274results about How to "Reduce the number of channels" patented technology

The patent discloses an intersection vehicle violation detection and identification method based on deep learning. The method comprises the following steps of 1, collecting intersection video detection images; 2, identifying traffic signs and vehicles; 3, tracking the vehicles; and 4, performing vehicle violation detection and identification. Based on automatic identification and tracking of the traffic signs and the vehicles in the intersection monitoring videos, the violation conditions of vehicles at the intersection can be automatically identified, and the system has few false detection and high efficiency.

The invention belongs to the technical field of signal processing, and particularly relates to an electroencephalogram emotion recognition method based on a graph convolutional neural network. The electroencephalogram emotion recognition method provided by the invention comprises the following steps: firstly, constructing an emotion-related function connection mode according to the correlation between electroencephalogram channel signals; then modeling each extracted single-channel electroencephalogram feature into a graph signal based on a function connection mode structure, and establishinga graph convolutional neural network model; inputting a training set graph signal to carry out iterative optimization training on the model, and optimizing network parameters by adopting a back propagation algorithm and a cross entropy loss function; and finally, inputting a test set graph signal into the trained graph convolutional neural network, and performing classification recognition on thetiter, wake-up and dominance three-dimensional emotion model. According to the electroencephalogram emotion recognition method, the spatial and functional relationship of the electroencephalogram datais restored, so that electroencephalogram emotion features with higher discriminability can be extracted; and the channel electrode most relevant to emotion can be selected according to the channel correlation, and processing time and hardware cost are reduced while the number of channels is reduced.

Tissue stimulation systems generally include a pulse generating device for generating electrical stimulation pulses, at least one implanted lead including at least one electrode for delivering the electrical stimulation pulses generated by the pulse generating device, and a programmer capable of communicating with the pulse generating device. In tissue stimulation systems, two or more electrical stimulation combinations may be delivered to a patient simultaneously through a reduced number of channels. Systems and methods described herein may combine the two or more stimulation combinations into a reduced number of new stimulation combinations for delivery of the stimulation combinations over a reduced number of channels.

The invention discloses a bar-cigarette self-picking craft, which is characterized by the following: opening the cigarette box automatically; fetching cigarette pile from box; feeding material automatically through over two picking equipments; picking different kinds of bar-cigarette on respective transmission equipment according to the need of order. The invention can improve material-supplement and picking efficiency and system disposal ability, which reduces the labor strength.

The embodiment of the present application discloses a face key point positioning model training method and device, an apparatus and a storage medium. The training method includes: constructing a CNN model for face key point positioning, wherein the number of convolution layers of the CNN model is greater than a first threshold, and the number of channels of the convolution layers is less than a second threshold; using the CNN model to perform face key point positioning on training samples, and obtaining prediction positions of face key points, wherein the face key points include n types, and the n is an integer greater than 1; calculating respective loss function values corresponding to the n types according to prediction positions and real positions of each type of face key points, and calculating a loss function value of the CNN model; and stopping training the CNN model and saving the CNN model if the loss function value of the CNN model is less than a preset threshold. The embodiment of the present application reduces the model size by constructing the elongated CNN model while ensuring that the positioning accuracy is not lost as much as possible.

The invention discloses a touch control display panel, which comprises a first substrate, a second substrate, a display medium layer and a touch control electrode layer.The touch control electrode layer is equipped with a plurality of induction areas and each of the induction areas is equipped with a plurality of sub-induction areas.The touch control electrode layer comprises multi-strip driving electrode serials and multi-strip induction electrode serials.Each driving electrode serial comprises a plurality of driving electrodes and each driving electrode comprises a plurality of sub-driving patterns.The induction electrode serials and the driving electrodes are serially interconnected and staggered.Each induction electrode serial comprises a plurality of induction electrodes and each induction electrode comprises a plurality of sub-induction patterns.Each of the sub-induction patterns is respectively equipped with an induction perimeter.The induction perimeter of the sub-induction pattern of the sub-induction area of each of the induction area is different from each other, or the driving perimeter of the sub-driving pattern of the sub-induction area of each of the induction area is different from each other.

