67 results about "Iterative strategy" patented technology

The invention discloses a knowledge graph multi-hop reasoning-based complex manufacturing process intelligent planning decision method, which comprises the following steps of: 1, constructing a manufacturing process knowledge graph to obtain a triple set, generating a query-path pair sequence, and mapping entities and relationships in the query-path pair sequence into low-dimensional vectors through an embedded layer; 2, performing expansion causal convolution calculation on the query-path pair sequence through a shallow feed-forward convolution network module, extracting key process features and potential correlation between processes, and generating a feature vector fusing the potential process correlation; 3, performing rotation position coding and causal mask processing on the feature vector output in the step 2 by adopting a parallel multi-query attention module, and realizing cross-process dependency relationship modeling through a multi-head attention mechanism; 4, generating a global optimal process scheme through beam search; and on the basis of a rule-guided dynamic aggregation iteration strategy, mixing a process path generated by the model with an original training set, and updating parameters.

The invention discloses a financial knowledge graph construction method and system based on artificial intelligence, and relates to the field of artificial intelligence data processing. The method comprises the following steps: performing multi-dimensional semantic analysis on a heterogeneous financial data source, and extracting a structured semantic fragment; constructing a financial entity perception unit, identifying a multi-granularity entity and generating a unique code; generating a preliminary relation graph based on the event cascade relation and the attachment structure, and injecting a semantic translation label; normalizing the atlas relationship through semantic separation and a label reconstruction mechanism to form a financial relationship network with consistent semantics; executing evolution increment iteration in combination with the newly added corpus, and dynamically updating nodes and edge sets; and performing semantic consistency and structural integrity evaluation on an iteration result, and outputting a stable financial knowledge graph structural body. By introducing a multi-factor semantic analysis model, a causal relationship modeling mechanism and a graph evolution iteration strategy, systematic improvement of the financial knowledge graph in the aspects of structural expression precision, semantic reasoning ability and dynamic adaptability is achieved.

The invention discloses a functional gradient piezoelectric material optimization method based on a physical information network and deep regression. The method comprises the following steps of: 1, constructing a physical information neural network embedded with a piezoelectric constitutive relation; 2, developing a deep regression network based on symbolic regression, wherein the deep regression network is used for converting the piezoelectric phase volume fraction distribution predicted by the neural network into an analyzable mathematical expression; step 3, carrying out joint optimization on the two networks by adopting an alternate iteration strategy; under the condition that parameters of the deep regression network are kept unchanged, a physical information neural network is trained preferentially to meet the precision requirements of a displacement field, a stress field, potential distribution and a piezoelectric phase volume fraction; and then fixing the physical information network, optimizing the expression generation capability of the deep regression network, and finally analyzing the piezoelectric phase volume fraction expression. According to the method, the physical information neural network and the depth symbol regression technology are fused, so that multi-physical field collaborative optimization of component distribution of the functionally graded piezoelectric material is realized.

The invention relates to the technical field of radar waveform optimization, in particular to a synthetic broadband waveform optimization method based on reinforcement learning. According to the method, a deep reinforcement learning framework is adopted, and end-to-end optimization is carried out on the center frequency and amplitude parameters of multiple sub-pulses. The method comprises the following steps: constructing a synthetic broadband waveform signal model, and parameterizing sub-waveforms to determine a search space of an optimization problem; converting a waveform optimization problem into a reinforcement learning form, and defining a state space, an action space, a reward function and a state transition mechanism; based on a deterministic soft action value iteration strategy algorithm, point spread function features are extracted in combination with a Transform encoder, and sub-pulse parameters are dynamically adjusted. Through the method, the main lobe resolution can be effectively suppressed and improved, the target detection performance of a radar system is improved, and the method has good global optimization capability and environmental adaptability and is suitable for cognitive radar and adaptive waveform design scenes.

