438 results about "Adaptation method" patented technology

The invention relates to an AI-based automatic adaptation method for a work approval process, and the method comprises the steps: collecting multi-source approval data, and carrying out the preprocessing of the multi-source approval data, and forming standardized approval data; analyzing the system rule text by using the pre-trained AI large model, and extracting a structured approval rule comprising a trigger condition, an approval role and a process node sequence; matching a basic examination and approval template according to the form field and the user permission information, and dynamically generating an adaptive process comprising multiple stages of examination and approval nodes, aging parameters and an additional examination and approval link; and through resource scheduling optimization and time domain correlation analysis, establishing an optimized mapping relation based on flow execution characteristics such as approval timeliness deviation and node skipping frequency, and outputting a visual flow chart and execution parameters. A business logic writing mode is replaced by automatic analysis of an AI large model on an unstructured rule; the dynamic template matching and continuous iterative optimization mechanism can quickly respond to business changes, and resource scheduling optimization and automatic process generation shorten the implementation period and reduce the operation and maintenance cost.

The invention discloses an enterprise scientific and technological achievement adaptation method based on big data accurate retrieval and query, and relates to the technical field of big data processing and knowledge maps, and the method comprises the steps: constructing an enterprise technical field knowledge map based on a standardized multi-modal feature matrix, receiving incremental technical data in real time by adopting an Apache Flink streaming processing framework, and carrying out the real-time retrieval and query of the enterprise technical field knowledge map. Dynamically updating a graph node relationship, removing expired nodes through a pruning algorithm, and outputting a dynamic knowledge graph with a timestamp; and receiving a query request of a user, extracting a query vector by utilizing the cross-modal embedding model, retrieving a Top-N candidate node list through a dynamic knowledge graph, calculating a comprehensive similarity score in combination with a graph relation weight, and outputting a sorted technical achievement list. Multi-source heterogeneous data is mapped to a unified semantic space through a cross-modal embedding model, and the intelligent level and the actual application effect of scientific and technological achievement adaptation are comprehensively improved in combination with construction and updating of a dynamic knowledge graph.

The invention relates to an automatic protocol adaptation method for a real-time access system of multi-source equipment. The data acquisition layer receives original protocol data sent by equipment and transmits the original protocol data to the protocol adaptation layer. And the protocol metadata analysis module extracts, analyzes, compares and matches the data, and automatically identifies the protocol type of the equipment. The rule engine converts original protocol data into a system unified internal data format according to a conversion rule and then transmits the original protocol data to the data processing layer, and after cleaning, integration, analysis and other operations are carried out, the original protocol data are transmitted to the application interface layer to be called by an upper-layer application system. And when new protocol equipment is accessed, the hot plug function allows the system to dynamically load the new protocol adapter module in a state of not stopping running. According to the method, automatic adaptation of a new protocol is achieved through rapid deployment of the containerized micro-service architecture in combination with newly added protocol rules and metadata of the protocol management module, the whole process does not need to compile and publish the system again, and efficient and stable operation of the multi-source equipment real-time access system is guaranteed.

The invention discloses an Internet of Things test platform adaptation method, system and device based on post quantum cryptography. The method comprises the following steps: constructing a quantum threat perception model; establishing a multi-dimensional adaptive decision matrix, and generating a candidate post-quantum cryptography algorithm set; dynamic injection of a cryptographic protocol is executed, and lossless replacement of a communication layer encryption suite is achieved; deploying a runtime performance monitoring engine, and collecting encryption and decryption delay, memory occupancy rate and electromagnetic radiation characteristic data in real time; and constructing an algorithm switching decision tree based on reinforcement learning, and when a quantum attack feature or an equipment performance threshold is detected to break through, triggering cryptographic algorithm dynamic migration. A quantum threat perception driven Internet of Things security adaptation system is constructed, and by introducing a dynamic password switching mechanism, a hardware acceleration optimization scheme and a trusted execution environment, the security upgrading problem of traditional Internet of Things equipment in the post-quantum era is solved. According to the invention, the execution efficiency of the PQC algorithm of the resource-constrained equipment can be improved, and the key updating delay is reduced.

