57results about How to "Effective interaction" patented technology

The invention provides a robust multi-modal image segmentation method and system based on instance perception query, and the method comprises the steps: obtaining original image data and text description data, and generating training data through semantic analysis and causal graph construction; extracting image features and text features by adopting a multi-layer coding network, and establishing a feature dependency relationship through causal reasoning; generating instance clustering features based on attention calculation, and obtaining fusion features in combination with a multi-scale fusion strategy; and calculating query confidence and generating candidate segmentation masks, and outputting a final segmentation result through post-processing optimization. According to the method, by introducing causal reasoning and instance awareness query mechanisms, the cross-modal feature alignment precision and the robustness of the segmentation result are improved, and the method has better segmentation performance and generalization ability in a complex scene.

The invention discloses a drug target interaction prediction method based on a protein sequence. The method comprises the following steps: constructing a drug target database; constructing a target feature extractor based on a pre-trained protein language model; constructing a drug feature extractor based on a pre-trained molecular model; constructing a feature learning method by adopting a self-attention mechanism, constructing a drug target interaction learning method by adopting a cross attention mechanism, constructing a predictor in combination with a linear layer, and constructing a prediction drug target interaction model; training and optimizing parameters of the drug target interaction prediction model to obtain an optimized drug target prediction model; inputting a drug target into the optimized prediction model, outputting a prediction result, and finally marking potential active sites and important amino acid residues of the drug and the target through attention score visualization. According to the method, the precision of predicting the interaction between the drug and the target spot based on the protein sequence is improved, and the interpretability of the deep learning model is realized.

The invention relates to the technical field of artificial intelligence, and provides an intelligent control method and device for a tool body, equipment, a storage medium and a program product. The method comprises the steps of obtaining image features and text features; inputting the image features and the text features into a high-order target planning network to obtain target hidden variables and multi-modal features; inputting the target hidden variable and the multi-modal feature into a low-order execution network to obtain an action sequence corresponding to the current subtask; the robot is controlled to execute the action sequence, an execution result of the action sequence is obtained, and whether the current subtask is completed or not is judged based on the execution result; and if the current sub-task is completed, generating an action sequence corresponding to the next sub-task based on the low-order execution network until all the sub-tasks are completed. Through the above mode, a hierarchical strategy of high-order planning and low-order execution is adopted, so that the robot can complete execution of complex tasks in a changeable real environment, and the intelligent control method is high in applicability and high in control efficiency.

The invention discloses a distributed large model training-oriented multi-granularity semantic tree construction system and a distributed large model training-oriented multi-granularity semantic tree construction method. According to an edge-cloud information fusion space, samples and text tags of a cloud server are embedded and collected into public data set information of edge equipment # imgabs0 # for multi-modal data fusion; the model information interaction space is used for embedding samples and text labels of a cloud server and obtaining a global initial cluster center Ol belonging to Rd * l by adopting a hypergraph clustering algorithm; the cloud server constructs a semantic space # imgabs1 # based on the initial cluster center Ol, and the cloud server distributes the initial cluster center Ol and a text embedding # imgabs2 # of the semantic space to each edge device to construct an edge-cloud cooperative interaction space; the cloud server generates a multi-granularity semantic tree; according to the edge-cloud collaborative interaction space construction global granularity generation model, appropriate data granularity is cooperatively generated in a construction system, and the matching requirement of a heterogeneous model is met.

The invention discloses a method for synergistically hydro-thermally treating sludge based on tea residue and organic acid, which comprises the following steps: mixing sludge with organic acid and tea residue, and performing hydro-thermal treatment and dehydration treatment on the obtained mixed slurry to obtain filtrate and a mud cake. According to the method for carrying out the synergistic hydrothermal treatment on the sludge based on the tea residues and the organic acid, the hydrothermal reaction temperature can be remarkably reduced under the synergistic effect of the tea residues and the organic acid, meanwhile, the sludge is subjected to the synergistic hydrothermal treatment through the organic acid and the tea residues, the dehydration performance of the sludge can be effectively improved, and meanwhile generation of nondegradable organic matter such as melanoidins can be effectively inhibited; therefore, efficient, environment-friendly, recycling and low-cost treatment of the sludge can be realized by fully utilizing the characteristics of the organic acid and the tea residues and cooperating with a hydrothermal treatment technology, and the method has the advantages of low reaction temperature, high treatment efficiency, low energy consumption, good dehydration effect, greenness, environmental protection and the like; wide application prospects and popularization values are realized.