The invention relates to an electric vehicle wireless charging system LPV-Hammerstein model identification method based on an alternating least square iteration strategy, and belongs to the technical field of power electronic system modeling and parameter identification. The method comprises the following steps: firstly, constructing an LPV-Hammerstein model structure based on a WPT system circuit topology and a working mechanism, and then collecting discrete data of a system input control signal U, an output DC voltage Vo and a scheduling variable Req; then, alternately fixing other parameters by adopting an alternate least square iterative algorithm, and converting a to-be-estimated parameter problem into a linear least square sub-problem to be solved; and when a convergence condition is satisfied, a final LPV-Hammerstein model parameter estimation result is obtained. According to the method, the common nonlinear and parameter change coupling characteristics in the wireless charging system of the electric vehicle can be effectively processed, a systematized and efficient way is provided for obtaining a high-precision system dynamic model, and the limitation that the system cannot be accurately described by a traditional linear model or a simple nonlinear model is overcome.

The invention discloses an asymmetric cross-modal hash learning method for potential structure information, and relates to the technical field of cross-modal retrieval. Analyzing a sample clustering structure, sample similarity information and semantic similarity information among the multi-modal samples to obtain a projection matrix of different modals, a clustering center of a projection space, a clustering indication matrix, a mapping matrix, a discrete hash code and a calculation formula of an auxiliary variable; updating the projection matrix of each modal, the clustering center of the projection space, the clustering indication matrix, the mapping matrix, the discrete hash code and the auxiliary variable by using an iteration strategy; determining a hash function of each modal according to the discrete hash code obtained by the last iteration and the multi-modal training sample; the hash function is used for performing similarity retrieval on the query sample in the multi-modal data. According to the method, the Hash code with higher discrimination can be obtained, so that the retrieval performance of cross-modal retrieval is improved.

The invention belongs to the field of coherent combination, and discloses a multi-strategy segmented fusion adaptive SPGD algorithm for a laser coherent combination system. The method is used for solving the problems that in a traditional SPGD algorithm, due to the fact that a fixed gain coefficient and disturbance amplitude exist, the convergence speed and stability of the algorithm are contradictory, and the optimal performance is difficult to achieve in different environments and system states. According to the method, the algorithm is divided into three stages in the iteration process, and self-adaptive adjustment of the gain coefficient and the disturbance amplitude is realized according to the characteristics of a hardware system in combination with three different iteration strategies. According to the method, the convergence speed of the algorithm can be remarkably improved, meanwhile, the stability of the algorithm can be effectively improved, the algorithm has higher robustness in consideration of hardware system characteristics, and the good application prospect and feasibility of the algorithm are shown.

The invention discloses a short-wave time difference positioning method based on particle swarm optimization, which is realized by aiming at a short-wave signal source positioning problem under the condition that ionized layer information is unknown through the following steps of: dynamically screening a receiving station combination with optimal geometric distribution by adopting a particle swarm optimization algorithm based on longitude and latitude and time difference observation data of a receiving station group; taking the main receiving station as a coordinate origin, and converting the geographic coordinates of the selected station into rectangular plane coordinates; constructing a time difference positioning equation set by using a Chan algorithm, and calculating a target position through pseudo-inverse solution and an iteration strategy; and filtering the ill-conditioned solution based on a geometric accuracy factor threshold, analyzing a positioning point cluster through sliding window density, and outputting median coordinates as a final positioning result. According to the method, the receiving station is innovatively selected and modeled as a combinatorial optimization problem, the positioning precision in a complex electromagnetic environment is remarkably improved through a collaborative architecture of particle swarm optimization and a Chan algorithm, and the problem of positioning drift caused by poor geometric distribution of stations in a traditional method is effectively solved.