The invention relates to the technical field of permission adaptation, in particular to a CRM-GIS-oriented multi-protection permission dynamic adaptation method and system. The method comprises the following steps: firstly, defining in a CRM-GIS platform and acquiring a multi-dimensional authority attribute related to an access request in real time; based on a preset permission policy rule in the platform, processing an access request by using a policy decision engine; then, based on the current spatio-temporal information and behavior pattern information of the user, an artificial intelligence risk assessment module is utilized to analyze the real-time risk level of the current operation of the user; thirdly, dynamically judging and adjusting a final permission adaptation decision aiming at the access request by integrating the multi-dimensional permission attribute, the permission strategy rule and the risk level; according to the final permission adaptation decision, the CRM-GIS platform compulsorily executes corresponding access control operation, and limits or allows access to data and functions in the platform; according to the invention, the reliability of multi-dimensional right dynamic adaptation can be improved.

The invention belongs to the technical field of intelligent control, particularly relates to a deep learning-driven intelligent home scene dynamic adaptation method, and aims to solve the problem that an existing intelligent home system is difficult to realize high-precision personalized scene adaptation in a multi-user and multi-device environment due to dependence on a static rule. The method comprises the steps of collecting multi-source heterogeneous user behavior data and performing semantic enhancement preprocessing, constructing a hierarchical time sequence behavior coding model to extract local time sequence dependence and cross-equipment long-range association features, clustering to generate a dynamic scene prototype and mapping the dynamic scene prototype into an executable condition-action rule, after the rules are deployed, a closed-loop optimization mechanism is constructed through explicit and implicit user feedback, and online incremental updating and self-adaptive evolution of the behavior model and the scene rules are achieved. According to the technical scheme, the user complex behavior mode can be deeply understood, the scene adaptation precision is continuously optimized, the individuation level, logic consistency and system robustness of intelligent services are improved, and meanwhile privacy safety and real-time response are guaranteed through edge calculation.

The invention provides a credential and credential environment adaptation method and system in combination with a cloud computing platform, and relates to the technical field of credential and credential environment adaptation, and the method comprises the steps: generating adaptation analysis basic features based on static structure information; performing interface adaptation analysis by using the API support database and the adaptation analysis basic features; after a virtual credential environment is established by utilizing the virtual machine, executing the operation record of the target software, and establishing an operation data set; behavior capture of the operation data set is carried out under the guidance of the risk hot area of the compatibility matching graph by utilizing a behavior abnormity identification channel, and clustering analysis is carried out on behavior capture results; uI presentation of the target software is recorded, presentation consistency verification is carried out based on the UI presentation, and a presentation consistency score is established; and establishing an adaptation result according to the adaptation risk score table, the behavior stability score and the presentation consistency score. According to the invention, the technical problem of poor adaptation stability in the prior art can be solved, and the technical effect of improving the adaptation stability is achieved.

The invention discloses a Java-based large language model dynamic adaptation method and system, belongs to the technical field of artificial intelligence, and aims to solve the technical problems of overcoming the interface difference of multiple model interfaces, simplifying the development process and providing standardized and high-expansibility large model management. Comprising the following steps: providing a model registration and dynamic loading service, a protocol conversion and parameter standardization service and a load balancing and failover service; a streaming transmission protocol service, a function call dynamic injection service and a global error processing mechanism are provided; when a user provides a document analysis and partitioning service, a vectorization index construction service and an RAG enhanced generation service to ask questions, relevant document blocks in the Pinecone are retrieved through the RAG enhanced generation service to serve as contexts to be injected into cue words of the large language model, and answers generated by the large language model are returned.

The invention discloses a dynamic form generation and cross-database adaptation method based on metadata driving, and belongs to the technical field of data architectures. The method comprises the steps that an original metadata attribute graph is established; on the basis of the original metadata attribute graph and user roles and equipment parameters for calling the forms, generating an optimal equipment form interaction data graph corresponding to each user permission; verifying an interaction path between the forms and a cross-database demand of a target database type to carry out cross-database mapping adaptation analysis, and generating an optimal form data adaptation database of the corresponding equipment of each user permission; and performing rule compiling based on multi-source heterogeneous data input by a user in real time and execution logic in the original metadata attribute graph, marking an incremental propagation path in the optimal form interaction data graph of the corresponding equipment of each user permission, and updating the optimal form data adaptation database of the corresponding equipment of each user permission. The method has the advantages that the data interaction efficiency is improved, and the interaction round-trip times and redundant data transmission are reduced.

The invention belongs to the technical field of artificial intelligence and localization basic platforms, and discloses a local large language model operation framework transplantation and adaptation method based on a domestic DCU environment. According to the method provided by the invention, DCU-based local operation framework adaptation is realized, and the hardware selection range of a mainstream large language model operation framework is greatly expanded; through the model quantification and compression technology, the hardware of the local deployment large language model can be quickly applied to the field of natural language processing, and the competitiveness of the domestic hardware in the large language model application is improved.

