170results about How to "Reduce false negative rate" patented technology

The invention provides an Internet of Things data acquisition and abnormity early warning method and system, and relates to the technical field of Internet of Things data processing, and the method comprises the steps: constructing a digital twin model for a physical entity at a cloud end, and issuing prediction data to an edge node; acquiring sensor data by an edge computing node at a basic frequency, and computing a residual error between real-time data and predicted data and an information entropy of a residual error sequence; dynamically adjusting the data acquisition frequency of the edge node based on a two-dimensional decision space formed by the residual error and the information entropy; and when the residual error and the information entropy both exceed the threshold values, potential abnormity is judged, an edge side abnormity analysis program is triggered, and abnormal event information is uploaded. According to the invention, through cloud edge cooperation and intelligent decision making, the acquisition frequency is reduced in a stationary period, resources are saved, the frequency is improved when an abnormal symptom appears, monitoring is enhanced, and optimal balance between resource consumption and monitoring precision is realized; double criteria are adopted to effectively filter noise interference, and the false alarm rate and the missing report rate are reduced.

The invention discloses a coal mill internal multi-source heterogeneous data fusion early warning method and system, and belongs to the technical field of industrial equipment fault diagnosis. The method aims at solving the technical problems that an existing coal mill early warning method is insufficient in multi-source data fusion, poor in working condition adaptability, low in early warning accuracy and lack of fault tracing and accurate control. According to the technical scheme, the method comprises the steps that structured process parameter data and unstructured physical field data of the coal mill are obtained; constructing a digital antigen vector representing a running state; constructing an autologous space based on the historical health data, and generating a dynamic immunodetector set covering a non-autologous space; calculating the Euclidean distance between the digital antigen vector and the detector and a danger signal index; early warning is triggered through cooperative judgment, feedback control is executed, and a fault source is positioned in combination with fault tracing. The device state can be comprehensively represented, the working condition adaptability and the early warning accuracy are improved, accurate fault positioning and control are achieved, and the device maintenance cost is reduced.

The invention belongs to the technical field of geological disaster monitoring and early warning, and discloses a slope geological disaster multi-source remote sensing image intelligent early warning method and system, and the method comprises the steps: a multi-source remote sensing data adaptive collection step: constructing a multi-source remote sensing data cube; a multi-scale spatio-temporal feature coupling extraction step: generating a spatio-temporal coupling feature vector based on an improved Swin Transform architecture and an adaptive spatio-temporal attention mechanism; a disaster risk dynamic assessment and feedback step: calculating a disaster risk level based on a multi-modal deep learning network, and dynamically adjusting a feature fusion weight according to a risk assessment result to form closed-loop feedback; and an intelligent early warning decision and response step: constructing a multi-stage early warning threshold system to trigger early warning and generate an early warning report, the slope geological disaster early warning method realizes multi-scale spatial-temporal feature adaptive fusion and a closed-loop cooperation mechanism, and significantly improves the accuracy and intelligent level of slope geological disaster early warning.

The invention belongs to the technical field of e-commerce logistics management, and particularly relates to an e-commerce product storage full-process monitoring management method, which comprises the steps of synchronously acquiring multi-source heterogeneous monitoring data in real time through a sensing equipment network; performing space-time alignment preprocessing on the original data, and converting heterogeneous data into a uniform data format; semantic fusion is carried out on the standardized data, and a job behavior knowledge graph with a complete spatio-temporal context is constructed; inputting the atlas into a pre-training hybrid model, and detecting an inventory abnormal event; and generating a strategy instruction in combination with the real-time order and resource data, carrying out joint analysis on an operator, and outputting a compliance score and a specific violation operation type. Through adoption of the technical scheme, complete reconstruction of the context of the operation behavior can be realized, the anomaly detection accuracy is remarkably improved, the abnormal event response delay is shortened, and the inventory loss rate caused by human errors is reduced.

The invention provides a vulnerability early warning method and system based on asset fingerprint driving. The method comprises the following steps: acquiring vulnerability information data from a plurality of vulnerability information sources; asset fingerprint information of assets in the target system is collected; based on semantic analysis, version comparison and a fingerprint matching algorithm, performing fine-grained association matching on the vulnerability information data and asset fingerprint information, and identifying target assets affected by vulnerabilities; performing influence range analysis, risk level evaluation and repair priority ranking on the target assets to generate a vulnerability influence analysis report; based on the vulnerability impact analysis report, sending an early warning notification to a related person in charge, and automatically generating an electronic work order; the processing state of the electronic work order is tracked, the repair progress is recorded, and full-life-cycle closed-loop management of vulnerabilities from recognition to repair is achieved. The missing report rate and the false report rate are effectively reduced, and compared with manual matching, the full-automatic mode can form a complete closed loop, and the early warning efficiency is improved.