The invention provides a high-dimensional medical feature selection method, system and device and a storage medium, and belongs to the field of medical data processing, and the method comprises the steps: obtaining a medical data set; the population parameters are initialized; the fitness value of the current population is calculated, the populations are sorted according to the fitness value, and a hierarchical iteration strategy is adopted for population individuals ranking the first three; searching other population individuals by adopting an aurora optimization algorithm, and updating positions through a probability chaos formula; and performing random selection strategy search on the updated population, performing random selection from the existing solutions, further updating the current solution, outputting the population, and obtaining a corresponding medical feature combination according to the output population. Iteration efficiency is enhanced, invalid search is reduced, medical feature selection is optimized, and subsequent classification tasks are facilitated.

The invention relates to the technical field of settlement monitoring, in particular to a method for predicting a building settlement trend value, which comprises the following steps of: testing the settlement amount of a building in each period through settlement observation points so as to obtain a period test value in-time ordinal sequence of each settlement observation point; establishing a traditional GM (1, 1) model for the time sequence sequence to predict so as to obtain a time response function; constructing an equal-dimensional metabolism GM (1, 1) model through an iteration strategy; obtaining an optimal dimension according to the development coefficient of the time response function; and according to the isometric metabolism GM (1, 1) model under the optimal dimension, obtaining the ash action amount and the development coefficient, and obtaining the settlement trend value of the building. According to the method, the settlement trend value of the building can be effectively predicted, the defects of subjectivity and randomness of parameter selection in a traditional method are overcome, and reliable technical support is provided for engineering practice. The invention further provides a device for predicting the settlement trend value of the building.

The invention discloses a robust robust sequence extraterrestrial celestial body terrain reference surface fitting method and system, and relates to the technical field of terrain reference surface fitting, and the method comprises the steps: collecting three-dimensional laser point cloud data to construct a reference plane geometric model, and building a joint error model of an observation value error and a design matrix error; adopting a sequence PEIV total least square framework to carry out grouping iteration adjustment on the point cloud data, and recursively updating parameter estimation by utilizing an error propagation matrix and a matrix inversion formula; introducing a two-factor robust mechanism, dynamically calculating a weight according to a residual error and estimating a robust scale; and circularly optimizing the parameters through an iterative strategy until a convergence condition is met, and outputting reference plane parameters. According to the method, different terrain complexity, data scales and error distribution scenes can be covered, the adaptability and engineering practicability of terrain datum plane fitting are remarkably improved, and reliable geometric reference is provided for subsequent obstacle detection and safe area decision making.

The invention provides a self-adaptive vacuum defoaming process optimization method based on an intelligent control system, and aims to solve the problems of micro-bubble residue, too high energy consumption and the like caused by the fact that a traditional defoaming technology cannot adapt to liquids with different viscosities, and accurate defoaming is realized through a multi-dimensional sensing technology and strategy generation and regulation. Presetting a high-precision vacuum parameter and a viscosity grading strategy; establishing a liquid characteristic vector model by adopting a YOLOv7 and Hough transformation algorithm, and matching a historical optimal strategy in combination with weighted cosine similarity; generating a defoaming strategy based on the pre-training network and a multi-objective optimization algorithm, and verifying the defoaming strategy in the virtual environment; defoaming is carried out in four stages, and efficient foam breaking is achieved through sectional PID control, resonance vibration and cooperation of a two-stage pump; a Kalman filtering algorithm is integrated, pressure, temperature and vibration data are fused, and vacuum pump power and a heating strategy are adjusted; and constructing a multi-index evaluation system, and driving strategy iterative optimization through a reward signal.