The invention provides an industrial large model adaptation method and device based on cross-mode linkage and comparative learning, and relates to the field of data processing. According to the method, historical operation data and production line working conditions in the industrial field are obtained; according to the historical operation data and the working condition of the production line, a digital twinning environment of the industrial large model is constructed, and the digital twinning environment comprises a first agent, a second agent and a third agent; injecting a mixed fault signal into the digital twin environment through the first agent; performing fault type identification on the mixed fault signal through a second intelligent agent to obtain an identification result; performing comparative learning on the recognition result through a third agent to obtain a chess playing result; and dynamically adjusting the type and intensity of the digital twin environment based on the chess playing result, generating an adaptation strategy, and adjusting the industrial large model according to the adaptation strategy. By implementing the technical scheme provided by the invention, the adaptation accuracy of the industrial large model can be improved.

The invention discloses an intelligent routing and unified adaptation method for a large language model, and the method comprises the steps: defining all access details of the model through a declarative configuration file, and achieving the zero-code access of the large language model without writing any code for a newly-added model; a completely consistent calling interface is provided for an upstream application, the isomerism of all downstream large language models is shielded, and a unified request and response abstraction layer is constructed; through a strategy engine and a JSON path technology, complex streaming response including content thinking is precisely processed, and intelligent analysis and content extraction are carried out; dynamic configuration and intelligent strategy hot update are supported; intelligent routing of the model is realized, and an optimal large language model instance is dynamically selected; and meanwhile, enterprise-level governance capability is provided, governance functions such as fusing, degradation, current limiting and monitoring are integrated, and stability guarantee is provided for model calling. Therefore, the maintainability, the expandability and the user experience consistency of the system are comprehensively improved.

The invention discloses a micro-front-end-oriented universal basic capability adaptation method and device and a storage medium, and the method comprises the steps: determining a sub-application which needs to be loaded at present, and obtaining the feature parameters of the sub-application; calculating the characteristic parameters through a preset adapter matching algorithm, and determining an adapter matched with the sub-application according to a calculation result; setting configuration items of the adapter according to the characteristic parameters of the sub-applications; and when the sub-application is activated, mounting the page content of the sub-application to a micro-front-end framework for display through the adapter. The universal micro-front-end adapter provided by the invention can automatically carry out front-end framework adaptation according to the sub-application framework and judge the adopted micro-front-end technology and base implementation according to the operating environment, the adaptation problem of different micro-front-end technology frameworks and bases is solved, the difference of gateway authentication, request headers and the like is processed through the adaptive requester, and the adaptability of the micro-front-end technology frameworks and bases is improved. The flexibility and expansibility of integration of the micro-front-end frame are improved.

The invention provides a training and pushing integrated scene-based neural network parameter adaptation method, a medium and equipment, and belongs to the technical field of neural network parameter adaptation. Comprising the following steps: selecting initial parameter configuration, carrying out initial evaluation on a neural network model, and recording the performance of the neural network model; according to an initial evaluation result, positioning the parameters in an optimal parameter region by adopting Bayesian optimization; in the optimal parameter region, local optimization of parameters is carried out by using a gradient descent method; and for the neural network model after parameter optimization, the neural network model is compressed to a smaller version through pruning and quantization. According to the method, the parameter adaptation capability of the super-large-scale model can be effectively improved, the calculation overhead in the adaptation process is reduced, and meanwhile it is guaranteed that the model performance is not affected.

The invention discloses a dynamic adaptation method and device for display content of an LED screen and a medium, and the method comprises the steps: collecting original feature data of audiences in a radiation range of the LED screen through a multi-mode biosensor, carrying out the preprocessing of the original feature data, and generating a desensitization feature vector through local anonymization processing; performing real-time analysis on the desensitization feature vector by using a multi-modal fusion model in an edge computing node to generate a primary emotion tag; uploading the primary emotion label and the associated historical behavior data of the audience to a cloud platform so as to predict potential interests of the audience through a preset emotion deep analysis model, and generating a dynamic content adaptation strategy in combination with the potential interests of the audience; on the basis of a dynamic content adaptation strategy, display content matched with the emotion label is called from a pre-stored material library, low-delay rendering is conducted on the display content, and color parameters and the content display sequence of the LED display screen are dynamically adjusted according to the emotion analysis result.

