57results about How to "Solve the loss" patented technology

The invention discloses a connecting piece welding quality detection method and system, a medium and electronic equipment, and relates to the technical field of industrial automation and machine vision. The method specifically comprises the following steps: controlling a moving mechanism carrying an imaging device to move to a detection position, and obtaining a to-be-detected image containing a connecting sheet; inputting the to-be-detected image into a pre-trained target detection model, and obtaining a prediction bounding box of a connection sheet region; performing size expansion based on the prediction bounding box, and generating a region of interest covering the deformation region of the connecting sheet; in the region of interest, extracting a target feature point coordinate of a connecting sheet and a centroid coordinate of a welding spot by using an image processing algorithm; and calculating a distance value between the coordinates of the target feature point and the coordinates of the mass center of the welding spot, and judging the welding quality of the connecting piece based on the distance value. The invention aims to realize high-precision and high-efficiency welding quality detection of the connecting piece under the conditions of complex background interference and uncertain target form.

The invention relates to a brain-computer interface system which comprises a flexible electrode array, a receiving device, a flexible connecting wire and external equipment, wherein the flexible electrode array is implanted into a meninx middle artery in a minimally invasive manner through a superficial temporal artery puncture and a vascular intervention procedure; the receiving device is embedded under the skin of a head and is provided with a signal transmitter; the external equipment is provided with a signal receiver. A flexible electrode array is implanted in an anterior branch or posterior branch of a main branch of a middle meninx artery through a superficial temporal artery intravascular minimally invasive technology, the number of electrodes and the array are arranged according to needs, and a receiving device buried in the subcutaneous part of the head is electrically connected with the flexible electrode array implanted in the middle meninx artery. Craniotomy is not needed, operative complications are avoided, minimally invasive implantation reaches the surface of the cerebral cortex, operation of a doctor is easier, the operation time is shortened, and the comfort level of a patient is improved.

The utility model discloses a ball hitting machine which comprises a main machine shell, and a plurality of ball hitting main machines are arranged in an inner cavity of the main machine shell. The ball hitting main machine comprises a direct-current rotating motor assembly, the direct-current rotating motor assembly is fixedly connected with a ball hitting machine main shaft, the ball hitting machine main shaft is further fixedly connected with a small ball releasing assembly, the small ball releasing assembly is connected with a small ball part, and the small ball part is always connected with the small ball releasing assembly; the direct-current rotating motor assembly is used for recovering the small ball releasing assembly from the released state to the to-be-released state. The small ball releasing assembly is fixedly connected with a spring assembly, and the spring assembly is used for pushing the small ball releasing assembly to rotate. Rotation of the direct-current rotating motor can enable the small ball releasing assembly to enter the releasing state and then restore to the to-be-released state of the small ball part, and the small balls or the small sticks released in the whole process do not need to be manually picked up and recycled. Therefore, the problems that small balls or sticks and other targets need to be manually picked up and recycled, so that recycling is inconvenient, and the targets are lost due to untimely recycling are solved.

This invention discloses a method for obtaining the wide-angle single-station RCS of a medium target based on a compressed sensing algorithm. The steps are as follows: setting a wide-angle uniform incident plane wave illuminating the medium target; defining RWG basis functions and calculating the excitation vector for each incident angle; performing sparsification and compression on the excitation vectors respectively; generating the PMCHWT equations for the target surface and solving to obtain the compressed induced electromagnetic current matrix; recovering the original induced electromagnetic current matrix from each column of the compressed induced electromagnetic current matrix using the OMP algorithm; and calculating the radar cross section of the medium target based on the induced electromagnetic current vectors. This invention can obtain the RCS result at any angle with fewer PMCHWT equation solutions and combined with the OMP recovery algorithm, reducing the solution time for the wide-angle RCS of the medium target while ensuring accuracy.

This invention relates to the field of intelligent control technology for dredged pipelines, specifically disclosing a low-concentration intelligent diversion and interlocking control method for dredged pipelines. The method includes: responding to real-time operational data reported by each vessel at a preset frequency, calling a global fault mapping model to perform topological analysis on the fault information of a single vessel to construct a fault transmission map; based on the fault transmission map and the real-time operational data, calculating the global risk value of the current collaborative vessel group, and determining the risk level to which the global risk value belongs. This invention, by introducing a hydraulic connectivity masking mechanism, enables the system to dynamically correct the topology of the graph neural network in real time based on valve status and flow difference, solving the problem that static models cannot adapt to sudden changes in pipeline physical connections, and ensuring the accuracy of fault impact range calculation.

