871results about How to "Improve robustness" patented technology

A dynamic cache system is configured to flexibly respond to changes in operating parameters of a data storage and retrieval system. A cache controller in the system implements a caching policy describing how and what data should be cached. The policy can provide different caching behavior based on exemplary parameters such as a user ID, a specified application or a given workload. The cache controller is coupled to the data path for a data storage system and can be implemented as a filter in a filter framework. The cache memory for storing cached data can be local or remote from the cache controller. The policies implemented in the cache controller permit an application to control caching of data to permit optimization of data flow for the particular application.

The invention relates to a convolutional neural network based crowd density distribution heat map generation method, which comprises the steps of dividing a crowd picture set into a training sample set and a test sample set, performing crowd label image segmentation by using convolutional neural network architecture, and carrying out number regression by using a convolutional neural network; correcting a density map through a multi-scale template operation, generating a crowd density distribution heat map according to the corrected density map and the regression number, and completing crowd density distribution estimation. According to the invention, deep characteristics of an image are extracted by using a powerful learning ability of the full convolutional neural network so as to perform accurate crowd segmentation, and low efficiency and blindness of density calculation of a traditional method for full image characteristics are overcome; a crowd near-far perspective effect is overcome to a certain extent through multi-scale template correction; and mapping is performed in allusion to the estimated number of people, lateral comparison can be performed on heat maps of different cameras, the method is applicable to various crowd scenes, and the crowd density distribution heat map can be acquired in real time.

The invention discloses a single target tracking method based on a convolution neural network. The method comprises the steps that training data are used to preliminarily train a six-layer convolution neural network classification model; when tracking test is carried out, Ground-Truth information given by the first frame of a video is used to extract some sample data; through parameter fine-tuning of the network model, the network model can better adapt to the currently tracked video sequence; a Bounding Box regression model is trained for a currently tracked target to optimize a tracking result; and for a correct tracking result, the Bounding Box regression model is used to optimize the tracking result to acquire the accurate position of the target object. When tracking is carried out, network model parameters are timely and appropriately updated, so that the model can better adapt to the currently tracked video sequence. According to the invention, a pooling layer in the network structure is improved; a detection module is added, which makes the performance of a tracker more robust; and the tracking precision is improved.

The invention discloses a method for identifying handwritten numbers based on a convolutional neural network (CNN) and a support vector machine (SVM). The combination of a convolutional neural network and a support vector machine increases the accuracy of identifying handwritten numbers. The method includes the following steps: enlarging a handwritten number image training set; conducting normalization; establishing two CNNS; training the two CNNs; establishing the SVM; remaining the alternating part of convolutional layers and pooling layers at the front edge of fully connected layers of the two CNNs, putting the fully connected layers of the two CNNs in serial connection and connecting the fully connected layers to the SVM to obtain a combined network; training the combined network; testing the handwritten number image testing set to obtain the result of the identification of the handwritten number. The method yields accuracy as high as 99.60%. According to the invention, the method obviates the need for complex pre-processing and is more adaptive and stable. The method also has high identification accuracy and is more reliable and robust. The method is applied to handwritten number identification in the fields of finance, postal service, data statistics, etc.

A system and method for detection, tracking, classification, and/or identification of a moving object from a moving sensor uses a three-dimensional (3D) model. The system facilitates generation of a 3D model using images from a variety of sensors, in particular passive two-dimensional (2D) image capture devices. 2D images are processed to determine viewpoint and find moving objects in the 2D images. Conventional techniques or an innovative technique can be used to find segments of 2D images having moving objects. Viewpoint and segment information is used for generation of a 3D model of an object, in particular using both object motion and sensor motion to generate the 3D model.

The invention relates to a cascade regression-based face key point positioning method. The cascade regression-based face key point positioning method includes the following steps that: 1) a large number of face image data are acquired, and an initial key point position is marked; 2) the face image data are trained, and a coarse regression machine can be obtained through learning, and then with the output of the coarse regression machine adopted as input, a fine regression machine can be obtained through learning; and 3) face image data to be recognized are given, and the initial shape of a face is regressed to the vicinity of a real shape through the coarse regression machine, and then, with the output of the coarse regression machine adopted as input, the precise coordinates of a face key point can be obtained through the fine regression machine. According to the cascade regression-based face key points positioning method of the invention, a coarse-to-fine cascade regression algorithm is adopted, and a large number of samples are learned, and multi-feature fusion and multiple-regression machine fusion are realized, and therefore, the speed and the robustness of the algorithm are improved greatly, and the face key point can be excellently positioned under the situations of occlusion, low lightness and poses such as side faces, and the accuracy and the speed of face key point positioning can be effectively improved.

