92results about How to "Reliable identification" patented technology

The invention provides an unsupervised continuous emotion electroencephalogram analysis method and device based on deep reinforcement learning, and the method comprises the steps: in a prototype learning stage, obtaining sample features of time sequence signals corresponding to sample objects, carrying out the clustering processing of the sample features based on a K-Means algorithm, obtaining the global distribution information of emotion information in the time sequence signals, and carrying out the clustering processing of the global distribution information; in the reinforcement learning stage, a distribution-prototype reward function is determined according to the global distribution information, an Actor-Critic framework is adopted, a near-end strategy optimization algorithm PPO is used, an Actor executes a current action for a given time sequence signal based on a current strategy, Critic determines a current score of the current action according to a distribution-prototype reward fed back by the environment, and the current score of the current action is determined according to the distribution-prototype reward fed back by the environment. And the Actor performs strategy learning based on the current score until a preset iteration termination condition is reached, and a model used for determining key signals in the target time sequence signals according to the target time sequence signals is obtained. Therefore, the accuracy and reliability of the finally obtained model are improved.

The invention discloses an arc risk situation awareness prediction method and system based on multi-characteristic dynamic mapping, and belongs to the technical field of electrical fault detection.The method comprises the steps that arc characteristics are extracted according to a current waveform of an electric power system, and the arc generation position is determined by combining the propagation rule of the arc characteristics; predicting the development trend of the arc form image and the energy signal according to the arc form image and the energy signal of the current waveform; according to the development trend of the arc characteristics, taking the arc as a heat source, reflecting the contact area of the arc and air by the arc form image, and combining with the heat transfer rule of the arc to obtain the temperature field distribution of the arc; and according to the arc generation position and the temperature field distribution, in combination with the distance perception between the ignition point of the evaluation object and the arc boundary, predicting the arc risk on the installation line of the evaluation object, and making corresponding emergency prevention measures according to the risk evaluation result. And accurate prediction, reliable identification and rapid cutting-off of the fault arc are realized.

The invention relates to the technical field of underwater static target object detection and processing, in particular to an unmanned cross-domain positioning and acoustic fingerprint processing method and system for an underwater static target. Comprising the following steps: performing wide-area scanning on a task sea area, and scheduling an AUV and unmanned aerial vehicle cluster to a target area after discovering a suspicious target; the unmanned aerial vehicle cluster receives an acoustic signal of the AUV and transmits the acoustic signal to the cooperative resolving center in combination with self-positioning data so as to provide accurate coordinate guidance for the AUV; the AUV sails to a target area, parallel processing map construction and accurate positioning, target identification and dual-mode fingerprint generation are carried out, and a target task package is packaged; the ROV receives and analyzes the task packet, autonomously plans a path and sails to a target area, a sonar is started to collect data and generate real-time feature fingerprints, and the real-time feature fingerprints are matched with fingerprints in the task packet; and after matching succeeds, a specific job task is autonomously executed. According to the invention, a complete automatic solution is provided for scenes such as deep and far sea detection, emergency salvage, pipeline maintenance and the like.

The invention provides an elevator personnel existence detection method and system based on a millimeter wave radar, and relates to the technical field of detection, and the method comprises the steps: orthogonally arranging a positioning radar and a motion detection radar at the top of an elevator car, collecting radar echo signals, and carrying out baseband demodulation; a positioning baseband signal and a motion detection baseband signal are processed through a dynamic threshold elimination method and a self-adaptive harmonic suppression algorithm respectively, and compensation enhancement is carried out in combination with a geometric structure of the lift car; performing fast Fourier transform on the enhanced signal in the dimensions of distance, speed and height, acquiring three-dimensional coordinate data of a target by adopting an adaptive spatial spectrum decomposition positioning algorithm, and performing non-uniform sampling Doppler processing to obtain target motion track data; a recursive state estimation model is adopted for time sequence tracking, human body height judgment is achieved, and the method is used for safety monitoring in the lift car, and early warning for children and pets entering the lift car independently and early warning for long-time staying in the lift car.