The invention discloses a pump turbine flow-induced vibration multi-scale coupling calculation method, and belongs to the technical field of pump turbine fluid-solid coupling dynamics. According to the method, the gap flow effect is included in a multi-scale unsteady fluid-solid coupling analysis framework of the pump turbine, the gap drainage basin is included in three-dimensional modeling of the pump turbine, the three-dimensional model is subjected to grid division by adopting a multi-scale grid technology, and a small transition region is arranged between a movable guide vane and a rotating wheel, so that the gap flow effect is obtained. The transverse direction is consistent with a main runner grid, and the longitudinal direction corresponds to a gap grid, so that data coordination of large-scale and small-scale regions is realized. A fluid-solid coupling solving method of a bidirectional iteration strategy is adopted, a flow field and a structure field are solved respectively, a system coupling module carries out load transmission and coordinated updating in a residual weighting and node mapping mode, the physical consistency and numerical stability of coupling simulation are effectively improved, and the system performance is improved. The method has remarkable advantages in the aspects of improving simulation precision, enhancing calculation stability, improving design efficiency and the like.

The invention belongs to the technical field of intelligent control and electronic circuits, and particularly relates to an intelligent peripheral control method based on an Android box, which comprises the following steps: acquiring sensor data and peripheral operation data of the Android box; performing feature data extraction through a time-frequency joint feature algorithm; a fault precursor prediction model is utilized to identify potential risks, a synchronous state is fed back to a user through voice, an equipment abnormal trend and a current operation condition are grasped, a federal learning cooperative training model is utilized to perform dynamic monitoring based on the potential risks, and a predictive pre-control instruction and an emergency instruction are generated; the method is executed after verification of the block chain, a standby power supply is started in an extreme scene, a user is informed of a real-time state through LED light flickering, a reinforcement learning iteration strategy is utilized, a maintenance scoring system is constructed, and point details are displayed to the user through an APP according to maintenance point redeemable services. Therefore, the problems of insufficient data processing and risk prediction capabilities, lack of user maintenance incentive mechanisms and the like in the prior art are solved.

The present application discloses the construction and application of an esophageal cancer survival period prediction model based on an improved salp swarm algorithm. The present application uses a partitioned iteration strategy to divide the iteration of the algorithm into two periods, namely the early iteration period and the late iteration period; in the early iteration period, a leader fusion mutation strategy is used to update the leader position to increase the exploration ability of the algorithm; in the late iteration period, an optimal individual leading movement is adopted to update the leader position to optimize the exploitation ability of the algorithm; then a dimension-by-dimension Gaussian mutation strategy is used to update all salp swarm individuals to improve the diversity of the population; before the end of the algorithm iteration, the optimal salp position is subjected to an optimal neighborhood perturbation and the fitness values are compared, and a greedy strategy is used to select the optimal fitness value to improve its ability to jump out of the local optimal solution. The IPSSA-BP model of the present application has a good fitting effect, high accuracy, high prediction precision, and good prediction stability.

The invention discloses a multi-agent cooperative grouping and income distribution method, which comprises the following steps of: introducing an improved HCS algorithm to reasonably initialize tasks and each agent, calculating sub-task priorities according to skill requirements and prediction workload distribution, optimizing cross-task resource distribution by combining income prediction and cost constraint in an inter-task competition stage, and optimizing the task resource distribution. And calculating the marginal contribution of the intelligent agent based on a Shapley value, carrying out income distribution in combination with a dynamic excitation coefficient, adjusting an income distribution proportion according to parameters, and carrying out iterative strategy updating on a formed initial grouping result until the initial grouping result is converged to Nash equilibrium after the individual rational constraint and the overall income maximization are met. According to the method, initial allocation is intelligentized, the resource utilization rate and the total task income are remarkably improved based on skill demand prediction and the workload priority, fair and efficient income allocation is ensured, and the method is suitable for industrial automation, smart cities and other scenes.

The invention discloses a multi-interpolation method and system for medical multi-view incomplete data, and relates to the technical field of machine learning, and the method comprises the steps: constructing a unified matrix and a binary observation indication matrix through feature level splicing, and achieving single interpolation in combination with deficiency perception standardization, adaptive neighbor selection and Lregularization condition regression; an iteration strategy of missing proportion sorting is adopted to improve convergence, random permutation and regression coefficient disturbance are introduced to generate m groups of candidate interpolation results, and uncertainty is described; and finally, outputting complete multi-view data based on missing rate adaptive weighted fusion, destandardization and view segmentation. According to the method, through multi-view feature splicing modeling, Pearson's adaptive neighbor regression and missing proportion driven iterative interpolation, accurate multiple interpolation of high robustness, quantifiable uncertainty and reducible view structure for medical multi-view incomplete data is realized.