The invention discloses an efficient optimization selection and model adaptation method and system based on LoRA and MoE technologies, and the method comprises the steps: firstly, constructing a LoRA module pool for various models and vertical field data sets according to different field models and data set features, so as to rapidly screen and call adaptation models in similar tasks; secondly, based on downstream task description, hidden layer features and confusion, performing coarse-grained LoRA selection from a LoRA module pool; and then, by using the performance characteristics and the single LoRA weight distance, further screening a fine-grained LoRA module from the LoRA modules selected from the coarse-grained LoRA modules based on diversity. And finally, activating the selected LoRA module by adopting a Top-k strategy based on the MoE architecture, and finally determining the LoRA module which is most adaptive to the downstream task in combination with performance feedback after integration with the basic model. According to the method, a large number of LoRA modules can be effectively constructed, the module most adaptive to the downstream task is selected, and the accuracy of the model in processing tasks such as mathematical reasoning is remarkably improved.

The invention relates to the field of artificial intelligence, in particular to a heterogeneous GPU-oriented large model reasoning platform adaptation method and system. The method comprises the following steps: continuously and dynamically collecting and maintaining all available GPU resources in a server cluster and attribute and state information of a deployed reasoning acceleration framework instance through proxy service deployed in each server node; providing a single entry point to receive a reasoning service request, and analyzing the request to determine a large model needing reasoning; according to the analyzed large model needing reasoning and the currently aggregated resource information, selecting a reasoning acceleration framework instance which is optimal in load state and compatible for the request through a decision algorithm, and routing the task to the reasoning acceleration framework instance; heterogeneous response formats returned by different inference acceleration framework instances are converted and packaged into a unified standardized format, and then the unified standardized format is returned to a requester. Scattered and diversified computing resources are effectively utilized, and dependence on a single hardware manufacturer is reduced.

The invention discloses an efficient large language model adaptation method based on server-free edge computing. The method comprises the steps that an LLM reasoning task is divided into four computing stages of lightweight preprocessing, shallow reasoning, deep reasoning and lightweight post-processing; carrying out compression optimization on the model through pruning, quantification and hierarchical optimization technologies; on the basis of an MOSEC server-free scheduling framework, a sensitive index CSI and a real-time load state are calculated in combination with a layer-device mapping table, and a scheduling decision is dynamically generated by taking minimization of total delay as a target; and finally executing cloud edge collaborative reasoning according to a decision result. According to the method, the calculation and communication overhead is effectively reduced, and the efficiency and the resource utilization rate of large language model reasoning in the edge environment are improved.

The invention relates to an industrial protocol adaptation method, system and device based on dynamic feature recognition and a medium. The method comprises the following steps: acquiring original communication data of industrial equipment; performing feature extraction on the original communication data to obtain a depth feature vector; performing feature matching on the depth feature vector in a known protocol feature library to obtain a protocol matching result; if the protocol matching result is the unknown protocol parameter, marking the depth feature vector as unknown to obtain an unknown feature vector; clustering all the unknown feature vectors based on the protocol matching result to obtain candidate protocol groups; and for each group of candidate protocol groups, generating a new protocol analysis rule set. The method can support parallel operation of multiple communication protocols, is suitable for different manufacturer devices, automatically classifies unknown protocols through a machine learning algorithm, and enhances self-learning and self-adaptive capabilities.

The invention relates to the technical field of computers, and provides a cross-platform installation package automatic adaptation method and system, and the method comprises the steps: obtaining a unified description file and an operating system fingerprint of each target platform; based on the operating system fingerprints of the target platforms and the installation package adaptation parameters, matching is carried out, and dependency trees of the target platforms are generated; and generating an adaptive installation package based on the dependency tree of each target platform and the installation package adaptation parameter. According to the cross-platform installation package automatic adaptation method, specific parameters and shared parameters of the platform are managed in a centralized mode through unified description files, environment intelligent perception is achieved and the dependency tree of the platform is generated in combination with dynamically obtained operating system fingerprints, and the adaptive installation package is generated through analysis of the installation package adaptation parameters of the platform and the dependency tree. The problems of low adaptation efficiency, high configuration error rate, high maintenance cost and the like are solved; in addition, a unified construction management mechanism realizes cross-platform centralized management and control, and the accuracy and efficiency of cross-platform installation package adaptation are improved.