The invention provides a joint design method and system for water flow estimation and ship waterline monitoring, and belongs to the technical field of ship monitoring, and the method comprises the following steps: firstly, a shore-based center deploys an underwater robot and plans a motion path according to the needs of a monitoring task, and builds a water gauge data set for model training; when the underwater robot arrives at the nearby area of the ship, the underwater robot measures the water flow speed and direction by using a Doppler velocimeter, and predicts a future water flow velocity field in combination with the trajectory information of the underwater robot; meanwhile, the underwater robot collects a waterline image through an airborne camera, pre-processes the waterline image, and calculates the credibility of the waterline image; and further, the underwater robot corrects the attitude in real time according to water flow velocity field prediction and waterline image credibility, and joint estimation of the water flow and the ship waterline is realized. According to the invention, water flow estimation and ship waterline monitoring are combined, and stable and efficient omnibearing ship waterline monitoring is realized through the underwater robot.

The invention discloses a vehicle identification method in the technical field of vehicle re-identification, and aims to solve the technical problems that the vehicle identification precision is low and visually similar vehicles cannot be distinguished. The method comprises the following steps: obtaining local fine-grained features by using a graph structure, and recognizing a vehicle through fusion features obtained by fusing the local fine-grained features and global features, thereby improving the recognition capability and robustness of a system, improving the effect of the system in coping with environmental inconsistency in the real world, ensuring reliable vehicle recognition, and improving the vehicle recognition efficiency. The accuracy and the key area identification capability are improved, and visually similar vehicles are distinguished.

The invention provides a water conservancy safety detection method based on AI edge calculation, and the method comprises the steps: deploying a sensor in a water conservancy system according to the installation and coding of the sensor, coding the sensor, and recording the geographic position information and measurement physical quantity of the sensor; setting measurement parameters, and setting a measurement interval value and a threshold value of each sensor; data linkage setting: defining data linkage selection data, and triggering pump IO output when the data reaches a specific interval threshold value; the method comprises the following steps: monitoring sensor data in real time, if the data is abnormal, starting a video AI to check, comparing whether the result is consistent, if the video AI confirms that the sensor data is abnormal, intercepting a real-time video as sample data, sending the sample data to a cloud server, triggering a corresponding alarm according to the abnormal condition, and notifying related personnel to process. The accuracy and timeliness of abnormal condition detection are improved, and a powerful guarantee is provided for safe operation of a water conservancy system.

According to the abnormal wear monitoring method for the disc cutter of the shield tunneling machine, tunneling parameters closely related to the wear state of the disc cutter are screened according to the abnormal wear monitoring requirement of the disc cutter in the tunneling process of the shield tunneling machine, and data missing caused by the collection problem is filled up. According to the long and short term memory self-encoder model based on the abnormal attention mechanism, prior correlation in an original sequence is obtained through a learnable Gaussian kernel, the distinguishability between normal and abnormal conditions is improved by obtaining correlation differences of the sequence, and accurate and reliable abnormal wear monitoring is conducted on the shield tunneling machine disc cutter in combination with the extreme value theory. According to the method, the tunneling parameters closely related to the running state of the disc cutter are identified more accurately, and the model training precision and the accuracy of abnormal wear monitoring of the disc cutter are improved. The method is of great significance to the application of abnormal wear monitoring of the disk cutter of the shield tunneling machine.

The invention discloses a tower climbing obstacle identification and positioning method based on machine vision. The method mainly comprises the following steps: S1, acquiring comprehensive environment and obstacle information and monitoring illumination conditions in real time; s2, developing an adaptive illumination compensation algorithm, and dynamically adjusting image processing parameters according to real-time illumination conditions; s3, performing semantic segmentation on the image, identifying different types of obstacles, and taking different countermeasures according to the characteristics of the obstacles; s4, constructing a high-precision real-time positioning system, accurately positioning the identified obstacle, and generating three-dimensional coordinate information of the obstacle; s5, tracking the position change of the obstacle in real time in the tower climbing process, and continuously updating the position information of the obstacle; and S6, designing an intelligent path optimization algorithm, planning a safe tower climbing path, and dynamically adjusting the tower climbing path in the tower climbing process. According to the method, misjudgment caused by insufficient illumination or overexposure is greatly reduced, and the safety of tower climbing operation is improved.