The application belongs to the technical field of millimeter wave radar anti-collision, and particularly relates to a correlation method for high-voltage line detection of a millimeter wave radar, aiming at the problem of large change of high-voltage line detection results of adjacent antenna rows in a millimeter wave radar anti-collision mode, the application avoids the problem of high-voltage line information loss by setting a survival period and converting to geographic coordinates, and correlates the high-voltage line detection results of adjacent antenna rows, preventing the same high-voltage line from being repeatedly recorded. The method can effectively solve the problem of millimeter wave radar high-voltage line information loss, improve the reliability of high-voltage line warning, and ensure the safety of helicopter flight.

The present application relates to a kind of forest leaf instance segmentation method and system based on deep learning, method includes: obtaining vegetation image, vegetation image is input into leaf instance segmentation model, obtains leaf instance segmentation prediction result;Leaf instance segmentation model is trained using training set;Training set includes: vegetation original image;Feature extraction and enhancement are carried out using backbone module in leaf instance segmentation model, and adaptive spatial fusion mechanism in progressive feature pyramid network is integrated to dynamically adjust feature weight, generate dynamic fusion feature;Through the dynamic asymmetric spatial perception mechanism built-in in dynamic anomaly regression head module, the corresponding multi-source deformation feature layer of dynamic fusion feature is obtained, and the feature fusion strategy of top-down cascaded decoding module is used to optimize multi-scale feature, obtain multi-source fusion feature layer, further using multi-source fusion feature layer, generate leaf instance segmentation prediction result.The present application solves the problems of data scarcity, poor adaptability and low efficiency.

Embodiments of the present application provide a kind of time sequence encoding method of space, three-dimensional reconstruction method and related product.Time sequence encoding method includes: sequentially projecting N line encoding diagrams to the space to be measured;For each pixel position of the space to be measured, based on the pixel gray value of the same pixel position in the N first shooting image, the time sequence value corresponding to the gray scale change peak value in the gray scale change curve is determined as the folded time sequence value;The M pixel interval is encoded by the preset encoding rule;Based on the encoding rule and the encoding value corresponding to the pixel interval of each pixel position, the folded time sequence value corresponding to each pixel position is unfolded, and the unfolded time sequence value corresponding to each pixel position in the space to be measured is obtained.The embodiments of the present application simultaneously utilize space information and time domain information to time sequence encode each pixel position of the space to be measured, effectively solve the problem of data loss in high reflection or high dynamic area reconstruction.

The utility model discloses a building electrical design multifunctional tool table belongs to building electrical technical field, including tool table body, the front side of tool table body top is provided with the storage groove, and the multiple baffle in storage groove divides the space into independent area, can be according to the accurate zoned storage of tool type, avoids all kinds of tools to mix and pile up, and sliding type tool box is used to store commonly used tool specially, effectively solves the tool loss problem of traditional tool table because of disorder, reduces resource waste, makes tool management more rationality, and the push-pull type switch is realized to the cooperation of sliding block and sliding groove of cover plate, and cooperation buckle assembly can complete locking and unlocking quickly, and tool box can be directly along mesa push-pull and use, shortens the tool access time, reduces work interruption, and the flat mesa can be formed after cover plate closes, further expands operation area, solves the mesa disorder problem of traditional tool table tool pile up, reduces the risk of accidental contact tool.

The utility model provides an organic waste plasma pyrolysis and cracking integrated device, relates to the technical field of garbage resource utilization, the organic waste plasma pyrolysis and cracking integrated device comprises a box shell, a plasma torch, a pyrolysis mechanism and a feed pipe, a cracking chamber is formed in the box shell, a hydrogen outlet is formed in the box shell, and the plasma torch is arranged in the cracking chamber. The hydrogen outlet is communicated with the cracking chamber; the plasma torch is arranged on the box shell, and the output end of the plasma torch is located in the cracking chamber; the pyrolysis mechanism comprises a pyrolysis chamber arranged in the cracking chamber and an exhaust pipe, one end of the exhaust pipe is communicated with the pyrolysis chamber, the other end of the exhaust pipe protrudes out of the box shell, a sealing plug is connected to the exhaust pipe, the exhaust pipe is further communicated with a gas guide pipe, and the other end of the gas guide pipe is communicated to the plasma torch; one end of the feeding pipe is communicated with the pyrolysis chamber, and the other end is connected with a feeding furnace door. The pyrolysis chamber is arranged in the cracking chamber, the high-temperature environment in the cracking chamber is fully utilized, and the high temperature in the cracking chamber can directly act on the pyrolysis chamber, so that the pyrolysis process is more efficient, the heat loss is reduced, and the energy consumption is reduced.

