability of traffic scheduling is improved.

The invention belongs to the technical field of parameter identification, and provides a thermal power generating unit electro-hydraulic servo system PID parameter identification method and system, and the method comprises the steps: obtaining the historical operation data of a steam turbine, carrying out the preliminary screening of the data, and constructing a parameter identification data set; according to the mean square error between the actual value and the calculated value of the opening degree of the valve, an objective function of PID parameter identification of the DEH electro-hydraulic servo system is established; setting population initial parameters, and generating a chaotic initial population; based on a parameter identification data set and a target function, introducing an initial population value of the generated chaotic initial population into an improved Bayesian optimization model, and obtaining a preliminary optimization result through Gaussian modeling and kernel function prediction; and taking the preliminary optimization result as input, and utilizing a whale optimization algorithm introducing a Levy flight disturbance mechanism to carry out iterative search for multiple times to obtain a final optimization result. According to the invention, the parameter identification precision and robustness are improved.

The invention discloses a wireless power transmission coil optimization method and a related system, which can realize automation of coil structure parameters and multi-target global optimization by acquiring a coil geometric parameter space, constructing a comprehensive optimization function and performing iterative search in the coil geometric parameter space. The problems that in traditional design, the number of simulation iterations is large, design efficiency is low, local optimum is prone to occurring, and multiple performance indexes are difficult to balance are effectively solved, coil design efficiency and precision are remarkably improved, and dependence on artificial experience is reduced. Therefore, the number of manual intervention and simulation is remarkably reduced, the coil design efficiency is improved, and the prototype development period of the WPT system is shortened. Besides, iterative search is carried out in the parameter space, so that the whole design space can be effectively explored, a locally optimal solution trap can be jumped out, and a globally optimal or approximately globally optimal coil structure parameter combination can be obtained more possibly.

An unmanned aerial vehicle assisted MEC network energy efficiency optimization method based on reinforcement learning belongs to the field of mobile communication, and comprises the following steps: firstly, carrying out system modeling and state information collection, inputting state information to a strategy model, generating flight and energy transmission actions, and carrying out unmanned aerial vehicle movement and service equipment selection; an energy transmission and task unloading model is defined, and a gradient-based equipment unloading time distribution algorithm is started; then initializing binary search of the equalization marginal utility gradient, executing single iterative search of the equalization marginal utility gradient, updating a search interval of the equalization marginal utility gradient, and completing search of the equalization marginal utility gradient; then determining a final time allocation scheme, executing energy transmission and task unloading, collecting performance data, calculating a reward function and storing experience; and finally, updating the Critic network, updating the Actor network and softly updating the target network, and obtaining an optimal strategy after iterative optimization. The overall energy efficiency and task processing throughput of the system are remarkably improved.

The invention discloses a model quantification method and system based on stability scoring, and the method comprises the steps: calculating the output features of each layer of an original model and the probability distribution of the output features of each layer of a quantification model, calculating the stability value of each layer of the quantification model according to the probability distribution, and evaluating the stability of each layer of the model through the stability value. Bit width is distributed according to characteristics of each layer, the method does not depend on labels, is only based on network forward activation information, and simultaneously measures output distribution difference and direction consistency before and after quantization, and a hierarchical stability measurement index can be constructed according to a stability value to accurately identify a steady-state layer and a high-sensitivity layer for guiding precision distribution. The bit width is adaptively reduced by adopting an iterative search strategy on the basis of ascending sorting of the stability value scores; on the premise that precision and resource constraints are met, the model compression ratio is increased to the maximum extent; retraining is not needed in the whole process, and large-scale network one-key quantitative deployment is supported.

