42135 results about "Operations research" patented technology

A method is provided for estimating the number of distinct tokens in a text stream using a modified text-to-text variational autoencoder (T5VQVAE) model. The method includes receiving a continuous input of a text stream; dynamically maintaining a buffer that stores a probabilistic subset of tokens from the text stream; calculating a sampling probability for each token based on a condition related to the current state of the buffer; updating the buffer based on the sampling probability to include or exclude tokens; encoding the buffered tokens into a latent space using the T5VQVAE model; and estimating the number of distinct tokens in the text stream based on the tokens in the buffer and the corresponding sampling probabilities.

The invention discloses a collaborative sharing method, system and device on an energy data chain fusing multi-party security calculation and zero knowledge proof and a storage medium, and belongs to the field of energy data security sharing and block chain privacy protection, and the method comprises the steps: executing desensitization processing on original data according to types, and generating a cryptographic commitment value and a data hash identifier; encrypting and fragmenting the desensitized data, registering a task structure body through an intelligent contract, and uploading the task structure body; and each node decrypts the ciphertext, executes calculation and generates a zero-knowledge proof, and submits the zero-knowledge proof to the block chain to be verified by the smart contract. According to the method, data privacy protection and result verifiability unification are realized, cross-mechanism cooperation efficiency is improved, an auditing system on a whole process chain is constructed, and safe sharing of energy data is guaranteed.

The invention relates to the technical field of intelligent data fusion deduction, in particular to a task planning method and system based on intelligent deduction, and the method comprises the steps: constructing a rule knowledge graph with weight annotation through integrating a real-time battlefield situation and a preset rule library, and carrying out the task planning through the real-time battlefield situation; a large-model intention understanding engine is used for analyzing a commander instruction to generate a five-dimensional intention vector, an initial scheme is generated by combining a double-engine mechanism of a rule hard constraint engine and a large-model soft constraint engine, and graph weights are dynamically adjusted through reinforcement learning to optimize a rule path. The system monitors node conflict frequency in real time, adopts a near-end strategy optimization algorithm to re-distribute weights for high-frequency conflicts, activates a new rule injection mechanism when the intention conformity is insufficient, and finally outputs a three-level structured scheme including a main scheme, an alternative scheme set and a risk early warning report, thereby realizing an intelligent closed loop from situation awareness to scheme generation. And the command decision-making efficiency and the combat collaboration are obviously improved.

The invention discloses an accounting data intelligent processing method and system for enterprise financial auditing, and relates to the technical field of accounting data intelligent processing, and the method comprises the steps: obtaining multi-mode enterprise financial data, and carrying out the preprocessing; carrying out multi-modal semantic understanding analysis on the unstructured text and image data; constructing an enterprise financial space-time knowledge graph containing time attributes; inputting into an anomaly analysis model, extracting spatial structure characteristics of the financial entity in the topological network, and extracting dynamic characteristics of the financial relationship evolved along with the time sequence; identifying an abnormal source, evaluating a systematic risk value and generating an abnormal propagation path; and integrating to generate a final audit report. According to the method, structured and bill images are fused, identifiers and time calibers are unified, abnormal source and propagation are positioned based on the space-time knowledge graph, closed-loop counter-knock and cross-period anomalies are identified, the auditing accuracy and coverage rate are remarkably improved, the workload of false report, missing report and manual recheck is reduced, and a traceable structured report is quickly generated.

The invention discloses a production process state monitoring scheduling optimization method based on real-time data acquisition, relates to the technical field of manufacturing process scheduling, and is used for solving the problem of insufficient real-time performance and stability of process scheduling. According to the method, a process monitoring mechanism based on real-time acquisition and closed-loop scheduling is constructed, a time sequence structured data frame is formed under a unified time reference, operation stability and quality offset characteristics are extracted, a data credible label and a current process state factor vector are generated, an optimized scheduling model is input, and task conflicts and resource bottlenecks are identified according to the data credible label and the current process state factor vector. According to the method, path compression and sequence adjustment are implemented in combination with scheduling priority mapping and a resource path diagram, scheduling deviation vectors are constructed through task response time delay and process blocking in operation, rules and parameters are triggered to be updated online and locally rearranged, and solidification is performed after verification in a prediction window, so that equipment idling and switching fragmentization in the production process are reduced, and the production efficiency is improved. And the real-time performance of the production process, the resource utilization rate and the system stability are improved.