The invention relates to a photoelectric converter, a detector and scanning equipment. The photoelectric converter comprises a silicon photomultiplier array and a light guide coupled with the silicon photomultiplier array, wherein the silicon photomultiplier array comprises i*j silicon photomultipliers spliced on a horizontal plane, and the i and the j are both integers greater than or equal to 2. The detector comprises a scintillation crystal, an electronic system, a light guide and silicon photomultipliers. The scanning equipment comprises a detection device and a frame, wherein the detection device comprises the detector, and the detector comprises the photoelectric converter. According to the photoelectric converter, the detector and the scanning equipment, a photoelectric conversion scheme of the photoelectric silicon photomultipliers is mainly employed, the silicon photomultipliers have small volumes and are closely arranged, the silicon photomultipliers in proper dimensions and in proper quantity are matched with the light guide in proper shape, a PET detector with high spatial resolution can be established, so the spatial resolution of the whole PET system can be improved, the PET detector having DOI and TOF performances is proper to establish, application to PET / MRI can be realized, and low cost is further realized.

The invention provides a shape detection device and method based on a spiral FBG (fiber bragg grating) sensor array, and the device comprises the spiral FBG sensor array, an FBG demodulator, data collection and shape reconstruction equipment, and display equipment. The device does not need other peripheral auxiliary equipment, is not affected by electromagnetic interference, is good in medical compatibility, does not emit rays, and can achieve the long-distance remote monitoring. Through the spiral configuration of the FBG sensor array, the device can achieve the multi-point deformation detection of a single optical fiber, and forms a quasi-distributed detection system. Through the packaging of the optical fiber on a nickel-titanium alloy filament, the device protects a grating point from being worn, is strong in anti-interference capability, and enlarges the detection range. The device can obtain the shape of a flexible robot in real time, and is high in real-time response. Meanwhile, the device detects the curvature and torsion, is simple in detection method, can be used for the shape sensing detection in a gastroscope, colonoscope and a flexible / soft robot, and plays an important role in the flexible robots in the fields of dibaster assistance, medical rehabilitation and national defense security.

A touch-sensing electrode structure includes multiple first electrodes and multiple second electrodes. Each first electrode includes a longitudinal part extending in a first direction and multiple branch parts connected to the longitudinal part. The second electrodes are disposed on at least one side of each longitudinal part, and each of the second electrodes at least spreads over a region between two adjacent branch parts of each of the first electrodes. Through the invention, the inductive capacitance of the touch-sensing electrode structure and the sensitivity of the touch control operation can be improved, and the effect of touch-sensing linearity can be improved, and channels required by a single-layer touch-sensing structure and line impdedance can be reduced.

The invention provides an interleaved parallel magnetic-integration bidirectional full-bridge LLC resonant converter and relates to the technical field of power and electronic applications. The interleaved magnetic-integration bidirectional full-bridge LLC resonant converter comprises two paths of an interleaved parallel bidirectional full-bridge LLC resonant converter. The first path of the bidirectional full-bridge LLC resonant converter comprises a first primary-side circuit, a first secondary-side circuit, a resonant transformer and a magnetic inductor thereof. The second path of the bidirectional full-bridge LLC resonant converter comprises a second primary-side circuit, a second secondary-side circuit, a resonant transformer and a magnetic inductor thereof. The primary-side circuit and the secondary-side circuit of each path of the bidirectional full-bridge LLC resonant converter are respectively in the form of a full-bridge LLC resonant circuit and are used for realizing the bidirectional energy flow. The two resonant transformers, the magnetic inductors thereof and four resonant inductors of the two paths of the interleaved parallel bidirectional full-bridge LLC resonant converter are magnetically integrated through integrated magnetic pieces. According to the technical scheme of the invention, the large-power, high-transformation-ratio, small-volume, bidirectional and high-efficiency operation of the electric energy is realized.

The invention belongs to the technical field of wireless communications, and discloses an asymmetrical shared hybrid beam forming transmitting-receiving device. A transmitting end adopts a shared digital-analog hybrid beam forming structure, and a receiving end adopts a digital beam forming structure. Compared with a system adopting a digital beam forming framework in a conventional MIMO (Multiple Input Multiple Output) communication, the device has the advantages that an RF (Radio Frequency) pre-coder constructed by an analog phase shifter is used at the transmitting end, so that the number of radio-frequency links can be reduced; cost is lowered; and total transmitting power is lowered. Compared with a system in which hybrid digital-analog beam forming structures used at the transmitting end and the receiving end, the device has the advantages that only the digital beam forming structure is used at the receiving end of a system model, so that the beam forming structure at the receiving end is simplified, and the overall system complexity is lowered.