The invention discloses a cable cabling control method based on process big data, and belongs to the technical field of cable cabling control. Comprising the following steps that S1, parameters are collected in real time and preprocessed, historical batch records are called, the tension instability and the arrangement offset trend rate are calculated, simulation coupling factors are constructed, and an optimized coefficient combination is obtained; s2, the current tension instability and the arrangement offset change rate are calculated respectively, and the adjusted take-up speed and the adjusted twisting torque are obtained; according to the method, historical batch records of cables of the same specification are called, the tension instability and the arrangement offset trend rate are calculated, a numerical optimization algorithm is adopted to iteratively search an amplification coefficient and a sensitivity coefficient, a process-oriented optimization objective function is minimized, an optimized coefficient combination is obtained, and the stability of the cable tension is improved. Therefore, self-adaptive parameter optimization based on historical data is realized, the high adaptability of a control strategy to different batches of process fluctuations is improved, and the variability of quality indexes is remarkably reduced.

The invention relates to the technical field of low-altitude safety, and provides an aircraft optimal trajectory screening method and system based on adjustable phase quantum search. Candidate trajectories are generated according to flight mission constraints, cost values are calculated, and Boolean markers are distributed based on cost threshold values to identify high-quality solutions; performing iterative search and measurement on the balanced quantum superposition state of the coding track index by utilizing a Grover quantum search algorithm based on the adjusted search phase to obtain an output index; and mapping the index into a track, obtaining an optimized track through classic local optimization, adaptively adjusting the phase parameter and the cost threshold value based on performance evaluation, and repeatedly executing until an end condition is met, thereby determining an optimal track. On-line reconfiguration of a strategy is realized by introducing an adjustable phase mechanism, a feedback closed loop is established to realize adaptive optimization of parameters, and the real-time requirements of scenes such as urban air traffic control and unmanned aerial vehicle clusters are met.

The invention discloses a dynamic hierarchical encryption system for multi-tenant e-commerce data, and relates to the technical field of multi-tenant data security, the system comprises an encryption management center, the encryption management center is in communication connection with the following modules: a data sensing and grading module, which is used for collecting multi-source context data of an e-commerce platform, and sending the multi-source context data to the encryption management center; performing basic sensitivity level division on the context data; and the multi-tenant knowledge graph construction module is used for constructing a multi-tenant knowledge graph based on the tenant-data-authority association network in combination with the multi-source context data. According to the method, dynamic self-adaptive adjustment of encryption strength is realized through cooperation of the fruit fly algorithm and the knowledge graph, the system dynamically optimizes combination of key length and the encryption algorithm based on data grading labels and real-time context data, and the fruit fly algorithm searches an optimal solution through iteration, so that the system has high encryption efficiency while ensuring compliance. And encryption delay is controlled within a real-time requirement range, so that the user experience in a multi-tenant scene is remarkably improved.

The invention discloses a pixelated matching circuit design and optimization method based on a large language model, and relates to the technical field of radio frequency circuits, and the method comprises the steps: obtaining a target demand and a physical constraint of a matching circuit; generating an initial topological structure based on the target demand and the physical constraint, and performing parameter estimation on the initial topological structure to obtain an initial binary matrix constrained by the initial topological structure and the parameter; obtaining an optimization problem based on the initial binary matrix and the target demand; wherein the optimization problem comprises distribution of optimization parameters, optimization objectives and solution spaces; and obtaining a key performance index and an optimization algorithm according to the optimization problem, and carrying out iterative search in the optimization algorithm according to the key performance index to obtain an optimal solution for solving the optimization problem so as to generate an optimal pixelated topology. According to the method, the optimal pixelated topology can be generated, and efficient global optimization is realized.

The invention provides a photovoltaic energy storage joint optimization scheduling method and system based on deep reinforcement learning, and relates to the technical field of energy management, and the method comprises the steps: collecting multi-source heterogeneous time series data, extracting energy storage health state features through a dual codec network and a recurrence plot convolution network, dividing an energy storage power interval, and constructing a feasible region; based on the power difference value sequence, extracting fluctuation characteristics in a segmented manner, and constructing a risk evaluation index; and generating a plurality of search paths in the feasible region, obtaining an energy storage charging and discharging power time sequence through iterative search, and generating a scheduling instruction. According to the method, light storage joint optimization scheduling under the energy storage safety constraint is realized, and the photovoltaic consumption rate and the system economy are improved.