The invention provides an intelligent short message scheduling method and device based on multi-dimensional dynamic optimization, and the method comprises the steps: obtaining the performance data of a plurality of short message channels, and calculating a channel health score based on a weight dynamic adjustment model; determining a scheduling strategy according to the priority identifier of the to-be-sent message, and performing channel screening and optimal matching; executing message sending and monitoring a sending state; terminal state detection is carried out on the failure message through operator base station signaling, and a decision tree model is applied to determine a retry strategy; and performing Huffman coding compression processing on the P2-level marketing messages which fail in retry, and performing batch sending in an idle window. According to the method, a comprehensive performance evaluation index and reward function model is also constructed, and parameter optimization is performed by applying a reinforcement learning algorithm. According to the invention, multi-dimensional dynamic channel scoring, intelligent retry decision making based on terminal state perception, batch processing with balanced cost-time efficiency and a closed-loop self-optimization system are realized, the short message delivery rate is obviously improved, and the invalid retry rate and the sending cost are reduced.

The invention provides a circuit board production yield root cause tracing method, which comprises the following steps of: acquiring process parameters, equipment states, environment variables and quality detection results of a whole production process, and constructing a multi-dimensional time sequence database; extracting a typical manufacturing process modeling unit through a sliding time window and dynamic time warping; establishing a cross-process dynamic causal relationship graph in combination with nonlinear Granger causal test, a structural equation model and a dynamic Bayesian network; an intervention and anti-factual reasoning method is applied, the causal effect and path stability under parameter disturbance of each process are evaluated, and the influence of a key causal path is quantified; according to the method, the accuracy of defect rate root cause positioning can be improved, and powerful support is provided for circuit board production process optimization and quality improvement.

The invention relates to the technical field of prediction and alarm, in particular to a school computer room data operation monitoring and early warning method and system, and the method comprises the following steps: obtaining real-time data, such as CPU utilization rate, memory occupancy rate and network delay, through a sensor, generating a parameter sequence after standardization, extracting a difference sequence through a sliding window, and recognizing the output trend risk of a continuous rising interval. And extracting a behavior frequency to generate an abnormal coefficient, carrying out weighted analysis on the coupling degree to obtain an early warning threshold, and generating an early warning instruction if the growth rate exceeds the threshold and lasts for three time slices. According to the method, difference characteristics are extracted through a standardized time sequence and a sliding window, behavior state association is analyzed in combination with a periodic frequency, coupling degree dynamic early warning is calculated, a continuous deviation trend is screened, a real-time threshold range is formed, abnormal confusion is reduced, occupation of irrelevant alarm resources is reduced, risk identification capability is improved, and stable operation of equipment is guaranteed. And hidden danger accumulation occurrence probability is reduced.

The invention relates to the technical field of communication data analysis, and discloses a communication network service health degree analysis method, system and device and a medium, and the method comprises the steps: obtaining real-time link state data, carrying out the quantitative analysis, and judging whether the link jitter exceeds a preset threshold value or not; when abnormality is detected, abnormal points can be positioned and the influence range of the abnormal points can be analyzed, so that the influence degree of nodes on the abnormal diffusion path can be identified; calculating a service quality index change value, and determining a starting point and a diffusion direction of chain degradation; by analyzing historical traffic data of the key nodes, a service traffic change trend and traffic dense area distribution characteristics are obtained. According to the method, the bandwidth allocation strategy is dynamically optimized, the bandwidth resources at the junction are reconfigured, and the regional service health degree is evaluated. And if the degradation is not effectively controlled, monitoring and detection mechanism parameters can be adaptively adjusted, and continuous analysis and control of link abnormity are realized, so that the network performance and the service quality are effectively improved.

The invention discloses a method for analyzing a matching degree between a demand and an output result based on text semantics. The method comprises the following steps: step 1, labeling a data set; step 2, technical document preprocessing; 3, training and predicting a single-parameter model; 4, integrating prediction results of the multi-parameter model; the method has the beneficial effects thatthe method is simple; deep learning and the NLP technology are applied to the field of project association degree calculation of enterprise project management for the first time. Calculating an association matching degree between the two projects according to project requirements and result description; the associated project positioning difficulty is effectively reduced; meanwhile, the demand side can be helped to quickly and efficiently locate high-quality projects adapting to the demand of the demand side; time and resource investment for achievement screening and matching are greatly reduced, the association matching degree between projects is calculated by means of text data of existing project achievement technical documents and project declaration guidelines, and then large enterprises are assisted in screening high-quality projects with the high matching degree in the project bidding and tendering link.