The utility model discloses intelligent sensing radar monitoring equipment, and relates to the technical field of monitoring equipment, the radar monitoring equipment comprises a shell, and the interior of the shell is hollow; the monitoring camera is fixedly arranged on the shell and is used for monitoring a preset monitoring area; the circuit board is fixedly arranged in the shell, and a radar sensing device and a main control device are arranged on the circuit board; the radar sensing device is used for sensing a moving object or a static object in the predetermined monitoring area, and the main control device is connected with the monitoring camera and the radar sensing device through different interfaces and is used for sending the moving object or the static object to the radar sensing device under the condition that the radar sensing device senses an object of a preset type. And controlling the monitoring camera to start video recording. The utility model aims to improve the sensing accuracy of the sensing monitoring equipment.

This invention relates to an IMU-assisted online GNSS anomaly detection and reporting method and apparatus. The method involves acquiring GNSS observation data and IMU measurement data and aligning them in time; generating a predicted motion state of the receiver at the current observation epoch based on the IMU measurement data; calculating the predicted GNSS observation value of at least one satellite at the current observation epoch based on the predicted motion state and satellite ephemeris information; constructing a residual sequence based on the predicted GNSS observation value and the corresponding GNSS observation value in the GNSS observation data, performing sliding time window statistics, and determining anomalies based on the statistical results; when an anomaly is determined, generating a time index and a spatial index corresponding to the anomaly, forming a spatial-level interference event, and reporting it. This invention effectively reduces the false alarm rate and missed alarm rate in scenarios involving obstruction, multipath interference, and other interference, exhibits higher sensitivity to observation anomalies caused by moving interference, effectively lowers the implementation threshold at the receiving end, and possesses high engineering application value.

The invention belongs to the technical field of nondestructive testing of concrete structures, and relates to an urban underground comprehensive pipe gallery concrete engineering quality evaluation method which comprises the following steps: acquiring an original acoustic emission signal and performing structured processing to generate a structured acoustic emission event data object containing multi-dimensional feature information; carrying out feature correction and verification on the data object in combination with the environment data, and generating a focusing scanning instruction containing a target three-dimensional coordinate and a risk confidence coefficient; generating a parameterized adaptive focusing scanning configuration packet through physical parameter reverse mapping based on energy spectrum characteristics, driving a phased array ultrasonic system to perform targeted imaging on a target area and extracting acoustic image characteristics of defects; and finally, comparing the correlation between the data object and the image features, executing closed-loop verification, and adaptively updating discrimination parameters for risk determination. The technical problems of insufficient evaluation accuracy and verification lag caused by mutual separation of passive monitoring and active detection in the prior art are solved.

The present application relates to a kind of laser radar and camera online self-calibration method and system without induction, the method includes: S1.in the process of vehicle driving, through multi-source trigger mechanism acquisition laser radar point cloud data and camera image data, simultaneously obtain vehicle motion state information and automatic driving perception system output result, the multi-source trigger mechanism includes time trigger, event trigger and cloud data mining trigger parameter degradation model;S2.the pre-processing and feature extraction of the multi-source data collected, obtain laser radar point cloud feature and camera image feature;S3.based on geometric constraint, motion consistency and perception result fusion, constructs multi-modal verification model, and carries out multi-dimensional verification to calibration parameter.The present application not only reduces the artificial maintenance cost, through automatic means maintenance vehicle state to achieve sensor calibration parameter degradation self-detection and self-calibration without induction, and improves the reliability and attendance rate of automatic driving vehicle.

The embodiment of the application discloses an industrial internet threat identification method and system, the method comprises the following steps: collecting multi-source data in the industrial internet scene in real time through a bypass deployed collection terminal; pre-processing and feature extraction are performed on the multi-source data to obtain first features and second features; the first features and the second features are fused to obtain fused features; the fused features are identified by using a threat identification engine to obtain a threat identification result. The application has the following advantages: 1. The single-dimensional detection limitation is broken, the coordinated attack can be effectively identified, and the false positive rate and the false negative rate are reduced; 2. The known threat and the unknown threat are fully covered, and the high safety requirement of the industrial system can be met; 3. The attack chain correlation analysis across domains and time is supported, the attack process can be completely restored, the attack influence range is clear, and accurate basis is provided for threat disposal; 4. The industrial production network is not invaded, the normal communication process is not disturbed, and the core requirement of the industrial scene, that is, low invasion and high availability, is met.

