6853results about How to "Guaranteed accuracy" patented technology

The invention discloses an electromagnetic transient simulation and mapping method and system oriented to multi-physics field coupling, and belongs to the technical field of substation digital operation and maintenance and electromagnetic simulation. Deploying multiple types of sensors in a transformer substation, preprocessing the acquired multi-physical field data, and fusing with the three-dimensional twinborn model; a coupling algorithm is adopted to carry out multi-physical field coupling analysis, multi-physical quantity display and thermodynamic diagram superposition are realized, and scene evolution change simulation under various typical working conditions is supported; a micro-service architecture is adopted to construct a digital twin system, double closed-loop verification of simulation and actual measurement is performed through actual measurement data and simulation data of a real-model test field, model optimization is completed, and real-time mapping of electromagnetic transient dynamic simulation and a transformer substation physical entity is realized. The method improves the precision of model and entity mapping, meets the requirements of real-time tracking and response of the electromagnetic transient change of the transformer substation, adapts to the digital transformation of different transformer substations, and provides a basis for the operation and maintenance of the transformer substations.

The invention discloses a knowledge question-answering method and system based on a large language model and a semantic abstract. The system comprises a tree structure abstract generation subsystem which comprises a knowledge base configuration module, a file management module, a document analysis module and a knowledge block management module and is used for automatically generating tree structure hierarchical knowledge blocks for large manual documents or multi-chapter manual documents; the dual-channel retrieval engine subsystem comprises a question rewriting module, a semantic abstract retrieval module, a detail knowledge block retrieval module, a father-son retrieval module and a prompt word construction module, and all the modules work cooperatively to ensure that semantic abstract knowledge blocks and detail knowledge blocks related to the question of the user are retrieved. According to the method, by introducing the semantic abstract based on the tree structure and the two-channel retrieval, perception of global knowledge and accurate extraction of local fine-grained knowledge can be achieved at the same time, and then the ability of a knowledge base question-answering system in processing the general and inductive problems is improved.

The invention provides a cross-platform fusion access method, system and device of equipment and electronic equipment, the method is applied to the technical field of Internet of Things, and the method comprises the following steps: a mobile terminal performs security authentication on target central control equipment; the mobile terminal sends a structured configuration file including the equipment information of the plurality of smart home equipment, the region division rule, the scene mode and the linkage rule to the target central control equipment; the target central control equipment receives and analyzes the structured configuration file, and updates the corresponding smart home equipment according to an analysis result; the target central control equipment receives a control instruction of the cloud platform, converts the control instruction into an internal protocol instruction and then issues the internal protocol instruction so as to control the corresponding smart home equipment to execute operation; the target central control device synchronizes the state information of the smart home device to the cloud platform; and when an upgrading instruction of the cloud platform is received, issuing an upgrading task to execute an upgrading process. According to the method, the real-time performance of remote control of the smart home equipment and the accuracy of state feedback are guaranteed.

The invention relates to the technical field of block chains and security technologies, and particularly provides a security protection detection method and system for a smart contract based on a block chain. The security indexes at least comprise a password economics security index, a dynamic behavior characteristic index, a formalized verification index and a consensus layer security index; and performing quantitative evaluation on the collected multi-dimensional security index to obtain a comprehensive security score of the target block chain, scoring the cryptoeconomics security index through a preset cryptoeconomics security index scoring model to obtain a cryptoeconomics security score, and sending the cryptoeconomics security score to the target block chain. The current comprehensive security score of the target block chain is compared with the adaptive dynamic risk threshold, and if the current comprehensive security score is higher than the adaptive dynamic risk threshold, an early warning signal is sent out, and the effect of comprehensively evaluating the block chain smart contract is achieved.

The invention relates to the technical field of image recognition, in particular to a red tide algae recognition method based on a hyperspectral image, and the method comprises the following steps: calculating a gradient value of each pixel in a spatial dimension and a spectral dimension according to an original hyperspectral data cube, setting a diffusion coefficient for diffusion, and obtaining a three-dimensional fidelity hyperspectral data cube. According to the method, the gradient change trends of the spatial dimension and the spectral dimension are extracted at the same time pixel by pixel, and the regional differentiation diffusion operation is matched, so that the expression ability of the alga plaque boundary and the consistency of the internal spectral characteristics are enhanced, and the fidelity of the hyperspectral data in fine-grained region identification is improved; upper convex hull calculation of a spectral reflectivity curve and construction of a continuous spectrum background baseline are introduced, and characterization of pigment absorption morphological characteristics is enhanced; and in combination with the characteristic absorption wave bands of chlorophyll and phycobiliary, the synergistic weighted response to the distribution of the algae pigment is realized, and the sensitivity and selectivity of pigment recognition are improved.

