60results about How to "Focus" patented technology

The application is suitable for the technical field of dismounting of torsion tube, and provides a dismounting method of torsion tube of airplane footrest. The dismounting steps of the torsion tube comprise: moving the inner tube towards the first bushing, cutting off the part of the inner tube exposed to the outer tube through the first bushing; moving the inner tube towards the second bushing, so that at least part of the inner wall of the first bushing is exposed to the inner tube; fixedly connecting the plug in the first bushing; applying force to the plug, separating the plug and the first bushing from the outer tube together, and taking out the inner tube; fixedly connecting the plug in the second bushing; applying force to the plug, separating the plug and the second bushing from the outer tube. In the application, the cooperation structure of the outer tube and the inner tube is used to cut off part of the inner tube, so that the first bushing is exposed to the inner tube, the plug is fixed with the first bushing, the force point of the first bushing is increased, the operator can separate the plug and the first bushing from the outer tube, the dismounting of the torsion tube is completed without damaging the structure of the outer tube, and the replacement cost of the torsion tube is reduced.

The invention relates to the technical field of welding spot detection, and provides a pin type welding spot detection method and system for light source scene change, and the method comprises the steps: firstly, employing a lightweight convolutional neural network MobileNetV3 to extract welding spot image features; then, a feature style mixing module is added to a middle layer and a low layer, image effects under different illumination conditions are simulated by randomly changing the style of features, and the adaptability of the model under different scenes is improved; a coordinate attention module is added to a high layer, and the recognition capability of the model on a welding spot fine defect area is enhanced; and finally, judging whether the welding spots have defects and defect types or not through a classification module. The method can effectively improve the accuracy and stability of contact pin type welding spot defect detection in a complex industrial environment.

The brake debugging platform for the yaw motor comprises a working bin and two test bases installed on the inner wall of the lower end of the working bin, driving motors are arranged in the outer walls of the upper ends of the two test bases, and the outer walls of the upper ends of the two driving motors are in transmission connection with detection gears. A driver main body is arranged on the upper side of the working bin, a connector is fixedly connected to the outer wall of the lower end of the driver main body, and a driver gear meshed with the detection gear is fixedly connected to the outer wall of the lower end of the connector. When the detection gear and the driver gear are screwed for detection, no matter where the detection is carried out, the continuous lubrication can be realized without frequent manual intervention, the gear is ensured to be always in a good lubrication state in the detection process, the continuity and the stability of the detection are ensured, and the detection does not depend on external specific lubrication facilities or conditions.

The present application relates to the field of optics, in particular to a broadband achromatic superlens with a large field of view. The superlens is composed of a cylindrical base embedded with a non-equal-height circular hole array and a nano-pillar array with a multi-ring structure. The non-equal-height circular hole array serves as the light entrance surface, and the nano-pillar array serves as the light exit surface. The circular hole array on the base has different depths and radii, and the nano-pillar array unit structure layer is composed of multiple ring bands, each ring band corresponding to a wavelength to achieve broadband achromatism. The nano-pillars arranged in the unit structure layer have various structures, such as cylindrical, annular, cross-shaped, etc. Different nano-pillar structures are used from the center to the outside, thereby reducing the mutual interference between adjacent ring bands. The materials of the unit structure layer and the base are transparent or low-absorption materials, and appropriate materials are selected according to different wavelength ranges. The superlens proposed in the present application can focus light beams of any wavelength range under a large field of view and a large spectral bandwidth.

This invention discloses an emotion analysis and feedback device based on visual and audio recognition in the field of psychological instruments and equipment technology. The device includes: a mobile base; a functional panel, vertically fixed to the top of the mobile base; a positioning ring, vertically suspended to the side of the functional panel, with an internal opening larger than the frontal contour of the human body, used for centering the user and restricting their sitting posture; a connecting arm, one end connected to the lower end of the functional panel and the other end connected to the positioning ring, used to support the positioning ring in a suspended state; and a flexible covering assembly located above the connecting arm, its surface forming a sloped structure facing the functional panel, used to guide the user's gaze to focus on the functional panel. This device features a hollow, large-sized positioning ring, which can center and align the user's torso and correct their sitting posture, avoiding blind spots caused by posture deviation. Combined with multi-directionally deployed visual and audio acquisition components, it significantly improves the accuracy of emotion analysis.