The application discloses a radio data processing method, device and equipment and a storage medium. The method comprises the following steps: acquiring a radio data set to be processed; wherein the radio data set to be processed has corresponding generation time information, the radio data set to be processed comprises radio data to be processed of a to-be-processed task, and the to-be-processed task comprises a plurality of subtasks executed in a preset order; determining a target memory matched with the radio data set to be processed according to the generation time information, a total number of memories and a pre-established memory management table, and storing the radio data set to be processed in the target memory; calling a processing thread corresponding to each subtask in the plurality of subtasks in the target memory according to the preset order, and processing the radio data to be processed by using the processing thread. The technical scheme of the application improves the calculation speed of radio data.

The invention relates to the technical field of rail transit simulation, in particular to a rail transit simulation scene modeling method for enhancing the sense of reality, and solves the technical problem that the rendering effect is poor in an illumination sudden change scene in the prior art. The method comprises the following steps: constructing environment shielding data of a track line; the environment shielding data comprises a sky opening coefficient and an instrument panel geometric shielding coefficient corresponding to each track position of the track line; performing space-time mapping synthesis on the sky opening coefficient based on the real-time running state data of the train, and determining a real-time environment illumination observation value of the rail position where the train is located; performing illumination change analysis on the rail position of the train based on the real-time environment illumination observation value, and determining the illumination sudden change intensity of the rail position of the train; and performing model brightness rendering compensation on the instrument panel area of the train based on the illumination sudden change intensity and the geometric shielding coefficient of the instrument panel.

The invention discloses an event identification method and system for pig house intelligent monitoring, a camera and a medium. The method comprises the following steps: collecting and sampling pig house video streams; performing target identification on the sampled image to obtain an identification result including categories and position coordinates of pigs, personnel, vehicles and the like; performing label matching on the identification result based on a pre-configured label rule, wherein the matching rule at least comprises a target appearance rule, a region intersection rule, a target continuous appearance rule and a target disappearance rule; matching a pre-configured event rule according to a label matching result, and generating hog house monitoring events such as entry of pigs into a corridor and entry of personnel into an effective area; and intercepting an event associated video clip, monitoring a 4G network signal state, and adaptively selecting to locally store or report the event data and the video clip to a cloud platform. According to the invention, accurate and flexible event identification aiming at a pig house scene is realized, and the reliability of monitoring data in an environment without a wired network or a weak network is guaranteed.

This invention discloses a backdoor attack method, system, and medium based on multi-object semantic coexistence in text-to-image diffusion models. This method utilizes the perspective of multi-object semantic coexistence in text-to-image diffusion models to develop MOBA backdoor attack schemes. First, by constructing a trigger alignment dataset and optimizing backdoor implantation and semantic preservation in parallel, the attack success rate and visual concealment are effectively improved, reducing the impact of semantic corruption on attack effectiveness. Second, during model training, an attention-based decoupling backdoor enhancement mechanism is adopted. By decoupling the attention regions of different objects, the semantic integrity of the input prompt is maintained while the backdoor is activated and the attention distribution is reasonably adjusted to reduce the impact of generation bias on attack concealment. The method of this invention ensures both visual concealment and model performance under benign input while achieving efficient backdoor attacks.

The application discloses a communication method based on a TTE dual-redundancy network and belongs to the technical field of satellite network communication, and comprises the following steps: a hierarchical guarantee mechanism of BE data cold backup and TT / RC data hot redundancy is designed; then, for BE multicast service, redundancy switching is carried out through remote control instructions, for BE unicast service, an ARP trigger mechanism is modified, and a redundancy attribute is introduced in the ARP protocol, so that data transmission is only carried out on an effective redundancy link where a target device is located. The state of the dual-redundancy path is dynamically monitored in combination with an SNMP protocol, so that the network resource utilization rate and the real-time performance and reliability of network data transmission are significantly improved.