This invention discloses a sensor-integrated multi-UAV beam trajectory design method for anti-UAV operations. First, a system model is constructed, defining the channel model, signal model, and power consumption model. Then, a mathematical model of the long-term Cramer-Rao bound (CRB) minimization problem is established, and the optimization problem is reconstructed, addressing uncertainties in the optimization objective. Next, the problem is broken down into three sub-problems: joint optimization of transmit and receive beams, joint optimization of user association and UAV scheduling, and UAV trajectory optimization. These sub-problems are solved using Lagrange relaxation, semi-definite relaxation, and penalized successive convex approximation methods, respectively. Finally, an alternating iterative strategy is used to coordinate the solutions to these sub-problems. This invention solves problems in existing technologies such as incomplete coverage, poor sensing performance, signal interference, and positional uncertainty caused by UAV maneuvers, improving cooperative sensing performance and meeting the dual requirements of flexible sensing and reliable positioning.

The invention discloses a shot domain-detection point domain multi-stage joint iterative separation method for simultaneous excitation of aliasing data on both sides of an offshore towing cable, and belongs to the technical field of efficient acquisition and processing of seismic exploration of oil and gas reservoirs, and the method comprises the following steps: S1, obtaining aliasing data to be separated, and preprocessing the seismic data; s2, acquiring a shot domain-detection point domain neural network seismic training data set {B, P}; s3, constructing a shot domain-detection point domain neural network model; s4, performing multi-iteration training on a pre-established shot domain-detection point domain neural network model by using the training data set; and S5, performing intelligent separation on to-be-separated aliasing data by using the trained shot domain-detection point domain neural network model to obtain separated seismic data. According to the method, multiple sets of multi-data-domain network models are trained by utilizing an iteration strategy, and mapping relationships between different aliasing degrees, different data domain features and label data are captured from one set of training data, so that the network model training process is remarkably improved, and the processed data quality is ensured while the efficient implementation of the algorithm is ensured.

The invention provides a curvature weighted 3D-SIFT and adaptive radius FPFH-based point cloud registration method and system. The method comprises the steps of obtaining point cloud data and performing preprocessing; carrying out point cloud key point extraction based on curvature weighted 3D-SIFT to obtain corresponding key points; constructing a point cloud feature descriptor, and calculating the feature descriptor by using an improved FPFH fast point feature histogram algorithm; an attitude estimation scheme in an IGSP algorithm is adopted, firstly, an energy function is constructed, then the energy function is optimized through a KM algorithm to determine global optimal correspondence, and transformation is calculated on the basis; and finally, continuously repeating the process of optimal corresponding matching and transformation calculation through an iteration strategy to obtain a globally optimal solution. The method is combined with density characteristics, the description capability of complex scenes (such as vegetation and irregular terrains) is remarkably enhanced, and mismatching is reduced.

The invention discloses a functionally graded material structure optimization method based on a physical information network and deep regression. The method comprises the steps of 1, building a physical information network, calculating strain and stress through automatic differential, and building physical constraints in combination with a constitutive equation and an equilibrium equation; the deviation between the network prediction and the physical law is quantified through a mean square error to ensure that the prediction result conforms to the mechanics principle; 2, building a deep regression network, and optimizing and training the deep regression network through the mapping relation; an optimal expression is searched through gradient descent, and an analyzable volume fraction function model is finally generated and used for material distribution optimization; and step 3, adopting an alternating iteration strategy, finally outputting an expression of VA (x, y, z), verifying Young modulus distribution accuracy through Voigt homogenization, and realizing automatic optimization design of the functionally graded material. According to the method, the optimal distribution of each component of the functionally graded material structure is automatically obtained by adopting a physical information neural network method, so that the performance of the functionally graded structure is optimized.