The application provides a traditional Chinese medicinal material hyperspectral anomaly detection classification method and system and a computer device, relates to the traditional Chinese medicinal material quality inspection field, and the method comprises the following steps: reading hyperspectral data, performing foreground detection on the hyperspectral data by adopting a multiscale spatial spectrum fusion method to obtain an abnormal area foreground mask; collecting N three-dimensional Patch arrays of the foreground mask area through three-dimensional Patch automatic sampling; setting a three-dimensional convolutional neural network of a wave band self-attention mechanism, inputting the foreground mask and the three-dimensional Patch arrays into the neural network to generate a high-level semantic feature vector; and finally outputting a classification result through a neighborhood spatial spectrum self-supervision aggregation attention module of the high-level semantic feature vector. The traditional Chinese medicinal material hyperspectral anomaly detection classification method, system and computer device are used, a new efficient, accurate and automatic traditional Chinese medicinal material mildew and insect damage detection method is realized, and the detection quality and the practical value of the system are significantly improved.

The application discloses a sound-absorbing and sound-insulating material and relates to the technical field of sound-absorbing and sound-insulating materials. The sound-absorbing and sound-insulating material is characterized in that PET hollow fibers are used as a framework, PP melt-blown fibers are used as filling materials to form a high-frequency sound-absorbing layer, the PET hollow fibers are used as a framework structure to ensure that the sound-absorbing and sound-insulating material has sufficient strength and stability, and melt-blown cloth fluff is added to the high-frequency sound-absorbing layer to prepare a low-frequency sound-absorbing layer. The PP melt-blown fibers are directly deposited and embedded into the PET fiber framework to form a material with small pores but many pores. The high-frequency sound-absorbing layer and the low-frequency sound-absorbing layer are layered and stacked, and non-woven fabric is arranged on the outer surface to form a novel sound-absorbing and sound-insulating material. The sound-absorbing and sound-insulating material not only has better sound-absorbing performance, but also has improved low-frequency sound-absorbing performance.

This invention relates to the field of blood glucose concentration prediction technology, specifically disclosing a method, system, and device for blood glucose concentration prediction based on a multi-branch attention mechanism. The method includes: acquiring multivariate time-series data related to blood glucose regulation; dividing the features into multiple feature subsets; inputting each feature subset into a corresponding independent recurrent neural network branch for time-dependent learning to generate a hidden state sequence; inputting each hidden state sequence into a corresponding attention submodule for processing, outputting a weighted feature representation; fusing the weighted feature representations to obtain a unified feature representation; and generating a blood glucose concentration prediction result through an output layer. This invention separates the temporal patterns of heterogeneous features through a multi-branch structure and focuses on key time steps using an attention mechanism, solving the problems of feature interaction interference and insufficient capture of key information in existing technologies, significantly improving the accuracy and clinical applicability of blood glucose prediction.

This invention discloses an anti-tangling cotton yarn processing jet dyeing machine. A guide wheel is fixedly connected to the inner wall of a fixed frame. A jetting device is fixedly connected to the inner wall of the fixed frame on one side of the guide wheel. The jetting device includes a feeding pipe. A heating ring is sleeved and fixedly connected to the side of the feeding pipe. The bottom of the heating ring is connected to the outlet of an electric water pump. A rotating pipe is rotatably connected to the side of the feeding pipe. A winding wheel is fixedly connected to the side of the rotating pipe. A drive shaft of a first motor is fixedly connected to the end of the rotating pipe away from the feeding pipe. The fixed end of the first motor is fixedly connected to the side of the fixed frame. The inlet of the electric water pump is connected to the side of a collection tank. The feeding pipe passes through the side of the fixed frame and is fixedly connected to the side of the fixed frame. The fixed frame provides overall support for the equipment. Cotton yarn is guided along the guide wheel after being introduced.

The utility model discloses a multifunctional foldable guqin rack capable of adapting to different playing scenes, which comprises a vertical rod A and a vertical rod B. The middle position of the front side of the vertical rod B is provided with a rotating bolt penetrating through the vertical rod A. The tail end of the rotating bolt is externally provided with a fastening nut. The tail ends of the lower sides of the vertical rod A and the vertical rod B are respectively provided with a support frame. The tail end of the upper side of the vertical rod A and the tail end of the upper side of the vertical rod B are each provided with a supporting rod, the outer portions of the front ends and the rear ends of the two supporting rods are each provided with an innovative structure composed of a limiting clamping ring plate, a lantern ring, a rotating rod, a limiting ring, a threaded rod, a pedal and a top rod, and the guqin frame can freely rotate by 360 degrees. In the playing process, a user does not need to move the whole position of the seven-stringed plucked instrument frame, the direction of the seven-stringed plucked instrument frame can be rapidly adjusted through simple operation according to stage layout, audience viewing angles or own playing habits, the adaptability of the seven-stringed plucked instrument frame to different playing scenes is greatly improved, and diversified playing requirements are met.