The application provides a traffic pavement crack dehazing and de-noising segmentation method based on FA-Diff-UNet, introduces a FA-Diff module in the encoding front end, fuses diffusion amplitude residual reconstruction and phase fine-tuning, and internally builds a super-resolution reconstruction head to recover high-frequency details suppressed by fog and noise without introducing pseudo texture; a double-branch down-sampling structure of a CDM module and a DBAT module is used in the encoder part to expand the effective receptive field and retain the characteristics of small cracks under the condition that the parameters are controllable; a DHLA module is introduced in the bottleneck layer to decouple and suppress noise for high-frequency cracks and low-frequency backgrounds; the decoder replaces the simple skip connection of the UNet module with a BFM module to realize bidirectional gating fusion and redundancy suppression of semantics and details, and a DBAT refinement block is connected in series after each level of up-sampling to further enhance cross-block connectivity and direction consistency. The crack edge definition and connectivity are significantly improved, the recall of small cracks is improved, the false detection is reduced, the accuracy and the calculation efficiency are considered, and the method is suitable for actual road scenes such as complex weather and weak light.

The application discloses a system for safe clipboard synchronization between an isolated environment and a host computer, and belongs to the technical field of computer security and virtualization. The system comprises: a protocol proxy core module for intercepting clipboard protocol requests between the isolated environment and the host computer; a proxy entity module for creating a proxy entity on the host computer side to replace an application in the isolated environment holding clipboard ownership; a synchronization state manager module for maintaining internal and external time sequence identifiers for tracking the update sequence of the isolated environment and the host computer clipboard; and a loop check list. When the system initiates synchronization, the system records the synchronization operation in the loop check list; when a clipboard change is detected, the system identifies and filters loop events triggered by self-synchronization by comparing the loop check list; and when responding to a data request, the system determines the latest data source by comparing the internal and external time sequence identifiers. The application realizes safe, stable and reliable cross-environment clipboard synchronization.

This invention discloses a power on / off control system and an in-vehicle product. The system includes a switch module, a switch state detection module, a first enable module, a second enable module, a power output module, and a control module. The switch state detection module outputs a first-level signal when the switch module is in a closed state and a second-level signal when the switch module is in an open state. The first enable module outputs a first enable signal when the switch state detection module outputs the first-level signal. The control module controls the second enable module to output a second enable signal when the switch state detection module outputs the second-level signal and a data saving completion signal is received or the duration of the second-level signal output by the switch state detection module reaches a preset duration. The power output module outputs a power supply voltage upon receiving the first enable signal and stops outputting the power supply voltage upon receiving the second enable signal. This invention enables the in-vehicle product to power off only after data saving is complete.

The application provides a method and device for quickly locating faults in an MGX system, and belongs to the field of server system management and fault diagnosis. The method comprises the following steps: determining an idle universal serial bus interface on a mainboard, and selecting a target USB interface and a fixed IP network segment configuration corresponding to the target USB interface; a BMC establishes a physical connection with a device where an HMC is located through the target USB interface, and constructs a virtual local area network channel; in the virtual local area network channel, the BMC and the HMC perform out-of-band communication; the BMC obtains inventory information of components in the MGX system from the HMC, and classifies the components according to component types; the HMC respectively pre-collects state information of each category and stores the state information according to the classification result; and the BMC periodically reads the state information based on a Redfish protocol, and identifies faults in the MGX system according to the state information. The method and device provided by the application can quickly and accurately locate faults.

The invention relates to the field of software performance monitoring, and provides an automatic performance monitoring method and device for a vehicle-mounted multimedia application and a storage medium. The method comprises the following steps: collecting multi-dimensional performance parameters of a target application during operation in real time; based on the multi-dimensional performance parameters of the target application during operation, judging whether a preset triggering condition is met or not; when a preset triggering condition is met, automatically starting a plurality of data acquisition channels in parallel to obtain multi-modal data related to the performance event; performing association analysis based on a unified timestamp on the acquired multi-modal data and the multi-dimensional performance parameters to construct a complete data evidence chain of the performance event; and automatically generating a performance analysis report containing performance bottleneck positioning information and optimization suggestions based on the performance event cause and influence range determined by the correlation analysis. According to the technical scheme, accurate and efficient positioning and automatic diagnosis of the vehicle-mounted application performance problem are achieved.

