582 results about "Decision-making" patented technology

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as agent autonomous decision making, financial science and technology and medical health, and discloses a multi-modal hierarchical feature fusion and decision making method, device, equipment and medium, and the method comprises the steps: obtaining vision, language and motion data, and carrying out the hierarchical feature extraction to generate a multi-modal initial feature set; feature importance is analyzed, screening and dimension reduction are carried out, and screened multi-modal features are obtained; performing semantic enhancement on the screened multi-modal features to generate multi-modal semantic enhancement features; executing cross-modal attention fusion on the multi-modal semantic enhancement features to obtain cross-modal fusion features; and inputting the cross-modal fusion features into a semantic reasoning network to generate a decision result. According to the method, through multi-level screening dimension reduction, semantic enhancement and cross-modal attention fusion, the model can accurately utilize key feature relationships among multi-modal data, redundant interference is reduced, and semantic reasoning accuracy and decision-making efficiency are improved.

The invention relates to a remote-control system of a stratospheric airship and a control method thereof. The control system is divided to four levels: a decision-making level for executing the selection and switch of a remote-control flight mode according to the actual working state data, and sending a corresponding mode command; a planning level for dynamically planning and generating an airship motor drive command according to the mode command sent by the decision-making level; an executing level for realizing the drive control to an airship motor according to the motor drive command of the planning level; and a perceiving level for finishing the real-time monitoring and data collection of the airship working state, and providing the original data of the airship working state to the decision-making level. The four-level control system architecture design of the control system is capable of simplifying the control device, improving the control efficiency, and fundamentally solving the problems of the function, efficiency and reliability of the near space airship ground and airborne control system.

The invention relates to the technical field of artificial intelligence, and discloses a multi-agent social network simulation method and system based on a cognitive inference chain, and the method comprises the steps: initializing a multi-agent system comprising a social environment engine, a user portrait engine and a cognitive inference engine, executing a multi-agent social network simulation cycle of a preset round of iteration, in each iteration round, the social environment engine pushes social information to the intelligent agent as external stimulation and activates the intelligent agent to execute an independent decision, the cognitive state of each dimension in the cognitive reasoning chain is updated through large language model reasoning, corresponding social behaviors are generated, and the social behaviors and corresponding cognitive state tracks are recorded; and periodically analyzing historical records to optimize influence coefficients among all cognitive dimensions of the cognitive inference chain, and adjusting an inference strategy of a preset large language model. The simulation of the cognitive process of the intelligent agent is a transparent and traceable evolutionary process, and the complete and understandable simulation of the'observation-cognition-behavior 'cycle is realized.

The invention relates to the field of bill auditing, and particularly provides a financial bill auditing decision-making method and system, a terminal and a medium, and the method comprises the steps: collecting multi-mode finance and tax data including texts, images and structured data, and carrying out the preprocessing and alignment; carrying out feature extraction on the preprocessed data by using a multi-modal large model, carrying out feature fusion by using a dynamic weight distribution algorithm based on the credibility of each modal, the service priority and historical feedback, and obtaining a risk probability through a risk identification neural network; the risk probability is compared with a preset threshold value and rule in an auditing rule base, automatic passing is triggered, after risk abnormity is recorded, passing is conducted, and auditing actions such as manual auditing or starting of a high-risk emergency plan are pushed; and finally, adjusting model weight parameters according to manual feedback information to realize system self-optimization. The auditing decision-making efficiency is improved, and the accuracy and the service adaptability are improved.

The invention discloses a customs-used domestic large language model retrieval enhancement generation method, which comprises the following steps of S1, constructing a GraphRAG, and implementing dynamic relationship deconstruction on multi-modal data such as customs announcements, enterprise customs declarations and international provisions based on the deep semantic understanding capability of a large language model; s2, dividing laws and regulations sub-communities based on a Leiden algorithm, combining a Leiden community discovery algorithm with laws and regulations effectiveness analysis, and constructing a triad effectiveness map of laws and regulations clauses-revision events-effective areas; s3, hierarchically retrieving a framework, and decomposing a retrieval process into three-level probability decisions; and S4, establishing a knowledge graph full-stack system architecture of the multi-source heterogeneous data. According to the method, a GraphRAG technology is taken as a core carrier, and breakthrough of a customs complex knowledge scene is realized through double innovation paths: firstly, a graph structure retrieval enhancement model adaptive to customs business characteristics is constructed, and secondly, a dynamic governance mechanism of a Leiden community discovery algorithm optimization regulation system is introduced, so that the decision reliability of a customs intelligent supervision system is improved.