The invention relates to a self-adaptive firewall rule generation method and device based on real-time threat intelligence, and the method comprises the steps: carrying out the semantic analysis of a threat intelligence text, extracting attack indexes and attack technique and tactics labels, and generating an intelligence triple set and an attack technique and tactics label set; calling a language model to generate candidate rules based on an intelligence analysis result, carrying out credibility scoring, intercepting a low-scoring rule as a high-risk illusion rule, and entering other rules into a to-be-verified set; constructing a digital twinborn shadow sub-domain to perform real-time traffic verification on the rule set to be verified, and dividing the rule set to be verified into an effective rule set and an invalid rule set according to a verification result; constructing a rule dependence graph based on the invalid rule set to perform pollution traceability, and adding nodes with excessive pollution into an isolation list; and cleaning the historical training samples based on the isolation list and the interception rule log, and performing incremental training on the language model to generate an updated model.

The invention provides a multi-process-mode intelligent switching method and system for vehicle-mounted sludge treatment, and the method comprises the steps: carrying out the time-space alignment and standardization of multi-dimensional sensing data such as sludge characteristics, environmental parameters and equipment states, carrying out the real-time reasoning of a process switching critical value through combining with a lightweight incremental regression model, generating a self-adaptive threshold value band with a confidence interval, and carrying out the real-time reasoning of the process switching critical value. Dynamic monitoring and mode switching pre-judgment of actual operation parameters are realized; in the switching process, a weighted fusion progressive transition control strategy is adopted, the transfer rate is adaptively adjusted according to real-time feedback, the operation stability is guaranteed, and the intelligence, reliability and switching smoothness of process switching of the vehicle-mounted sludge treatment system are remarkably improved.

The invention provides a non-power-cut test method for mutual inductance parameters of a power transmission line under excitation of different-frequency signals and a related device, and solves the problems of difficulty in accompanying stop and high safety risk of adjacent lines. The method comprises the following steps: selecting one circuit in the same corridor for power failure, connecting the head end with an induced electricity suppression device, and performing three-phase short-circuit grounding at the tail end; injecting a different-frequency sinusoidal current of 45-55Hz into the power failure line; the synchronous measurement device collects head and tail voltage and current of a running line based on a satellite synchronous clock; extracting pilot frequency signals through two-stage notch filtering and Fourier transform; and different-frequency zero-sequence self-impedance and mutual impedance are calculated and converted into power-frequency impedance. According to the invention, accurate test without power failure can be realized, and safety and power supply reliability are guaranteed.

The invention discloses a cisatracurium besilate content testing method which comprises the following steps: respectively weighing a test substance and a reference substance in a measuring flask, adding hydrochloric acid for dissolving and metering a volume to a scale, shaking up to prepare a corresponding solution, finally, measuring the solution according to an ultraviolet-visible spectrophotometry, measuring absorbance at a place of a maximum adsorption wavelength of 280nm, and calculating the content of cisatracurium besilate according to an anhydrous substance. According to the cisatracurium besilate content testing method, the solution is simply and conveniently prepared, no organic solvent is used, and the pollution to the environment and the harm to a human body are small. The cisatracurium besilate content testing method is higher in testing speed, and the time of testing each batch of products in average only needs 30 min. A place of a maximum adsorption wavelength of 280 nm is selected, so that the testing accuracy is ensured. The cost of a reagent used in a process of checking each batch of samples does not exceed 20 yuan; the testing cost of each batch of samples by adopting a high-performance liquid chromatography method is not less than 200 yuan, so that the cost of enterprises is greatly lowered.

The invention discloses a calibration method of a polarization focal plane imaging system and an imaging system, relates to the field of photoelectric imaging and remote sensing detection imaging, and solves the problems of low correction efficiency and difficulty in correcting a system with a changed polarization state in the conventional polarization error correction technology. According to the calibration method, a plurality of coding patterns and corresponding polarization modulation matrixes are obtained based on the collected uniform light field and serve as calibration results. The polarization error correction method comprises the steps of obtaining an observation intensity vector based on a target scene; a corresponding polarization modulation matrix is obtained based on the coding pattern according to a coding polarization modulation matrix lookup table, and then a down-sampling Stokes vector after polarization error correction is obtained; and completing the correction of the polarization error through an imaging reconstruction method. A polarization focal plane imaging system comprises an imaging lens, a DMD digital micromirror array coding module, a focal plane DoFP polarization detector and an imaging reconstruction module. The method is suitable for the fields of remote sensing detection, target identification, environment monitoring and the like.