The application relates to a hand-off detection method and device. The method comprises the following steps: acquiring a hand force torque value corresponding to a steering wheel and a response torque value; acquiring a zero offset compensation value, which is used for offsetting the influence of zero point bias of a torque sensor on a resultant torque of the hand force torque value and the response torque value; determining a target resultant torque according to the hand force torque value, the response torque value and the zero offset compensation value; judging whether the target resultant torque is greater than a first threshold value; and if not, determining that a driver has taken off the steering wheel. The scheme provided by the application can overcome the difference between different vehicles, improve the accuracy of hand-off detection, and reduce the false positive rate and the false negative rate.

The application discloses an AI fault diagnosis and auxiliary decision method and system for a lithium battery, and relates to the technical field of lithium battery safety monitoring.The application continuously collects transient stress waves generated by internal bubble rupture through a piezoelectric acoustic emission sensor, extracts the main ridge line center frequency, energy attenuation time constant and duration of each acoustic emission event by using adaptive time-frequency analysis and time-frequency ridge line tracking, constitutes a feature vector, inputs the feature vector into a pre-trained deep learning AI event classification model, accurately distinguishes normal electrochemical reaction bubble events from electrolyte boiling bubble events, counts the boiling bubble event density in a sliding time window, combines an adaptive safety threshold and a fault severity factor to determine whether a local hot area and a gas film have been formed inside the battery cell, and divides the fault level, captures the acoustic emission characteristics of electrolyte boiling, realizes early warning of thermal runaway, and takes into account the timeliness of safety protection and the controllability of vehicle operation.

The application relates to the technical field of hydrogeological monitoring, in particular to a hydrogeological dynamic monitoring and analysis system and method based on big data; specifically, hydrogeological data is collected in real time, data denoising, space-time calibration and error correction are carried out in combination with edge computing, noise and drift error are eliminated by adopting variational mode decomposition, geological model constraint and wavelet transform, and the storage and transmission efficiency is optimized by increment encoding and Huffman encoding, and a quadratic form check code is generated to ensure data reliability; a space-time deep learning model and a self-attention mechanism are fused to predict a hydrological evolution trend, adaptive anomaly detection is realized by combining DBSCAN clustering and prediction error analysis; water resource allocation strategies are optimized through reinforcement learning, risk levels are evaluated in layers, and management schemes are generated, and finally, monitoring data, abnormal scores and decision results can be visually displayed. The application integrates multi-source data fusion, space-time modeling and intelligent decision-making, and improves hydrological monitoring precision and response efficiency.

The application discloses a kind of high-altitude operation personnel rule violation identification method based on gesture capture, including the following steps: generating high-altitude operation image frame sequence;Scene reference reconstruction is carried out, and high-altitude operation scene reference data is generated;Personnel region anchoring is generated by carrying out human contour segmentation;Personnel gesture observation sequence is generated;Human posture capture is carried out, and operation personnel gesture joint sequence is generated;Gesture trajectory condenses, and personnel gesture trajectory data is generated;Coupling data of gesture boundary is obtained by gesture-boundary coupling projection;Using improved CleverCatch model, candidate rule violation data is generated;Temporal continuity verification and gesture stability check are carried out, and effective rule violation data is generated, and high-altitude operation personnel rule violation identification result is obtained, improve the accuracy and stability of high-altitude operation personnel rule violation identification, reduce artificial patrol omission and ordinary video identification false alarm.

The invention provides a credible analysis method, device and equipment for an electric meter image and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: determining a toughness spectrum energy field of an electric meter sub-region image according to the intensity difference of the edge or texture of the electric meter sub-region image in different directions, the toughness spectrum energy field is used for measuring the acceptance of the electric meter sub-region image to different disturbance degrees, and the electric meter sub-region image is obtained by dividing the electric meter image according to different components; acquiring a multi-path anomaly detection result of the electric meter sub-region image, and acquiring a directional disturbance strategy of the electric meter sub-region image according to the multi-path anomaly detection result; generating a disturbance bubble image set of the electric meter image according to the toughness spectrum energy field and / or the directional disturbance strategy of the electric meter sub-region image in the electric meter image; and performing time evolution rationality analysis on the electric meter image according to the disturbance bubble image set to determine whether the electric meter image is credible, and according to the scheme of the invention, whether the electric meter image is counterfeited can be accurately identified.