The invention relates to the technical field of Bluetooth keyboard control, and discloses a link control method and system for a Bluetooth keyboard and multiple devices, and the method comprises the steps: obtaining the current user behavior data of each device; inputting the current user behavior data into a pre-trained user use habit model for prediction to obtain prediction equipment which is most likely to be operated by the current user; searching from a connection device list according to the prediction device to obtain a link instruction of the corresponding device; wherein the connection device list is a database storing link instructions of each device; and in response to a selection link operation triggered by a user, realizing link between the corresponding equipment and the Bluetooth keyboard according to the link instruction. According to the method, the problem that when the Bluetooth keyboard is connected with a plurality of devices, the target device can be controlled only through a plurality of steps, and the user operation is troublesome can be solved.

The invention discloses a personalized teaching method and system based on multi-dimensional learning style analysis, and relates to the technical field of teaching. The personalized teaching system based on multi-dimensional learning style analysis comprises a learning style analysis module, and the learning style analysis module is connected with a teaching content management module. According to the system, a personalized learning plan can be made according to the learning style and progress of students through a personalized learning path and an intelligent recommendation engine, learning is flexible and efficient, in addition, rich teaching resources such as videos, interactive courses and textbooks in the system enhance the diversity and interestingness of learning, and the learning efficiency is improved. The real-time feedback and multi-dimensional evaluation mechanism ensures continuous support in the learning process and helps students to adjust strategies in time to optimize the learning effect, the system also combines online and offline mixed teaching, provides wider learning choices, enhances interaction and communication opportunities between teachers and students, and improves the learning efficiency. Therefore, the teaching effect and the satisfaction degree of students are effectively improved.

The invention discloses an automatic ICD coding method and system based on hierarchical feature optimization and prediction correction, and relates to the technical field of natural language processing, and the method comprises the steps: inputting a current electronic medical record text into a pre-trained automatic ICD coding model for coding, and obtaining a corresponding ICD code; the automatic ICD coding model comprises a text representation module, a coding representation module, a synonym fusion attention module, an initial prediction module and a prediction correction module which are connected in sequence; the text representation module processes input medical record text features by using a cascade convolutional network, the synonym fusion attention module introduces synonym fusion attention to enable ICD coding synonym description and electronic medical record text deep semantic representation to interact, and the prediction correction module based on a mask auto-encoder is further introduced. According to the method, the initial prediction result can be optimized and reordered, and the coding combination which most accords with the semantic and dependency relationship is selected, so that the prediction accuracy is improved, the coding efficiency is improved, and the automatic ICD coding problem is solved.

The invention relates to the field of plant growth regulators, in particular to an ultramicro-granulated energy-gathered humic acid composite preparation which is prepared from the following raw materials in parts by weight: 50-80 parts of ultramicro-granulated energy-gathered humic acid, 10-30 parts of nanoscale metal oxide, 5-20 parts of a microbial agent, 1-10 parts of a plant growth regulator and 5-15 parts of a slow release agent. According to the invention, ultrafine particle shaped humic acid, nano-scale metal oxide, a microbial agent, a plant growth regulator and a slow-release agent are organically combined. The effect of the soil conditioner is far better than that of a single component or a traditional soil conditioner.

The utility model discloses a circulating powder supply system of SLM printing equipment. A lower connector of a residual powder transfer powder tank is connected with a gas circuit powder conveying pipeline; an outlet of the blower is connected to an inlet of the gas circuit powder conveying pipeline; the cyclone separator is connected with an outlet of the gas circuit powder conveying pipeline; a powder outlet of the cyclone separator is connected with an inlet of the powder drying chamber, the powder drying chamber is connected with an inlet of the ultrasonic screening device, and the ultrasonic screening device is connected with the main powder storage tank. An output port in one side of the air bag is connected with an inlet of the arch breaking pipeline, an outlet of the arch breaking pipeline is internally connected with the main powder storage tank, the material supplementing tank, the top powder storage tank and the transfer powder tank, the top powder storage tank is arranged at the top of the SLM printing equipment, and an outlet of the top powder storage tank is connected to an inlet in the top of the SLM printing equipment; the material supplementing tank is connected to the gas circuit powder conveying pipeline; and the SLM printing unit is arranged on the gas circuit powder conveying pipeline. The problem that in the prior art, the efficiency is low due to the fact that powder is manually added or carried to complete the powder supply link of the printing equipment is solved.