A method of determining fatigue load of at least one operative wind turbine, comprising: providing a transfer function that associates an obtained at least one measurement value of a first sensor to an obtained at least one measurement value of a second sensor the at least one measurement value of the first sensor and the at least one measurement value of the second sensor being obtained by use of a reference wind turbine at which the first sensor and the second sensor are located; obtaining at least one measurement value of a third sensor, wherein the third sensor is located at the at least one operative wind turbine, and wherein the third sensor corresponds in type and position to the first sensor at the reference wind turbine; calculating at least one transfer function value corresponding to the obtained at least one measurement value of the third sensor by use of the provided transfer function; calculating the fatigue load of the at least one operative wind turbine based on the calculated transfer function value.

The invention discloses a target counting method and system based on a double-attention multi-scale cascade network. The target counting method comprises the following steps: inputting an image subjected to graying processing into an initial module of the double-attention multi-scale cascade network for initial feature extraction to obtain an initial feature map; inputting the initial feature mapinto a first branch network of the cascade network to obtain a low-level detail feature map and a high-level semantic feature map; performing channel attention transformation on the feature map to generate global feature information; inputting the initial feature map into a spatial attention pyramid structure of a second branch network in the cascade network to generate multi-scale features of theimage; fusing the multi-scale features and the global feature information to obtain a fusion result; performing feature extraction on the fusion result to generate an estimated target distribution density map; and carrying out pixel summation on the target distribution density map to obtain an estimated target counting result. The target counting method can effectively realize accurate target counting in a complex scene, and has good robustness and generalization.

A vehicle is equipped with a camera (which may be a stereoscopic camera) and a computer for processing the image data acquired by the camera. The image acquired by the camera is processed by the computer, and features are extracted therefrom. The features are further processed by various techniques such as object detection/segmentation and object tracking/classification. The acquired images are sometimes contaminated by optical occlusions such as raindrops, stone-chippings and dirt on the windshield. In such a case, the occluded parts of the image are reconstructed by optical flow estimation or stereo disparity estimation. The fully reconstructed image is then used for intended applications.

A computer-implemented method is described for authenticating an identity of a user requesting execution of a computerized transaction via a first client computing device. The first device and a second client computing device in proximity each execute applications for communicating with a server. The first and second devices contemporaneously capture voice sequences including at least one audible sound vocalized by the user, and a timestamp indicating when the sound is captured. The identity of the user is validated based upon a determination that the user vocalized the pass phrase, and a difference between the first timestamp and the second timestamp is below a predetermined threshold. The first device executes a computerized transaction with the server computing device upon receiving validation of the identity of the user.

The present invention relates to a quick-change tool holder system, comprising a base part with a quick-change tool holder receptacle and a quick-change tool holder having a cutting tool receptacle configured for insertion into the quick-change tool holder receptacle.

A system, method and computer program product is provided for mitigating the effects of burst noise on packets transmitted in a communications system. A transmitting device applies an outer code, which may include, for example, a block code, an exclusive OR (XOR) code, or a repetition code, to one or more packets prior to adaptation of the packets for transmission over the physical (PHY) layer of the communications system, wherein the PHY layer adaptation may include FEC encoding of individual packets. The outer coded packets are then separately transmitted over a channel of the communications system. A receiving device receives the outer coded packets, performs PHY level demodulation and optional FEC decoding of the packets, and then applies outer code decoding to the out6r coded packets in order to restore packets that were erased during transmission due to burst noise or other impairments on the channel.

The invention relates to an automatic parking method based on fusion of vision and ultrasonic perception. The method comprises the following steps that visual parking space detection is carried out, ultrasonic radar parking space detection is carried out, two parking space detection results are fused, and parking space tracking and parking space display are carried out; a target parking space is determined, and a parking speed and a parking track are planned in combination with a current scene map; according to the visual detection and the ultrasonic radar detection, obstacles at a driving area around a vehicle and an edge of the driving area are detected, and the obstacles are subjected to motion estimation; the probability of collision of the vehicle itself and the obstacles is calculated, and if the collision probability is greater than a preset threshold value, an obstacle avoidance function is triggered; and parking track tracking is carried out through a low-speed controller, gear control, vehicle speed control, steering wheel steering control and obstacle avoidance control along the parking track are achieved, and automatic parking is completed. According to the method, thedetection effect of the parking space and the obstacles around the vehicle is improved, the robustness is improved, a motion obstacle track estimation module is added, and the obstacle avoidance reliability is improved.