The present application relates to the technical field of gene mutation detection, and particularly relates to a system, method and application for detecting exon skipping mutation.The present application constructs a self-catalytic system based on specific double recognition induction of CRISPR-Cas12a, which is used for sensitive and reliable detection of MET 14 exon skipping mutation of lung cancer patients.The system can effectively release the fluorescence signal through the Cas12a auxiliary cutting, and further realize signal amplification through the self-catalytic mechanism.The present application can solve the technical problems of the existing exon skipping mutation detection system, such as complex detection process, long time consumption, and unsatisfactory detection accuracy.The detection system of the present application has high sensitivity, stability and good reproducibility, and the method has low cost, simple operation, short detection period, good specificity, and low false positive rate and false negative rate.The present application is suitable for clinical sample determination and has an ideal application prospect.

The invention discloses an insulator for a direct current converter transformer and an online monitoring method thereof, and relates to the technical field of insulators. According to the insulator for the direct current converter transformer, the insulator body and the connection structure are internally integrated with a multi-stage state monitoring system comprising a first-stage sensing unit, a second-stage sensing unit and a third-stage sensing unit; according to the invention, three types of key state data of internal mechanical stress of the insulator, the connection state of the first metal accessory and the second metal accessory and the overall temperature distribution can be synchronously acquired in real time, real-time monitoring of the insulator and multi-dimensional comprehensive information acquisition are realized, early warning is realized based on the multi-dimensional comprehensive information, and the reliability of the insulator is improved. According to the method, preliminary early warning can be carried out according to single data, meanwhile, time-space alignment and comprehensive judgment can be carried out on multi-dimensional data, high-level fault early warning is carried out, the stability and reliability of the insulator are improved, and sudden faults are reduced.

The invention discloses intelligent online monitoring equipment applied to a cable joint, which belongs to the technical field of cable detection and comprises a transmission optical fiber, a fiber bragg grating sensor, an optical fiber demodulator, a data processing unit and an intelligent early warning unit. The transmission optical fiber is used for monitoring signal transmission; the sensor is installed at a key temperature measurement part of a cable joint and is connected in series through optical fiber fusion to form a sensing network. The optical fiber demodulator realizes multi-joint positioning monitoring and wavelength offset acquisition through a wavelength division multiplexing or time division multiplexing technology; the data processing unit converts the wavelength offset into temperature variation; and the intelligent early warning unit adopts a convolutional neural network, a bidirectional long-short-term memory network and a self-attention mechanism mixed model to realize fault grading early warning and temperature rise trend advanced pre-judgment. The cable joint fault early warning system does not need field power supply, is high in anti-electromagnetic interference capability, can realize all-weather online monitoring and accurate positioning, and effectively improves the accuracy and timeliness of cable joint fault early warning.

This invention relates to the field of power equipment monitoring technology, and discloses a method and system for assessing the insulation status of medium and high voltage switchgear based on data fusion. The method includes: collecting partial discharge pulse data, slowly changing state data, and reference environmental data for each assessment zone; performing time-series correlation processing on the partial discharge pulse data to obtain transient candidate events; and conducting a first-stage judgment based on temporal proximity, spatial consistency, propagation order, and phase convergence to obtain co-source discharge events and source assessment zones; retrieving slowly changing state data and performing delayed closure processing; conducting a second-stage judgment on the processing results to obtain valid degradation events; and performing zoned sequence analysis on the valid degradation events to obtain degradation evolution results; and outputting the insulation status assessment results of the source assessment zones based on the degradation evolution results. This application improves the accuracy, stability, and targeted maintenance of insulation status assessment and graded early warning.

The invention relates to a machine room ceiling condensation and leakage monitoring and processing method, device and equipment and a medium. The method comprises the steps that environment temperature and humidity data and leakage detection signals in a machine room are collected in real time; determining a condensation risk index corresponding to each cabinet in the machine room through the temperature and humidity data and preset equipment arrangement parameters; and determining a corresponding dehumidification strategy and / or drainage strategy based on the condensation risk index and the leakage detection signal, and correspondingly processing the machine room. The device has the effect of improving the machine room ceiling accumulated water and leakage treatment efficiency.