The invention discloses a behavior tree generation method and device, electronic equipment, a storage medium and a program product, and relates to the technical field of computers.The method comprises the steps that an initial cue word corresponding to a virtual object and a target game environment where the virtual object is located are obtained; inputting the initial cue word into a pre-trained decision model to generate a first behavior tree corresponding to the virtual object; deploying the first behavior tree to a target game environment to obtain game match data; inputting the game playing data into a pre-trained cognitive model, and generating data description information corresponding to the game playing data; and optimizing the first behavior tree based on the data description information to obtain a target behavior tree of the virtual object in the target game environment. By implementing the technical scheme of the invention, automatic generation and optimization of the game virtual object behavior tree can be realized based on collaborative closed-loop iteration of the decision model and the cognitive model, and the intelligent level and the generation efficiency of the game virtual object are improved.

This application relates to a leisure bean product and its preparation method. The preparation method includes the following steps: weighing the prescribed amounts of fried tofu puffs, filling slurry, braising broth, and mixed seasoning oil; rapidly freezing the selected fried tofu puffs; deep-frying the frozen fried tofu puffs at high temperature to obtain square-shaped fried tofu puffs; injecting the filling slurry quantitatively into the first-fried fried tofu puffs, and then rolling and kneading to obtain a semi-finished product; deep-frying the semi-finished product at high temperature, removing it, draining it, and cooling it to room temperature; immersing the second-fried semi-finished product in braising broth for braising, and then draining the liquid; placing the drained semi-finished product in an oven for baking; coating the baked semi-finished product with mixed seasoning oil for seasoning; and packaging, high-temperature sterilization, cooling, and air-drying the seasoned semi-finished product to obtain the leisure bean product. The leisure bean product obtained by this preparation method has a rich and complex taste and a sandwich structure, and is suitable for industrialized and automated production.

This application provides an optical system and a laser cutting device. The system includes a first focusing lens, a concave lens, and a focusing lens group arranged coaxially along a first direction. The first direction is the initial transmission direction of the light emitted by the target light source. The optical axis of the first focusing lens passes through the virtual focal point of the target light source. The virtual focal point is the point where the reverse rays of the light emitted by the target light source converge. The distance between the first focusing lens and the virtual focal point of the target light source is greater than the focal length of the first focusing lens. The concave lens is configured to change the distance between itself and the first focusing lens to adjust the aberration of the light emitted by the target light source. As the distance between the first focusing lens and the concave lens in the collimating lens group changes from far to near, the optical system exhibits the following trends: negative defocus beam convergence and tending towards a flat top, positive defocus dispersion and tending towards Gaussianization, then the distribution of positive and negative defocus beams becomes equal, and finally, negative defocus dispersion and tending towards Gaussianization, and positive defocus beam convergence and tending towards a flat top.

This invention relates to a method, apparatus, and electronic device for displaying polysemous content in an electronic learning device, comprising: establishing and storing a mapping relationship between at least two meanings of a target word and corresponding example sentences, wherein each meaning is uniquely associated with at least one corresponding example sentence; displaying a selectable list containing at least two meanings in a first area of ​​the display interface of an electronic paper display, the list having a currently active visual focus indicator; displaying example sentences corresponding to the meaning currently identified by the visual focus indicator in a second area of ​​the display interface of the electronic paper display; switching the visual focus indicator to the target meaning in response to a received selection instruction for the target meaning in the selectable list; and synchronously and uniquely updating the content displayed in the second area in response to the switching of the visual focus indicator. This invention eliminates the cognitive burden on users between different displayed contents, improving focus, comprehension accuracy, and operational efficiency in word learning.

