45results about How to "Low computing power requirements" patented technology

The invention discloses a multi-scale image semantic segmentation method. The method comprises the following steps: obtaining a to-be-segmented image and a corresponding label; constructing a full convolutional deep neural network, wherein the full convolutional deep neural network comprises a convolution module, a hole convolution module, a pyramid pooling module, a 1 * 1 * depth convolution layer and a deconvolution structure; setting hole convolution as channel-by-channel operation, and utilizing low-scale, medium-scale and high-scale characteristics in a targeted mode; training the full convolutional deep neural network, establishing a loss function, and determining parameters of the full convolutional deep neural network by training the sample image; and inputting the to-be-segmentedimage into the trained full convolutional deep neural network to obtain a semantic segmentation result. By means of the method, the image semantic segmentation problem with complex details, holes andlarge targets can be well solved while the calculated amount and the parameter number are reduced, and the consistency of category labels can be reserved while the target edges can be well segmented.

A robot direct teaching control method based on moment balance is characterized in that current values of joint motors of a robot under some specific poses and moving states are calculated through measurement and modeling, and an analytic expression of the current values required by balancing of the gravity and the friction of joints under any pose and any moving state of the robot is obtained. During robot teaching, the current values calculated through the analytic expression are directly utilized to control the joint motors to enable the robot to be located under the environment similar to the environment without gravity and friction, and direct teaching can be conducted on the robot at the moment. The method has the advantages of being simple in control, low in system cost, convenient to operate and the like.

The invention provides an inertial measurement system multi-sensor redundancy method. The method comprises the steps that through multiple measurement values output by multiple sensors, an output result of an inertial measurement system and projection values of the output result on all the sensors are calculated; absolute values of differences between output values of all the sensors and the projection values are calculated according to the output values and ordered; iterative judgment is performed according to the maximum absolute value of the differences and a preset precision deviation threshold value until a correct failed sensor is judged, and the output result of the inertial measurement system is a calculation result of the rest sensors. Compared with an inertial measurement system multi-sensor redundancy method achieved through a parity vector algorithm, the problem that the multi-sensor redundant inertial measurement system independently judges the failed sensor without relying on preorder data or subsequent data is solved, not only can the failed sensor be accurately judged, but also the sensor which is occasionally failed or of which the output precision does not meet the design requirement can be judged, and the reliability and the output precision of the inertial measurement system are improved.

The invention discloses a mobile advertisement optimized delivery method. The method includes the steps that existing click logs and existing conversion logs of delivered advertisements are acquired, and historical advertisements relevant to the advertisements to be delivered are found; the correlation coefficient of the historical advertisements and the advertisements to be delivered is graded; users are graded according to user click and conversion conditions of the historical advertisements; two times of grading is counted, the user with the highest grade is screened, and the advertisements to be delivered are pushed to the user with the highest grade. Based on the correlation between the historical advertisements and the advertisements to be delivered and the user click and conversion conditions of the historical advertisements, most appropriate crowds are found, and therefore the purpose of accurate delivery is achieved, and the requirements of the method for computing power of a server are low.

The invention discloses a network public whiteboard system based on cloud computing, and the system comprises terminals and a network public whiteboard platform, wherein the terminals communicate with the network public whiteboard platform through a browser, the network public whiteboard platform further comprises a transmitting unit, a foreground U1 unit and a network public whiteboard processing unit, the foreground U1 unit at least comprises a display subunit and an interaction subunit, the display subunit is used for displaying the data in the browser of the terminals, and the interaction subunit is used for intercepting and capturing input operation on the browser of the terminals through a message hook and sending the input operation to the network public whiteboard processing unit through the transmitting unit; and the network public whiteboard processing unit is used for receiving operation data on the browser of the terminals, processing requests of different users, coordinating the parallel processing of the users, transmitting the processing result through the transmitting unit and displaying the processing result on the corresponding terminal browser through the foreground U1 unit.