The invention provides a radiotherapy plan generation method and device, equipment and a medium, and the method comprises the steps: obtaining a target function image of a target object; the target function image comprises a target region, an organ at risk and a target part; determining target dose information of the target part based on the target function image, the target function partitioning result, the target region segmentation result and the endangered organ segmentation result; the target function partitioning result is used for indicating spatial distribution of a plurality of functional regions of the target part divided according to the functional state, and the target dose information comprises doses of different regions of the target part; and generating a radiotherapy plan based on the target dose information. By applying the embodiment of the invention, the set dose information of the target part is more reasonable, so that the radiotherapy plan generated based on the target dose information can ensure the accuracy of tumor control, reduce the risk of radioactive injury and realize targeted protection of the target part.

The invention discloses a matrix type temperature detection system and method based on temperature-sensitive diodes, the problems of complex wiring and large temperature drift of a traditional multi-node temperature measurement system are solved through M * N matrix layout, row and column time-sharing gating and constant current driving technologies, the system adopts an independent column constant current source to ensure the measurement consistency, and a high-precision ADC and a dynamic calibration algorithm are combined, so that the measurement accuracy is improved. And the precision of + / -0.2 DEG C is realized within the range of 50-150 DEG C. The method is particularly suitable for high-density temperature measurement scenes such as battery thermal management, electronic equipment overheating monitoring and the like.

The invention relates to a vibration excitation method and system based on impedance identification, and the method comprises the steps: carrying out the frequency domain conversion of force-speed data obtained by a displacement closed-loop system, and obtaining the mechanical impedance of a drive point; performing flexibility compensation on the mechanical impedance of the driving point to obtain the real impedance of the structure end; extracting an imaginary part and constructing a matching residual error by using the real impedance of the structure end, correcting the excitation frequency of the current vibration exciter through gradient frequency, and iteratively determining the optimal excitation frequency of the vibration exciter meeting a convergence condition; mechanical active power generated by the vibration exciter is calculated under the optimal excitation frequency, dual-target optimization is carried out on the amplitude of the vibration exciter, and the optimal amplitude with the maximum mechanical active power and the minimum equivalent output of the actuator is obtained; based on the real impedance of the structure end, the matching criterion and the optimal amplitude, variable parameters are introduced to construct a source-side impedance model, a source-side impedance matching frequency domain interval is expanded through slow time mark setting parameters, and mechanical active power generated by a vibration exciter is coupled to a tested structure.

The invention discloses an electric arc dynamic suppression method for hot plugging of an electric meter connector, relates to the technical field of electric energy metering equipment, and aims to solve the problems of electric arc generation, overproof metering error and the like caused by passive protection and lack of dynamic cooperative control in the prior art. The core of the method is that gap distance and power grid voltage phase signals are acquired in real time through a microcontroller, and the plugging speed, the signal injection and the position fine adjustment are dynamically regulated and controlled in three stages of pre-contact, low-current transition and full connection; and a machine learning model is combined to optimize a parameter threshold value, a redundancy safety mechanism and a multi-phase electric meter cooperative control strategy, so that double management and control of arc peak energy and metering errors are realized. The method can actively inhibit arc generation, avoid connector ablation and insulation deterioration, guarantee metering continuity and accuracy, adapt to complex power grid and environment working conditions, and meet the precision requirement of the intelligent electric meter.

The invention relates to the technical field of time length detection, and discloses a news simulcasting time length detection alarm method based on voice recognition, which comprises the following steps: S1, noise reduction: integrating an advanced digital signal processing technology and a professional noise reduction algorithm to form a refined processing system adaptive to'news simulcasting 'audio, according to the system, aiming at collected useless components such as environment noise and equipment noise, noise characteristics are captured through a multi-dimensional signal analysis model, an effective information core identifier is defined, a differentiated characteristic database of signals and noise is constructed, and a technician sets scientific noise reduction parameters according to the differentiated characteristic database. And accurate noise identification and stripping: integrating an advanced digital signal processing technology and a professional noise reduction algorithm, capturing noise characteristics through a multi-dimensional signal analysis model, accurately stripping environmental noise (such as studio echo and external interference sound) and equipment noise (such as sound recording equipment floor noise and transmission link electromagnetic interference), and simultaneously retaining the original texture of effective audio.