This invention relates to the field of data reasoning technology, and discloses a message push strategy optimization method and system based on deep reinforcement learning. The method includes: jointly encoding the static attributes, observation sequence, and selection identifier of the agent to generate a decision state vector and an action feature set; obtaining a preliminary value scalar by performing a dot product operation on the decision state vector and the action embedding vector; normalizing the scalar to obtain a policy probability distribution, and determining the action identifier to be executed accordingly; retrieving environmental interaction data based on the identifier to obtain a multi-dimensional feedback vector; updating the state representation weights based on the feedback, and constructing a prediction vector by combining the action features; finally, iterating the policy parameters based on the prediction vector to obtain an optimization index. This invention can improve the efficiency of message push strategy optimization based on deep reinforcement learning.

The invention discloses a navigation method, device and equipment based on a GRPO algorithm and a medium, and relates to the technical field of reinforcement learning, and the method comprises the steps: calculating the average similarity between a current strategy and a plurality of previous iteration strategies based on KL divergence; updating a step length factor through an average reward change rate and an average similarity determined based on a plurality of iterated rewards; determining a gradient estimation correction item based on the gradient estimation of the sampling trajectory, determining target gradient estimation according to the gradient estimation correction item and the original gradient estimation, and updating the current strategy through the target gradient estimation and the updated step length factor; when the current strategy is updated, the importance weight is cut, the target function of the GRPO algorithm is corrected according to the cut weight, the GRPO algorithm is trained based on the corrected function and the updated strategy, so that the intelligent agent learns the optimal strategy based on the trained GRPO algorithm, and the outlet of the labyrinth is determined according to the optimal strategy. Therefore, the stability of the algorithm is improved.

The invention relates to the technical field of wireless communication, in particular to a low-complexity signal detection method of an MIMO-OTFS system, which comprises the following steps: setting a grouping symbol number according to a channel condition, and calculating a maximum iteration number; in each iteration: calculating a filtered received signal through the channel matrix and the received signal; constructing a matrix psi, processing a structure matrix psi S of the matrix psi by adopting an RCM algorithm to obtain a permutation matrix, and calculating a reordering matrix according to the permutation matrix; performing Cholesky decomposition on the reordering matrix to obtain a banded matrix; calculating the SINR of each symbol which is not detected at present; all symbols which are not detected at present are arranged in a descending order according to the SINR size, the first S symbols in the sorting result are taken to be detected one by one and output, and the channel matrix and the received signal are updated; judging whether the maximum iteration number is reached or not, and if not, continuing the next iteration; according to the method, the time complexity is reduced while the bit error rate performance is ensured through an iterative strategy of reliability symbol detection and interference elimination.

The invention provides a time-of-flight mass spectrometer resolution optimization method and device, a medium and computer equipment. The method comprises the following steps: S1, obtaining a target function and a PSO particle swarm; s2, based on the objective function, performing PSO iteration on the PSO particle swarm; judging whether a termination condition is met or not, and if yes, terminating iteration and generating a final result; if not, executing the step S3; s3, judging whether an NM iteration condition is met or not, if yes, executing NM iteration on the PSO particle swarm after PSO iteration, correcting the PSO particles based on an NM iteration result, and executing the step S2 on the corrected PSO particle swarm; if not, executing the step S2; and S4, outputting a final result. By the adoption of the method, the optimization model with the mass resolution as the objective function is constructed, and the iterative strategy of mutual fusion of the particle swarm optimization (PSO) and the Nelder-Mead simplex (NM) algorithm is adopted, so that organic combination of global search and local fine optimization is achieved.