The application belongs to the technical field of computer software, and discloses a dynamic binding and analysis method and system for multi-source heterogeneous data in a low-code platform, which comprises the following steps: using a recursive path serialization algorithm, a lexical rule is disassembled into an ordered attribute access sequence, and a business field identifier is extracted based on the ordered attribute access sequence to output a listening queue with a dependency relationship; the listening queue is used as a driving basis to perform multi-dimensional layered retrieval and data aggregation operation, and an evaluation strategy is selected according to the execution result and the complexity of the reference mode, and a memory isolation cloning result is output to complete data binding and analysis processing. The application constructs a self-describing protocol with conflict avoidance capability to solve the problem that static JSON configuration cannot directly express dynamic logic in a low-code environment, and can statically extract field-level fine-grained dependencies and establish accurate responsive listening.

The application provides a negative sample enhancement APT attack detection method based on graph structure learning, mainly solving the problem that the existing data set compatibility, edge information utilization and insufficient negative samples of the data set lead to poor detection effect. The scheme comprises the following steps: 1) obtaining a heterogeneous data set and constructing a visual traceability graph; 2) preprocessing the traceability graph, dividing it into snapshots according to the time stamp, and constructing a training set and a test set; 3) designing an encoder and a decoder of the graph neural network, taking the time window, the node feature, the adjacency matrix and the edge feature as the input of the encoder, and generating the reconstructed adjacency matrix through the decoder; 4) designing a loss function, and training the graph neural network through RNN; and 5) inputting the to-be-detected data into the trained model, identifying abnormal attack traffic, and completing the detection. The application can effectively improve the accuracy and robustness of APT attack detection, and can be used for the development and deployment of the APT attack detection and defense system in the network security field.

The application relates to the field of artificial intelligence and image processing technology, and discloses a multi-modal large model driven complex scene image data processing method and system. The method obtains target image data, extracts multi-scale semantic features, performs feature coding based on a scene context dependency graph, generates optimized feature representation through attention guided feature fusion and multi-level semantic reasoning, and executes a target processing operation. The application can effectively improve the complex scene image understanding accuracy, enhance the feature expression capability, improve the performance of the model in spatial relationship reasoning and semantic analysis, and is suitable for complex visual scene analysis and decision-making.

The invention provides a high-missing-rate feature processing method and device, and the method comprises the steps: carrying out the preprocessing of original features in original feature data, obtaining the preprocessed transformation features, carrying out the multi-path embedding transformation of the original features based on a plurality of embedding transformation rules, and obtaining the transformation features of the original features; fusing a plurality of embedded vectors obtained by conversion into a target embedded vector as a first derivative feature; obtaining a comprehensive similarity matrix based on comprehensive similarity obtained by fusing multiple similarities between the target embedding vectors of any two original features; for each original feature, selecting the original feature of which the comprehensive similarity with the original feature meets a preset similarity condition, and generating at least one second derivative feature; and splicing the transformation features, the first derivative features and the second derivative features to obtain a feature set, thereby deeply mining the relationship of the high-missing-rate features, improving the quality of the feature set, and enhancing the model training effect.

The application discloses an active landslide identification method, and relates to the field of disaster identification, which comprises the following steps: acquiring image data of a region to be identified, and constructing an active landslide data set; constructing a context-aware adaptive fusion model; the context-aware adaptive fusion model comprises an encoder and a decoder; the encoder comprises a plurality of feature processing layers and a convolution layer connected in sequence; the feature processing layer comprises a convolution layer and a wavelet transform down-sampling WDB module connected in sequence; the decoder comprises a plurality of multi-branch scale adaptive aggregation MSA modules, and a plurality of feature extraction refinement layers and a convolution layer connected in sequence; the feature extraction refinement layer comprises a convolution layer and a convolution up-sampling CUB module connected in sequence; based on the active landslide data set, the context-aware adaptive fusion model is trained to obtain an active landslide identification model; the active landslide data set to be detected is input into the active landslide identification model to obtain an active landslide identification result.