The invention discloses a territorial space planning method based on big data, and belongs to the technical field of territorial space planning. Comprising the following steps: step 1, constructing a territorial space knowledge graph; step 2, constructing a multi-dimensional territorial space planning solution space, identifying key constraints and determining an optimization path; 3, searching and generating an optimal planning scheme in a solution space; 4, performing quantification and grading processing on hard constraints and soft constraints in territorial space planning to realize multi-objective comprehensive optimization; step 5, performing multi-dimensional confidence evaluation and validity verification on the planning scheme generated by optimization; and step 6, realizing continuous optimization and adaptive evolution of territorial space planning.

The invention discloses a zero-trust network dynamic access control method based on an AI behavior portrait, and the method comprises the following steps: 1, collecting multi-source real-time behavior data during an access request; 2, constructing an AI behavior portrait engine based on historical data, inputting the integrated multi-source behavior data, calculating a behavior deviation degree through the AI behavior portrait engine, and outputting a risk score; 3, dynamic strategy decision making, wherein a decision making engine executes hierarchical control according to the risk score; 4, continuous session monitoring and real-time adjustment are carried out; step 5, when risk upgrading is detected in the session, degrading the session authority, limiting high-risk operation, terminating the session, and retaining evidence obtaining data; step 6, audit event generation and portrait updating; and step 7, strategy optimization closed loop. According to the method, the risk score is calculated in real time based on the AI behavior portrait, transition from static authorization to dynamic permission adjustment is realized, and internal threats such as voucher stealing and the like are effectively blocked.

The invention relates to a power distribution system optimization method considering a space-time game under vehicle-station-network interaction, and the method employs a double-layer optimization framework integrating the space-time game and dynamic electricity price, an upper decision maker is a power distribution system operator, and the power distribution system operation cost is minimized and the renewable energy consumption rate is maximized. Generating a dynamic electricity price signal in a time-sharing and partition manner; the lower-layer main body comprises an electric vehicle user and a charging station operator, and the electric vehicle user optimizes a charging time period and a charging station and forms charging demand distribution through a non-cooperative game based on a dynamic electricity price signal and a charging service charge by taking the maximum self-charging decision utility as a target; and a charging station operator optimizes and adjusts a charging service fee decision by taking charging demand distribution as input and taking revenue maximization as a target. Compared with the prior art, the method can collaboratively optimize the economic operation of the power grid and the consumption of renewable energy sources, promotes the efficient flow and balanced distribution of resources among regions, and provides decision support for the dispatching of the power distribution network.

The invention relates to a low-altitude logistics unmanned aerial vehicle path planning method, which comprises the following steps of 1, discretizing an urban low-altitude region into computable grid units, a dynamic-static obstacle classification modeling technology is combined to endow a static obstacle with a basic risk degree, a height attenuation factor is introduced to quantify the influence of a high-altitude obstacle on a low-altitude path, and a collision risk of a dynamic obstacle is dynamically evaluated through a time dimension risk degree prediction model; 2, based on an improved genetic algorithm, comprehensively considering flight time, height change, risk degree and dynamic risk cost through a multi-objective optimization objective function, and generating a globally optimal task allocation scheme; 3, generating a local optimal three-dimensional path considering meteorological conditions and dynamic obstacle influence by adopting an improved particle swarm algorithm; and 4, performing path planning to feed back actual flight time and energy consumption to task allocation for optimization.