The present invention discloses a finite time convergence time-varying sliding mode attitude control method, relating to a finite time convergence time-varying sliding mode reentry aircraft attitude control method based on a high order sliding mode observer, belonging to the technical field of aircraft attitude control. The invention discloses the finite time convergence time-varying sliding mode attitude control method, on the basis of aircraft attitude model feedback linearization, high-frequency vibration is weakened based on a low-pass filter, the arrival segment of slip mode control can be eliminated through introducing a time-varying item, the robustness of a system is enhanced, a jump problem caused by time-varying function first-order derivative discontinuity can be eliminated, and the time-varying sliding mode attitude control of finite time convergence is realized. According to the method, the reentry aircraft attitude control tracking error can be converged to zero in a finite time, the jump phenomenon caused by the first-order derivative discontinuity of a time-varying item can be eliminated, the high frequency vibration is weakened, and the robustness of a reentry aircraft attitude control system can be improved.

The invention discloses a fatigue driving detection method and a fatigue driving detection method system based on a convolutional neural network and belongs to the technical field of image processingand mode recognition. The method comprises the steps of firstly, collecting a two-dimensional face image of a driver in a driving state, classifying stepwisely according to the fatigue degree, and establishing a fatigue driving image library; secondly, constructing the convolutional neural network containing a data layer, a convolutional layer, a pooling layer, a connection layer and a classification layer; thirdly, iteratively training the constructed network with image data and labels in the fatigue driving image library as input of the convolutional neural network by use of a back propagation algorithm so that loss function values output by the network decrease gradually and are restrained; and fourthly, inputting a test sample of the face image of the driver in the driving state, identifying the test sample by use of a trained convolutional neural network model so as to implement detection classification of the fatigue degree of the face image of the driver. According to the fatigue driving detection method and the fatigue driving detection method system based on the convolutional neural network, compared with a conventional machine learning method, identification and classification effects are obviously improved, and a feasible concept is provided for real-time monitoring of fatigue driving.

The invention discloses a graph convolutional neural network model and a vehicle trajectory prediction method using the same. The model is composed of an encoder module, a spatial information extraction layer module and a decoder module. The method comprises the following steps: firstly, sampling a predicted vehicle and surrounding vehicles in a traffic scene at a frequency of 5Hz, and collectingposition coordinates and kinetic parameters of each vehicle sampling point, including horizontal and longitudinal coordinates, horizontal and longitudinal vehicle speeds and accelerations; calculatingcollision time TTC between the predicted vehicle and surrounding vehicles according to the coordinates and speeds of the predicted vehicle and the surrounding vehicles, and judging vehicle behaviors;inputting each historical track of the vehicle containing the information into the model, encoding time sequence interaction features in the track, extracting spatial features, summarizing the features into context vectors, and inputting the context vectors into an LSTM decoder to generate future track coordinates of the vehicle. According to the method, the problem that feature information generated by vehicle interaction cannot be obtained by using a traditional recurrent neural network is solved, and the prediction precision of the vehicle trajectory is greatly improved.

The invention discloses a scene-based hybrid invasion detection method and a scene-based hybrid invasion detection system. The method comprises the following steps that: 1, an invasion detection system extracts the characteristics of acquired audit data according to an access control strategy and related application scene information; 2, the invasion detection system forecasts a suspicious invasion event by a combined forecasting method according to the characteristics of the audit data; 3, the invasion detection system determines an invasion event by combining the suspicious invasion event and alarm associated information; and 4, if a certain invasion event occurs, the invasion detection system generates alarm information, and transmits the alarm information to a security management platform for visual display. By the method and the system, the accuracy of invasion detection can be improved, the security of a physical layer of the system can be protected, and internal attacks of the system can be prevented.

The invention discloses a method and a device for protecting neighbor tables in an IPv6 network; the method comprises the following steps of: monitoring a distribution message used for distributing IPv6 address prefix; recording the IPv6 address prefix carried wherein; monitoring a DAD message used for conflict address detection DAD; recording to the address allocation information carried in the DAD message corresponding to the recorded IPv6 address prefix; carrying out a validity judgment on the learning message used for learning neighbor table according to the recorded address allocation information; and stopping carrying out normal neighbor table operation according to an invalid learning message. The method and the device can improve the robustness and safety of the neighbor table, can be simply realized simultaneously, need not to update the IPv6 protocol stack, are easy to be deployed and have small maintenance quantity.