The present invention belongs to the sonar small target detection technology field, in particular discloses an anti-frogman sonar transducer array. The anti-frogman sonar transducer array comprises an electronic storehouse shell, an emission transducer array, a reception transducer array and a pedestal. The pedestal is arranged on the electronic storehouse shell, the emission transducer array is a plurality of column structure arrays which are distributed in the circumference direction of the lower part of the electronic storehouse shell, the reception transducer array is a horizontal direction cross array arranged on the pedestal, and the column structure arrays of the emission transducer array are vertical with the reception transducer array. According to the present invention, by a reception and emission separated mode, the emission transducer array covers a circumference 360-degree direction via the plurality of column structure arrays which are distributed in the circumference direction of the electronic storehouse shell, and a vertical directivity open angle of the emission transducer array is controlled by the number of the array elements of each column of the column structure arrays or the phases of the signals added at each array element, so that the narrow beam scanning of a vertical plane can be realized, and the reverberation interference coming from the sea surface and the sea floor is suppressed effectively.

The invention discloses a quantum error correction decoding method and a quantum error correction decoding device based on a neural network, and a chip, which relate to the technical field of artificial intelligence and quanta. The quantum error correction decoding method comprises the following steps of: acquiring error symptom information of a quantum circuit; performing block feature extractionon the error symptom information through using a neural network decoder to obtain feature information; and subjecting the feature information to fusion decoding processing through using the neural network decoder to obtain error result information, wherein the error result information is used for determining error data quantum bits and corresponding error types in the quantum circuit. According to the quantum error correction decoding method and the quantum error correction decoding device, a block feature extraction mode is adopted, the number of channels of the feature information obtainedby each feature extraction is reduced, and the input data of the next feature extraction is reduced, thereby being conductive to reducing the number of feature extraction layers in the neural networkdecoder, further reducing the depth of the neural network decoder, shortening the decoding time of the neural network decoder correspondingly, and satisfying the requirement of real-time error correction.

A touch-sensing structure includes a substrate and a conductive layer. The conductive layer spreads over a surface of the substrate and includes a plurality of first electrodes, a plurality of second electrodes, a plurality of first conductive lines, and a plurality of second conductive lines. The surface is divided into a plurality of regions. The second electrodes are divided into multiple second electrode groups, and each second electrode group is formed by at least one of the second electrodes in each of the regions. Each of the first conductive lines is connected to one of the first electrodes, and each of the second conductive lines is connected to one of the second electrodes. The second conductive lines connected to the second electrodes in the same second electrode group are electrically connected with each other.

The invention provides a controllable high-speed multi-channel signal acquisition control circuit system. The system comprises an FPGA (field programmable gate array), wherein the FPGA is connected with an FOT (fiber optical transceiver) to realize communication with a host computer; the FPGA is connected with a DAC (digital-analog converter) chip to output analog signals for control or driving; the FPGA is connected with a plurality of AD (analog-digital) conversion chips to realize analog data conversion of multiple channels; the FPGA is connected with a reserved control port for an extending function. The controllable high-speed multi-channel signal acquisition control circuit system meets the work requirement for synchronous continuous sampling of a large quantity of analog signal channels, solves the problem of resource waste caused by repetitive design of a common high-speed multi-channel signal acquisition control circuit system, shortens the design time for a multi-channel signal acquisition circuit and is applicable to signal acquisition work on various occasions.

The embodiment of the invention provides a target detection method and device, and the method comprises the steps: detecting the size of a preset to-be-detected target, and determining the size of a first convolution kernel; carrying out the convolution operation of an original image through the first convolution kernel, and obtaining an initial characteristic pattern of the original image; determining a characteristic extraction model comprising a plurality of preset characteristic extraction layers, wherein the characteristic extraction model is used for outputting the effective characteristic information, according with a preset task, of the image according to the initial characteristic pattern through the convolution operation of the plurality of preset characteristic extraction layers; obtaining the type of the to-be-detected target and the original information of the to-be-detected target in the original image according to the characteristic information through a preset multi-class regression algorithm and a preset position regression algorithm. The method can improve the actual operation efficiency of target detection.