Embodiments of the present disclosure provide a cumulative data updating method and device, electronic equipment and storage medium. The method comprises: obtaining an event message based on a trigger event; determining the data category according to the event message; generating first data cumulative information according to the event message in response to the first data category; updating the event message staging queue according to the event message in response to the second data category, and writing into the database; periodically obtaining the event message staging data within a preset time period, reading the corresponding event message, and generating second data cumulative information; and updating the cumulative data of the corresponding event type according to the first data cumulative information and the second data cumulative information. Through the method, the accuracy and consistency of the heavy index data accumulation can be ensured while ensuring the system performance, and the problem of data loss caused by network jitter is solved.

This invention relates to an automatic vehicle weighing system based on a dynamic transportation process, belonging to the field of vehicle load detection technology. The system includes: a data acquisition unit for real-time acquisition of deformation angle data from measuring points on the vehicle's suspension system and generation of timestamps and vehicle identifiers; a data processing and communication unit for data preprocessing, local caching, obtaining vehicle location and operating conditions, and enabling local storage to resume transmission when communication is interrupted; a background service and calculation unit, where a dynamic benchmark management module automatically identifies periods of no-load, stationary, and horizontal state, stores the acquired deformation angle data as dynamic no-load benchmark values, and automatically triggers benchmark updates when the deviation between the new benchmark value and the initial benchmark value exceeds a threshold; and a load calculation module calculates the vehicle's load weight based on real-time deformation angle data and dynamic no-load benchmark values, combined with a load calculation model. This system achieves high-precision, maintenance-free dynamic measurement of the vehicle-mounted weighing system and improves its adaptive correction capability against zero-point drift.

The invention belongs to the technical field of autonomous underwater vehicle control, and particularly relates to an actuator saturation fault-oriented AUV fault-tolerant control method, a program, equipment and a storage medium. According to the method, the saturation control item is introduced into the fault compensation of the AUV, the redundancy characteristic of the hybrid actuator can be fully utilized, the force and torque redistribution under the fault is realized through the dynamic fault distribution matrix, and the saturation of the actuator is inhibited by means of the saturation error item of the Lyapunov function. On the premise that the trajectory tracking precision is not lost, the recovery time after a fault is greatly shortened, the output fluctuation of the actuator is reduced, the saturation risk is avoided, the control reliability and stability of the AUV under the fault working condition are remarkably improved, and the situation that the AUV is lost due to the fault generated by saturation of the actuator can be effectively reduced.

The application provides a construction worker unsafe behavior intelligent identification method and early warning system, the method comprises the following steps: obtaining a to-be-detected image of a construction site; inputting the to-be-detected image into a construction worker unsafe behavior identification model to obtain an unsafe behavior identification result of a construction worker; wherein the unsafe behavior at least includes an unworn state, a wrong wearing state or a blocked state of personal protective equipment; and the construction worker unsafe behavior identification model is a neural network model for target detection. The detection accuracy is significantly improved, and the fine-grained identification capability is enhanced. Since a learnable position coding component is introduced into the backbone network, the model can explicitly model the spatial position relationship of the feature map, enhance the spatial position perception capability of the personnel (especially small targets and blocked targets) in the construction scene under complex background interference, solve the problem that the traditional convolutional network is not sensitive to absolute position information, and improve the feature extraction accuracy in a complex background.

The invention discloses a skeleton action recognition method based on a partition space-time collaborative graph convolutional network, and relates to the technical field of computer vision and pattern recognition, and the method comprises the steps: firstly obtaining human skeleton sequence data through a depth sensor or a posture estimation algorithm, and carrying out the preprocessing; then, constructing a partition association graph, dividing part partitions based on a human body natural structure, and establishing in-part stable topology and cross-part semantic topology; then introducing a part dynamic collaborative convolution (PDC-GC) module, and cooperatively extracting global semantic and local dynamic features of the action through part association modeling and dynamic neighborhood convolution; introducing a space-time sub-domain coding (STSE) module, capturing multi-scale space-time dependence by dividing four sub-domains, and reducing calculation complexity in combination with sparse coding; and finally, constructing a partitioned space-time collaborative graph convolutional network (PC-GCN) network, fusing a PDC-GC module, an STSE module and multi-level features, and realizing high-precision skeleton action recognition.