The invention provides an AI-based laboratory equipment scheduling optimization method and system, and the method comprises the steps: firstly obtaining a state monitoring data set containing the characteristics of equipment operation power consumption, idle time length, environment interference factors and the like in real time, and then carrying out the multi-dimensional analysis of the state monitoring data set; generating an availability evaluation index set containing characteristics of equipment load fluctuation, maintenance period prediction, compatibility matching and the like, and an experiment task priority queue, performing cross decision analysis on the availability evaluation index set and the experiment task priority queue based on a preset dynamic resource allocation model, and obtaining a scheduling strategy set containing a task allocation path, a cooperative operation rule and a conflict resolution mechanism; scheduling strategy parameters are calibrated according to experimental task operation log data, an optimized execution instruction set is generated, instructions are fed back to an equipment control system to adjust the equipment state, a dynamic resource allocation model is iteratively updated according to execution feedback data, and efficient scheduling optimization of laboratory equipment is achieved.

The invention relates to the technical field of energy scheduling, and discloses a smart city energy dynamic scheduling system and method based on big data analysis, and the method comprises the following steps: the operation state of a transformer substation, the basic parameters of a charging pile and regional load prediction data are collected in real time, data cleaning and abnormal value filtering are carried out; constructing a complete power grid-charging facility dynamic information base; calculating the power supply margin of each region based on the capacity loss of the faulty transformer substation, establishing a weight scoring system of charging pile power adjustment in combination with the charging demand urgency, and determining the reduction or recovery priority of each charging load; and generating a charging pile power adjustment instruction through a multi-target optimization model, and iteratively correcting a power distribution scheme and generating a final scheduling instruction set by taking minimization of user satisfaction loss as a target while meeting the power grid capacity. According to the invention, by constructing a dynamic response mechanism and a multi-target collaborative optimization model, accurate and rapid regulation and control of the traffic load are realized in the scene of sudden power shortage of the power grid.

The invention relates to the technical field of granary ventilation and energy consumption optimization, in particular to an intelligent granary ventilation and energy consumption optimization decision-making method based on reinforcement learning, and the method comprises the steps: obtaining the internal and external multi-dimensional state data of a granary, respectively constructing a basic state vector, a space correlation feature vector and a grain condition change trend vector based on the multi-dimensional state data inside and outside the granary; combining the basic state vector, the space correlation feature vector and the grain condition change trend vector into a space-time feature tensor, inputting the space-time feature tensor into a space-time topology perception decision model based on a graph attention network, and outputting an optimal composite action vector for controlling ventilation equipment; and analyzing and converting the optimal composite action vector to drive ventilation equipment of the granary to execute control. According to the method, through fusion of space-time topology perception and deep reinforcement learning, accurate prediction and active regulation and control of the granary state are realized, and energy consumption of a ventilation system can be reduced to the greatest extent while grain storage safety is ensured.

A system and method for integrated optimization of autonomous electric vehicle fleets across transportation and electricity networks which employs artificial intelligence to dynamically allocate autonomous electric vehicles between mobility services and electricity grid services based on real-time conditions. The platform acquires data including energy mix forecasts, earth observation measurements, vehicle owner schedules, and emission-based route penalties to generate coordinated allocation decisions. Vehicle owners specify availability through a scheduling interface. The system optimizes vehicle utilization through a hierarchical optimization approach implementing mobility demand-side flexibility and electricity demand-side flexibility simultaneously. Multi-objective genetic algorithm optimization balances revenue generation, energy costs, emissions reduction, and battery health. The integrated approach maximizes value creation across both transportation and energy domains, reducing urban emissions while enhancing grid stability through coordinated management of distributed energy resources in autonomous electric vehicle fleets.

The invention relates to a data processing method and device based on big data and advertisement pushing, and the method comprises the following steps: obtaining the historical behavior data of a user on a multi-channel platform, constructing a dynamic interest label map according to the historical behavior data, and depicting a user interest evolution process. Combining with a social relation network to analyze an interest propagation path, forming a user social interest diffusion trajectory, and introducing a time decay weighting mechanism to generate a dynamic interest decay curve. According to the method, user interests and advertisement materials are subjected to semantic similarity matching, a personalized advertisement recommendation list is generated, an optimal advertisement putting strategy is determined through multi-target optimization configuration and comprehensive consideration of display positions, opportunities and forms, accurate and efficient advertisement pushing is achieved, and the problems that a traditional user portrait method often depends on a static label system, and the user experience is poor are solved. The dynamic characteristic that the user interest changes along with time is difficult to reflect, so that the advertisement recommendation content lags behind the real intention of the user.

