42results about How to "Good real-time detection" patented technology

The invention relates to a method for detecting Android malicious software by means of a random forest classifier in real time. The method comprises the steps of 1 collecting network data, 2 calling data flow through an application program interface for grouping, 3 extracting data flow minimum unit, 4 extracting calling sequence characteristics, 5 training a hidden Markov model, 6 training a random forest model, 7 extracting application program interface data flow characteristics of a to-be-detected sample, 8 inputting feature vectors of the to-be-detected sample into a random forest detection model and judging whether or not output of a field network characteristic detection model is the malicious software category and 9 outputting the malicious software category corresponding to the to-be-detected sample. Accordingly, the malicious software can be detected in real time, software transmitted in the network can be detected, and the good detection accuracy rate is achieved.

The invention discloses a fault online diagnosis method of uninsulated track circuit tuning zone, which relates to the online diagnosis of a track circuit. The method comprises the steps of obtaining cab signal inductive voltage amplitude envelope, shunt point positions, train speed and track circuit signal carrier frequency information, pretreating the data to obtain the length of the track circuit, the quantity and position of compensation capacitors and cab signal inductive voltage normalized amplitude envelope, performing segmented fitting on the cab signal inductive voltage normalized amplitude envelope and establishing corresponding feature vectors, establishing a neural network, training the neural network by utilizing the feature vectors and performing fault diagnosis on the uninsulated track circuit tuning zone equipment, and sending diagnosis results of the tuning zone equipment to a corresponding ground monitoring client-side through a cab signal remote monitoring system. The method has the characteristics of being complete in functions, high in detection accuracy, high in applicability, simple in method, good in real-timeness and the like, no any hardware equipment is added in the method based on the existing cab signal system, and the cost is low.

The invention discloses a double-threshold energy detection method based on history sensing information. According to the method, probabilities that states of occupying a licensed band by a licensed user are changed or not in adjacent sensing frames are estimated through an energy detection method; a high threshold and a low threshold satisfying constraint of a target false alarm probability are set according to a theoretical value of a false alarm probability and the probabilities; spectrum detection is carried out; and when an energy statistic value of a current sensing frame is greater than the high threshold, it is judged that the licensed user occupies the licensed band in the current sensing frame, otherwise, secondary judgment is carried out on the state of occupying the licensed band by the licensed user through utilization of the energy statistic values of the current sensing frame and a history sensing frame according to the high threshold and the low threshold. The method has the advantages that the constraint of the target false alarm probability is satisfied, a relatively high detection probability is obtained, the spectrum detection performance is improved, moreover, the complexity is low, the detection timeliness is relatively good, and the occupied system bandwidth is low.

The invention discloses an oil delivery pipeline leakage detection method based on information fusion of an extreme learning machine. The oil delivery pipeline leakage detection method comprises the following steps that S1, historical data of a pipeline leakage experiment are collected as an initial training sample; S2, after flow information and pressure information in the historical data are subjected to characteristic value extraction and calculation, information fusion characteristic data are obtained; S3, the obtained information fusion characteristic data serves as input information, anda pipeline leakage detecting model is established through an extreme learning machine method; S4, the accuracy of the detecting model is verified; and S5, the information fusion characteristic data obtained after the flow information and pressure information of a to-be-detected pipeline are subjected to characteristic value extraction and calculation are input the final detecting model, and the state classification result of the to-be-detected pipeline is output. According to the oil delivery pipeline leakage detection method, the flow information and the pressure information of the pipelineare collected and subjected to characteristic value extraction and calculation, thus information fusion characteristic data are obtained to serve as the training sample, the main characteristics of asignal are covered, and the computation amount of the extreme learning machine method is decreased.

The invention discloses a voltage fast detection algorithm suitable for a grid power system low voltage ride through function. The voltage fast detection algorithm suitable for the grid power system low voltage ride through function comprises the following steps: (1) acquiring amplitude value information under the normal operation of a power grid based on a phase-locked loop of a synchronous reference coordinate frame, (2) separating the voltage of a sampled power grid into three-phase of A, B, C, conducting a period of delay to the sampled power grid, and acquiring the absolute value of an difference value, (3) setting a voltage threshold value and comparing the difference value in the step (2) with the voltage threshold value, (4) if the difference value is smaller than the voltage threshold value, then judging that the voltage is normal, if the difference value is larger than the voltage threshold value, then turning to step (5), (5) sending real-time voltage signals into a maximum value tracking module and an over zero point tracking module through each of the phrases of voltage, after judging that the system enters into a low voltage status, and (6) recording the maximum value tracking module and the over zero point tracking module to acquire an amplitude value after the voltage of the power gird drops.