The invention provides a multi-modal information fusion feeding decision-making system and method for chicken breeding behavior recognition, and the method comprises the steps: collecting a multi-source heterogeneous data set, and extracting a primary fusion feature vector; constructing a triple knowledge graph; mining implicit association rules of the ingestion frequency and the body temperature; performing secondary fusion on the primary fusion feature vector and an implicit association rule to generate an intermediate decision feature; and generating a feeding decision instruction through the pre-training decision model and the expert rule base. According to the method, the knowledge graph is constructed, GNN reasoning is utilized, and a manual preset rule static mode is replaced; performing secondary feature fusion, generating intermediate decision features by combining primary fusion features and implicit rules, and then combining a pre-training model and an expert rule base, ensuring decision real-time performance, integrating domain knowledge, outputting accurate adjustment parameters, realizing full-link intelligence, improving the accuracy and adaptability of breeding chicken feeding decisions, and improving the accuracy and adaptability of chicken feeding decisions. Therefore, dynamic requirements of complex breeding scenes are met, and chicken flock health and breeding efficiency improvement are promoted.

The invention relates to the technical field of semantic decision management and control, in particular to a multi-modal semantic alignment intelligent decision and risk management and control system, which comprehensively and accurately captures cross-modal semantic association through multi-modal semantic alignment processing so as to generate a plurality of possible reasoning chains with reliability and interpretability. And combining dimensions such as knowledge conflicts and historical risks, calculating reasoning overlapping values to evaluate inter-chain association, realizing multi-dimensional and multi-angle analysis of potential risks, bringing risk fingerprint values and the reasoning overlapping values into a quantitative calculation framework of decision response values, dynamically setting a decision response threshold value, and realizing quantitative calculation of the risk fingerprint values and the reasoning overlapping values. The system can flexibly trigger emergency, early warning or monitoring response according to different risk levels, and refinement and differentiation of risk management and control are realized. The mechanism not only ensures timely disposal in a high-risk scene, but also avoids resource waste in a low-risk scene, and effectively balances risk prevention and control and execution efficiency.

The invention provides a large model thinking chain knowledge distillation method and system based on anti-factual reasoning, and belongs to the technical field of artificial intelligence and natural language processing. Comprising the steps that minimum semantic disturbance is applied to a key position of an original reasoning task data set, and an anti-fact problem is constructed; based on the original problem and the anti-fact problem, calling a large language model to generate a multi-view reasoning chain; a high-loyalty distillation data set is screened out through causal intermediary analysis, and thinking chain consistency optimization training is carried out on the large language model; and deploying the large language model after optimization training to a downstream task. According to the method, by analyzing the anti-fact space of the input problem and quantifying the causal contribution of the inference chain to the final decision, the causal reliability and logic consistency of the large model thinking chain can be improved, and then effective data support is provided for downstream inference application.

The invention discloses a coordinated control method and system for comprehensive energy multi-agent coordinated group control and autonomous decision making. The method comprises the following steps: establishing a relaxation strategy of an agent corresponding to a distributed power supply; the main coordinator iteratively solves to determine a global coordination signal according to the global operation data of the integrated energy system and the operation data of each agent by taking the lowest total cost, the highest system energy efficiency, the minimum carbon emission and the minimum global penalty coefficient meeting the global relaxation constraint as a relaxation global optimization target; the minimum operation cost of each intelligent agent, the minimum response deviation to a global coordination signal and the minimum deviation between an actual state and a reference state are taken as control targets; establishing an electricity price autonomous decision-making model of each agent and a response power boundary autonomous decision-making model for a global coordination signal; and based on the control target of each agent, the electricity price autonomous decision-making model and the response power boundary autonomous decision-making model, predicting to obtain a control target value of each agent, and realizing balance between a global optimization target and local autonomy.