The invention relates to a voltage sampling method and a voltage sampling device. The sampling method comprises the following steps of: acquiring, namely continuously acquiring a plurality of voltage values according to a preset time period within a set period of time and storing; calculating, namely calculating the difference value of the voltage values acquired in adjacent time periods; comparing, namely comparing the difference value with a preset value, sending an abnormity identification signal if the difference value is greater than the preset value and sending a normality identification signal if the difference value is not greater than the preset value; outputting, namely, when the normality identification signal is received, outputting a plurality of stored voltages; and clearing, namely when the abnormity identification signal is received, clearing the plurality of stored voltages. Therefore, a voltage acquisition error in a voltage rising or falling process is avoided and the requirement on the operational capability of a processor is relatively low.

The invention discloses a human face living body detection method and device, equipment and a medium. The human face detection method comprises the steps of obtaining a training set; inputting the training data into a teacher network for first training to obtain a trained teacher network model; inputting the training data into a student network for second training to obtain a trained face living body detection model; and inputting the to-be-detected face image into the face living body detection model to obtain a detection result whether the object in the face image is a living body or not. According to the invention, the knowledge distillation learning device is used to guide the student network to carry out training, the student network model with fewer model parameters is obtained as the trained face living body detection model, and the calculation speed of the model and the speed of face living body detection are improved; the training process of the human face living body detection model is assisted by updating and optimizing the model parameters through the meta-learning device, and the universality is better. The method and device can be widely applied to the field of artificial intelligence.

The invention relates to the field of electric vehicles and discloses an electric vehicle CAN network anomaly detection method. The method comprises a step of configuring a vehicle control unit with multiple CAN networks which comprise a first CAN network, and configuring a controller related to power to the node of the first CAN network, a step of judging whether message transmission is the message transmission between the vehicle control unit and the first CAN network or not when a message is detected, a step of detecting whether external equipment transmits a virtual message by using a mode of receiving a transmission message ID if so, and a step of judging whether the vehicle control unit has a message transmission fault if not, disconnecting a data terminal power supply if the vehicle control unit has a message transmission fault, judging that a data terminal is invaded by an external network if the fault is recovered after the power supply is disconnected, and judging that the CAN networks are connected to invasion equipment if the vehicle control unit does not have a message transmission fault. According to the scheme, the requirement of the computing ability of a chip is low, the first CAN network can immediately emit fault information, and the real-time requirement of power equipment is satisfied. The invention also discloses an electric vehicle CAN network anomaly detection device.

The invention relates to a method for realizing a digital copyright management system and a digital copyright management system. The method for realizing a digital right management system comprises the following steps: receiving a media content encrypted by a back-end digital copyright management system; decrypting the encrypted media content according to a special license transmitted by the back-end digital copyright management system; re-encrypting the decrypted media content; transmitting the re-encrypted media content to a media content server; receiving a license request message transmitted by a service platform to generate a license; and transmitting the license to a user terminal corresponding to the media content. The invention realizes interactive operation between the DRM systems of diversity and heterogeneity, is convenient for users to operate, reduces the requirement on the computation capacity of the user terminal and simplifies the complexity of the user terminal.

The invention discloses a power battery insulation resistance online monitoring device. The monitoring device comprises a controller unit, a signal conversion module, a relay switching module and a signal amplifying module; the signal conversion module is accessed into one output end of the controller unit, the relay switching module is accessed into the other output end of the controller unit, the relay switching module and the signal amplifying module are accessed into the output end of the signal conversion module separately, the controller unit is accessed into the output end of the signalamplifying module and generates a PWM signal, and the signal conversion module converts the PWM signal output by the controller unit into a sinusoidal signal with fixed frequency; the controller unitcontrols switching of the relay switching module; the sinusoidal signal is injected into a power battery output circuit after passing through the relay switching module; the signal amplifying moduleadjusts an obtained insulation resistance detection signal to be within a detectable range of the controller unit; the controller unit conducts calculating to obtain a power battery insulation resistance value through the collected insulation resistance detection signal.