The invention discloses a new energy model data processing method, a related device and a storage medium, and the method comprises the steps: obtaining multi-source heterogeneous data of each stage of equipment unit of an energy storage power station, and constructing a digital twin simulation model corresponding to each equipment unit through the multi-source heterogeneous data; associating the digital twinborn simulation models, corresponding equipment entities in the energy storage power station and data acquisition equipment on the equipment entities, and performing scene construction by using the digital twinborn simulation models based on layout information of the equipment entities to obtain a digital twinborn scene of the energy storage power station; performing data association on each digital twinborn simulation model in the digital twinborn scene of the energy storage power station and corresponding data acquisition equipment, and establishing a data transmission channel; and updating the digital twinborn scene of the energy storage power station in real time and monitoring and managing each equipment entity according to an equipment management method by using the data of each equipment entity acquired by each data acquisition equipment in real time.

The invention provides a target tracking method and system based on space channel summation attention, and the method employs a dual-branch feature extraction network based on a space channel summation attention module to carry out image feature extraction, and carries out the processing of a space attention mechanism and a channel attention mechanism in the extraction process. Information interaction between a space domain and a channel domain is achieved, a large amount of structured space information and channel information are obtained, final template features and intermediate search features are obtained, the final template features and the intermediate search features are fused and sent to a head prediction network, and the maximum response position of a tracking target in a search area can be obtained. And target tracking is carried out. According to the method, the advantages of space channel summation attention are fully utilized, so that the tracker can well deal with the difficulties of target shielding, rapid movement, complex background and the like in the tracking process, and more accurate tracking can be realized.

The invention discloses a method for preparing a graphene dispersion liquid by using a gemini surfactant in combination with a sanding process, and belongs to the technical field of graphene. Adding graphene into the gemini surfactant solution, carrying out pre-dispersion, and sanding the obtained graphene aqueous solution to obtain the graphene dispersion liquid. The method provided by the invention can realize efficient dispersion of graphene, is good in stability, safe and environment-friendly, remarkably simplifies the subsequent processing technology of the material, and has important practical significance and industrial potential.

The invention discloses an image fusion method based on a double-branch feature decoupling auto-encoder, and belongs to the field of computer vision and image processing. According to the method, for the problems of modal pollution and structural distortion in infrared and visible light image fusion, structural semantic information and high-frequency texture details of a source image are extracted respectively by constructing a content feature coding module and a detail feature coding module, and feature decoupling is achieved. And performing deep fusion on the decoupled features by using an adaptive feature weighting mechanism, and reconstructing a fused image with infrared target saliency and visible light detail definition through a shared decoder. According to the method, an end-to-end two-stage training strategy is adopted for optimization, complex prior or post-processing is not needed, the effects of improving the fused image contrast, edge preservation and target recognition performance are remarkable, and the method is suitable for the fields of weak light monitoring, intelligent perception, unmanned system navigation and the like.

The invention discloses a front and rear buffer assembly of a dune buggy and a design method thereof, and belongs to the field of dune buggies. The front and back buffer assembly of the all terrain vehicle and the design method of the front and back buffer assembly of the all terrain vehicle comprise a vehicle frame, buffer structures are arranged on the two sides of the vehicle frame, each buffer structure comprises supporting assemblies arranged on the two sides of the vehicle frame, the supporting assemblies are connected with the vehicle frame through dampers, and the supporting assemblies and the dampers can weaken impact force; the buffering structures are arranged at the two ends of the frame, when the frame is impacted, impact can be weakened through the supporting assemblies on the two sides of the frame, redundant impact force is transmitted to the dampers to be controlled, the influence of the impact force on the frame is greatly reduced, and the service life of the frame is prolonged. And meanwhile, the impact force suffered by the frame can be effectively judged, and the shielding structure is released to shield and protect the driving position of the vehicle when excessive impact is sensed, so that the harm of external chippings to a user is reduced.