The invention discloses an automatic wavelength tuning method and system for a tunable laser based on particle swarm optimization, and is applied to the technical field of wavelength tuning of semiconductor tunable lasers. According to the method, a particle swarm optimization (PSO) algorithm is taken as a core, multi-target iterative search optimization is carried out on initial current-wavelength data, and in a tuning initialization stage, a system firstly carries out coarse scanning on three paths of current of a left reflecting grating region, a right reflecting grating region and a phase modulation region of a sampling grating distributed Bragg reflection laser; manual intervention is not needed, the system can complete efficient configuration screening in a given target wavelength range, the intelligent level of the tuning system is remarkably improved, good automation capacity and expansibility are achieved, scanning of a large number of invalid data and tedious operation procedures in the traditional tuning process are avoided, and the tuning efficiency is improved. And meanwhile, a high-quality current-wavelength lookup table can be quickly constructed, and the method has wide engineering application prospects in optical communication, sensing and other scenes.

The invention provides an MIG fragment rearrangement method and system based on topology awareness and a double-layer agent, and belongs to the technical field of artificial intelligence. The method comprises the steps of collecting calculation of each partition and video memory occupation and remaining execution time, generating partition embedding through Kronecker multiplication, Hadamard product and PCIe neighborhood aggregation, constructing an address adjacency matrix, generating a global condition vector through Sigmoid gating and fusion pooling, and inputting a diffusion network to generate a fragment evolution trajectory and confidence. And calculating continuous splicable capacity based on confidence, constructing a graph structure, evaluating candidate migration actions by a lightweight graph strategy network, and selecting an optimal sequence through risk tensor and Bayesian calibration iterative search. And in the execution stage, the migration duration is controlled through an NVML lock table, RDMA straight pulling and a sliding window, and the device tree is updated after the migration duration is completed. Real evolution is continuously collected on line, deviation is monitored through divergence, the model is updated through distillation, and the optimal action and pause duration are recorded through an experience pool.

The present disclosure describes methods, systems, apparatus, and media for object identification and classification, utilizing multi-feature and multi-modal data. This includes shape, material, brand, price, odor, taste, tactility, and sound. The system integrates a server space for data processing, a querying device for iterative searches, and a data interface module for refining results. It features AI-driven image optimization, feature extraction, and pattern recognition, employing novel techniques for fusing multi-feature and multi-modal embeddings utilizing multi-head attention. Additionally, a linker module powered by two active learning with feedback loops AI models consolidates scattered data into a unified object information database. The system also employs novel AI algorithms for isolating the object of interest through a saliency map and semantic analysis, as well as for enhancing raw images with a GAN-autoencoder.

The embodiment of the invention provides a different-address machine production and distribution cooperative scheduling optimization method and system and a storage medium, and belongs to the technical field of scheduling optimization. Comprising the following steps: constructing an objective function of service time of a parallel machine and a constraint function of the objective function, and generating an initial solution meeting the constraint function through a sorting scoring function based on variable machine lease cost; intelligently selecting a disturbance operator based on a deep reinforcement learning algorithm, and disturbing the initial solution in a variable neighborhood search mode based on the disturbance operator to generate a disturbance solution; performing local optimization based on the disturbance solution through a local search mode to determine a current optimal solution and a current optimal solution of the target function; and performing iterative optimization on the current optimal solution in an iterative search mode to determine a global optimal solution of the target function. According to the method, production and distribution are collaboratively optimized while the cost constraint existing in actual production is considered. The convergence efficiency is high, and the probability of falling into local optimum can be reduced.