The invention relates to the technical field of computer vision and image processing. The invention provides a large model-fractal combined assisted mineral image intelligent segmentation and identification method, which comprises the following steps: acquiring mineral image data, and preprocessing to obtain a mineral image data set; constructing an initial joint model based on the SAM and the fractal model, and training the initial joint model to obtain a joint model; performing optimization processing on the joint model through the mineral image data set to obtain an optimized joint model; segmenting the to-be-processed mineral image, and marking boundaries of mineral particles in the to-be-processed mineral image; and fractal analysis is conducted on the labeled mineral particles through the optimized joint model, and segmentation masks and fractal quantitative information of the mineral particles are output. The problems that a traditional mineral image analysis method depends on manual observation and manual labeling, time and labor are consumed, and analysis results are inconsistent and uncertain due to the fact that the traditional mineral image analysis method is prone to being affected by subjective factors are solved.

The application discloses a baby crying sound recognition method for white noise equipment, relates to the field of audio processing and intelligent acoustic recognition, and comprises the following steps: collecting pure reference signals, first signals and second signals for synchronization and preprocessing, and obtaining mixed audio signals based on the first signals; calculating residual signals through a self-adaptive echo cancellation algorithm; calculating acoustic masking parameters based on the residual signals, the pure reference signals and environmental noise estimation; constructing a three-level recognition processing path comprising light feature detection, registered voiceprint comparison and multi-modal information fusion; selecting a recognition processing path based on the numerical range of the acoustic masking parameters, confirming the crying event of the target baby from the residual signals; and triggering corresponding graded alarms based on the crying event confirmation result. The three-level recognition path is scheduled through the acoustic masking parameters, multi-modal information is fused, and accurate and low-power baby crying monitoring under strong interference is realized.

The invention discloses a puncture tissue recognition and transmembrane event control method and system based on force sense and image fusion, and the method comprises the steps: collecting a force signal at the tail end of a puncture needle, a real-time ultrasonic image sequence and the pose data of a mechanical arm, and carrying out the timestamp alignment to construct a synchronous data set; on the basis, tissue stiffness, a force change rate, extreme value features, texture features, envelope boundary features and optical flow displacement features are extracted, the tissue level where the puncture needle is located is recognized through multi-modal fusion, and a membrane penetrating event is detected in combination with the force sense extreme value features and envelope boundary fracture; according to the tissue level and the transmembrane event, event driving control strategies such as constant-speed propulsion, deceleration early warning, rotation assistance, transmembrane braking and path compensation are automatically triggered, so that the puncture process is accurately regulated and controlled. The accuracy of puncture tissue recognition and the reliability of transmembrane event judgment can be remarkably improved, and the safety and stability of puncture operation are improved.

The invention relates to a method and a radar arrangement (12, 18) for radar-based dimensional classification of an object (14). The invention also relates to a correspondingly designed motor vehicle (10). In the method, radar signals (24) reflected by an object (14) are detected in a plurality of receiving channels (20). The detected reflected radar signal (24) is then evaluated with respect to its phase profile over the plurality of receiving channels (20). In the event of a phase jump in the phase curve, the respective object (14) is classified as an extended object (14).

The application provides an anti-malicious anchor node target node positioning method, device and equipment, and belongs to the field of network security and signal processing. The method comprises the following steps: acquiring the received signal strength measurement data between nodes, establishing a received signal strength nonlinear measurement model containing an unknown attack strength parameter, further converting the multiplicative noise item in the established model into an explicit additive equivalent noise item, constructing an optimization model with the weighted residual square sum minimization as an objective function, introducing an L1 norm regularization term for the attack strength parameter, coupling the identification of the malicious anchor node and the position estimation of the target node into the same optimization problem to obtain an initial optimization model, introducing an auxiliary variable for relaxing the non-convex item, converting the non-convex constraint into a convex constraint and solving the same to determine the target node position, and providing a reliable basis for subsequent network security maintenance and node trust management.