The invention relates to the technical field of integrated supply chain plan collaboration, and discloses an integrated supply chain plan collaboration intelligent decision-making method and system, and the method comprises the steps: decomposing a single prediction value into a commitment layer signal and an option layer signal, generating a differential signal, generating an agility index through issuing a periodic probe signal, and carrying out the calculation of the agility index; the method comprises the following steps: establishing an agility index of each option signal, calculating a collaborative entropy index based on the time value of each option signal, and finally, carrying out closed-loop dynamic adjustment on a decision activation threshold according to the agility index and the collaborative entropy index. Therefore, the problem of plan vulnerability caused by the fact that a traditional planning system depends on a single deterministic predicted value is solved, the whole collaborative system internally accommodates uncertainty at an information source, the operation stability of the collaborative system does not excessively depend on prediction accuracy any more, and mode conversion from passive prediction execution to active preparation response is achieved.

The invention discloses a management decision-making method and system based on a knowledge base construction technology. The method comprises the following steps: performing sequential relationship extraction on multi-source financial data to obtain a sequential relationship set related to query content; constructing an event-entity incidence matrix corresponding to the time sequence relation set; according to the time sequence relation set and the event-entity incidence matrix, constructing a dynamic knowledge graph; determining causal effect parameters in the causal graph structure by adopting a dual machine learning model; constructing a structural causal model according to the causal graph structure and the causal effect parameters; and generating an anti-fact prediction result by using the structural causal model, and generating a decision scheme corresponding to the query content based on the anti-fact prediction result. The technical problem that decision information including accurate causal basis and prospective simulation information cannot be generated due to the fact that the causal relationship between financial data is difficult to determine and the intervention effect cannot be dynamically deduced in a related management decision method is solved.

The invention discloses a personalized computer-aided decision-making method and system fusing multi-modal data, and relates to the field of personalized computer-aided decision-making, and the method comprises the steps: mapping multi-source heterogeneous modal data to a unified semantic embedding space, and obtaining a multi-modal unified representation vector set; carrying out three-layer progressive fusion on a feature layer, a situation layer and a decision layer of the multi-modal data to generate a global decision context vector; based on a cross attention mechanism, outputting a fused context sensing personalized vector; based on the behavior cloning model, outputting probability distribution on all decision options; according to the user feedback operation data, generating a user personalized decision strategy and performing dynamic optimization; and generating a structured decision report containing visual traceability information based on the hierarchical fusion process and decision reasoning logic. End-to-end intelligent generation from multi-source heterogeneous data to personalized decisions is realized, and a standardized process is converted into personalized customized decisions.

The invention discloses a reasoning optimization method and device for a code generation large model, equipment and a medium, and relates to the technical field of model reasoning, and the method comprises the steps: in the reasoning process of a code generation task of a target code generation large model, executing a multi-granularity uncertainty quantification step in parallel every time a new Token is generated, obtaining a multi-granularity uncertainty score; constructing a state space vector, and utilizing a preset reinforcement learning strategy network to evaluate the selection probability of a plurality of preset reasoning optimization strategies based on a preset smooth decision mechanism so as to determine a target reasoning optimization strategy; if the strategy is a preset reasoning acceleration strategy, optimization processing is carried out through speculation decoding; if the strategy is a preset exploration optimization strategy, performing optimization processing by using a preset multi-path sampling technology and a preset knowledge enhancement technology; if the strategy is the preset fuzzy processing strategy, taking the plurality of candidate outputs as target reasoning outputs for optimization processing; and evaluating the decision effect according to the reasoning result to optimize the preset reinforcement learning strategy network.

The invention relates to the technical field of artificial intelligence, and discloses a reinforcement learning decision optimization method, system and device based on a causal large language model, and the method comprises the following steps: initializing an intelligent agent and a strategy network thereof; obtaining track information of a historical sequence decision generated by interaction; extracting causal variables from the trajectory information by adopting a large language model, and constructing a structural causal model; obtaining an agent strategy-driven causal intervention mechanism, and dynamically correcting a causal relationship in the structural causal model; according to a task-related causal chain extracted from the corrected structural causal model, generating a semantic sub-target corresponding to a causal relationship; designing a multi-modal reward function fused with semantic similarity; and updating the strategy network by adopting the obtained sub-targets and rewards. According to the method, the problems of low learning efficiency, insufficient adaptability and lack of effective reasoning ability of the reinforcement learning agent in a complex environment in the prior art are solved, and the method has the characteristic of high decision-making efficiency in a dynamic environment.

The invention provides a multi-modal big language model inference engine which comprises a pre-training big language model adaptation layer, a pollution remediation knowledge graph enhancement module, a multi-modal information understanding and fusion unit, a technology implementation optimization module and an inference chain and decision interpretation generator. The system adopts a hierarchical modular design, all components communicate through a standardized API interface, efficient and stable data circulation is ensured, and the problems of knowledge application limitation, opaque reasoning process, difficulty in professional knowledge fusion, knowledge updating lagging, limited multi-modal data processing capability and the like in the existing contaminated site remediation decision process are solved.