The invention discloses an optimization and control integrated design method for tail gas waste heat recovery organic Rankine cycle. The method comprises the steps of building a mechanism model of a tail gas waste heat recovery organic Rankine cycle system; designing a control structure of the organic Rankine cycle system, designing a controller and a corresponding control equation for the control structure, and forming a closed-loop control system by the designed controller and the mechanism model; selecting tail gas working conditions in a power change range of tail gas waste heat; according to the selected tail gas working conditions and the closed-loop control system, building an integrated optimization model comprising tail gas working condition uncertainty analysis and solving an optimal operation point and an optimal control parameter of the integrated optimization model under each tail gas working condition; and according to the optimal operation point and the corresponding control parameter obtained by optimization solving, obtaining optimal operation points and control parameters corresponding to other tail gas working conditions by utilizing a regression method. The method can improve the waste heat recovery efficiency of the organic Rankine cycle system and is small in online calculation amount and good in robustness.

The embodiment of the invention provides a method and device for analyzing a human body in an image and electronic equipment, and the method comprises the steps: obtaining a to-be-detected image whichis an image comprising a character; Inputting the image into a human body analysis model, the human body analysis model being used for performing feature extraction on the image, performing prediction of at least two analysis types on the extracted features, and outputting prediction results of the at least two analysis types corresponding to the image; Obtaining a human body analysis result corresponding to the image according to an output result of the human body analysis model; According to the embodiment of the invention, the human body analysis model is used for extracting the image, andthe extracted features are used for predicting a plurality of analysis types at the same time, so that compared with the prior art, the number of times of feature extraction on the image is reduced,therefore, the calculation amount of the human body analysis model is greatly reduced, and the calculation capability requirement on electronic equipment is reduced.

The invention discloses a train operation control method and system. The method comprises steps of building a virtual vehicle-mounted equipment model and a virtual trackside equipment model in groundcontrol equipment; based on the virtual vehicle-mounted equipment model and the virtual trackside equipment model, performing safety logic control on trains within the jurisdiction range of ground control equipment. The method is advantaged in that the virtual vehicle-mounted equipment model and the virtual trackside equipment model are arranged in the ground control equipment, so functions of thevehicle-mounted equipment are simplified, the number of safety equipment of the train operation control system and interfaces between the equipment are greatly reduced, system cost is reduced, and compatibility and safety of the system are improved.

A method for obtaining the traffic parameters in video mode includes such steps as drawing a rectangular region (1-7m X 1-6m) in a video image, drawing a white line or putting a long rod on a road close to the designated region, calculating the ratio of the marker length to the length of the marker in the image, searching the characteristics of image corner in the region, and obtaining the traffic volume, car speed and average car speed according the the number of the said corner characteristics and the change in position.

The invention relates to the technical field of positioning, and concretely relates to a vision and inertia combined positioning method. A visual positioning result is combined with an inertia component attitude result to reckon the predicted positioning result, and when the new visual positioning result appears, the consistence of the new visual positioning result with the predicted positioning result is judged; and when the new visual positioning result is consistent with the predicted positioning result, the new visual positioning result, a new predicted positioning result reckoned in a subsequent period of time and the predicted positioning result undergo attitude fusion, and the fused attitude is output as a positioning result. Consistence judgment and attitude fusion make the real-time property and the accuracy of the positioning high and the positioning result gentle.

The invention discloses a mobile network user perception analysis system and a mobile network user perception analyzing method based on big data; the system comprises a collection layer, a data sharing layer and an application layer; the collection layer comprises an intelligent collection module used for collecting a signaling and message flow of a current RNC (radio network controller) generated by a GPEH (general performance event handling) platform from an OSS (open source software) according to a preset frequentness; the data sharing layer comprises a data sharing module used for processing the signaling and message flow obtained from the module of the collection layer and then providing the flow for a customer perception multidimensional data module; the application layer comprises a customer perception multidimensional data module and a network real-time monitoring module; the customer perception multidimensional data module is used for modeling a customer perception and providing a perception model with multiple dimensions for a customer; the network real-time monitoring module obtains interface data from the customer perception multidimensional data module and carries out real-time monitoring, tracking and warning. The system and the method disclosed by the invention have the big data analysis function, thereby creating a condition for fine network planning in a global scope; a network structured analysis and a network difficult problem depth analysis can be realized.