According to the dialogue analysis method and device, the glasses, the equipment and the readable storage medium, the dialogue audio is collected in real time and converted into the dialogue text, the dialogue text is labeled on the basis of the speech function type of the speech round, deep information support can be provided for understanding of the dialogue type, and the dialogue analysis efficiency is improved. Key information and structural features in the dialogue are accurately identified by using a sequence detection rule of a preset dialogue type and a machine learning algorithm, so that feedback information is immediately generated and displayed, a dialogue guider quickly knows the real-time condition of the dialogue in the dialogue process, dialogue adjustment and dialogue process optimization are conveniently and quickly performed, and the dialogue efficiency is improved. And the dialogue efficiency and effect are improved.

This invention discloses a low-light scene analysis method based on multimodal feature fusion and clustering, belonging to artificial intelligence technology. It constructs a single-branch feature extraction network based on Transformer; a multimodal feature interaction and fusion module to achieve multimodal feature interaction and fusion; a multimodal fusion feature clustering module to cluster the features after multimodal interaction and fusion, utilizing the semantic and spatial distances between different features to achieve clustering while simultaneously downsampling the features, mitigating the problem of loss of detailed edge information in structured downsampling; and a multi-scale feature aggregation and decoding module to receive feature information from the encoding network and classify each feature pixel according to the semantic distance of the multi-scale features. This invention can fully utilize visible light and thermal image information and can be applied to scene analysis and navigation of unmanned systems in diverse and complex low-light scenes.

The invention relates to the field of sodium ion batteries, in particular to a composite sodium ion battery solid electrolyte and a preparation method thereof, which are used for solving the problems of low ionic conductivity, poor interfacial compatibility and poor chemical stability of the existing solid electrolyte. According to the preparation method, the high-performance polymer is used as a main raw material, the dispersed and modified carbon nanotubes are added into the high-performance polymer, the advantages of all the components are fully played, the ionic conductivity of the solid electrolyte is effectively improved, the electrochemical performance of the sodium-ion battery is effectively improved, and the sodium-ion battery has excellent chemical stability; compared with the prior art, the all-solid-state sodium-ion battery has the advantages that the cycle stability and safety of the all-solid-state sodium-ion battery are remarkably improved, the service life is prolonged, the potential safety hazards of volatile, explosion and the like of the traditional liquid electrolyte are avoided, and the preparation method is simple, feasible, easy to operate, low in cost and beneficial to large-scale commercial application.

The invention belongs to the technical field of biological genes, and discloses a method for identifying RNA interaction protein based on a Turbo ID technology and an MCP / MS2 system, the method can be used for identifying protein interacting with RNA in vitro or in vivo, and the method specifically comprises the following steps: firstly, respectively constructing RNA expression vectors inserted with MS2 tags, then constructing fusion protein vectors expressing MCP-Turbo ID, and finally constructing fusion protein vectors for expressing MCP-Turbo ID; the method comprises the following steps of: inserting RNA (Ribonucleic Acid) into an MS2 tag, transgenosis into a plant receptor by utilizing an agrobacterium tumefaciens mediated genetic transformation method to obtain a transgenic plant material expressing MCP-Turbo ID, combining the RNA inserted with the MS2 tag with the MCP-Turbo ID in vivo / in vitro by utilizing a hybridization or in-vitro incubation method, further enabling the Turbo ID to be adjacent to the RNA, biotinylating the RNA binding protein, and obtaining a transgenic plant material expressing the MCP-Turbo ID. And then enriching biotinylated protein by adopting streptavidin magnetic beads, and identifying and analyzing through Western Blot and mass spectrometry. The method not only solves the problem that the interaction protein of specific RNA in a living cell environment cannot be effectively identified in the prior art, but also can capture a transient or low-affinity RNA-protein interaction relationship.

This invention discloses a pet drying system with an interactive module and its usage method. The pet drying system includes a housing with a pet-containing cavity inside. The housing has an opening that connects to the pet-containing cavity, and a door for opening and closing the opening. It also includes a human-pet interaction module and a pet autonomous entry / exit module. The human-pet interaction module includes an interactive window, and the pet autonomous entry / exit module includes a pet autonomous entrance / exit. The interactive window and the pet autonomous entrance / exit are located on the housing and are respectively connected to the pet-containing cavity. This pet drying system with an interactive module allows for effective interaction between the user and the pet, and also enhances safety.