This invention discloses a slurry sensing and adaptive control method for the homogenization process of lithium-ion batteries, relating to the field of lithium-ion battery production technology. The method includes the following steps: obtaining the electrolyte concentration matrix and slurry flow velocity field of the target object; performing differential fusion on the electrolyte concentration matrix and the slurry flow velocity field to generate a local coupling index; performing weighted autoregression calculation on the electrolyte concentration matrix, and correcting the weighted autoregression calculation result based on the local coupling index and the slurry flow velocity field to generate an electrolyte concentration prediction result. This scheme extracts the electrolyte element set at the most recent A time points along the time axis for each grid point and performs weighted autoregression calculation. In the time dimension, it can differentiate the weighting of data according to the importance of different time points, further performing flow field diffusion calculation on the slurry flow velocity field and concentration gradient field, and using the local coupling index to spatially correct the weighted autoregression result.

The present application relates to the technical field of automobile battery safety monitoring, in particular to an electric vehicle battery module operation safety real-time monitoring and early warning system based on the Internet of Things, comprising a multi-dimensional sensing module, a data preprocessing module, an Internet of Things transmission module, an analysis and decision module, an early warning execution module, a safety protection module and a storage and traceability module; the multi-dimensional sensing module collects battery operation state data; the data preprocessing module is used to solve the problems of large noise, chaotic time sequence and insufficient precision of original sensing data; the analysis and decision module analyzes the battery operation state, identifies hidden troubles and determines the risk level; the early warning execution module executes differentiated and graded early warning operations; the safety protection module executes active safety protection operations; the storage and traceability module realizes full-time sequence data retention and fault traceability; the system relies on a multi-algorithm fusion artificial intelligence model to realize accurate estimation of the battery state, early prediction of thermal runaway and analysis of accumulated fault risks, and greatly reduces the false alarm rate and the missed alarm rate.

The application provides a fault prediction method and related device of an oil-immersed transformer, and belongs to the technical field of transformer fault prediction. According to the fault stage division result of the oil-immersed transformer, the fault stage in which the current oil-immersed transformer is located is determined; the first data is retained to evolve the law, the second data is extracted to reflect the sensitivity feature of the oil-immersed transformer fault, the third data is extracted to reflect the physical essence feature of the oil-immersed transformer fault, and the first feature vector, the second feature vector and the third feature vector are obtained correspondingly; according to the fault stage in which the current oil-immersed transformer is located, the first feature vector, the second feature vector and the third feature vector are fused to obtain a fused channel feature vector; and the fused channel feature vector is input into an improved DCCNN model with determined parameters to obtain a fault prediction result of the oil-immersed transformer. The application solves the problem that the accuracy of the fault prediction of the oil-immersed transformer is not high.

The present application relates to the technical field of aerospace artificial intelligence, and provides a space debris collision risk probability prediction method based on deep learning, which takes ESA Kelvins real rendezvous data as a benchmark, and uses Kessler to generate a synthetic CDM sequence to expand the training sample to improve the coverage of high risk and complex situations; at the same time, a plurality of update records and their associated related events are expressed as a dynamic heterogeneous graph, a dynamic graph neural network is used to learn the law of risk evolution over time and the similarity between events can be optionally used for knowledge transfer, so that the continuous regression prediction result of the final collision risk probability of the rendezvous event can still be stably output under the condition of missing and noise, and the prediction is quickly refreshed when new CDM arrives through the local graph incremental update mechanism, which significantly improves the accuracy, robustness and online availability of risk prediction.

This invention discloses an early warning method and system for high and low voltage switchgear, belonging to the field of power system automation. The method includes: constructing a multi-dimensional sensing network to collect multi-source heterogeneous data such as temperature rise, partial discharge, mechanical characteristics, and environmental conditions in real time; preprocessing the data and aligning it with spatiotemporal features, and dynamically correcting the sensitivity threshold of the partial discharge criterion based on humidity or pressure change rate; extracting multi-dimensional fault feature vectors and dynamically assigning weights to each component according to operating conditions and environmental conditions; inputting the weighted feature vectors into a long short-term memory network integrating an attention mechanism to generate a multi-timescale predicted trajectory of the operating state; comparing the predicted trajectory with preset multi-level state thresholds in real time, and executing a three-level early warning judgment (attention, abnormal, and severe) and corresponding linkage protection strategies. This invention can reduce the risk of false alarms and missed alarms, providing reliable support for the intelligent operation and maintenance of high and low voltage switchgear.