The invention relates to a system and a method for media multipath relay transmission business quality-of-experience collaborative evaluation, and belongs to the field of communication. In the system, a QoE (quality of experience) monitoring and evaluation management server is added on a server-side, and a QoE monitoring and evaluation management client-side provides parameter information required by business QoE evaluation for collaborating with the server to complete business quality-of-experience evaluation and management. The QoE monitoring and evaluation management server is regarded as a core, and a calculability mode of QoE evaluation on the server-side is achieved, calculation capability requirement of the client-side is effectively lowered. The system and the method are beneficial to business QoE calculation mode database maintenance in media transmission, a problem that business QoE evaluation can not be carried out due to database missing caused by single client-side maintenance is effectively avoided, and accordingly, calculating pressure of the client-side is reduced, an evaluation mode with the server as a center and other nodes collaborating is achieved, and QoE evaluation feedback efficiency in media transmission is improved, and guidance of transmission path and media load distribution is further completed.

The invention relates to a static SOC estimation method based on dichotomy iteration, and belongs to the field of new energy measurement and control; according to the method, an approximation iteration method is obtained according to a dichotomy theory, and intermediate value comparison processing is carried out by continuously reducing the interval, so that the effective estimation of the staticSOC value of a lithium ion battery is realized; a fitting polynomial is obtained directly by the lithium ion battery via an OCV-SOC curve, so that transformation or approximation processing is avoided, and the system processing error is reduced to the maximum extent; according to the method, the OCV approximation error range is set, so that the SOC error can be manually adjusted, and the controllable precision is realized; the method is low in algorithm complexity, the number of iterations is small, the response speed is high, the function similar to a universal meter can be realized, and thestatic SOC can be directly measured from the outside; on the basis of a traditional table look-up method and an interpolation method, and the method is improved to be a method for directly utilizing characteristic fitting expression calculation, so that the precision is improved, the pre-data processing work is reduced, and the operation capacity requirements of the processing unit are reduced, the static SOC of the lithium ion battery can be effectively measured and calculated.

The present invention relates to a task package optimization method, system and device. The method comprises: S1, obtaining attribute information of a plurality of packages; S2, establishing corresponding multi-dimensional vectors according to the package attribute information; and S3, performing classification of the packages through inputting the k-means clustering algorithm into the multi-dimensional vectors, wherein the packages with the same classification set are taken as a whole body to input the large-scale neighborhood search algorithm for path planning. order of magnitudes of the packages can be compressed to 10%-20% of the original so as to greatly reduce a search space and reduce the calculation capacity requirement for the anaphase algorithm, the packages processed through the k-means clustering algorithm satisfies the requirements of accurate, stable and rapid path planning of large-scale neighborhood search with no need for extra cost investment.

The invention relates to an automatic generation method of an interface document. The method comprises the following steps: generating analog request information; performing analog request on a browser according to the generated analog request information, and collecting a return result of the browser; and automatically generating the interface document according to the return result of the browser. Compared with the prior art, the automatic generation method of the interface document provided by the invention has the advantages of being capable of reducing the time for developers to write the interface document, high in degree of automation, and convenient to realize and so on.

The invention provides a method and a system for simulating touch instructions by shielding a camera shooting area. The method comprises collecting the image sequence of a shielding operation captured by a camera; identifying the collected image sequence to separate the area with a shield and the area without a shield in every image, and converting every separated image into a binarized image; analyzing the change of the shield in every image of the image sequence to identify operational events happed in the image sequence; determining whether the identified operational events confirm to preset events in a preset event sequence; if so, generating corresponding simulative touch instructions; if not, not generating any simulative touch instructions; performing corresponding simulative touch operations according to the generated simulative touch instructions. The invention also provides the system corresponding to the method. The method and the system for simulating the touch instructions by shielding the camera shooting area greatly reduce the requirements on operational capability of the system compared with existing hand touch identification, and save keys or touch input elements compared with existing touch input manners.