The invention provides a method and a device which are used for realizing the separation of the data in an upper row from the data in a lower row. The method mainly comprises that a load balancer obtains the relative information for communication between the load balancer and a client and between the load balancer and a server by a transmission layer protocol; wherein the relative information includes IP addresses and port information and the load balancer transmits the relative information to the server; the load balancer receives the data packets transmitted by the client through the transmission layer protocol and then transfers the data packets to the server according to the relative information, and the server directly transmits the data packets to the client according to the relative information. The invention can realize the separation of the data in the upper row from the data in the lower row in cluster systems such as IPTV using the load balancer, reduce the pressure of the load balancer and greatly improve the toughness of the systems.

The invention provides a system and a method for online recognition of road unevenness. The system comprises a power source module, an acceleration sensing system, a vehicle speed information collecting system, a signal preprocessing module and a road unevenness calculating module. The method includes that road tests are performed on a sample vehicle on two roads different in grade, a linear relation between a quotient of an unsprung mass vertical acceleration mean square value of the vehicle model and vehicle speed and a road unevenness coefficient is demarcated; when the vehicle travels, each unsprung mass vertical acceleration of the vehicle and vehicle traveling speed information are collected; the collected information is compared with a demarcation result of the sample vehicle of the vehicle model, and current road unevenness can be acquired in real time. Compared with current similar technology, the system has the advantages of high transportability, quickness in operation and the like, and is nonsensitive to stiffness damping coefficient change of a suspension and especially suitable for active and semi-active suspension control systems variable in stiffness damping, and robustness of the system can be guaranteed.

The invention is suitable for the software field, and provides a method, device and equipment for controlling software for auto-updating. The method comprises the following steps: dividing a storage space to two code regions mutually reserved for each other; updating codes in the code region in the idle state during software updating, and making a start-up mark point to the codes in the newly updated code region. By dividing the storage space to two code regions mutually reserved for each other, the method can update the codes in the code region in the idle state during software updating without influencing the normal operation of the equipment, thereby not influencing the normal use of the equipment by users. Therefore, the robustness and the fault tolerance of the updated module can be effectively improved while obviating the damage to the system by the abnormal conditions during the updating process. In addition, the updating does not take the user operation time because the updating may run in the background.

The invention discloses a disturbance observer-based underactuated ship path following control method. A virtual ship kinematics model is built; a target reference path is planned; a position time series signal of the target reference path is generated; the position (xd, yd) and the azimuth angle psid of the tracked target at the current time are acquired; the relative position error ze between areal ship and the virtual ship, an LOS heading angle command signals psir of the real ship, and first and second-order derivatives thereof are calculated; a nonlinear disturbance observer is designed;disturbance is estimated; a control law is designed according to a robust adaptive control strategy; a host propulsion control moment command signal tauu and a yaw control moment command signal taurare calculated; and path following control is realized. Dynamic estimation signals for unknown ocean disturbances under severe sea conditions are acquired through a nonlinear disturbance observer method, the control moment of driving propulsion and a steering actuator can be effectively reduced, and the robustness and the economy of underactuated ship path following control under severe sea conditions can be improved.

The invention discloses a method for controlling a nonlinear robust position of an electro-hydraulic servo system with time-varying output constraints, and the method comprises the following steps of: selecting an electro-hydraulic position servo system as a research object, creating a nonlinear system model and considering external interference of the system and modeling uncertainty; estimating unmodeled interference and uncertainty through an extended state observer and performing feed-forward compensation by combining a backstepping control method, and flexibly performing time-varying asymmetric constraints to output position tracking errors by adopting a nonlinear robust controller designed based on time-varying asymmetric barrier Lyapunov functions; the control voltage of the nonlinear robust controller is continuous and more beneficial for practical application in engineering.

The invention relates to a voltage-current double closed-loop terminal sliding mode control method of a Buck converter. The voltage-current double closed-loop terminal sliding mode control method of the Buck converter aims to solve the problems that the traditional linear sliding mode control method is slow in response speed and low in stability precision, and includes the following steps: 1 mathematics model building of the Buck converter; 2 load observer design according to output voltage VC of the Buck converter and the loaded current iL; 3 voltage terminal sliding mode controller design: a voltage controller outputs an inductive current demand signal iL* according to a demand input direct-current voltage tracking error ev; 4 current linear sliding mode controller design: a current controller outputs a control signal mu of a controllable switch tube V of the Buck converter according to the inductive current demand signal iL*. The voltage-current double closed-loop terminal sliding mode control method is applied to the electricity field.

