40results about How to "Improve results" patented technology

The invention discloses an automatic causal structure generation method based on semantic representation and logical reasoning of a large language model. The method comprises the following steps: acquiring an input text; performing semantic coding and clustering on the obtained input text by utilizing a large language model, and establishing a candidate causal variable set; causal relationship detection is carried out on the established candidate causal variable set based on anti-fact intervention and do-calculation; performing causal direction judgment, and generating a directed acyclic causal graph meeting logic consistency; and on the basis of the generated directed acyclic causal graph, natural language interpretation is generated by using a large language model, and logic consistency closed-loop verification is carried out. According to the method, automatic generation from the natural language to the causal structure is realized, the causal variable set is automatically extracted and constructed from the unstructured natural language text, the defects that variables need to be manually defined and modeling depends on field experts in the existing causal modeling process are avoided, and the labor cost and professional threshold of causal structure construction are remarkably reduced.

The invention discloses a CVE-ATTamp (Cascade Vector Enhanced ATTamp) for a network security scene. The invention discloses a CK association fusion RAG construction method, and relates to the technical field of network security. The method comprises the following steps: (1) carrying out structured analysis and vectorization on CVE vulnerability information data, extracting features and fusing CVSS scores to carry out semantic embedding; (2) ATTamp; semantic modeling of a CK threat matrix is carried out, and a threat semantic space containing tactical level information is constructed; (3) heterogeneous knowledge association and graph construction: calculating cross-domain semantic similarity and establishing logic mapping between vulnerabilities and tactics in combination with expert rules to form a heterogeneous knowledge graph; and (4) attack logic-oriented RAG retrieval and generation: retrieving a key attack path sub-graph based on a hybrid scoring mechanism, and driving a large model to generate a security analysis report with logic depth. According to the method, attack logic is stored through mapping, so that the problems that the traditional RAG lacks inference ability and security data fragmentation are solved, and the accuracy and tactical value of automatic threat analysis are remarkably improved.

This invention discloses a method, equipment, and medium for super high-rise construction monitoring based on digital twins, relating to the field of digital twin technology. The method includes: acquiring construction operation data from the super high-rise construction site and constructing a digital twin construction model; updating the boundary conditions of the digital twin construction model through a set of construction disturbance events to generate a disturbance-driven twin model; verifying the consistency between the virtual and real data of the model simulation state records in the construction state set and the disturbance-driven twin model to generate a twin credibility evaluation set; improving the local monitoring resolution of high-risk areas to generate twin resolution configuration results; performing local high-precision monitoring of the credible construction twin based on the twin resolution configuration results; and outputting super high-rise construction monitoring results. This invention, by improving the local monitoring resolution and generating twin resolution configuration results, avoids the waste of computational resources caused by global high-precision monitoring, thereby improving the efficiency of monitoring resource utilization and the response accuracy of high-risk areas.

This invention discloses a data management method and electronic device for a storage device, belonging to the field of storage technology. The data management method includes: when the lifetime of the hold table is less than the lifetime of the hold data area, writing data from the data cache to non-volatile memory, and the storage device entering a hibernation mode; when the lifetime of the hold table is greater than the lifetime of the hold data area, determining whether the data size in the data cache has reached a flush threshold; if the data size reaches the flush threshold, writing the data to non-volatile memory, and the storage device entering a hibernation mode; if the data size does not reach the flush threshold, saving the data to the hold data area, recording the boundary marker information of the data in the hold data area, and writing the system management data table to be migrated from the hold table to the non-volatile memory, and the storage device entering a hibernation mode. The data management method and electronic device for a storage device provided by this invention can extend the lifespan of the storage device.

This invention discloses a multi-channel electromagnetic induction detection method and system based on uniform magnetic field constraint. A uniform magnetic field is constructed within the detection area of ​​the structure under test. Electromagnetic induction response signals are acquired through multiple channels under the constraint of the uniform magnetic field, and the spatial coordinates of each acquisition channel are marked. The raw electromagnetic induction signals acquired by each channel are preprocessed, including noise reduction filtering, envelope extraction, and amplitude normalization. A deep learning model is then used to extract structural features from the preprocessed multi-channel electromagnetic response signals. Based on the spatial position information of each acquisition channel, magnetic field uniformity parameters, and response characteristics, a channel correlation weight matrix is ​​constructed to describe the spatial consistency and physical coupling relationships between channels. A multi-channel joint optimization model is constructed, and the equivalent electromagnetic parameter distribution inside the structure under test is reconstructed through iterative solution. This invention improves the reliability, repeatability, and engineering applicability of electromagnetic induction detection results.