The invention relates to the technical field of multi-agent systems, and discloses a multi-agent and multi-mode alignment interaction method and system for a review scene, and the method comprises the steps: mapping real-time multi-mode data into a local belief vector through a heterogeneous agent cluster, and calculating the spatial divergence to determine a global cognitive entropy; interaction time sequence gating logic based on entropy flow differential is established, and active alignment is triggered when the global cognitive entropy is larger than a convergence threshold value and a first-order derivative is smaller than a damping threshold value; according to the method, through entropy flow differential control and an active semantic disturbance closed loop, decision oscillation of the multi-agent system in the non-cooperative environment is eliminated, and the reliability of the multi-agent system in the non-cooperative environment is improved. Mathematical convergence of the belief state in an adversarial disguise scene is realized, and the system identification precision in a complex interactive game is improved.

The invention relates to a multi-competency intelligent scoring method and system, and belongs to the technical field of intelligent evaluation and talent evaluation. The method comprises the following steps: acquiring multi-source response data of a target evaluation object, executing evidence sufficiency and consistency analysis for each competency dimension, generating a judgment state identifier, and constructing a structural description; performing responsibility distribution on the evaluation evidence, mapping the evaluation evidence into a corresponding evidence model and generating a priority constraint relationship; performing time sequence consistency and stability analysis on the evidence model, and constructing a time sequence constraint rule to prohibit unconstrained score backtracking; and carrying out resolution processing on the continuous undetermined dimension, generating a scoring result through a preset risk scoring rule and a conservative determination strategy, and synchronously generating a scoring basis path and confidence information. According to the method, standard management and control of multi-source evidences and accurate constraint of the whole scoring process are realized, the scoring accuracy and traceability are effectively improved, and reliable support is provided for multi-scene talent evaluation decision.

The invention relates to the technical field of intelligent decision rule base construction, and discloses an intelligent decision framework construction method and device based on a dynamic ontology. The method comprises the following steps: acquiring original data streams from a plurality of heterogeneous data sources in real time, and constructing an initial dynamic ontology structure through a dynamic ontology modeling unit; performing semantic annotation and relation extraction on the original data flow by using the structure to generate semantic enhanced data; generating a candidate decision rule set based on the semantic enhancement data, and obtaining a verified rule set through consistency verification and conflict detection; calculating the adaptability score of the verified rule set in combination with the real-time environment data, and screening out an optimal decision rule subset according to the score and a preset threshold value; the method is integrated into an intelligent decision rule base, and an optimization loop is triggered based on an update state of the base. According to the method, the data utilization efficiency and the rule quality are improved, the rule base can dynamically adapt to the environment, and decision effectiveness and reliability are enhanced.

The invention discloses an automatic driving safety operation system integrating environmental perception and decision reasoning. The system generates and dynamically updates a security risk map covering an operation area by fusing real-time environment perception, historical operation data and traffic management information. On the basis, the collaborative safety decision-making module further carries out behavior modeling and intention prediction on other traffic participants, and in combination with map risks and traffic instructions, a driving strategy is actively adjusted under a dynamic game framework. According to the invention, the safety control is improved from passive response to an active mode of behavior pre-judgment and game dominance, and the safety and reliability of the operating vehicle in a complex environment and the cooperative capability of the operating vehicle and traffic management are obviously enhanced.

The invention discloses an auxiliary decision-making system for cognitive competence assessment of old people, which relates to the technical field of cognitive auxiliary decision-making and comprises an environmental noise acquisition and feature extraction module, a time sequence fluctuation analysis module, a dynamic threshold reconstruction and judgment module, a cross-domain consistency calibration module, a multi-modal synchronous verification module and a dynamic threshold regulation and control module. According to the method, the environmental noise is dynamically analyzed, so that the defects of a fixed noise suppression threshold and a static feature extraction model in the traditional technology are overcome. A sound field energy distribution model is constructed in real time and time sequence fluctuation analysis is combined so that a noise fluctuation interval can be identified, a voice starting point detection threshold value is dynamically adjusted and evaluation precision is enhanced. Through multi-mode synchronous verification combining a reaction time curve and a facial movement track, a time offset error is corrected, a judgment standard is adjusted in real time according to noise changes, misjudgment and virtual high risk early warning are avoided, and therefore the accuracy and reliability of cognitive ability assessment of the old are improved.