The application provides a kind of based on feature enhancement and composite model's ultra-short-term wind power prediction method, belongs to ultra-short-term wind power prediction technical field, this method includes: by calculating the mutual information of multiple meteorological characteristics and wind power output and normalizing, screening to obtain key features, and constructing wind power prediction fitness function, by minimizing wind power prediction fitness function, obtain enhanced input features;By introducing one-dimensional convolutional neural network and bidirectional long short gate recurrent unit, construct the residual network of bidirectional long short gate recurrent unit and convolutional neural network fusion, and introduce space-time attention mechanism, construct to obtain composite residual network ultra-short-term wind power prediction model;And use multiple meteorological characteristics and enhanced input features to predict to obtain ultra-short-term wind power prediction result;The application solves the problem that the influence of meteorological multi-scale fluctuation on output cannot be accurately grasped, prediction is disconnected, disturbance is not considered and prediction accuracy continues to decline with step increase.

The present application relates to the technical field of bracket auxiliary instrument, more particularly to a bracket tweezers with positioning structure. The bracket tweezers with positioning structure comprises a tweezers head, a first tweezers arm and a second tweezers arm connected to two sides of the tweezers head, the first tweezers arm is connected to a first tweezers neck, the second tweezers arm is connected to a second tweezers neck, the first tweezers neck and the second tweezers neck intersect in space; the first tweezers neck is connected to a first tweezers tip, the second tweezers neck is connected to a second tweezers tip, each of the first tweezers tip and the second tweezers tip is provided with a pincer part, the pincer blades of the two pincer parts are oppositely arranged; a positioning mechanism is arranged between the first tweezers arm and the second tweezers arm, between the first tweezers neck and the second tweezers neck, and between the first tweezers tip and the second tweezers tip, the positioning mechanism comprises a driving assembly and a scale blocking assembly connected to the driving assembly, the scale blocking assembly abuts the dental and maxillofacial surface, and the end of the first tweezers tip and the second tweezers tip is provided with a scale for measuring the bracket position. The present application can provide bracket preset force and position the bracket on the tooth surface.

The utility model discloses a high-voltage fuse mounting structure convenient for rapid replacement, which relates to the technical field of power equipment and comprises a mounting base, insulator bodies are mounted on two sides of the top end of the mounting base, a fixed shell is fixedly connected to the top end of each insulator body, and a push rod is slidably connected to the inner wall of each fixed shell; according to the utility model, synchronous action of the two clamping assemblies is realized by utilizing the double-station clamping assembly, clamping and loosening of the fuse are completed, the working steps are simplified, and only one worm needs to be focused on screwing, so that not only is the operation efficiency improved, but also the possibility that an operator moves and operates back and forth among a plurality of positions during high-altitude operation is reduced, and the labor intensity of the operator is reduced. And the replacement work of the fuse can be completed more intensively, the risks of fatigue, poor physical strength, misoperation and the like caused by long-time hang are reduced, and the personal safety of operators is guaranteed.

A bath pedal convenient to install comprises a pedal body, and a containing area capable of containing a lifting rope is arranged at the bottom of the pedal body. A hollow pipe is installed at the bottom of the pedal body, and the interior of the hollow pipe is hollow to form a containing area. And the hollow pipe is fixed at the bottom of the pedal body through an adhesive. The bath pedal has the advantages that the containing area capable of containing the lifting rope is innovatively arranged at the bottom of the bath pedal body, and the core advantage of the structure is that an acting point is provided for the installation process. During installation, installation personnel can enable a lifting rope to penetrate through the hollow flat pipes arranged at the bottom of the pedal. Afterwards, the two persons can respectively pull the two ends of one lifting rope, and the whole bath pedal can be horizontally lifted easily and stably like a lifting handle. The problem that in the prior art, due to the fact that pedals are heavy and lack suitable grabbing points, force is difficult to apply especially in narrow spaces such as three faces leaning against the wall is thoroughly solved.

The application discloses a SAR small target detection network based on coordinate perception attention and spatial semantic context. Target detection is an important task of remote sensing image intelligent processing. Through a series of algorithms, the target on the image can be automatically positioned and recognized. Due to the imaging mechanism of SAR, there are many speckle noise points in the image, which makes it difficult to distinguish the target from the background, so compared with optical remote sensing images, the processing of SAR images is more difficult. The application can significantly improve the detection effect of small-scale targets of SAR remote sensing images, especially for the SAR remote sensing small target detection scene with complex background and serious noise interference, on the basis of effectively suppressing background noise, the detection rate and detection precision are significantly improved, the robustness and generalization ability of the model are strengthened, and the adaptability of the model to different scenes is improved.