The invention discloses an enterprise risk data determination method and device, a storage medium and an electronic device, and the method comprises the steps: constructing an enterprise relation graph corresponding to an enterprise set according to the risk of each enterprise in the to-be-predicted enterprise set and the incidence relation between the enterprises, each enterprise serves as a node in the enterprise relation graph, and the nodes of the enterprises with the incidence relation are mutually connected; nodes with self risks larger than a first threshold value in the enterprise relation graph serve as risk nodes; respectively spreading the risk signal of the risk node to all non-risk nodes which are connected with the risk node and meet the spreading condition; and determining and updating the risk data of the non-risk node according to the risk data of the non-risk node and received risk signals transmitted by all risk nodes connected with the non-risk node. Accurate prediction of enterprise risks is realized by simulating diffusion of risk signals in an enterprise network.

The invention discloses an engineering safety video generation method and system based on a generative large model. The method comprises the following steps of obtaining an application field and a theme requirement of a required engineering safety video; on the basis of application fields and theme requirements, a retrieval enhancement generation mechanism is adopted to retrieve in the multi-modal knowledge base to obtain related data, and a large language model is utilized to generate a training script copywriting; extracting cue words from the training script copywriting, and based on the cue words, generating a corresponding image by using a diffusion model; and converting the training script copywriting into an explanation audio, and finally synthesizing a virtual explanation video with synchronous mouth shapes by an audio driving model. The method can be suitable for a plurality of engineering scenes such as buildings, mining industry, tunnels, high-altitude operation and the like, has the characteristics of process automation, high content specialty, high generation efficiency and high cross-scene adaptability, and effectively improves the production efficiency and training effect of engineering safety education videos.

The invention provides a clamping device for a thin-wall whole pipe fatigue test and a test method, and belongs to the technical field of clamping devices. The device comprises clamping cushion blocks which are fixedly connected to two ends of a thin-wall whole tube sample main body and are used for transmitting test force; the high polymer material film adhesive tape is wound on a transition area of the clamping cushion block and the thin-wall whole pipe sample main body; the chock plugs are used for being plugged into the two ends of the thin-wall whole-tube sample main body, and packing paper is installed between the chock plugs and the inner wall of the thin-wall whole-tube sample main body and used for supporting the inner wall of the sample and evenly transmitting test force. The clamping cushion blocks are fixedly connected to the two ends of the sample and matched with the high polymer material film adhesive tape to reduce friction, the chock plugs are wound around the packing paper to support the inner wall, a stable clamping structure is formed, and it is ensured that fatigue fracture occurs in the working section. The problems of clamping stress concentration and fracture of a non-working section can be effectively avoided, the accuracy of a fatigue life evaluation result of the whole thin-wall pipe is improved, and the method can be widely applied to high-cycle and low-cycle fatigue performance tests under different conditions.

The invention relates to a credit distribution method in multi-agent cooperation, and the method comprises the following steps: obtaining joint data of a multi-agent system, the joint data comprising observation information and execution actions of each agent; the joint data are input into a discriminator model, and the discriminator model carries out modeling on an interaction dependency relationship among multiple agents based on an interaction dependency relationship modeling module of an attention mechanism; generating an auxiliary reward signal corresponding to each agent through a discriminator model; generating a fused reward signal based on the auxiliary reward signal and a global reward signal from the environment; and training and updating the multi-agent strategy model by using the fused reward signal so as to optimize the cooperation strategy of the multi-agent system. By introducing a credit distribution mechanism based on a discriminator model, automatic decomposition and optimization generation of individual reward signals are realized, so that the learning efficiency and stability of a multi-agent system in a complex cooperative task are remarkably improved.

The invention relates to the technical field of underground water simulation, in particular to an underground water numerical simulation method coupled with vegetation ecology. The method realizes bidirectional coupling between groundwater and vegetation ecology through iteration, and specifically comprises the following steps: inverting vegetation coverage based on groundwater burial depth; inversion of actual evapotranspiration is carried out based on underground water burial depth and vegetation coverage; and finally, updating the net supply amount of the groundwater model based on evapotranspiration until convergence. According to the method, a statistical function relation is established to replace a complex physical model, and drainage and overflow cell water levels are corrected in iteration to ensure calculation stability.