The invention discloses a manufacturing task autonomous negotiation and execution method based on a large language model agent, and the method comprises the steps: constructing a production scheduling agent, an equipment management agent, a material distribution agent and a quality control agent, analyzing a natural language task instruction through the production scheduling agent, and decomposing the natural language task instruction into subtasks; each agent calculates a utility value based on the load rate, the resource matching degree, the estimated completion time and the historical success rate, and performs structured negotiation to achieve a task allocation consensus; a prediction-check-rollback architecture is adopted to generate an action instruction sequence, the sequence is compiled into a time Petri network transition sequence, and reachability verification is carried out based on hard security constraints; production environment data is collected in real time to trigger anomaly detection and re-negotiation, and a formalized security verification and causal anti-factual reasoning parameter updating mechanism is introduced. According to the method, unstructured instruction understanding, autonomous task planning, multi-agent collaborative decision and closed-loop optimization are realized, and the problems of real-time performance, safety and interpretability of a large language model in manufacturing control are solved.

The invention relates to the technical field of multi-agent cooperative control, and discloses a multi-agent autonomous decision-making method based on deep reinforcement learning. The method comprises the steps of synchronously detecting an initial collaborative state of a cluster, performing joint situation assessment, and judging a collaborative operation mode according to a quantitative situation. And analyzing the capability of each agent and the real-time task load, and constructing a distributed task knowledge graph. And utilizing the atlas to drive a deep reinforcement learning network, coupling computing resource allocation and a task path, and generating a preliminary behavior strategy of each agent. And performing cluster-level conflict detection and iterative negotiation adjustment, and finally issuing an executable action instruction sequence. According to the method, integrated optimization of resource allocation and action paths is realized, the situation understanding and negotiation mechanism is enhanced through the knowledge graph to guarantee the collaborative consistency, and the collaborative decision-making efficiency and task execution robustness of a multi-agent system in a dynamic environment are improved.

The invention discloses a multi-agent collaborative visual navigation reasoning enhancement method and system, and the method comprises the steps: an inference agent analyzes input data, and generates structured output containing a reasoning process and an initial answer; independently analyzing the input data by the evaluator agent, generating an evaluation answer and reviewing the output of the inference agent; comparing whether the initial answer is consistent with the evaluation answer, and if yes, outputting a final answer; if not, calling a bifurcation feedback agent, and generating a core bifurcation point abstract and a visual evidence verification plan; and calling a final decision maker agent to directionally retrieve the video evidence according to the visual evidence verification plan, and outputting a final answer. Four roles of an inference person, an evaluator, a divergence feedback person and a final decision maker are introduced, a complete'generation-evaluation-feedback-optimization 'decision link is constructed, and the cognition and inference accuracy in a complex scene is remarkably improved.

The invention relates to the technical field of rural space governance and intelligent decision making, in particular to a rural space planning method and related equipment, and the method comprises the steps: obtaining multi-source data of a target rural, the multi-source data comprising land data, population data and rural economic data of the target rural; performing fusion processing of space-time alignment on the multi-source data to obtain fusion data; constructing a spatio-temporal knowledge graph based on the fused data, and extracting entity attributes and spatial relationships; and inputting the entity attributes and the spatial relationship into a pre-trained planning optimization model, iteratively optimizing a rural space matching strategy according to rigid constraints, and outputting an optimal planning result, the technical problems that in the prior art, rural planning faces data splitting, element analysis isolation, planning scheme staticization and actual development requirements are disjointed, the decision making process depends on subjective experience, and economic-ecological benefit tradeoff is difficult to quantify are effectively solved.

The invention provides a hierarchical ethical adaptive method and system based on a development stage, and the method comprises the steps: collecting the multi-modal behavior data of a user, and recognizing the cognitive development stage of the user; constructing an ethical rule knowledge graph with a hyponym inheritance and context activation mechanism; fusing user acceptability and ethical conflict strength, and performing strategy balance based on a game model; calling the generative language model and the hierarchical template library to generate matched ethical feedback content; different culture expression styles are adapted through a culture migration network; and a user stage transition window is predicted based on the cognitive evolution trend, and an intervention mechanism and strategy adjustment are triggered. The system supports personalized ethical guidance strategy configuration, is compatible with various user types and cross-culture situations, realizes adaptive optimization and expression intellectualization of ethical decision, and improves user understanding degree, acceptability and ethical guidance effect. The method is suitable for multiple scenes such as education guidance, value intervention and man-machine ethical interaction.