The invention provides a spatial transcriptome tissue specific expression gene identification method and system based on a multi-dimensional statistical index, and belongs to the technical field of gene identification, the method comprises the following steps: obtaining gene expression data corresponding to a plurality of genes in different cells, different cells belonging to different types of tissues; the gene expression data is data processed based on a space transcriptome; for each gene, determining a basic statistical feature and a differential expression feature corresponding to the gene based on the gene expression data corresponding to the gene in different cells; determining specific index characteristics of the gene based on the basic statistical characteristics of the gene; and determining a tissue specific expression gene from the plurality of genes based on the basic statistical characteristics, the differential expression characteristics and the specific index characteristics corresponding to each gene. According to the invention, the accuracy of tissue specific expression gene recognition can be improved.

The application discloses a protease catalytic type attribution method based on protein cleavage site embedded vectors, comprising the following steps: constructing a cleavage site reference set of known protease substrates; obtaining N reference site vectors and M attribution site vectors; performing standardization processing on the N reference site vectors and the M attribution site vectors; calculating M*K first similarities and M*N second similarities; obtaining the nearest neighbor reference sites corresponding to each attribution site; obtaining the nearest neighbor support labels and the nearest neighbor support degrees of each attribution site; obtaining a credibility threshold required for attribution of the catalytic type of the attribution site; comparing the first similarity, the similarity difference and the nearest neighbor support degree with the credibility threshold required for attribution of the catalytic type of the attribution site; generating a corresponding credibility level or marking the attribution site as an uncertain candidate site; and the application can output attribution results of each attribution site according to the same data structure.

The application provides an aluminum pad structure and a manufacturing method thereof. The manufacturing method of the aluminum pad structure comprises the following steps: sequentially forming a first aluminum layer, a protective layer and a passivation layer on a substrate; performing dry etching on the passivation layer by using a plasma containing fluoride ions to form a passivation layer opening, wherein the passivation layer opening is located above a preset region of the first aluminum layer and the bottom surface of the passivation layer opening stays on the surface of the protective layer; then forming a second aluminum layer on the protective layer and removing the part of the second aluminum layer which is higher than the passivation layer, thereby obtaining a sandwich-type aluminum pad structure composed of the first aluminum layer, the protective layer and the second aluminum layer. The manufacturing method of the aluminum pad structure can avoid the direct contact of the etching process with the aluminum pad, thereby eliminating the generation of aluminum fluoride compound residues and fluoride ion residues on the surface of the aluminum pad from the root. The aluminum pad structure is in a sandwich type, has less aluminum fluoride compound residues and fluoride ion residues, is beneficial to improve the bonding and yield test results, and reduces fluorine precipitation.

The invention provides a strabismus screening system based on multi-dimensional dynamic eye movement tracking. The strabismus screening system comprises a data acquisition module, a data preprocessing module, a data analysis module and a result feedback module. A cartoon experimental normal form is adopted for an experiment through a domestic Pupil.IO AIO eye tracker, so that the eye movement track is monitored, the stability of a fixation point is dynamically quantified, the dynamic change during jump reaction is captured, multi-dimensional eye movement feature calculation such as sight line difference analysis is carried out, and finally dichotomy result judgment is carried out. The problems of inherent limitation, high equipment price, insufficient clinical suitability of a data analysis model and lack of a standardized process of a traditional inspection method are effectively solved.

The embodiment of the application provides a kind of based on three-dimensional point cloud's corrosion depth distribution extraction method and related device, method includes: first, the three-dimensional point cloud data of the outer surface of the steel wire to be measured is obtained, and curved steel wire is converted to straight coordinate system, the radial distance and ring angle of each surface point are calculated;Then segmented point cloud is carried out according to preset interval along axial direction, in each segment, based on the median and median absolute deviation of outer deviation, threshold is set, the abnormal point located outside the cross section fitting circle is removed, and the real point in the corrosion pit is retained;Then, the uncorroded section at both ends is set as reference section, the radius sequence is fitted, and the median value is taken as global reference radius;Finally, for the retained real point, the absolute corrosion depth is calculated, and the corrosion depth distribution probability density diagram and two-dimensional heat map are output, the interference problem of scanning noise, registration error and outside fly point to corrosion depth statistics can be solved, and the data purity and result reliability of corrosion depth calculation are significantly improved.