The invention provides an energy optimal excavator obstacle avoidance trajectory planning method based on quintic NURBS. The method comprises the steps that a kinetic model of an excavator working device is constructed; based on the dynamic model and the target obstacle data, generating collision-free path points by using an improved fast expansion random tree algorithm, and describing a tooth tip movement track of the excavator bucket by using a quintic non-uniform rational B-spline curve; further constructing an energy optimal obstacle avoidance trajectory optimization model of the excavator working device; and applying a particle swarm algorithm to model optimization, and searching an optimal solution through iteration to determine an optimal obstacle avoidance track of the unmanned excavator. According to the invention, the autonomous obstacle avoidance operation of the unmanned walking type excavator is efficiently and accurately carried out under the working condition of vehicle-shovel cooperative unloading, and the method has important significance for ensuring the safety of the excavator in the operation process and improving the operation quality.

The invention discloses an accelerator architecture design method and electronic equipment, relates to the technical field of computers, and performs dynamic fractal space coding on a design parameter matrix of an accelerator to obtain an analog coding vector. And performing meta-learning dynamic weight entropy search on the analog coding vector by using a historical task data set to determine a new design scheme. And when the difference between the sensitivity parameters in the new design scheme and the sensitivity parameters in the neighbor set meets a difference condition, performing multi-objective optimization simulation on the new design scheme. And when the difference does not meet the difference condition, multiplexing the simulation result of the neighbor set. And performing iterative search according to Bayesian optimization, and determining a final accelerator architecture design scheme in the selected design scheme and the simulation result thereof. Dimension reduction is performed on a high-dimensional discrete space, and a search direction is guided by means of an efficient meta-learning dynamic weight entropy acquisition function. And rapid exploration of an accelerator architecture design space under multi-objective optimization is realized.

The invention discloses an asphalt mixing process dynamic optimization method based on multi-data fusion and digital twinning, and relates to the field of road engineering, and the method comprises the steps: collecting multi-source sensor data in the production process of an asphalt mixing station, carrying out the fusion processing of the multi-source sensor data, and obtaining a state parameter feature set, the multi-source sensor data comprises temperature, humidity, flow, vibration signals, dust concentration and asphalt viscosity; inputting the state parameter feature set into the intelligent optimization model, and outputting an optimized process parameter set value through iterative search and prediction feedback; and the optimized process parameter set value is issued to an asphalt mixing station control system for execution, and a digital twin model is driven based on real-time production data to carry out visual monitoring and closed-loop control. By collecting multi-source data and combining an intelligent optimization algorithm, automatic adjustment and real-time quality prediction of process parameters are achieved, the intelligent level of the production process is improved, stable product quality is ensured, and dynamic working condition changes are coped with.

The invention discloses a natural gas hydrate reservoir layered balanced mining regulation and control method based on simulation optimization. The method comprises the steps that S1, a three-dimensional geologic model is constructed according to logging, earthquake and core data of a target area; s2, establishing a heat-flow-force-chemical multi-field coupling numerical model, and performing initialization; s3, defining a multi-dimensional engineering decision parameter space; s4, establishing a comprehensive objective function; s5, an optimal engineering decision parameter combination Xoptimal is obtained through iterative search; and S6, well drilling and well completion operation is carried out according to the Xoptimal, and dynamic verification and feedback adjustment are carried out on the model and the mining system based on real-time monitoring data in the production process. A heat-flow-force-chemical multi-field coupling numerical model is established based on a three-dimensional geologic model, an agent model and multi-objective optimization are introduced, and dynamic verification and feedback adjustment are performed in combination with monitoring data during production, so that interlayer contradictions are effectively inhibited, balanced utilization of reservoir energy is realized, and the production efficiency is improved. And the overall recovery efficiency and the development economy of the multilayer hydrate reservoir are improved.