The invention provides an advertisement monitoring method, device and system, electronic equipment and a storage medium. The advertisement monitoring method is applied to a server and comprises the following steps: acquiring a screenshot image of content displayed by display equipment within preset time; extracting advertisement features from the screenshot image; matching the advertisement features with a sensitive advertisement feature library; and when it is determined that the matching rate of the advertisement features and the sensitive advertisement feature library exceeds a preset valuewithin a preset time, determining that a provider of the display content plays the sensitive advertisement. Advertisement monitoring is executed by the server, and compared with manual screening andrecognition, the advertisement monitoring strength can be improved to a certain extent.

The invention relates to the field of vehicle intelligent perception, and discloses a vehicle behavior identification method. The method comprises the steps of obtaining the current transverse distance of a front vehicle relative to a main vehicle and the first duration of transverse distance change of the front vehicle; wherein the front vehicle is a vehicle which runs in front of the host vehicle and is adjacent to the host vehicle, and the front vehicle and the host vehicle run on different lanes; determining a first lane changing two-degree-of-freedom chi-square value of the front vehicle based on the current transverse distance and the first duration, and determining a first lane changing probability of the front vehicle relative to a lane where the host vehicle is located based on the first lane changing two-degree-of-freedom chi-square value; and based on the first lane changing probability and a first set lane changing probability threshold value, identifying the lane changing behavior of the front vehicle. According to the method, the cut-in and cut-out behaviors of the vehicle are identified by using the chi-square distribution of two degrees of freedom, and the vehicle behavior identification efficiency is improved on the basis of ensuring the vehicle behavior identification accuracy.

The embodiment of the invention discloses a workshop real-time scheduling method and device based on a composite intelligent agent relates to the technical field of workshop scheduling control, the requirements and quantity of operation carriers can be reduced, a central-node-free structure is adopted, a center node with high stability and high computing power is not needed. The method comprises astep of selecting a workpiece in a workpiece buffer area when an equipment intelligent agent detects that physical equipment is idle, a step of reading data in an RFID chip of a workpiece tray wherethe selected workpiece is located, activating a workpiece intelligent agent and generating a workpiece intelligent agent program corresponding to the selected workpiece according to the read data in the RFID chip, a step of allowing the generated workpiece intelligent agent to update the data in the RFID chip of the workpiece tray where the selected workpiece is located after the processing negotiation of a next process is completed, and a step of destroying the workpiece intelligent agent program corresponding to the selected workpiece and switching the workpiece intelligent agent into a non-activated state. The method and the device are suitable for a multi-intelligent-agent manufacturing workshop.

The invention provides a block motion real-time detection method based on feature point identification. The method comprises the following steps: S1, calibrating a camera; S2, shooting a digital image of a bulwark surface block through a camera, and sending the digital image to a digital signal processing system based on a field programmable logic gate array; S3, carrying out feature point detection on blocks in the image by the digital signal processing system ; S4, carrying out coordinate transformation after feature point detection is completed; S5, comparing the changes of the positions of the feature points of the block before and after the test, and subtracting the coordinates of the two images before and after the test to obtain a feature point change value so as to obtain the displacement of the block; and S6, displaying a displacement calculation result on a data processor. According to the block motion real-time detection method based on feature point recognition, the problem that the existing detection method cannot qualitatively analyze the motion and displacement of the block to realize quantitative measurement and real-time measurement is solved.

The invention provides a DSE optimization method based on a K-NN algorithm and a device thereof. The method comprises the steps of performing scheduling priority configuration on a traffic level in aTSN network model; training a K-NN algorithm, which replaces a traditional algorithm to execute schedulability analysis, in the DSE according to the configured scheduling priority to obtain a node proportion value meeting requirements; and comparing the node proportion value with a preset threshold value, if the node proportion value is smaller than the preset threshold value, using the K-NN algorithm, and if the node proportion value is greater than the preset threshold value, using a traditional algorithm in DSE to adjust the size of the preset threshold value. The DSE optimization method isrealized by combining a K-NN algorithm. A K-NN algorithm in a machine learning algorithm is used to replace an original schedulability analysis algorithm in traditional DSE. When the K-NN algorithm is judged to be infeasible, a traditional algorithm is still used, the calculation speed of schedulability analysis is increased, meanwhile, the false alarm rate is reduced, and the DSE optimization method and device capable of keeping high prediction accuracy and reducing false alarms are provided.