The invention provides an intelligent supply chain management method and system for a front warehouse of a chain drugstore. The method comprises the following steps: collecting full-process event data of each medicine batch, and respectively marking unique batch identification codes for different storage positions; performing data cleaning and time sequence normalization processing on the collected original data; based on drug data of different batches and storage areas, respectively constructing digital twin models, and recording a life cycle trajectory of each batch of drugs; gathering and serializing supply chain key behavior events related to each storage area; and by using a life cycle mapping algorithm, carrying out real-time pairing on the medicine validity period curve and the dynamic inventory consumption curve, and respectively calculating the dynamic change trend of the critical intersection point of the remaining inventory and the validity period for different storage areas and medicine types. According to the method, preposition, accurate early warning and efficient visual traceability of the inventory-period-of-validity linkage risk can be realized, and the intelligence and compliance level of preposition warehouse medicine management of chain drugstores can be improved.

The invention relates to the technical field of e-commerce data processing, in particular to an e-commerce sales platform background data management method and system. The method comprises the following steps: S1, collecting a commodity dynamic data stream, a user behavior event stream and a promotion strategy stream in real time, and generating a standardized data stream through time sequence alignment; s2, constructing a dynamic coupling data cube; s3, executing a real-time decision: in response to the payment request, selecting an inventory distribution or replacement commodity pushing strategy based on a commodity-user coupling matrix value; in response to the resource overload state, triggering a resource scheduling strategy; and S4, dynamically adjusting calculation parameters of the coupling matrix according to decision execution feedback. By solving the problems of real-time standardization processing of multi-source heterogeneous data and real-time coupling of dynamic inventory and user behaviors, the data processing capacity, inventory distribution efficiency and recommendation accuracy of the platform are remarkably improved.

Disclosed are a demand-driven restock order generation and restock plan scheduling method and system. On the basis that the nature of a supply chain restock principle is revealed, a decoupling point restock order generation and restock plan scheduling method and system based on a supply and demand contract are provided. A proper order issuing time and a proper order quantity can be realized on the basis of future real demand and the fulfilment situation of past actual orders; it is ensured that a supply chain is not interrupted; decoupling point inventory is controlled to vary within the most rational range; three threshold value control lines, i.e., order forecast, order tracking early alarm and interruption alert, which are used by the supply chain to execute monitoring are also provided; the identification and measurement of a supply capacity gap are provided; on-hand inventory and an open order are evaluated on the basis of a supply and demand contract and future demand; and delivery-time review is performed on unconfirmed future demand on the basis of supply capacity.

The invention relates to the technical field of network security, in particular to a dynamic access blocking method based on zero trust. Comprising the following steps: step 1, collecting whole network flow data in real time in a bypass monitoring mode through a flow mirroring function of a network switch, performing deep packet inspection analysis on the collected original flow data, and extracting network flow characteristic parameters; 2, maintaining a dynamic identity information base; 3, performing real-time behavior analysis on each network session; 4, according to the risk assessment result and the real-time security context, generating a dynamic access control strategy based on a minimum permission principle; 5, implementing access control at the network execution point; and 6, continuously monitoring the network flow and the strategy execution effect, collecting feedback data, optimizing the risk assessment model and the strategy generation algorithm based on the feedback data, and forming closed-loop control. By dynamically updating the identity and asset information, the system can identify new assets or changes in real time, so that the adaptability and response capability of a network environment are improved.

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

The invention relates to a federated learning system based on multi-key homomorphic encryption and adaptive differential privacy, and belongs to the technical field of privacy computing. According to the system, on the premise that no trusted third party exists, a multi-client collaborative key generation and threshold decryption mechanism is achieved, it is ensured that model parameters are always in an encrypted state in the aggregation process, and leakage of a single node is prevented. By introducing a parameter sensitivity analysis and selective encryption strategy, the system only encrypts high-risk parameters, and the encryption burden is effectively reduced. Meanwhile, in combination with an adaptive privacy budget allocation mechanism, the system dynamically adjusts noise intensity according to a model training state, and model performance and convergence speed are maintained while privacy protection capability is improved. According to the method, high robustness and collusion resistance are realized, stable operation under the condition that part of clients are offline is supported, and the method is suitable for application scenes such as medical treatment and finance with high data sensitivity and strict performance requirements.