The invention discloses a large model reasoning hardware accelerator based on data flow execution, and belongs to the technical field of large model reasoning hardware acceleration. Each node comprises a linear calculation kernel used for sequentially completing query / key / value projection and attention output projection in multi-head attention calculation, dimension raising projection and dimension reduction projection calculation in a feedforward neural network layer, and inverse quantization operation after each projection is output; the multi-head attention calculation kernel is used for completing loading and quantization of query / key / value vectors, multi-head division and cache management, attention score calculation, Softmax normalization, context vector generation and KV cache updating; the RLA kernel is used for completing residual addition, layer normalization, quantization / inverse quantization and nonlinear activation calculation of each layer; the three cores in the same node are located in the same SLR area, and data transmission is carried out through a hardware queue and an HLS blocking mechanism. According to the invention, high hardware utilization rate and low-delay reasoning can be realized.

The invention discloses an inspection agent collaborative perception system based on semantic driving, and relates to the technical field of intelligent inspection, and the system comprises a prototype mapping module which collects inspection target category information and inspection target monitoring indexes, and generates an inspection task semantic prototype set and an inspection task semantic mapping table through semantic conversion; the semantic map module is used for collecting inspection image data and inspection video data through an inspection agent, performing pixel-level visual feature matching based on an inspection task semantic prototype set, and generating a semantic confidence map and a semantic request map; the coupling mutual sending module is used for executing sparse selection and directional mutual sending of semantic supply and demand coupling under the common constraint of the semantic confidence graph and the semantic request graph, and generating a multi-source sparse semantic feature queue; and the fusion remarking module is used for executing position-level semantic attention fusion and measurable sensitivity recalibration on the multi-source sparse semantic feature queue to generate a fusion probability graph and a fusion instance table.

The invention discloses an airborne equipment acceleration monitoring and outrange control method and system, which cover acceleration monitoring of the whole flight process, combine dynamic scene modeling and a hierarchical response mechanism, and can timely identify an abnormal state and start corresponding measures. Through system calibration, noise identification, dynamic filtering, mistaken touch prevention parameter setting and a collision algorithm, false alarm and missing alarm caused by vibration, impact or sensor faults are effectively avoided, double modes of automatic collision detection alarm and manual triggering alarm are supported, the flexibility and safety of the system are improved, and once it is confirmed that the system is in distress, the system is safe and reliable. Rescue signals including position, time, height and other key information can be automatically sent, and the emergency response time is remarkably shortened.

The invention discloses a manufacturing industry production data real-time management system and method based on Internet of Things edge computing. Comprising a data acquisition and preprocessing module, a lightweight agent module, an edge computing module, a network transmission module, a platform management module, an application service module, a security protection module, a self-adaptive resource configuration module and a monitoring and diagnosis module. According to the invention, nearby acquisition, preliminary cleaning and real-time analysis of multi-source heterogeneous data from a sensor, a controller and industrial equipment are realized, the cloud burden is effectively reduced, and the overall processing efficiency is improved. A real-time anomaly detection mechanism based on a sliding window and machine learning is constructed, parameter deviation, equipment faults or process anomalies in the production process can be recognized within millisecond-level time, and early warning or control feedback is triggered.

The utility model belongs to the technical field of lubricating oil additive filling, and particularly relates to a lubricating oil additive finished product filling device which comprises a base, supporting legs are arranged at the four corners of the bottom of the base, a filling mechanism is installed on the top of the base, and the rear end of the filling mechanism is communicated with a stirring mechanism. And a conveying mechanism is arranged below the filling mechanism. According to the filling mechanism disclosed by the utility model, the lubricating oil additive can be quickly filled through the cooperative operation of the liquid pump, the hose, the connecting pipe and the filling head. The electric telescopic rod can adjust the height of the filling head to adapt to filling containers of different specifications, and the universality of the equipment is improved; the electromagnetic valve can accurately control opening and closing of each filling head, and the accuracy of the filling amount is ensured.

The invention relates to the technical field of electrical detection, in particular to a multi-stress coupling environment reliability testing method for an electricity consumption information acquisition terminal. Comprising the steps of applying a combined stress field to an electricity utilization information acquisition terminal through an environment test box; acquiring power utilization information acquisition terminal parameters; the method comprises the following steps of: capturing metering pulse response and carrier communication interaction signals of an electricity utilization information acquisition terminal under the excitation of a combined stress field by utilizing an electrical sensor, extracting carrier signal-to-noise ratio data and metering reference drift data, and generating a reference voltage compensation coefficient and an impedance matching threshold value; projecting the carrier signal-to-noise ratio data to an impedance matching threshold value, and constructing a stress failure envelope surface according to a hardware overload protection threshold value; and mapping the communication quality distortion residual vector to a stress failure envelope surface, and calculating a reliability life prediction value. According to the method, a uniform excitation environment is constructed, and performance degradation of different dimensions is uniformly quantified into accumulative stress damage integral, so that the comprehensive anti-interference capability of the terminal in a multi-stress coupling environment can be truly reflected.