The invention discloses a roadside parking intelligent monitoring method on the basis of vacant parking space image detection. The roadside parking intelligent monitoring method includes an offline setting portion and an online monitoring portion. Corresponding irregular regions are enclosed for each parking space in manual modes on the basis of photographed single-frame photos of roadside parking places during offline setting, and unique parking space numbers are attached to the irregular regions; an image is acquired by a camera at the intervals of a time period t during online monitoring, the images are fed to control centers via wireless networks, and then states of the parking spaces are detected. The roadside parking intelligent monitoring method includes preliminarily processing the images in detection procedures and acquiring edge graphs of the images; acquiring corresponding regions of edge images according to predefined region templates and clipping the corresponding regions; judging whether the parking spaces are vacant parking spaces or not on the basis of the irregular regions; recording current parking space states and parking start time and end time of the parking spaces according to change of the states of the parking spaces.

The invention discloses an LED (Light-Emitting Diode) display module and an LED display screen. The LED display module comprises a base plate and one or a plurality of groups of RGB (Red Green Blue) LED chips packaged on the front face of the base plate through packaging glue, and further comprises a signal data acquisition module and a signal processing module which are electrically connected, and the signal data acquisition module is arranged on the base plate or is directly connected with anodes and cathodes of the LED chips. The LED display screen disclosed by the invention is formed by splicing the LED display modules; the LED display module and the LED display screen disclosed by the invention overcome the defects that the junction temperature detection of the LED chips is inaccurate and on-site real-time continuous monitoring and prewarning of LED products cannot be realized, and have the advantages of accurate temperature detection and good detection real-time property.

A method for detecting irregular defects of industrial products comprises the following steps: step 1, image enhancement processing, specifically comprising histogram equalization and histogram matching; step 2, constructing a network model; and step 3, setting relevant parameters of the network model, specifically including setting of a default box, a default box matching mode and composition ofa loss function. According to the invention, image enhancement processing is carried out on a collected sample image, so that defects are more obvious. Then, on the basis of a convolutional neural network CNN, a defect detection network model is constructed in combination with the SSD target recognition model, and model parameters are reasonably designed, so that the detection problem of irregulardefects can be effectively solved.

The invention discloses an estuary suspended sediment concentration real-time measuring method based on sound attenuation. The estuary suspended sediment concentration real-time measuring method basedon the sound attenuation comprises the steps that an underwater sound signal sending end and a receiving end are arranged in an estuary, sound attenuation is calculated, acoustic attenuation disturbance caused by suspended sediment is calculated, and the suspended sediment concentration is iteratively calculated. According to the estuary suspended sediment concentration real-time measuring methodbased on the sound attenuation, covering areas are wide, the expandability is high, obtained parameters are more continuous in space, the detecting instantaneity is good, and the measuring precisionis high. Sound propagation technology is adopted for measuring, and the performance is better and development and application potentiality are more achieved compared with light, electromagnetic wave and other modes.

The invention discloses a vehicle detection method with granularity-level multi-scale characteristic based on asymmetric convolution. The method comprises the following steps: S1, obtaining a plurality of pictures containing vehicles through a video stream, and making a vehicle target detection data set which comprises a training set and a test set; S2, designing an asymmetric convolution AC unit;S3, embedding the asymmetric convolution AC unit into a Res2Net module to construct an asymmetric convolution ACB module with a granularity level multi-scale characteristic; S4, embedding an ACB module into a lightweight model ShuffleNetV2 to construct a vehicle target classification network ACBNet based on a convolutional neural network; S5, training a vehicle target classification network ACBNet by using the manufactured vehicle target training set; and S6, putting test set pictures into the classification network ACBNet for classification, then processing the test set pictures through a residual error prediction module to obtain category and position information of vehicles in the predicted images or videos, and framing out vehicle targets.

The invention discloses a converter system and a control method thereof. The converter system comprises a converter, a motor, a speed sensor and a control device. When the converter system works normally, the control device controls the converter by employing a rotor position signal sampled by the speed sensor; and the state of the speed sensor is monitored in real time according a rotor current error of the motor. When the signal of the speed sensor is abnormal, the control device employs a rotor position estimation signal to replace the rotor position signal sampled by the speed sensor to control the converter. According to the invention, with the system and the control method, a technical problem that the converter system shuts down due to fault occurrence after abnormity of the speed sensor signal can be solved; the shutdown time of the converter system before maintenance can be reduced; and the system stability can be improved under the circumstance that the speed sensor signal is abnormal.