The invention provides a monoclonal antibody for detecting alpha1-antitrypsin and application of the monoclonal antibody. The monoclonal antibody is any one of a monoclonal antibody A1 P39-5 and a monoclonal antibody A1 P40-6 or a combination of the monoclonal antibody A1 P39-5 and the monoclonal antibody A1 P40-6. The Serpin A1 antibody prepared by the invention has extremely high affinity, so that the detection sensitivity of a detection kit based on the antibody to Serpin A1 in serum reaches a pg / mL level which is far higher than a microgram / mL level of a conventional immunoturbidimetry. Meanwhile, the detection linear range is wide, a high-concentration sample does not need to be diluted for multiple times, and the operation error is reduced.

The application relates to the security check technical field and discloses a door plate applied to a security check device, the security check device has a detection channel, the door plate comprises a door plate body and a detection module, and the detection module is arranged on the door plate body. The detection module comprises a fundamental wave emission unit for emitting a fundamental wave signal, a first receiving unit and a second receiving unit, and the first receiving unit and the second receiving unit are used for receiving a harmonic wave signal. The first receiving unit and the second receiving unit are located at different positions in a first direction, wherein the first direction is parallel to the running direction of the detection channel. The door plate and the security check device provided by the application can reduce the range of the detection blind area in the running direction and reduce the false negative rate.

The invention provides an intelligent identification method for an illegal opening behavior of a box body, and relates to the technical field of data processing, the method comprises the following steps: monitoring physical state data of the box body in real time, the physical state data comprising a vibration signal, an opening and closing state signal and position information; performing feature extraction through the physical state data to obtain an initial behavior feature sequence; based on the behavior characteristic sequence, three specific structure points on the surface of the box body are dynamically selected as a group of reference positions, and the specific structure points comprise a box body lock installation point, a box body hinge connection point and a box body seam monitoring point; according to the selected three specific structure points, constructing a dynamically changing area range; performing partition analysis on the dynamically changed region range to obtain a plurality of sub-regions; and obtaining a behavior characteristic correction coefficient according to the dynamic change characteristic of each sub-region. The illegal opening behavior recognition precision is improved, and the false alarm rate and the missing report rate are reduced.

The invention discloses a wind turbine generator gearbox state monitoring and fault early warning method and system. The method comprises the following steps: defining a key part of a gearbox of a wind turbine generator as a flexible body, and constructing a multi-body dynamic model based on the flexible body; the multi-body dynamics model receives input multi-source time sequence data, then performs feature extraction and multi-index fault early warning algorithm processing, and outputs signal abnormal information for the current multi-source time sequence data; the signal abnormal information is gearbox fault data; constructing a mixed data set guided by a physical mechanism and performing data enhancement; and inputting a spatial-temporal feature coding fault diagnosis model based on the enhanced mixed data set, and then outputting a health index. According to the intelligent early warning system for the wind turbine generator gearbox, through the technical innovation of making up for data missing through modeling simulation, improving diagnosis precision through an intelligent algorithm and guaranteeing real-time early warning through edge calculation, a set of intelligent early warning system for the wind turbine generator gearbox which is high in precision, low in time delay and strong in generalization is successfully constructed.

The application discloses a kind of gas-liquid two-phase leakage ultrasonic safety detection method and device, it is related to mechanical engineering and nondestructive testing technical field, the method includes: deployment multichannel ultrasonic sensor array;Dynamic noise baseline based on adaptive Gaussian mixture model is constructed;Extract amplitude attenuation rate, center frequency offset and other multidimensional acoustic characteristics and calculate difference vector;Combining dynamic threshold and multi-probe time delay-amplitude information carries out leakage preliminary judgment and positioning;Through convolution-long short-term memory hybrid neural network, identify leakage mode and implement three-level risk early warning according to leakage rate.The system is composed of multichannel ultrasonic sensor array, signal processing module, feature extraction unit, positioning solution module and intelligent early warning engine.The application can realize high sensitivity detection, strong working condition adaptability, accurate positioning and quantification and closed-loop intelligent response, significantly improve the intrinsic safety level of gas-liquid two-phase flow system.