The invention relates to the field of new-generation information systems, and discloses an operating system adaptation method for accessing an IMS system terminal, which comprises the following steps of: acquiring network state information and hardware resource state information of the IMS system terminal, constructing a game model of bandwidth resource allocation on the basis of the network state information and the hardware resource state information, and determining a video transmission strategy adapted to a WebRTC communication application; and on the basis of the obtained video transmission strategy, further combining with the hardware resource state information to establish a multi-resource constraint optimization model, and calculating a hardware resource optimal allocation scheme in a terminal operating system for supporting the implementation of the video transmission strategy. By adopting a cross-layer state sensing and scheduling cooperation mechanism, the technical effects of multi-level information synchronization and dynamic scheduling in the video transmission process are achieved. Compared with a scheme of performing strategy adjustment only based on single-layer network feedback in the prior art, the problems of low response speed and single adaptive scene are solved.

The invention provides an AI flow chart driven test step dynamic adaptation method, device and equipment, and relates to the technical field of software testing. The AI flow chart is subjected to semantic analysis to obtain a plurality of executable operation tuples, a dynamic equipment capability library is updated in real time through technical parameters and real-time states of all the matching and testing instruments, then dynamic matching and mapping are carried out, a first mapping relation between the executable operation tuples and the matching and testing instruments is established, and the dynamic equipment capability library is updated in real time through the technical parameters and the real-time states of all the matching and testing instruments. According to the second mapping relation between the signal features and the port of the matching and testing instrument, the AI flow chart is executed based on the executable mapping scheme, automatic matching and mapping of all the testing steps in the AI flow chart are achieved, manual participation is not needed, the problem that the testing efficiency of AI flow chart driving is low is solved, and the equipment testing efficiency of AI flow chart driving is improved.

The invention provides a gear tooth surface precision detection method based on a deep learning network, a medium and equipment, and mainly solves the problem of insufficient model generalization ability caused by scarcity of real labeled data and domain difference in an existing method. The method comprises the following steps: collecting a plane workpiece image marked with a roughness level as a source domain data set, and pre-training an intelligent detection model; based on a gear shaping machining meshing motion relation, establishing a tooth surface morphology geometric simulation model, and generating virtual gear tooth surface images of different roughness grades; extracting a real gear tooth surface area image as a target domain sample; a domain adaptation method based on clustering optimal transmission is adopted, the feature distribution difference between a source domain and a target domain is calculated and minimized, model parameters are optimized, and self-adaptive migration of a feature space from a plane image to a gear tooth surface is achieved. According to the method, virtual sample generation and a domain adaptation technology are combined, so that the data acquisition cost is effectively reduced, and the detection precision and the model generalization ability under the small sample condition are remarkably improved.

The invention discloses a multi-mode identification industrial protocol adaptation method and a terminal, and the method comprises the steps: extracting a first multi-mode feature of a to-be-identified message, and determining a protocol type of the to-be-identified message according to the first multi-mode feature; whether a protocol with the same protocol type as the message to be identified exists in a preset protocol library or not is judged, if yes, a preset protocol template corresponding to the protocol with the same protocol type is selected, protocol conversion configuration is conducted on the message to be identified through the preset protocol template, and if not, candidate protocol templates corresponding to the message to be identified are created; and judging whether information that the candidate protocol template completes manual verification and correction is received, if so, obtaining a final protocol template according to the information, and performing protocol conversion configuration on the to-be-identified message through the final protocol template. According to the method, dynamic identification and matching of various industrial protocols can be supported, the workload of manual configuration is reduced, multi-source heterogeneous data in a multi-protocol scene can be rapidly and accurately identified and converted, and convenience is provided for interaction between devices.

The invention provides a low-delay video stream code rate adaptation method and system based on off-line element reinforcement learning. The method comprises the following steps: constructing an off-line expert track data set; constructing a code rate adaptation strategy network, and outputting a downloading code rate selection result and a network throughput correction value by taking a video stream transmission state as input; and adopting a meta-implicit Q learning algorithm to perform double-layer alternate optimization of an inner layer and an outer layer on the policy network, the inner layer updating policy parameters of the policy network based on the offline expert trajectory data set, and the outer layer further adjusting the policy network parameters through meta reinforcement learning to improve the adaptability of the policy network to unseen tasks. According to the method, multiple expert algorithm experiences are integrated through offline reinforcement learning, high generalization and adaptive ability can be realized in multiple target time delays and complex network environments without online trial and error exploration, and finally, high-quality and smooth playing of video streams in a low-time-delay live broadcast scene is supported.