The invention discloses a satellite-borne three-array-element monopulse two-dimensional direction finding method. The satellite-borne three-array-element monopulse two-dimensional direction finding method comprises the following steps: firstly, maximum baseline limit is calculated according to the minimum wavelength of received signals and the amplitude comparison angle measurement accuracy, the minimum baseline length requirement is calculated according to the required angle measurement accuracy and the maximum wavelength of the received signals, the maximum and minimum baseline requirementsare compared, and the length of baselines is determined; an amplitude difference tablet and a phase difference tablet which correspond to incident signals in different directions are calculated according to the antenna interval, the antenna wave beam directional angle and an amplitude and phase directional diagram; then the amplitude difference of the same signal is received through three array elements to be related to the amplitude difference tablet, and rough finding values of the signal incident directions are obtained; and finally, the search range of the signal incident directions is determined through the rough finding result and direction finding accuracy of amplitude comparison direction finding, the phase differences between three channels are related to the phase difference tablet to be interpolated, and then the fine finding angle result is obtained. By decreasing the number of the array elements and the system channels and the quantity of system equipment, the direction finding accuracy of a satellite-borne warning system is effectively improved.

The invention discloses a vehicle-mounted detection method fusing vehicle speed information and axle box vertical acceleration information for wheel flats, which mainly solves the problems that in the prior art, the existing detection method for wheel flats has the disadvantages of limited application scope, inconvenience in use and maintenance, and higher cost. The method comprises the following steps of: detecting vehicle operating velocity v and vertical vibration acceleration a of a vehicle axle box; carrying out information fusion on the velocity v and the vertical vibration acceleration a of the axle box according to a fusion rule, carrying out feature extraction on the fused data to obtain the vibration performance of the vehicle axle box, and then analyzing the vibration performance of the vehicle axle box when wheel flats exist; and digesting and comparing the obtained vibration performances, building a judgment rule of wheel flats, and then carrying out dynamic detection on wheel flats. Through the scheme, the aims of broader application scope, convenience in implementation and maintenance, and low cost are achieved, and higher practical value and popularization value are realized.

The invention provides a multilevel series LTD driving source and a synchronous trigger method of the multilevel series LTD driving source. An external trigger pulse is introduced into each of front n level modules of an LTD pulse source, synchronous trigger of other branch switches of n level modules is achieved by means of a branch circuit and an angular orientation transmission line in each module, meanwhile, a trigger pulse is introduced from each trigger branch circuit high voltage terminal of the n level modules, the trigger pulses are sequentially introduced into trigger pulse input ends of follow-up series modules at corresponding positions to trigger switches of trigger branch circuits in the follow-up modules, the remaining series modules are carried out according to the previous steps, and all the series modules of an LTD are synchronously triggered. The requirement of the multilevel series LTD pulse source for a trigger system and a trigger timing sequence can be obviously simplified, and module trigger reliability is improved.

The invention discloses a method for forming an ash removing water film for a wet electric dust remover and a dust collecting pole plate used by the same. The invention solves the problem that a water film for completely covering the pole plate can not be formed on a super-high pole plate higher than 10 meters, thus the wet electric dust remover can be used for removing dust in super-large-flow flue gas such as flue gas in coal burning boilers of thermal power plants and the like, and the invention has the advantages of reducing floor space, saving dust removing water, and the like. The method for forming the ash removing water film comprises the following steps: dividing the dust collecting pole plate into a plurality of vertical strip-shaped regions along the airflow direction, supplying water to the pole plate from the upper space of the dust collecting pole plate, limiting the dust removing water in each narrow strip-shaped region to flow downwards, and covering each small strip-shaped region, thereby forming the water film for completely covering the whole dust collecting pole plate.

The invention relates to a method of using sum and difference channels of a radar antenna for restraining radar side lobe interception. According to the method, antenna patterns of the sum channel and the difference channel of the radar main antenna are optimized, self-correcting signals of a radar system are used for obtaining amplitude and phase errors with the sum channel as the datum, and the amplitude and phase errors are used for error correction of the sum channel and the difference channel; according to the target position in the sum channel, the corresponding position, in the difference channel, of a target is searched for, and amplitude and phase correction is carried out on the complex number value of the position, where the target is located, in the difference channel by means of the amplitude and phase errors of the sum channel; the complex number modulus value of the position, where the target is located, in the sum channel is compared with the corrected complex number modulus value of the target in the difference channel to judge whether target echoes enter from main lobes or side lobes of the main antenna. Therefore, the radar system using the method can directly use the main antenna for recognizing side lobe targets without depending on an auxiliary antenna, the design difficulty of the radar antenna is effectively lowered, the number of the channels of the antenna is effectively reduced, and the reliability of the antenna is improved.