This application discloses a method for training a recognition model, a recognition method, a computer device, and a computer-readable storage medium. The training method includes the following steps: controlling a single-photon lidar to emit photon pulses towards a target scene and receiving one-dimensional single-photon echoes reflected from the target scene, where the target scene includes at least one object to be recognized; constructing a training sample set based on the one-dimensional single-photon echoes; constructing an initial model and inputting the training sample set into the initial model for iterative training; validating the initial model after each iteration, marking the validated initial model as the recognition model, and outputting it. The application method involves inputting the echo signal to be recognized into the trained recognition model, and obtaining the recognition result based on the processing of the recognition model. The recognition result includes the pose and type of the object in the target scene. Therefore, this application can achieve high-precision recognition of long-distance targets.

The present invention discloses an LED flash control circuit and control method capable of detecting water ingress and egress, comprising a power conversion circuit electrically connected to a power source, a single-chip computer system electrically connected to the power conversion circuit, a water ingress and egress detection circuit and a power control circuit, the water ingress and egress detection circuit electrically connected to the single-chip computer system, the single-chip computer system electrically connected to the power control circuit, the power control circuit electrically connected to the light detection circuit, and also comprising a light detection circuit electrically connected to the single-chip computer system, an LED control circuit and a flash frequency setting circuit. The control circuit disclosed in the present invention has the advantages of simple circuit structure, low cost, long effective working time, adjustable LED flash frequency, high reliability, etc. The control method disclosed in the present invention can enable the control circuit to stably and reliably identify water ingress and egress status, thereby extending the service life and effective working time of the equipment.

The invention relates to the technical field of radar calibration, and discloses a radar calibration method in a mixed environment. The method comprises: collecting radar echo signals and filtering the collected echo signals to obtain filtered echo signals, performing adaptive signal separation on the filtered echo signals to identify target echo signals and clutter signals, tracking the identified target echo signals to estimate target states, calibrating target position trajectories through sequential analysis, performing error estimation based on the target position trajectories obtained through analysis and the actual target position estimation, calibrating radar antennas based on the estimated errors, and improving the accuracy of radar calibration.

This utility model discloses a beauty head recognition circuit and a beauty device. The beauty device includes a main unit and a beauty head detachably mounted on the main unit. The main unit is equipped with an electroacupuncture assembly for interfacing with the beauty head. The beauty head recognition circuit includes a microcontroller module and several recognition modules. One end of each recognition module is connected to an electroacupuncture pin in the electroacupuncture assembly, and the other end is connected to the input port of the microcontroller module. When no beauty head is installed on the main unit, none of the electroacupuncture pins connected to the recognition modules are grounded. When the beauty head is installed on the main unit, at least one electroacupuncture pin connected to the recognition module is grounded, so that the level signal provided by the recognition module to the microcontroller module is different from that when no beauty head is installed. The microcontroller module can thus reliably identify the access status of the beauty head based on the level signals provided by the several recognition modules.

This invention discloses a method for rapidly detecting conventional and hybrid varieties of industrial chili peppers and its application, relating to the field of molecular breeding technology. The method includes: extracting DNA from fresh leaves of the industrial chili pepper to be tested and performing whole-genome resequencing at a sequencing depth of 8×–12×; aligning the sequencing data to the chili pepper reference genome (Capsicum chinense cv. 'PBC932') to screen for high-quality heterozygous SNP loci; using a 1 Mb sliding window to statistically analyze the distribution density of heterozygous SNPs across the entire genome; and determining the variety type based on density thresholds: an average density <300 SNPs / Mb indicates a conventional variety, and >2000 SNPs / Mb indicates a hybrid variety. This invention eliminates the need for phenotypic observation in subsequent generations, requiring only leaves from current-generation plants to rapidly and accurately distinguish conventional and hybrid varieties at the molecular level. It features a short detection cycle, objective and reliable results, and is not limited by the number of molecular markers, providing strong technical support for efficient identification and early screening of industrial chili pepper germplasm resources.