The invention discloses an extensible standardized micro optical array module multi-scene adaptation method, and particularly relates to the technical field of array module multi-scene adaptation, and the method comprises the steps: carrying out the periodic structure calibration of a micro optical module, and precisely obtaining the space imaging deviation and time sampling delay information between the modules; and obtaining a microscopic parallax deviation accumulation coefficient and a signal synchronization delay out-of-control coefficient, constructing a dynamic dislocation fusion obstacle evaluation model based on the coefficients, and outputting a dynamic dislocation fusion failure obstacle index, thereby realizing quantitative evaluation and failure grade division of the collaborative stability of the multi-module system. A reference basis with extremely high operability is provided for subsequent control and scheduling; furthermore, by recording obstacle trend sequences under different operation situations, clustering and attribution modeling are carried out in combination with environment state vectors, a scene label-obstacle trend mode association library is established, and a scheduling strategy mapping mechanism is introduced, so that cross-scene rapid adaptation and stable operation are realized.

The invention discloses an advertisement creativity dynamic generation and adaptation method and device, equipment and a medium. The method comprises the following steps: constructing a media characteristic knowledge graph for storing context characteristics, content format specifications and user portraits of different media platforms; carrying out multi-modal semantic analysis on the input initial creative material and extracting semantic elements; inputting the knowledge graph and the semantic elements into a generative adaptive model for processing, and outputting adaptive creative content which is matched with a target media platform in context and format and retains core semantics; putting the adaptive creative content on the target media platform and collecting deep behavior data generated by interaction between the user and the adaptive creative content; and performing iterative optimization on the generative adaptive model based on the depth behavior data, and reversely updating the media characteristic knowledge graph according to the optimization. According to the invention, deep and real-time intercommunication between creative design and media characteristics is realized, and the propagation effectiveness and brand consistency of advertisement creativity are improved.

The invention discloses a multi-mode graphical interface self-adaptive adaptation method and system based on environment perception, and relates to the technical field of graphical interface adaptation. The method comprises the steps that environment detection is conducted on a target display platform; rendering engine matching and rendering optimization are carried out based on the system environment information, a rendering engine module is constructed, and rendering optimization is completed based on balance of performance and power consumption; and seamless switching of the graphical interface is carried out by utilizing the rendering engine module according to the input and output modes of the user, and abstract compatibility of interface elements is carried out by adopting a virtualization technology in the switching process. The technical problems that in the prior art, a graphical interface is poor in compatibility under different platforms and input modes, and performance and power consumption are difficult to balance are solved, and the technical effects that multi-mode graphical interface self-adaption is achieved, and the compatibility and performance of the graphical interface under the different platforms and input modes are improved are achieved.

The invention relates to the technical field of computer vision, in particular to an adaptation method and system for continuous testing for differences between construction scene domains. According to the method, the similarity between Gram matrixes between adjacent domains is calculated, an elastic adjustment factor is set, the elastic adjustment factor is utilized, different weights are given to strong data enhancement and weak data enhancement, an elastic data enhancement strategy is provided, and an enhanced target domain image data set is input into a teacher model; updating the pseudo-tag by combining the elasticity regulation factor to obtain an elastic pseudo-tag; and inputting the target domain image data set into the student model to obtain a prediction result, constructing a global elastic symmetric cross entropy loss function based on the elastic adjustment factor, the cross entropy loss and the reverse cross entropy loss, updating student model parameters through the loss function, updating teacher model parameters, and finally obtaining a target model. According to the method, the construction scene monitoring model can adapt to complex domain changes when continuously learning test data, and the prediction result precision of the model in different environments is improved.

The invention discloses a rail transit construction safety identification model dynamic adaptation method and system. The method comprises the following steps: initializing an urban rail transit engineering construction potential safety hazard identification model and new scene training parameters; reading an urban rail transit engineering construction potential safety hazard identification image and a corresponding label in a new scene, and performing data enhancement operation; selecting a historical scene similar to the current new scene in feature, and setting a weight parameter of a new scene model by taking a corresponding encoder weight as an initial value; carrying out image coding and category prediction on the enhanced image data; calculating a total loss value comprehensively considering classification accuracy and model parameter stability through a multi-task optimization objective function; selecting specific parameters in the image encoder to update through adaptive methods such as a dynamic expert hybrid mechanism and dynamic low-rank adaptation; the problems that an existing continuous learning method is poor in cross-scene adaptability and insufficient in sustainable learning ability in urban rail transit engineering potential safety hazard identification can be effectively solved.