The application discloses a kind of distribution outside data delay calibration method and system, the method is first obtained historical layout training set and is based on line length threshold value identification distribution outside sample subset, then utilize inverse probability weighting method to calculate the calibration weight of each sample in training set, wherein distribution outside sample obtains higher weight, then using the weight to training set is weighted and re-trains machine learning model, obtains the delay prediction model after calibration, after calibration model deployment, directly used in interconnection delay prediction in engineering change order layout;The application is intelligently weighted and weighted training to internal sample of historical training set, without relying on any new tape data, effectively corrects the systematic underestimate of model to distribution outside long interconnection delay, so as to greatly improve the prediction accuracy, avoid the buffer insertion omission and timing violation caused by prediction deviation, enhance the efficiency and reliability of timing convergence.

This invention discloses a 3D drawing similarity retrieval method based on a spatial assembly structure vector tree, belonging to the field of 3D information retrieval technology. The method includes: constructing a spatial assembly structure vector tree based on a set of node spatial attributes and a set of assembly constraint attributes, according to the parent-child relationship of the original assembly hierarchy tree; performing normalization and standardization processing to generate a constraint-enhanced assembly vector tree; extracting hierarchical features from the constraint-enhanced assembly vector tree to generate hierarchical retrieval fingerprints, and matching these fingerprints with the hierarchical retrieval fingerprints of historical 3D drawings to establish hierarchical candidate mapping relationships and generate a hierarchical candidate retrieval spectrum; based on the hierarchical candidate retrieval spectrum, performing hierarchical flexible alignment to generate a flexible alignment mapping set, and calculating the comprehensive similarity between the 3D drawing to be retrieved and each historical 3D drawing. This invention achieves continuous organization of hierarchical correspondences and fine discrimination of similar features, improving the accuracy and usability of the retrieval results.

The present application relates to the technical field of equipment fault positioning, and more particularly to a power system equipment fault positioning method based on chain penetration technology, and the present application proposes the following scheme: obtaining a target alarm event and process variable data in a corresponding time window; constructing an industrial knowledge graph based on equipment topology relationship, process flow direction relationship, control interlocking relationship and historical fault knowledge; performing multi-level correlation traversal, correlation screening and priority evaluation around the target alarm event to determine a candidate measurement point node set; combining real-time measurement value and prediction value deviation, correlation distance, historical co-occurrence frequency and propagation direction consistency to calculate reason contribution scores, and outputting a potential fault reason set and troubleshooting suggestions. The present application improves the positioning accuracy and interpretability of abnormal alarms of complex industrial equipment.

The application discloses an optimization method of a rolling wing blade connecting rod mechanism, and belongs to the field of rolling wing aircrafts.The blade swing angle is represented by structural parameters of the blade connecting rod mechanism, and the structural parameters of the blade connecting rod mechanism, including the lengths of the connecting rods, the coordinates of key hinge points and configuration bias, are defined as a variable vector; multiple target functions are constructed for multiple key working conditions such as hovering, forward flight and transition, the multiple target functions not only quantify the performance of the connecting rod mechanism itself, such as kinematic fitting error with a target motion law, dynamic stability index and force transmission efficiency index, but also cover aerodynamic efficiency index and aerodynamic acoustic index which are directly related to the performance of the whole machine; and the optimal structural parameters of the blade connecting rod mechanism are obtained by solving the multiple target functions under constraints, so that the global optimization of the structural parameters of the blade connecting rod mechanism under multiple working conditions, multiple targets and multiple constraints can be realized.

This invention provides an intelligent inspection method, system, equipment, and medium for railway freight inspection, belonging to the field of railway freight inspection technology. The method includes: acquiring images of the freight cars using a line-scan camera array on both sides and the top of the track; performing spatiotemporal synchronization and adaptive parameter adjustment based on an encoder and a precise clock protocol; performing geometric correction, flat-field correction, motion compensation, and stitching on the images to generate a two-dimensional unfolded image of the freight car; using a deep learning-based freight car segmentation model to segment and number freight car instances, and extracting images of individual freight cars; finally, using a damage detection model integrating severity assessment to identify damage, outputting a structured result including damage category, location, confidence level, and severity score. This invention achieves non-contact, full-coverage, and high-precision intelligent inspection of freight cars, effectively improving inspection efficiency and automation levels.