The invention relates to a signal detection method, device and equipment, a medium and a product. The method comprises the following steps: acquiring a target receiving signal to be detected and a Monte Carlo tree corresponding to a transmitting signal; under the condition that the current node in the Monte Carlo tree is completely expanded, for each child node of the current node, shifting processing is carried out based on the number of access times of the child node, the exploration value of the child node is determined, and the optimal child node of the current node is determined based on the reward value and the exploration value of the child node. Adding the optimal child node into a current search path corresponding to the current iterative search process, taking the optimal child node as a new current node until the current search path is searched, and determining a reward value corresponding to each node in the current search path; and determining a target transmitting signal corresponding to the target receiving signal based on the reward value of each node in the Monte Carlo tree when a preset iteration end condition is satisfied. By adopting the method, the hardware implementation complexity can be reduced.

A heterogeneous neural processing system includes a first processor configured to perform encoding and decoding operations of an autoencoder, and a second processor configured to perform neural network operations of a particular task, the operations having iterative processing. The processors perform compute tasks through synchronous data exchange to implement a generative AI model. The first processor processes a feature map divided into data rows, caches the data into active memory using row-based depth-first scheduling, and selects deeper operations in the network hierarchy when processing branch inputs, outputs, and residual connections. And H reuses boundary pixels between the cache storage space segments, so that convolution and element-level operation can be executed simultaneously. A neural network adjusts the device analysis model to identify layer dependencies, applies search space constraints, performs an iterative search to generate a fusion plan, and selects an optimal plan based on external memory access and execution delays.

The invention belongs to the technical field of resource scheduling, and discloses an intelligent emergency resource scheduling system for quick response. Comprising the following steps: acquiring state information of a terminal and a network element, analyzing alarms and service changes, and generating an affected area; generating a topological graph model, and constructing an initial configuration vector for each topological node in the topological graph model; rasterizing the affected area, and determining serviceable nodes of the grid area; performing iterative search on the initial configuration vector, and determining an emergency configuration vector of each grid region; determining candidate interconnected node pairs by analyzing signal quality among different serviceable nodes in the affected area; and according to constraints of different edge elements in the topological graph model, constructing an optimal forwarding path in the affected area, and carrying out emergency resource scheduling. According to the invention, the affected range can be accurately determined, so that the configuration adjustment better fits the joint change condition of alarm and service, and the response efficiency and the resource matching degree of emergency resource scheduling are improved.

The invention relates to the technical field of microbial fermentation, in particular to a biological fermentation monitoring method and system based on multi-source data fusion. The method comprises the following steps: acquiring multi-source monitoring data in a fermentation tank and constructing a multi-dimensional state vector; on the basis of data of the multi-dimensional state vector in a preset time window, constructing a metabolic entropy of the current fermentation process; a simulated annealing algorithm is adopted to optimize control parameters, the temperature of the simulated annealing algorithm is positively correlated with the metabolic entropy, and minimization of a second energy function is taken as a target; a catastrophe reignition mechanism is set in the optimization process, fermentation abnormity is diagnosed based on triggering of the catastrophe reignition mechanism, and abnormal state characteristics are matched with a preset abnormity database through iterative search of an optimal weight distribution scheme so as to determine the abnormity type with the highest matching degree. By introducing the metabolic entropy and the catastrophe reignition mechanism, the sensitivity, the accuracy and the robustness of monitoring the biological fermentation process are remarkably improved.

The invention discloses a complex environment path planning method based on a PSO-SSA algorithm, and the method comprises the following steps: initializing, executing an iterative search process, dynamically adjusting the proportion of a PSO algorithm and an SSA algorithm in the PSO-SSA algorithm in the process, and carrying out the iterative optimization of the PSO algorithm at the same time; dynamically judging whether to switch an algorithm or not by utilizing an intelligent algorithm switching mechanism, and introducing a state stabilization mechanism while switching the algorithm; designing a comprehensive fitness function; sorting and interpolating the path control points to generate an executable track; introducing an elitist retention mechanism after each generation of iteration is finished; and when any termination condition is met, terminating the algorithm and outputting a robot path. According to the method, the quality and efficiency of path planning are remarkably improved, the method has good universality and expansibility, and an efficient and reliable solution is provided for intelligent navigation of the autonomous mobile robot.