The invention provides a nucleic acid probe, a primer group, a kit, a detection method and application, and relates to the technical field of biology. The nucleotide sequence of the nucleic acid probe is as shown in SEQ ID NO. 1; wherein the 3'end of the nucleic acid probe is modified with MGB, and locked nucleic acid is modified at a base position corresponding to the difference between the swine fever wild strain and the swine fever vaccine strain in the nucleotide sequence. According to the nucleic acid probe, a specific sequence is utilized to precisely target a swine fever wild strain, and locked nucleic acid modification is introduced at a differential site, so that the recognition capability on single-base mismatch is remarkably enhanced, non-specific binding to a vaccine strain is thoroughly blocked, and false positive interference is eliminated; and 3 '-terminal MGB modification is matched to improve the binding stability and sensitivity, so that the efficient and accurate identification of the swine fever wild strain under the immune background is ensured.

The invention provides a roof-based subsidence monitoring method for a middle-deep underground mining mine, which comprises the following steps: S1, a deep anchor type anchoring datum point is arranged in a stable rock mass, far away from a mining disturbance area, of a mine roadway or a goaf, and the depth of the deep anchor type anchoring datum point needs to penetrate through a fissure zone and go deep into an original rock stress area; s2, a displacement sensor is installed at the key part of the top plate of the area to be monitored, and the displacement sensor is configured to take the deep anchor type anchoring datum point as a physical zero point; s3, acquiring a distance change original signal of the roof relative to the deep anchor type anchoring datum point through a displacement sensor, and directly acquiring roof subsidence data; and S4, performing signal noise reduction, anomaly detection, nonlinear feature analysis and parameter inversion on the roof subsidence data by using a computer system, and triggering an early warning mechanism when the roof is judged to be in an unstable state or a subsidence parameter exceeds a critical threshold. And an unstable bottom plate is fundamentally abandoned as a measurement basis, and the authenticity and the accuracy of the subsidence data of the top plate are ensured.

The invention discloses a collaborative reasoning method and system for adaptive model segmentation in a heterogeneous computing power environment. The method comprises the following steps: establishing a numerical calculation precision difference model between a source computing device and a target computing device; at a source equipment end, carrying out sensitivity analysis on a state tensor to be migrated according to the difference model, and asymmetrically dividing the state tensor into a core sensitive part and an edge part; performing dimension reduction processing on the edge part to generate a low-precision basic tensor, extracting precise residual information from the core part, compressing to generate a residual compensation vector packet, and sending the two to target equipment; and at a target equipment end, main calculation is carried out by using the low-precision basic tensor, a residual error compensation vector packet is decompressed in a bypass manner, and recovered residual error information is injected into an intermediate result of a main calculation stream to complete precision compensation. The problems of reasoning state semantic drift and error accumulation caused by dynamic migration between heterogeneous devices are solved, and the robustness and efficiency of a collaborative reasoning system are improved while the model reasoning accuracy is guaranteed.

The invention relates to the technical field of computer vision, and discloses a camouflage target detection method based on cascade frequency domain perception and refinement feature guidance, comprising the following steps: a computer device obtains a camouflage image and extracts a multi-scale feature map; performing frequency domain separation by using a frequency domain sensing module, generating low-frequency and high-frequency characteristic sub-bands, and performing cross-hierarchy aggregation from bottom to top through a cascaded frequency domain sensor to obtain frequency fusion characteristics; receiving the frequency fusion features from top to bottom by using an aggregation guide decoding unit, performing guide refinement in combination with the deep decoding features, adapting to irregular edges by using a variable kernel convolution strategy in the refinement process, and generating a prediction feature map in combination with a partial convolution strategy; and outputting the predicted feature map as a detection result. According to the invention, through frequency domain information complementation and dynamic convolution refinement, the problem of difficult detection caused by highly similar texture of the camouflage target and the background is solved, and the positioning accuracy and edge segmentation integrity of the camouflage target are improved.