This invention relates to a reversible and reapplicable land parcel mapping method, comprising: loading target land parcel data; binding a map canvas; establishing map event listening and location status reception; determining the point type, status, and region component corresponding to the current operation; constructing a corresponding edit command object; executing the edit command object to generate modified data; writing the executed reversible edit command object into a reversal history queue, and maintaining the reversal history queue and the reapply queue; and extracting the target edit command object to realize the reversal or reapply of the land parcel mapping results. The reversible and reapplicable land parcel mapping method designed using this invention can solve the problems of existing map mapping processing, such as lack of stable operation rollback capabilities, high maintenance difficulty, easy occurrence of untraceable operations, difficulty in state recovery, and unstable maintenance of historical records.

The invention discloses an alarm root cause analysis system and method, and belongs to the technical field of cloud computing alarm analysis. Comprising the steps that a process arrangement platform responds to an alarm event and matches a data docking template according to an alarm type; the data docking module collects original operation and maintenance data from a specified operation and maintenance data source and performs standardization processing to obtain standard operation and maintenance data; the process arrangement platform generates a root cause reasoning prompt word according to the standard operation and maintenance data, and inputs the root cause reasoning prompt word into the large language model for root cause reasoning to generate candidate root causes; the process arrangement platform generates a verification query instruction for each candidate root cause and sends the verification query instruction to the data docking module; the data docking module queries in the original operation and maintenance data based on the instruction to obtain verification evidence and feeds back the verification evidence; and the process arrangement platform screens the candidate root causes of which the verification evidences meet a preset confidence coefficient condition to obtain a root cause analysis result. According to the method, a closed loop from model reasoning to data verification is constructed, and the defect that a traditional method lacks a verification mechanism for root cause analysis of cloud computing alarm is overcome.

The application discloses a mine self-rescuer training behavior evaluation method based on multi-modal fusion, comprising the following steps: step one: acquiring visual behavior data of a training object and sensing behavior data of a mine self-rescuer; step two: aligning the visual behavior data with stage actions and encoding interactive semantics, and analyzing key step states of the sensing behavior data; step three: detecting human key points through an improved RTMPose model; step four: determining spatial relationships through a motion determination algorithm; step five: determining consistency correlation, generating a training behavior fusion recognition result; step six: performing evaluation processing on the training behavior fusion recognition result through a hierarchical time sequence constraint evaluation mechanism; and step seven: generating an evaluation output result and completing result storage. Through the improved RTMPose model and the motion determination algorithm, the objectivity and result reliability of the mine self-rescuer training behavior evaluation are improved.

The application relates to a hip joint ultrasound standard surface screening and measuring method and device based on key point detection and time sequence consistency, and the method comprises the following steps: acquiring a hip joint ultrasound image sequence; detecting a plurality of anatomical key points in the ultrasound image, outputting a multi-channel response heat map, and obtaining anatomical key point coordinates; constructing a comprehensive consistency scoring function based on the anatomical key points and the time sequence change of adjacent frames of the hip joint ultrasound image, and calculating a consistency score; when the consistency score exceeds a preset threshold, the corresponding frame is determined as a candidate standard section frame; the development state of the candidate standard section frame is reviewed and distinguished to obtain a development state classification result; the alpha angle and the beta angle are calculated, the angle line and the key points are superimposed and displayed on an original drawing, a visual result is generated, and a structured report containing the visual result, the anatomical key point coordinates, the angle parameters and the development state classification result is output. Compared with the prior art, the application significantly improves the automation degree and the result interpretability while maintaining high-precision measurement.

The invention provides a multi-modal loyalty illusion detection method and system based on closed-loop multi-dimensional verification, and belongs to the technical field of large visual language models.The method comprises the steps that S1, a user instruction input into an LLM and an answer of the LLM are decomposed into structured units; s2, through a collaborative mode of LLM and a visual tool, calculating consistency scores of assertions from three dimensions of text-instruction consistency, text-image consistency and logic consistency, binding all verification results with a unique evidence ID, and determining that the assertions without evidence support are unknown; s3, executing type exclusive calibration on the consistency score, and generating a unified illusion score H through weighted fusion; and S4, providing a traceable evidence support for the detection result, and triggering a discarding mechanism when the evidence is insufficient, so as to ensure the credibility and practical value of the detection result. According to the method, accurate detection of three types of illusions of instruction inconsistency, text-image separation and logic inconsistency is realized.