The invention discloses an urban flood toughness evaluation method, apparatus and device, a medium and a computer program product. The method comprises the steps of obtaining hydro meteorological data of a to-be-evaluated region and geographic information data of an urban spatial form element layer; setting boundary conditions of a plurality of flood scenes according to the hydrometeorological data, constructing a multi-scene urban flood disaster simulation model in combination with geographic information data, and respectively generating flood inundation depth data of urban spatial form element layers of the plurality of flood scenes; and calculating a first toughness index, a second toughness index and a third toughness index of each flood scene so as to calculate a flood toughness comprehensive evaluation result of each flood scene of the to-be-evaluated region. According to the method, an evaluation system of multiple composite flood disaster scenes is established, different disaster factors and interaction thereof are comprehensively considered, and the effect of key space parameters in urban flood toughness is fully considered, so that the capability of coping with flood disasters in cities can be more comprehensively and accurately evaluated.

The invention discloses a railway station facility full life cycle information query method based on digital twinning. The method comprises the steps that a unified facility object with a unique logic identifier is generated based on multi-source heterogeneous data aggregation; constructing a facility tense version model containing time sequence nodes by using an event-driven rule; in combination with an engineering constraint knowledge base and statistical characteristics, evaluating physical feasibility and statistical association strength between events, introducing life cycle stage risk factors, and constructing a facility full life cycle causal chain model; and based on the model, responding to the explanatory query, executing a constrained path search in the facility full life cycle causal chain model, and generating a life cycle query result set containing a causal path. According to the method, the pseudo-correlation problem generated by pure data mining can be effectively solved, the cumulative influence of early-stage risks in the design and construction stage on later-stage operation is quantified, and accurate tracing and explanation of cross-stage fault causes are realized.

To predict the working time of sorting work for taking out merchandise and sorting it for each shipping destination.SOLUTION: The prediction device 100 includes a prediction unit 113 that predicts a work time of sorting work, which is work of taking out and sorting articles from a storage that stores the articles, based on a plurality of orders related to the articles (commodities), and calculates the work time as a predicted work time. The prediction unit 113 calculates the predicted working time based on the input information including the type of article to be taken out from the storage, the number of articles to be taken out from the storage, and the number of articles included in each order. The storage may be a storage shelf that the transport robot carries to a work place of the sorting work. The item may be transported along a predetermined path to a work location of a sorting operation. Further, a worker who performs the sorting work may move to a place where the storage is placed and perform the sorting work.SELECTED DRAWING: Figure 3

The invention discloses an agent collaborative optimization data center management system based on knowledge graph driving, and relates to the technical field of data center management. The system specifically comprises the following modules: a modeling and entity management module, a cross-regional global scheduling optimization module, an agent game negotiation optimization module, an agent trust management module, a game negotiation conflict identification module, a reasoning evidence management module and a cross-regional consistency verification module. By establishing a complete knowledge graph model, systematic modeling of resource attributes, constraints, historical decisions and strategy preferences is realized, a unified data basis is provided for negotiation among multiple agents, the problem of a suboptimal solution caused by information asymmetry is effectively solved, a hierarchical reasoning mechanism is adopted, and the probability of resource disruption is reduced. In combination with global-region-node three-level reasoning and multi-round game negotiation, recursive optimization from global to local is realized, the reasoning complexity is effectively reduced, and the negotiation efficiency is improved.

The invention discloses a storage checking method based on a multi-modal sensing technology, a robot and a storage system, and belongs to the technical field of storage management and intelligent sensing fusion. Multi-modal data such as a visual image, space depth, radio frequency sensing and infrared temperature are collected, and an image feature vector, a three-dimensional point cloud model, a radio frequency response matrix and a temperature map are constructed; generating a fusion recognition vector through a multi-channel fusion network based on an attention mechanism, and dynamically adjusting a modal weight; constructing an article space distribution map, and marking a perception missing region; automatically complementing low-confidence region data based on a priority scheduling algorithm; performing joint verification on the original fusion result and the completion result to form a final inventory list; if the confidence coefficient of a certain article is lower than an early warning threshold continuously for multiple times, triggering an abnormal alarm and generating a traceable sensing sequence; the method is suitable for a high-precision inventory task in a complex storage scene, and has the advantages of high recognition robustness, intelligent completion mechanism, traceable abnormity and the like.