The invention discloses a software-defined satellite-ground convergence network QoE (Quality of Experience) sensing routing architecture based on deep reinforcement learning, which utilizes the centralized control and programmable capabilities of a software-defined satellite-ground convergence network, organically combines the self-learning capability of deep reinforcement learning, and takes user QoE as a routing planning basis. A super controller, a domain controller and a slave controller which are physically distributed are adopted to realize collection and updating of network states, calculation of inter-domain routing, mapping and convergence of service flows, distribution of routing instructions and configuration of satellite switch flow table entries; and deploying a multimedia service QoE evaluation module and a deep reinforcement learning agent on the intra-domain controller, and calculating an intra-domain forwarding path by using a deep reinforcement learning algorithm by taking a user QoE value fed back by the QoE evaluation module as an award. According to the method, the calculation load between the controllers can be balanced, the accuracy requirements on the calculation capability of the controllers and the dynamics analysis and service flow modeling of the satellite-ground fusion network are reduced, and the user experience quality of the whole network is improved.

The invention provides a lightweight anonymous authentication method suitable for resource limited equipment, the method is realized by adopting an information system model, and the information systemmodel comprises three entities of a network management center, the limited equipment and an information center, and specifically comprises the following steps: setting system parameters, registering,signing, verifying and confirming a secret key. The invention can be suitable for equipment with limited computing resources. The invention discloses an anonymous authentication method. Only one elliptic curve point multiplication operation needs to be executed at the authentication end, and compared with a previous elliptic curve-based anonymous authentication method, the method has the advantages that the overhead of computing resources and communication resources is further reduced while the security is ensured, so that the method is more suitable for resource-limited equipment, and the requirement of the equipment on the computing capacity is reduced.

The invention relates to a cryptographic technology security control method and system for core table data, and the method comprises the steps: transforming a data table stored in a database, extracting a user table and an authority table, integrating the user table and the authority table into an access control table, carrying out the statistics of user access information based on the user tableand authority table data, carrying out the statistical classification according to the multiplying power of access frequency, and setting n levels of access frequency, establishing a frequency table,and establishing a corresponding multi-level database index table according to the frequency levels in the frequency table structure corresponding to the pure data class table; forming ciphertext forthe key data in the access control table and the frequency table; after user identity authentication, intercepting keywords from user login information and retrieval information, searching records containing the keywords from the frequency table and the access control table, obtaining corresponding ciphertext and carrying out decryption, and accessing a corresponding data table in the database. According to the method and the system, under the condition that all data of the database table cannot be subjected to full-amount encryption and decryption protection, safety control and plaintext confidential storage of the structural data of the database core table can be realized.

The invention provides an encryption method and system of a geological survey instrument based on a chaos sequence. The encryption system comprises a seed key pool arranged at an encryption end, wherein the seed key pool consists of a pre-set number of 32bits seed keys and is numbered. The encryption method comprises the following steps: randomly selecting a plurality of seed keys from the seed key pool during encryption; sending the selected seed keys into an extension coupling integer tent to generate a key sequence by mapping; and encrypting pre-set data. According to the encryption method and the encryption system provided by the invention, the calculation difficulty of chaos mapping and consumption are reduced through an integer method, so that the requirements on the power consumption of the geological survey instrument and ground disaster monitoring equipment can be met, and the excellent security property can be realized through expansion and coupling. Furthermore, the load calculation such as a division method and a power method can be avoided in the process; the requirement on the calculation capacities of hardware is low, so that the encryption method and the encryption system are applicable to the existing geological survey instrument and ground disaster monitoring equipment without redesign and redevelopment of the existing equipment.