The invention provides a neural network multi-objective optimization and FPGA hardware acceleration collaborative design method, and belongs to the field of deep learning model compression and hardware collaborative design. The method comprises the following steps: constructing a joint optimization space containing a neural network compression parameter and an FPGA hardware design parameter; a multi-target Bayesian optimization search strategy is adopted, iterative search is carried out in the joint optimization space, model precision, FPGA resource occupation and reasoning delay are synchronously optimized, and optimal candidate configuration is obtained; matching the compressed network structure with the FPGA parallel architecture by using a hardware-perceived pruning and quantification strategy; a multi-task performance prediction model is adopted to quickly predict the precision, resource occupation and delay of the optimal candidate configuration so as to accelerate the search process; according to the optimal configuration, a hardware accelerator code facing the target FPGA is automatically generated, and integration and implementation are completed. According to the method, collaborative optimization of neural network compression and hardware design is achieved, FPGA resource occupation can be remarkably reduced, the reasoning speed can be increased, and meanwhile the model precision is kept.

The invention relates to the technical field of industrial process automatic control, and discloses an automatic control method for a water sample analysis process, which comprises the following steps: establishing a unit step response reference model for representing a normalized standard track of a controlled object; iteratively searching an optimal amplitude scaling and time elastic factor in the two-parameter search space to minimize the fitting residual error of the reference model and the real-time data; the optimal amplitude scaling factor is utilized to determine a projection steady-state value to drive an execution unit, and the sampling interval is adaptively adjusted according to the optimal time elastic factor, the time elastic factor is introduced to realize closed-loop compensation of system kinetic parameter drift, a steady-state target can be accurately locked in an unbalanced state, and the stability of the system is improved. And the problem of control model mismatch caused by device aging or environment fluctuation is solved.

The invention discloses an IOS image and CBCT image registration method based on particle swarm and single-tooth optimization, and belongs to the field of image data processing, and the method comprises the following steps: respectively obtaining tooth semantic segmentation results and point cloud data of an IOS image and a CBCT image through preprocessing and segmentation steps; introducing a particle swarm optimization algorithm, and performing iterative search and optimization on six-degree-of-freedom rigid transformation parameters between the IOS image data and the CBCT image data so as to complete global initial registration of the dental arch scale; and on the basis of a result after global registration, a registration object is divided into a plurality of local units based on a single tooth, and local iteration alignment is performed on each unit, so that tiny dislocation on the single tooth is eliminated, and the overall registration effect is optimized. According to the method, on the basis of processing cross-modal difference and realizing global registration, the registration precision of a single tooth level is further ensured through a local optimization mechanism, and reliable automatic registration support is provided for clinical scenes such as orthodontics and implantation.

The invention provides a cross-database approximate nearest neighbor search method and system and a calculation framework, and the method comprises the following steps: constructing a graph index, storing the graph index in a relation table, and obtaining a graph index table; acquiring and storing a data set and a query set in a structured relation table form; the dismantling approximate nearest neighbor search process comprises a plurality of SQL operation stages including a candidate node screening stage, a neighbor expansion stage, a distance calculation stage, a result combination stage and a priority queue maintenance stage; and based on the graph index table, executing an iterative search process of the plurality of SQL operation stages on each query point in the query set, and finally outputting an approximate nearest neighbor search result of the query set. According to the method, the graph index is combined with the relational database, and the approximate nearest neighbor search is realized by adopting a plurality of SQL operation stages, so that the high-dimensional vector retrieval efficiency and the cross-database compatibility are remarkably improved, and the large-scale application of the vector data in a multi-element scene is promoted.