The application discloses a manufacturing method of an ultrasonic catheter, the ultrasonic catheter and a forming die thereof. The manufacturing method of the ultrasonic catheter comprises the following steps: preparing an array stack; selecting a lead-out wire; coating a polymer material at a preset distance from one end of the lead-out wire to form a preformed tube; electrically connecting one end of the lead-out wire close to the preformed tube to the array stack; adopting split pouring to manufacture a sound-transparent protective layer coated on the periphery of the array stack and one end of the preformed tube close to the array stack; electrically connecting the other end of the lead-out wire away from the preformed tube to a circuit connector; and forming a Parylene coating layer on the outer surface of the sound-transparent protective layer. The Parylene coating layer enables the sound-transparent protective layer to better adapt to the vascular environment of the human body and meet the requirements for use in blood vessels. Therefore, the selection range of the manufacturing material of the sound-transparent protective layer is wider, and a material with a lower pouring temperature can be selected to manufacture the sound-transparent protective layer, so that the array stack is prevented from being separated due to heat, and the acoustic performance, consistency and qualification rate of the ultrasonic catheter are improved.

The utility model discloses a kind of paste formula can end gland devices, including workbench (1), the top of workbench (1) is equipped with electric drive's carousel (2), the top of carousel (2) is equipped with guide channel (3), the outlet side of guide channel (3) is equipped with electric drive's feeding wheel (4), the outer circumferential surface of feeding wheel (4) is equipped with multiple feeding grooves (5) of uniform distribution, the rear side of guide channel (3) is equipped with gland mechanism;The gland mechanism includes upper die (6), the bottom surface of upper die (6) is equipped with with can (7) cooperation's die cavity (9), the sidewall of upper die (6) is equipped with the vacuumizing device (8) of intercommunication die cavity (9), the top of upper die (6) is equipped with the lifting mechanism of connection workbench (1), the bottom of die cavity (9) is equipped with pressing plate (10), lifting mechanism is connected pressing plate (10) by rotating mechanism.The utility model has the advantages of high production efficiency and low production cost.

The application relates to the technical field of electric power transaction data processing, and discloses a power spot transaction data time sequence mining method capable of representing market fluctuation characteristics, which comprises the following steps: performing variational mode decomposition on electric price data, locking a probability density trough by using kernel density estimation to determine a dynamic shunt threshold, routing components to a trend or fluctuation channel according to the dynamic shunt threshold, respectively generating prediction values by using a linear extrapolation model and a multi-head attention mechanism with a causal mask introduced and superimposing the prediction values, the application realizes adaptive boundary locking in a non-stationary environment by using data topological morphology, and blocks invalid feature interference by mutual information causal gating, solves the contradiction between mean regression and peak capture of a traditional model, and improves the effectiveness of a prediction result.

The invention belongs to the technical field of market information visualization analysis, and discloses a market information visualization and analysis system based on a self-adaptive equivalent amplitude reference. The system comprises a data preprocessing fluctuation calculation module, a dynamic equivalence amplitude reference generation module, an equivalence line generation maintenance module, a chip distribution dynamic calculation module and a market form analysis module, and is characterized in that based on an original market data set, an equivalence line amplitude is adjusted to carry out adaptive calculation on an object market volatility to obtain a volatility data set; the method comprises the following steps: calculating and adjusting an equivalent amplitude reference value based on a volatility data set to obtain a multi-amplitude equivalent line sequence data set, obtaining an equivalent line sequence data set and a chip distribution data set according to a multi-amplitude equivalent line, and comprehensively analyzing data and visually displaying. The method has the remarkable advantages of high market analysis accuracy, high chip distribution timeliness and good market noise data filtering effect.

This application provides a laser head-assisted focusing device and a laser device, belonging to the field of laser device technology. The laser head-assisted focusing device includes a housing; a movable part, movably installed within the housing, the movable part having a stepped surface; a probe, inserted and fixed to the movable part; and a locking part, including a rotating shaft and a cantilever, the rotating shaft being rotatably installed within the housing, and the cantilever fixed to the rotating shaft; the cantilever, as the rotating shaft rotates, switches the locking part between a locked state and an unlocked state; when the locking part is in the locked state, the cantilever abuts against the stepped surface, locking the movable part in a first position within the housing, and the probe is in a retracted state; when the locking part is in the unlocked state, the cantilever separates from the stepped surface, the movable part can move relative to the housing, and the probe is in a released state. In the locked state, the probe is retracted, with a short extension length beyond the housing to avoid damage; when the probe is needed, the rotating shaft drives the cantilever to rotate, separating the cantilever from the stepped surface, and the probe extends outward and is released.