The application discloses a low-code platform variable type automatic inference method and device. The method comprises the following steps: screening a variable with an automatic type identifier in a low-code platform to obtain a target variable; obtaining a default value inference result based on a type inference rule; traversing a structured model corresponding to a business logic and declaration information to obtain a node assignment inference result; obtaining a dependency relationship inference result through a multi-round iteration strategy recursive derivation; integrating the default value inference result, the node assignment inference result and the dependency relationship inference result to determine an initial type of the target variable; storing the initial type based on inference trace information and realizing change notification to finally obtain a target type. Through multi-stage inference, expression type calculation, chain dependency processing and other designs, the application does not need to manually specify the type, supports complex scene inference, maintains type safety, reduces use and maintenance costs, and improves the development efficiency and ease of use of the low-code platform.

The invention discloses an information active group sending method, which belongs to the technical field of information distribution, and comprises the following steps: triggering a task and analyzing to generate task metadata; capturing real-time features of users, generating dynamic tags, and screening effective users; generating compliant personalized single instance content in combination with the dynamic tag matching material; an optimal channel is matched and sent through a high-concurrency mechanism, and full-process compliance monitoring is executed; according to the method, the accuracy is improved through dynamic label and personalized content generation, the sending efficiency is guaranteed by means of high-concurrency scheduling and channel optimization, the risk is avoided through full-process compliance monitoring, the effect is continuously optimized by means of automatic iteration, and the method has the advantages of being high in practicability and high in practicability. The method solves the problems of low precision, poor efficiency and insufficient compliance of an existing group sending method, and is suitable for various large-scale information active distribution scenes.

The invention relates to the technical field of radar signal processing, and particularly provides a radar signal processing method and system based on a dual-mode adaptive clutter map. An iteration strategy emphasizing historical data and an iteration strategy emphasizing current data are respectively designed for the static clutter map and the dynamic clutter map, so that the static clutter background is kept stable, and meanwhile, the dynamic clutter can be quickly tracked; and further combining a static clutter weighting curve fitted based on an MTD improvement factor and a moving clutter weighting curve fitted according to environmental noise to generate adaptive detection thresholds, thereby effectively avoiding over-suppression of a real target while remarkably suppressing various clutters. The method can greatly reduce the false alarm rate in a complex environment, guarantees the stable detection of a real target, and has good system real-time performance.

The invention discloses an automatic inference method and device for low-code platform variable types. The method comprises the following steps: screening and matching variables with automatic type identifiers in a low-code platform to obtain target variables; obtaining a default value inference result based on the type inference rule; traversing the structured model and the declaration information corresponding to the service logic to obtain a node assignment inference result; recursive derivation is carried out through a multi-round iteration strategy to obtain a dependency relationship inference result; integrating the default inference result, the node assignment inference result and the dependency inference result, and determining the initial type of the target variable; and storing the initial type based on the deduced traceability information and realizing change notification, and finally obtaining a target type. Through multi-stage inference, expression type calculation, chain dependency processing and other designs, a user does not need to manually specify a type, complex scene inference is supported, type safety is kept, the use and maintenance cost is reduced, and the development efficiency and usability of a low-code platform are improved.

This invention provides a method and apparatus for image generation based on a priori guidance using a diffusion bridge model. The method includes: obtaining an image generation result from a source image using a diffusion bridge image generation model according to a preset iterative strategy; the preset iterative strategy includes: sampling the source image to obtain the image state at the initial time step and constructing a weak prior image state; inputting the image state and the weak prior image state into the diffusion bridge image generation model respectively to obtain a predicted image and a weak prior predicted image, and determining the guidance residual; performing frequency domain modulation processing on the guidance residual to determine its frequency domain representation, and combining it with the weak prior predicted image to obtain a target predicted image; sampling the target predicted image to obtain the image state at the next time step and repeating the iteration until a preset maximum number of iterations is reached to obtain the image generation result. This invention retains the powerful generation capability of the pre-trained model while achieving flexible guidance with zero training cost, reducing computational cost and time overhead.