The application discloses a kind of based on intravascular imaging optimization three-dimensional vascular model interframe interpolation method;Obtain the sequence image group collected by intravascular imaging technology;And each frame image in sequence image group is carried out denoising processing;Using segmentation algorithm extracts the lumen boundary in each frame image, obtains the two-dimensional profile sequence image group of vascular lumen with frame data;Interpolation times are calculated;According to interpolation times, difference image group is generated, and according to the frame data of each image in difference image group, it is inserted into two-dimensional profile sequence image group;According to the last obtained two-dimensional profile sequence image group, generate high-resolution three-dimensional vascular model.The application is obtained by the frame interpolation technology of image collected by intravascular imaging technology, and the frame distance of adjacent image is reduced, and the continuity and smoothness of three-dimensional vascular model are improved.

The application relates to the field of multi-level network data processing, and discloses a transparent data calling system and method based on a multi-level network, wherein the transparent data calling method based on the multi-level network comprises the following steps: constructing a data forming path table based on multi-level business data, combining node data mapping relationship, result generation rules and transmission dependency relationship; segmenting and slicing associated data participating in the same result generation process according to node levels, time continuity and result dependency relationship to form a health prediction curve; determining abnormal responsibility nodes around the abnormal fluctuation section in the health prediction curve; calling equipment operation records, communication transmission records and intermediate processing records of the abnormal responsibility nodes in the abnormal period to form a node abnormal cause list; and generating substitute data by combining historical effective data of normal associated nodes, same-level result constraint relationship and reverse checking rules of superior results. The application has the advantage of improving data calling traceability.

The application discloses a network protocol test method and system and a storage medium, and the method comprises the following steps: obtaining test messages of a system under test from a gateway; performing protocol identification on the test messages, extracting protocol information of a target protocol layer, and determining a protocol processing node according to the protocol information; when the protocol processing node exists, performing field attribute extraction on the protocol information, obtaining attribute data, and obtaining message verification data of a target object; performing data retrieval according to the protocol information, and determining a target processing message in combination with the attribute data; obtaining message processing data of the target object, and performing self-defined processing on the target processing message; taking a next protocol layer of the target protocol layer as the target protocol layer, then returning to the step of extracting protocol information of the target protocol layer, and determining the protocol processing node according to the protocol information until the protocol processing node does not exist. The application can simplify network protocol test environment deployment, and increase process control and result inspection control of network protocol test.

This application belongs to the interdisciplinary field of computer application technology and aerospace equipment technology, and specifically relates to an intelligent technical document compilation system and method based on evidence constraints. The system includes a data governance module, a compilation constraint graph construction module, an automated generation module, an intelligent review module, an intelligent retrieval module, a change impact analysis module, and an interface integration module. The system converts multi-source equipment technical data into source record objects, chapter unit objects, and evidence objects, constructs a chapter constraint graph, and generates chapter task packages containing template skeletons, mandatory entities, mandatory parameters, and evidence sets. Based on this, it outputs initial drafts of the document with evidence identifiers, and forms candidate release versions through rule review, graph consistency review, and semantic review. When the technical status changes, it identifies affected chapters and recompiles them accordingly. This invention improves the traceability, consistency, and timeliness of technical document compilation, review, retrieval, and updating.

The invention provides a gas turbine multi-parameter operation curve simplification method and system, and relates to the technical field of gas turbines. The method specifically comprises the steps of performing normalization processing on a plurality of parameter operation curves; setting the number of key working condition point targets; calculating the vertical distance from each data point to the end-to-end connection line of the curve based on the thought of a Douglas-Peucker algorithm; performing weighted summation on the vertical distances of all the curves at the same moment to obtain a comprehensive feature distance; selecting a time point corresponding to the maximum value of the comprehensive distance as a key working condition point record; recursively dividing the curves by taking the key working condition points as boundaries, repeatedly calculating until the number of the key working condition points reaches the standard, and connecting the key working condition points of the curves to form a simplified curve. According to the method, the problems that manual screening is low in efficiency and poor in consistency, and an existing single-curve simplification method cannot keep the coupling relation among multiple parameters are solved, automatic and collaborative high-fidelity curve simplification is achieved, and the efficiency and precision of gas turbine operation data analysis and simulation are remarkably improved.