The invention belonging to the field of face recognition discloses a yolo-based face detection and face alignment method. The method comprises steps of network training and network verification. At the step of network training, a face data set is established; an image in the face data set is marked; and a face detection and alignment database is reconstructed. Therefore, problems that the efficiency is low and joint tasks can not be carried out because face detection and alignment are carried out by means of stage division during MTCNN face recognition are solved; the robustness of face recognition and the generalization ability of the network are improved; and a problem of over fitting caused by a few of samples is solved.

The invention relates to an intelligent performance analysis and testing bench for batteries of motorized automobiles, which allows the studying and analysis on the charging and discharging characteristics of various batteries in a lab, so as to reduce the studying cost, shorten the studying time, simplify the studying procedures and improve the analysis accuracy. The invention comprises a main system interface module comprising a hi-voltage charging parameter setting module and a hi-voltage charging display module. The hi-voltage charging comprises hi-voltage programmed charging and hi-voltage constant-current charging. The main system interface module additionally comprises a low-voltage charging parameter setting module and a low-voltage charging display module. The low-voltage charging comprises low-voltage programmed charging and low-voltage constant-current charging. The main system interface module additionally comprises a discharging parameter setting module and a discharging display module. The charging comprises programmed discharging and constant-current discharging.

This invention discloses one rapid test method, which comprises the following steps: first network part receives second part to keep connection status message to send rapid detection message to second network part; if the first network part receives feedback rapid message in set time then to judge whether linkage is normal and to judge fault. This invention also discloses one rapid test device composed of receive module and rapid test module.

The invention relates to a unified power quality control approach and the device thereof. The device comprises a parallel converting-bridge, a series converting-bridge, a DC power source as well as a third converting-bridge. The approach comprises the steps that: the compensation for a variable value to gain a power grid; the compensation for a voltage distortion as well as a target current value and the change ratio of both the voltage distortion and the target current value are input taken as a controller; the fuzzy control rule base and membership function parameter to be adjusted are selected as antibodies; the optimized important parameter of the fuzzy controller is realized through an immunity genetic algorithm; a SVPWM drive signal is generated and drives the series/parallel active filter power component; the difference value of the required capacitor voltage value and the actual capacitor voltage tested by a Hall sensor is applied to a PI algorithm; the SVPWM drive signal is generated again to drive the third converting-bridge power component. The invention combines the genetic algorithm, the immunity algorithm and the fuzzy algorithm; therefore, the robustness and the dynamic-static performance of the controller have been remarkably enhanced.

The invention belongs to the field of image processing and computer vision, and provides a video image rain removing method in order to improve the visual effect of video images. The video image rainremoving method based on a convolutional neural network includes the following steps: selecting several consecutive frames of images; extracting the brightness component of each frame and the corresponding high-frequency component of the image; inputting the high-frequency component images to a constructed and trained convolutional neural network to get high-frequency non-rain component images processed by the convolutional neural network; and finally, synthesizing the non-rain component images and the retained low-frequency components to get rain-removed video images, wherein the specific relation of the convolutional neural network is D(P)=||hp(I)-J||<2><2>, hp(.) represents the convolutional neural network, P represents network parameters, I represents the original rain images, J represents the rain-free images, and the convolutional neural network is trained to minimize the value of D(P) so as to get an optimal parameter value P* and get rain-removed images J-^=hp*(I). The method is mainly applied to the occasion of image processing.

The invention provides an outdoor scene recognition method fused with sensor information. By using the method, an outdoor scene can be quickly recognized and the recognition correct rate is high. The method comprises the specific steps of 1, a mobile terminal collects an image to be recognized of the current scene, and gravity information which is produced when the mobile terminal collects the image is acquired; 2, the image to be recognized and the gravity information are packed into a compressed file and the compressed file is sent to a server; 3, the server calculates the gravity direction theta g of the image to be recognized according to the gravity information in the compressed file; 4, lines in the image to be recognized are detected; 5, wrong lines in horizontal and vertical lines are removed; 6, horizontal vanishing points and vertical vanishing points are calculated; 7, perspective distortion correction is carried out on the image to be recognized; 8, image matching is carried out, and the most similar sample image is searched; and 9, the scene information corresponding to the sample image is returned to the mobile terminal, so as to realize scene recognition.