The invention provides an expanded reality man-machine interaction method for a road engineering operation and maintenance stage, and relates to the technical field of intelligent construction and operation and maintenance of road engineering. The method comprises the following steps: in a road construction stage, arranging a sensor system based on road types, structures and functions, and marking spatial positions of various sensors; constructing a road digital asset system based on the road structure physical information and the real-time response data collected by the sensor system; based on the constructed road digital asset system, road digital asset information visualization is realized by using an extended reality technology, and a distributed man-machine interaction system is constructed based on industry regulations and processes of road engineering operation and maintenance work; in a road operation and maintenance stage, road digital asset information and a man-machine interaction system are anchored in a road physical environment based on space calculation. By adopting the method, the problems of data integration visualization, time-space integration mapping of data and interaction of engineering field people, environment and computers in the operation and maintenance process of existing road engineering can be solved.

The invention provides a path-based K-hop hypergraph neural network convolution method, which comprises the following steps of: S1, creating a hypergraph neural network model for a graph downstream task, and initializing the model; s2, acquiring shortest paths of different distances between nodes (hyperedges) and central nodes or hyperedges of the shortest paths; s3, aggregating node features on the super edges directly connected with the nodes; after the hypergraph is subjected to dual transformation, aggregation of hyperedges is carried out in the same way; step S4, acquiring node features updated after low-order aggregation as path features for high-order HyperGINE to perform node K-hop neighborhood aggregation; after the hypergraph data is subjected to dual transformation, obtaining corresponding aggregation representations of different hops; step S5, updating the neighborhood aggregation result of the respective K hops; and S6, completing extraction of hypergraph features by stacking the KHGNN layers formed in the steps S3, S4 and S5, and performing task prediction through an output network matched with a downstream task. According to the invention, efficient prediction of downstream tasks can be realized.

The invention relates to the field of computer vision, in particular to an image semantic segmentation method. According to the method, a feature graph corresponding to each image is extracted through an encoder, each image serves as a node, a connection relation is established between the nodes, and a completely directed graph is constructed; inputting the completely directed graph and all the feature graphs into a graph neural network, and extracting semantic similarity feature vectors; inputting the feature map corresponding to each image into an intra-class reasoning module, extracting class features of each class in each image, inputting the class features into a co-occurrence relation reasoning module, and extracting a co-occurrence feature vector of each image; and respectively splicing the semantic similarity feature vector and the co-occurrence feature vector of each image, calculating an adaptive weight through a full connection layer and a Softmax activation function, carrying out weighted calculation on the spliced feature vector according to the adaptive weight, and outputting a semantic segmentation feature vector of each image through the full connection layer. According to the method, the context information in the image is fully utilized, and the semantic segmentation precision is improved.

A communication method, an apparatus, and a system are provided and relate to the field of communication technologies, to perform whiteboard interaction during a call. A call terminal transmits a call video stream between the call terminal and a peer call terminal through an established video call media transmission channel; detects a first touch operation of a first user on a first whiteboard picture presented on a call interface, and generates first touch trace data used to describe a first touch trace generated by the first touch operation; sends the first touch trace data to a media server; receives second whiteboard media data from the media server through the video call media transmission channel, and presents, on the call interface based on the second whiteboard media data, a second whiteboard picture including the first touch trace and a second touch trace, where the second touch trace is a touch trace generated by performing, by a second user, a second touch operation on a whiteboard picture presented on a call interface of the peer call terminal.

The application discloses an interactive method and device, a terminal device and a computer readable storage medium. The method comprises: acquiring video stream data played by a user; identifying the video stream data and extracting a key scene; when the key scene is not a predetermined scene, sending scene video data corresponding to the key scene to a cloud end, and interacting with the user based on first interactive content data returned by the cloud end, wherein the cloud end is used to generate the first interactive content data based on the scene video data. The method of the application can reduce the computing pressure of the client through the functional division and collaborative cooperation of the client and the cloud end, improve the accuracy of scene identification and the relevance of interactive content, realize efficient and intelligent interaction in a complex video playing scene, and improve the user experience.