The invention provides an intelligent decision-making method and system based on adaptive cross-modal attention, and relates to the technical field of data processing, and the method comprises the steps: obtaining multi-modal data; performing preprocessing on the multi-mode data; extracting original semantic features in each mode from the preprocessed multi-mode data through a deep neural network; calculating the cross-modal attention of each original semantic feature; according to the cross-modal attention, performing cross-modal fusion on each original semantic feature to obtain a fusion feature; and according to the fusion features, intelligent decision making is carried out through a full-connection network. According to the cross-modal attention mechanism, information of different modals can be dynamically weighted according to the importance of each modal, and when the data quality of a certain modal is poor or is completely missing, the weight of the modal can be adaptively reduced, so that the interference of irrelevant or repeated information is avoided, the feature fusion effect is enhanced, and the decision accuracy is improved.

The invention relates to an irrigation decision-making method and system based on a crop model and deep reinforcement learning. The method comprises the following steps: creating a virtual environment for simulating crop growth through a decision support system crop growth model after parameter calibration; then constructing a time sequence state matrix, and encoding the time sequence state matrix into a context vector concentrated with historical dynamic information; outputting irrigation actions of each decision-making day through a strategy network of a deep reinforcement learning agent based on a soft strategy-value algorithm; therefore, a decision support system crop model and a deep reinforcement learning technology are deeply fused, an intelligent irrigation decision system with biological rationality is constructed, and the problems of insufficient training data and environment distortion of an existing deep reinforcement learning model in agricultural application are solved.

The invention discloses a knowledge graph enhanced reasoning method and a knowledge graph enhanced reasoning system for high-risk field medical decision, which are used for strictly limiting the reasoning of a large language model within a high-quality knowledge graph range, so that the reliability is improved, and factual errors and'model illusion 'are fundamentally eliminated. Through an innovative constraint generation and traceability mechanism, each conclusion can be traced to a direct evidence in the knowledge graph, the interpretability is enhanced, and the complete transparency of the reasoning process is realized. The mixed query driven sub-graph retrieval method can deeply understand the intention of a user, accurately position and extract structured knowledge related to complex problems, improves the retrieval precision, and has a retrieval effect far better than that of traditional keyword or vector retrieval.

The invention provides a long-range visual question and answer and multi-modal reasoning task agent construction method. For input images and natural language questions, generating answers through orderly calling of an external tool set; the construction process of the intelligent agent is divided into three stages: a global planning stage: analyzing user requirements through a global planning navigator, and screening tool subsets; in the tool autonomous execution stage, visual details are supplemented, background knowledge is retrieved and accurate operation is carried out in the multi-step process by surrounding three types of tools through a tool autonomous executor, and a returned tool result is used for assisting the next step of decision making; in the response synthesis stage, the reasoning process is combed through a model response synthesizer, and user-friendly and clear-format responses are sorted; and finally, the model response synthesis module extracts and obtains answers from the reasoning track and outputs the answers. The problems of global planning insufficiency and visual context forgetting generally existing in complex task reasoning of an existing multi-mode large language model are solved.

The invention discloses a clinical intelligent decision-making method based on proxy workflow and a storage medium. Comprising the following steps: acquiring and preprocessing clinical information of a patient, and constructing a candidate disease set; and constructing a proxy directed workflow. In the retrieval stage, the diagnosis criteria corresponding to the candidate diseases are retrieved and aggregated from the diagnosis criteria library rechecked by the experts to form working memory. The preliminary diagnosis stage model node generates a preliminary candidate diagnosis set in combination with work memory and patient medical history and physical examination. And the final diagnosis stage generates a final diagnosis result based on the preliminary candidate diagnosis and the complete clinical information. And when the global confidence is lower than a threshold value, the model node pointedly checks an information source and updates reasoning and confidence. A successful reasoning track forms a demonstration set after manual auditing, the demonstration set is used for supervising a fine tuning model to obtain an initial strategy, multiple structured outputs are generated through grouping sampling, relative strategy updating is carried out in combination with reward signals and reference strategy regularization constraints, and optimization and stable improvement of the model diagnosis capability are achieved.