The invention discloses a flame identification algorithm of low-rank analysis. First, video images are acquired by an infrared camera and preprocessed by Gaussian smoothing filter, and grayscale images are binarized. The external rectangle of a suspicious area of the current frame is extracted and zoomed in. The data of the external rectangle is used as a column vector. The data of external rectangles of a flame area of the continuous n-1 frame behind the current frame is extracted and zoomed to the same size, and the data of the frame is taken as column vectors frame-by-frame. All column vectors form a matrix D which his decomposed into the sum of a low rank matrix A and a sparse error matrix E. The singular value decomposition for the low rank moment A is obtained, and respective contribution rate is calculated by the square of the singular values. The number of singular values is determined by that the cumulative contribution rate is greater than the given threshold. The determined number is new rank rank(D) of the matrix D, and the boundary point theta* is obtained through experiments. An alarm is given off when the number is greater than or equal to theta*. A lot of experiments show that the algorithm has high recognition rate and robustness, and the method is proved to be effective.

The invention is suitable for the technical field of artificial intelligence and behavior detection, and provides a smoking behavior real-time detection method based on deep learning. According to themethod, a data set is constructed through a data acquisition standard, and a set of smoking convolutional neural network for smoking behavior detection is constructed. The method promotes research ofdeep learning on smoking behavior detection, and has good detection accuracy, and by means of deep learning, the method has very high real-time performance and robustness and can adapt to the environmental change of an application scene.

The invention provides a coal plough planing blade mechanical characteristic on-line testing device, and fundamentally solves the problem that mechanical characteristic data under a coal plough condition fails to be detected. According to the technical scheme, a coal plough cutter head comprises a cutter seat, a cemented carbide cutter head arranged on the front end of the cutter seat, a protection cover assembled on the cutter seat, and a cutter rod arranged on the bottom part of the cutter seat; the middle part of the cutter seat is provided with a square hole, and the vibration sensor is placed in the square hole through a strong magnetic base; a gap exists between the tail part of the cutter seat and the cutter rod; the square hole and the front end of the cutter seat form a tetragonal column, strain gage grooves are machined in the four surfaces of the tetragonal column, and the strain gage grooves are respectively equipped with strain gages I-III; and the front inner wall of the square hole is provided with the strain gage III; the protection cover is a U-shaped housing provided with a hook-shaped structure, the protection cover is embedded in the outer wall of the cutter seat through the front part of the cutter seat, and the hook-shaped structure is clamped to the rear part of the cutter seat and is fastened through a sunk screw. The coal plough planing blade mechanical characteristic on-line testing device has the advantages of simple structure, convenient usage, easy control, visual data detection, reliable detection result and the like.

The invention relates to the technical field of operation monitoring of coal mine belt conveyors, in particular to a mining belt conveyor conveying belt state detection method and system based on image recognition, and the method comprises the steps: installing a video sensor on a portal frame of a conveyor, firstly collecting a video of a belt surface area of a conveying belt in work in real time through the video sensor, screenshot is carried out on the video frame by frame to obtain multiple frames of continuous mining conveyor belt images; then, the video sensor analyzes and processes the images of the mining conveying belt, and finally, the video sensor displays the processed images of the mining conveying belt, and the processed images of the mining conveying belt comprise whether coal exists on the conveying belt or not and whether the conveying belt runs or not. The mining belt conveyor conveying belt state detection method based on image recognition is reliable in detection, good in detection real-time performance, convenient to maintain, low in cost and convenient for practical application in coal mine scenes.

The invention discloses a fresh concrete thixotropic property evaluation device which comprises a collection hopper fixed on a fixed support, the collection hopper is connected with a diaphragm pump through a first stainless steel braided tube, the diaphragm pump is further connected with a horizontal conveying steel tube through a second stainless steel braided tube, and a pressure sensor is arranged in the horizontal conveying steel tube. The horizontal conveying steel pipe is connected with a discharging nozzle through a stainless steel bent pipe, a flowmeter is arranged on the discharging nozzle, a stainless steel frame is arranged below the discharging nozzle, and the stainless steel frame is placed on a glass plate. The invention also provides a method for evaluating the thixotropic property of the fresh concrete. By simulating the pumping and construction process of the fresh concrete, the pumpability of the fresh concrete can be evaluated, the thixotropic performance of the fresh concrete can be reflected in real time, and the evaluation method and device have the advantages of being easy and convenient to operate, efficient, good in detection real-time performance, wide in application range and the like.