The invention discloses a multi-dimensional physical simulation method and system based on large construction equipment, and relates to the technical field of digital assembly type construction. The method comprises the following steps: acquiring dynamic parameters of large-scale construction equipment and physical attribute parameters of prefabricated parts, constructing a multi-dimensional physical simulation model based on MATLAB, and constructing a high-fidelity virtual construction environment containing a construction site model, an equipment model and a component model through IsaacSim; a simulation model is used for calculating a hoisting track and load data, and the hoisting track and the load data are transmitted to IsaacSim through ROS standardization to complete virtual simulation; real data are collected through a multi-mode sensor and transmitted to a digital twin platform through a cloud server, the platform fuses simulation and real data, and a deviation calculation and dynamic calibration algorithm is adopted to carry out comparative analysis to generate a safety assessment result and early warning information. The system correspondingly comprises a multi-dimensional physical modeling unit, a virtual simulation unit, a sensor unit, a cloud server unit, a communication transmission unit and a digital twin platform unit, precise simulation, real-time monitoring and intelligent early warning of hoisting operation are achieved, and support is provided for construction safety and fine management.

The invention provides a joint relation triple extraction method based on table filling. The joint relation triple extraction method comprises the following steps of obtaining text data and generating subject related features and object related features; based on the subject-related features and the object-related features, carrying out two times of interaction association to obtain relationship types and combined table features of the entity pairs; processing the features of the combined table to obtain a subject-related maximum feature and an object-related maximum feature; interacting the maximum characteristics related to the subjects and the objects with the global information of the text data to obtain reasoning path information between the relation triples; iterating the steps for multiple times to output a final feature vector, obtaining a head-tail entity relation set based on the final feature vector, and generating a relation triple by using a decoding strategy; according to the method, the association between the host and the guest and the association between the relation triads are fully considered, so that the interference of relation overlapping and entity nesting on the accuracy of table filling is effectively reduced, and the extraction of the relation triads is more accurate and efficient.

The invention relates to a ship safety information intelligent interaction method based on a shipborne terminal, which comprises a shipborne terminal and a mobile terminal, and is characterized in that the shipborne terminal comprises a shipborne automatic reporting device and a communication module; the communication module integrates a 4G communication module, a Beidou No.3 short message communication module and a satellite broadband network module to realize communication connection among the mobile terminal, the shipborne terminal and the shore end platform; the mobile terminal is connected with the shipborne terminal through a local area network to realize data interaction, and the mobile terminal realizes interaction with the shore end platform through the communication module; the shipborne automatic report device automatically generates a navigation report according to the ship position and stores the navigation report in a local storage of the shipborne automatic report; and the shipborne automatic reporting device sends the navigation report to the shore end platform through the communication module. The 4G communication module, the Beidou No.3 short message communication module and the satellite broadband network module are integrated through the shipborne terminal, switching of different communication modes is achieved, and real-time, stable and efficient shore ship information interaction is achieved.

The invention relates to the technical field of paint mist flocculants, in particular to a paint mist flocculant for water-based paint and a preparation method thereof, functional groups such as amino and hydroxyl in modified chitosan can more effectively interact with paint mist particles in the water-based paint after being specifically treated, so that the agglomeration and sedimentation of the paint mist particles are promoted, and the water-based paint mist flocculant is prepared. The enhanced flocculation effect is beneficial to reduction of suspended matters and turbidity in the wastewater and improvement of wastewater treatment efficiency and quality; moreover, the modified chitosan is used as a part of a flocculating agent, so that the removal process of paint mist particles in the water-based paint wastewater can be accelerated, and the wastewater treatment time is shortened; secondly, aldehyde groups in the modified carboxymethyl cellulose and an introduced melamine structure can more effectively interact with paint mist particles in the water-based paint, and agglomeration and sedimentation of the paint mist particles are promoted in modes of chemical adsorption, physical entanglement and the like.

The application discloses a fault diagnosis method based on an adaptive time-frequency fusion gated attention network and belongs to the technical field of equipment fault diagnosis. The method comprises the following steps: an adaptive time-frequency domain fusion network model (ATFFNet) is constructed, the model mainly comprises three modules of double-branch feature extraction, time-frequency information fusion and pooling fusion, a time-domain branch and a time-frequency branch cooperative feature extraction framework is constructed to realize joint modeling of multi-source information; an adaptive frequency band division and multi-scale feature enhancement mechanism is introduced in the time-frequency branch to improve the representation ability of complex frequency components and local degradation characteristics, and the representation ability of the model to time-frequency characteristics and weak fault characteristics in a non-stationary signal is improved; meanwhile, a gated attention fusion module is designed to realize effective interaction and adaptive aggregation of cross-branch and cross-band information, so that the diagnosis precision, robustness and generalization ability of the model in a multi-sensor scene are improved.