This invention discloses a lightweight method and system for identifying smoke areas in photovoltaic (PV) plants based on an improved YOLO framework. It relates to the field of image recognition technology and includes: multi-point monitoring feature mining based on historical smoke detection logs of the PV plant; real-time monitoring of a multi-sensor array controlled by a monitoring attention-configured graph network to obtain visual and thermal image streams of the plant area; introducing an improved YOLO multi-level learning architecture for visual and thermal smoke probability detection; interference tracing correction and time-dependent aggregation of the first and second smoke detection graph networks; and performing smoke area identification based on the reliable smoke detection graph network to obtain a smoke area identification graph network. This invention solves the technical problems of low accuracy, poor real-time performance, and insufficient reliability in smoke area identification in complex and vast PV plant environments, achieving accurate, reliable, and adaptively focused real-time smoke identification in PV plant areas.

The application discloses a detection method for a QUIC protocol denial of service attack (QDA) in a backbone network. The method can realize detection of a composite denial of service attack under the QUIC protocol. The detection method comprises an offline training stage and an online detection stage. In the offline training stage, according to the attack characteristics of the composite QDA, representative one-way traffic feature data is extracted, and corresponding feature groups are constructed. These features can effectively deal with the case that there is a large amount of one-way traffic in the backbone network node. Finally, a specific machine learning method is used for training to obtain an attack detection model. The application can be used for QDA attack behavior detection and early warning in a massive traffic scene such as a backbone network, and provides a basis for maintaining the security of the network.

The invention discloses a voice interaction graphene AI intelligent tea table control method, and relates to the technical field of home voice interaction, and the method comprises the control steps: S1, monitoring a surrounding user instruction based on a built-in microphone of a tea table, and after the tea table receives the voice instruction, synchronously carrying out the preliminary noise reduction processing, and retaining the voice segment of the instruction; s2, training a data set containing a noise scene, enabling a built-in model to learn features for distinguishing voice and noise, and filtering non-voice signals in real time; and S3, for dialects and accents in different regions, collecting a scale data set, training a dialect recognition model, and inputting the obtained voice signal into the dialect recognition model to recognize and judge the current dialect and accents. Through preliminary noise reduction and microphone array layout, the voice acquisition quality is improved, and instruction omission is reduced; secondly, by training a multi-noise scene data set, the adaptability of the model to a complex environment is enhanced, and reliable recognition is guaranteed; and aiming at dialects and accent optimization, the product universality is improved.

The invention discloses a numerical control lathe workpiece positioning method, device and system based on magnetization characteristics, and the method comprises the steps: carrying out the pre-scanning of the surface of a to-be-transferred workpiece on a station, obtaining the magnetic field intensity distribution data, and obtaining an optimal clamping point; processing the magnetic field intensity distribution data to obtain workpiece surface unstable area distribution data; carrying out magnetic moment analysis to obtain an optimal placement angle; selecting a three-point supporting placement scheme based on the optimal placement angle; and performing mechanical simulation analysis to obtain a judgment result of the stable state of the workpiece. According to the technical scheme, active pre-judgment and control over the magnetization state in the placing process are achieved through accurate recognition and analysis of the magnetization characteristics of the surface of the workpiece, the problem that the workpiece displaces on the measuring platform due to the magnetization effect is effectively solved, and the measuring precision is guaranteed.

This invention provides an open-set encrypted traffic classification method and system based on a structured fine-tuning pre-trained model, comprising: preprocessing the original encrypted traffic into a lexical sequence that can be processed by a language model; fine-tuning the pre-trained language model using the lexical sequence: performing structured guidance operations on the lexical sequence such as header field rearrangement and / or flow packet segmentation and representation to guide the model to learn the structured semantics of the traffic; extracting aggregated feature vectors of the traffic sequence to be classified using the fine-tuned model; and based on the feature vectors, using an open-set classification algorithm based on distance metrics to determine whether it belongs to an unknown category, and if not, further determining its specific known category. This invention can guide the model to deeply understand the protocol and packet sequence structure of encrypted traffic, thereby achieving both high-precision classification of known traffic and effective identification of unknown traffic in open-set scenarios.