The invention discloses a dynamic hand attitude reconstruction method and device based on millimeter wave radar signals, and the method comprises the steps: carrying out the preprocessing of hand ADC data received by a radar, then transmitting the data into a learnable beam forming layer, and outputting a spatial feature vector containing all preset azimuth angle and pitch angle direction response intensities; dimension rearrangement and stacking are carried out on the spatial feature vectors, and a 4D millimeter wave cube is constructed; inputting a 3D convolutional neural network to perform feature extraction, including an initial trunk layer and a plurality of feature extraction stages, aggregating features of different scales by using a path aggregation feature pyramid network after multi-attention enhancement, and adjusting the output features into a frame-level spatial feature sequence; and simultaneously capturing short-term and long-term motion modes of a frame-level spatial feature sequence through two parallel branches to realize hybrid time sequence dynamic modeling, carrying out multi-strategy feature aggregation on output, and outputting final 3D hand joint point coordinates through an attitude regression head.

The application relates to a small target chicken positioning method based on double-domain enhancement and dynamic priori, which comprises the following steps: performing multi-scale feature extraction on a breeding image, a chicken image and a background image respectively, performing double-domain enhancement on shallow features and generating breeding coding features; prototype construction is performed on chicken image features and background image features to generate chicken feature prototypes and background feature prototypes, and similarity enhancement and background suppression are performed; based on foreground scores and foreground k A content embedding and refined position priori of an object query are generated by selecting a high response feature point, and are input into a DETR decoder to output the boundary box coordinates and the category score of the chicken target. The application can alleviate the small target missing detection problem caused by the smoothing of shallow features, reduce the false detection risk caused by background interference, improve the small target positioning ability of the traditional DETR model which is limited by the lack of target area position priori in object query initialization, and effectively improve the positioning detection precision.

The application provides a space target recognition method and device based on an attention mechanism adaptive enhancement. The method comprises the following steps: acquiring a space target image; inputting a tensor of the space target image into a pre-trained space target recognition model to obtain a recognition result of the space target. For any one processing stage submodel, a tensor processing module set unit and a convolution block attention module are included. In the tensor processing module set unit, a plurality of tensor processing modules are connected in sequence, the tensor processing module is used for convolution and channel transformation of the input tensor, and / or transformation processing of the tensor element value. According to the technical scheme of the application, the tensor processing module is used for convolution and channel transformation of the input tensor, and / or transformation processing of the tensor element value, the accuracy of the final recognition result can be significantly improved, the convolution block attention module realizes channel screening, is conducive to focusing on key features, and improves the feature utilization rate.

The application discloses a remote sensing image segmentation method based on a multi-scale feature fusion and index selection mechanism, and belongs to the technical field of remote sensing image processing. The method comprises the following steps: different scale feature information in a remote sensing image is extracted by using a multi-scale convolution network, and a fusion feature map of the remote sensing image is obtained; a feature aggregation is performed on the fusion feature map of the remote sensing image by using an index selection enhancement module based on a sparse attention mechanism, and an enhanced feature map is obtained; and the enhanced feature map is decoded by using a decoder, and a remote sensing image segmentation result is obtained. The application can better focus on important areas in the remote sensing image by introducing multi-scale feature extraction and a sparse attention mechanism, so that the segmentation precision is improved.

This invention discloses an intelligent detection method for chain-penetrating faults based on AI algorithms, belonging to the field of intelligent fault diagnosis technology. The method includes: employing a multi-scale temporal neural network computation model to segment, aggregate, and encode events in a structured multi-rate temporal data stream, generating a multi-scale abnormal event set; constructing a hierarchical dynamic causal graph based on the multi-scale abnormal event set and applying scale-causal consistency constraints to generate a multi-time-scale fused causal graph; and using a reduced-order physical simulation method to perform counterfactual intervention and experimental alignment on the root causes at the ends of each path, obtaining a simulation-experimental deviation sequence for each chain-penetrating path. This invention further utilizes semantically aware reverse traversal and scoring pruning in the fused causal graph to select a set of candidate chain-penetrating paths that conform to the fault semantic logic and causal intensity distribution from a massive pool of potential causal paths.