This application discloses a digital management method and system for hospital nursing quality indicators, relating to the field of medical and nursing quality information technology. The method includes the following steps: setting assessment rules for evaluating nursing process quality indicators; acquiring patient nursing records and constructing a patient-centered event time series; based on the assessment rules, filtering nursing events related to the corresponding nursing process from the event time series; dynamically adjusting the strictness of the assessment rules based on the identity information of nursing staff and the clinical status information of patients; verifying the event time series according to the adjusted assessment rules and obtaining verification results; determining the compliance of the nursing process based on the verification results, and generating a compliance conclusion and audit report for the nursing process. This application can intelligently distinguish between genuine nursing quality defects and discrepancies in record appearances caused by differences in personnel and context, solving the problems of fragmented processes, isolated data, and delayed assessment in traditional nursing quality management.

The invention discloses a compressor health state prediction method, a prediction model training method and a related device. The method comprises the following steps: acquiring basic data of a compressor in different load states, and dividing the basic data into a training set and a verification set; inputting the training set into a constructed neural network model, training the neural network model, and if a loss value of a loss function of the neural network model is greater than an expected loss value, adjusting hyper-parameters of the neural network model until the loss value is not greater than the expected loss value, so as to obtain a trained neural network model; inputting the verification set into the trained neural network model, and comparing a prediction result with a health state in the verification set to obtain a prediction rate; if the prediction rate is smaller than the expected prediction rate, the neural network model continues to be trained until the prediction rate is not smaller than the expected prediction rate, a finally trained neural network model is obtained, and the finally trained neural network model serves as a compressor health state prediction model. And the health state of compression can be accurately predicted.

The application discloses a network security risk dynamic evaluation and protection system based on a multi-cascade type, relates to the technical field of network security, and comprises a multi-cascade data acquisition and preprocessing module, a dynamic data traceability graph construction module, a distributed security knowledge graph cooperation module, a multi-cascade risk cause-effect evaluation module and a dynamic protection and closed-loop optimization module. The application constructs a cooperative architecture of multi-cascade data acquisition and preprocessing, dynamic data traceability graph construction, distributed security knowledge graph cooperation, multi-cascade risk cause-effect evaluation and dynamic protection and closed-loop optimization, and continuously optimizes system operation parameters through a closed-loop feedback link, compared with the prior art, significantly improves the dynamic adaptation capability of network security risk evaluation to the multi-cascade architecture and the comprehensiveness of evaluation results, and can solve the problems in the prior art that network security evaluation mostly depends on single-dimensional data, lacks cross-module deep cooperation, and is difficult to adapt to dynamic changes of the multi-cascade architecture.

The application relates to the technical field of glass fiber cotton detection, and discloses a glass fiber cotton air permeability detection method, which comprises the following steps: clamping a glass fiber cotton sample between an air inlet cavity and an air outlet cavity, and making the sample cover an effective ventilation area; and arranging a nine-square grid shunt structure in the air outlet cavity, which divides the air outlet cavity into nine mutually independent detection units; and arranging a light-weight mark sliding plug in each detection unit. According to the scheme, the airflow passing through different areas of the glass fiber cotton sample is introduced into the nine mutually independent detection units respectively by arranging the nine-square grid shunt structure in the air outlet cavity, the moving distance of the light-weight mark sliding plug in the transparent measuring tube is used to represent the air permeability of each area, the mixing of the airflow in different areas in the air outlet cavity is avoided, the detection result is not averaged, and the accuracy and result traceability of the glass fiber cotton air permeability detection are improved.

The application relates to the technical field of sealing test equipment, and discloses an air tightness experimental device and method controlled automatically. The air tightness experimental device controlled automatically comprises a testing machine main body, a test platform, an end sealing mechanism and a control end. A supporting piece for supporting a measured pipeline is arranged on a test station on the test platform. The end sealing mechanism is arranged on the test station correspondingly and is used for mechanically sealing two ends of the measured pipeline. The end sealing mechanism comprises two sealing clamps, a displacement driving unit, a position adjusting unit, a pressure sensing unit, a posture recognition unit and a sealing property pre-checking unit. Through the arrangement of the end sealing mechanism, intelligent and accurate centering of the sealing clamp and the pipeline port is realized. In combination with pressure sensing and servo closed-loop control, the clamping pressure can be self-adaptively adjusted according to the actual contact state, the uniformity and reliability of the sealing interface are ensured, and the test result is more accurate.