The invention discloses a vehicle control method and system, a storage medium and a vehicle. The method comprises the steps that in the process that a vehicle executes an operation task in a target area, image information around the vehicle is obtained, and the image content of the image information at least comprises obstacles around the vehicle and an operation object corresponding to the operation task; a recognition model is utilized to recognize contour features of the operation object in the image information to obtain a recognition result of the operation object, and the recognition model is obtained by training a convolutional neural network model by utilizing an image information sample containing an operation object sample and a recognition result sample of the operation object sample; based on the identification result, determining a control strategy for the to-be-executed job task of the job object; and in response to a control instruction corresponding to the control strategy, controlling the vehicle to execute an operation task on the operation object at the target operation position. The technical problem that the control accuracy of the vehicle is low is solved.

The invention discloses a Hash and feature selection-based software and hardware equipment fingerprint fine-grained generation method, and belongs to the technical field of power network software and hardware equipment fingerprint biomarker generation. Based on the information entropy and the self-correlation coefficient, the features used for generating the device fingerprints are screened out from the numerous features, representative and stable features are extracted from the device fingerprints, and selected feature parameters are converted into hash codes used for the device fingerprint generation and matching step; different devices are identified and distinguished hierarchically and accurately through a fine-grained device fingerprint generation step based on feature selection, a hash code and a Bayesian network.

The application provides a computer vision defect detection system and method on an electric control board production line, and relates to the technical field of data processing.The method comprises the following steps: according to collected multi-angle images, combining illumination conditions and visual angle information, the overall area of the electric control board is partitioned; image block processing is performed to generate a plurality of image sub-areas, and texture features, brightness distribution features and geometric edge features are extracted in each image sub-area; rare defect feature enhancement processing is performed, low-frequency abnormal textures are repeatedly superimposed in multiple scales, and the edge fracture is directionally extended; the corresponding normal area combined features are compared and differenced, and normalized correction is performed in combination with the illumination conditions and the visual angle information; multi-angle recurrence degree analysis is performed, when the same suspected defect is detected under different angles, a reliable defect area is formed, otherwise, an interference area is removed, and an electric control board defect detection result is generated; and the application improves the accuracy of defect detection.

The invention belongs to the cross technical field of block chains and machine learning, and provides an abnormal account detection method based on graph continuous learning, and the method comprises the specific steps: 1, obtaining normal and abnormal account data with labels, and extracting account features; dividing the obtained accounts into a plurality of groups of tasks; 2, performing model training on the graph neural network model by using a group of tasks; 3, obtaining the importance score of the model parameters to the loss of the current task and the importance score of the topological structure, and calculating the final importance score of all the parameters to the current task; and 4, when the tasks are not trained completely, updating the loss function based on the final importance score, returning to the step 2 to continue training of the next group of tasks, and when the tasks are trained completely, using the obtained model for detection of abnormal accounts.

The invention relates to a strategy identification system for smart city management. The strategy identification system comprises a dynamic triggering mechanism, a high-angle shooting imaging mechanism, bilateral filtering equipment, maximum value filtering equipment, data sharpening equipment, strategy identification equipment and the like. The strategy identification system for smart city management is compact in design and intelligent in control, and the reliability and effectiveness of smart city management can be enhanced.

The application discloses a photovoltaic module defect detection method and a detection device, and the method comprises the following steps: receiving a current environment signal detected by an environment monitoring module; controlling a power inverter and an image acquisition module to work according to the current environment signal, so that the power inverter applies a preset mode of power disturbance to the photovoltaic module, and the image acquisition module scans the photovoltaic module according to a preset cruise route when the power inverter applies the disturbance to the photovoltaic module, and a plurality of thermal imaging images of the surface of the photovoltaic module are acquired; the starting time of the power inverter and the image acquisition module is less than 1ms; each thermal imaging image is received, and the defects of the photovoltaic module are determined according to each thermal imaging image. By using the above method, the temperature signal of the early defect can be actively enhanced, the subtle defects such as hidden cracks and virtual welding can be reliably identified, the millisecond-level accurate synchronization of excitation and acquisition is realized, the interference such as environmental temperature drift is effectively inhibited, and the signal-to-noise ratio of the detection signal and the result repeatability are greatly improved.