The invention relates to the emotion recognition field and the EEG channel selection technology, in particular to an EEG channel selecting method based on standard mutual information. The method includes the following steps of collecting data, extracting EEG through a public data set, conducting the short-time Fourier transform on the EEG to obtain a time-frequency graph, establishing an incidencematrix by calculating standard mutual information values between all channels in the time-frequency graph, analyzing the incidence matrix, setting the threshold selection channel, determining the optimum channel, and classifying data through a support vector machine classifier after the channel is selected to obtain the emotion recognition rate. The method has the advantages that the high emotionrecognition rate is still kept under the condition of greatly reducing the number of the channels, the selected channel provides an applicable scheme for the design of a wearable EEG device, the design complexity problem of hardware is solved, the classification accuracy of EEG can be effectively improved, and the data redundancy and computation complexity caused by using signals of all the channels are overcome.

The invention discloses a sparse projection ultrasonic CT image reconstruction method based on CS(compressed sensing), comprising the steps of: first scanning an object for times much less than times required the Nyquist theorem to obtain the projection data in a finite direction; performing one-dimensional Fourier transform on obtained projection data based on the Fourier diffraction projection theorem to obtain the space frequency information of an object fault cross section; allowing the object to possess sparsity in a transform domain through selecting a transform radix, and constructing the inverse problem based on 11 norm optimization according to the CS principle; and finally reconstructing an object image through a conjugate iterative algorithm. The method can substantially reduce projective ultrasonic CT projection times, system control complexity, equipment data memory space, and system scanning time, and minimize artifacts caused by patient breathing clinically; through iterative solution, the method can obtain high reconstruction quality, and can be greatly applied in the medical imaging field.

The invention discloses a device and a method for detecting feeble signals of a global navigation satellite system. The method comprises the following steps of: determining respective signal power and noise threshold by each detecting channel and outputting the signal power and the noise threshold to a multi-resident detector; comparing the signal power with the noise threshold by the multi-resident detector, if the signal power is higher than the noise power, the detection is successful, and then one is added on a counter; if the signal power is lower than the noise power, the detection is failed, and then one is added on a counter; repeating the steps, and judging whether the feeble signals of the global navigation satellite system are successfully detected or not by the multi-resident detector, if the detection is successful, transmitting the current code phase and a Doppler frequency point to a subsequent dynamic tracking module on the unified time point of the system to carry out dynamic synchronization; and if the detection is failed, transmitting the state information to a searching module of a unit to be detected on the unified time point of the system. By adopting the method and the device, the channel number of a detection correlator is reduced and the detection on the feeble signals of the global navigation satellite system is realized.

The invention relates to a multi-stage series-connected linear transformer driver (LTD) and a trigger method of the LTD. The LTD comprises multiple series-connected stacked LTD modules, wherein all the LTD modules are disposed in a common non-magnetic metal housing. The technical problems of complicated trigger circuits, high manufacturing cost and difficulty in guaranteeing a trigger sequence by use of a conventional multistage series-connected LTD are solved. The multi-stage LTD modules are connected in series and stacked in a common housing, 3 to 4-stage LTD modules can share one trigger pulse, and all the LTD modules can share one charging cable, such that the number of introduced trigger cables and the number of charging cables are substantially reduced, conventionally quite complicated line arrangement is simplified, and the equipment operation reliability is improved.

The invention provides an organic light emitting diode pixel compensation circuit used for driving an organic light emitting diode. The organic light emitting diode pixel compensation circuit is characterized by that the compensation circuit comprises a multipath multiplexer and a plurality of sub-pixel compensation circuits. The connection relationship between any one switch in the multipath multiplexer and the corresponding sub-pixel compensation circuit is as follows; a first connection terminal of the switch receives a data voltage, a control terminal of the switch receives an enable signal to make the switch in response to an effective electrical level of the enable signal received by the control terminal for conducting conduction, and a second connection terminal of the switch is connected with the corresponding sub-pixel compensation circuit so as to provide the data voltage for the corresponding sub-pixel compensation circuit. By adopting the organic light emitting diode pixel compensation circuit, the multipath multiplexer can be used for providing the data voltage for each organic light emitting diode sub-pixel, the number of data signal transmission channels is greatly reduced, and the production cost of an integrated circuit is reduced.

The invention discloses a rail local defect vehicle-mounting detection method merging with vehicle speed information and front and rear axle box acceleration information, which belongs to the traffic information security technology. The method comprises the following steps of: collecting the vehicle speed information and the front and rear axle box vertical vibration acceleration information of at least one vibration acceleration sensor at one position by a collector, and then merging the speed information with the front and rear axle box vertical vibration acceleration information, extracting the features of the merged information to obtain a time frequency energy distribution function of the vibration acceleration, then finding a displacement coordinate position corresponding to the time frequency energy distribution function peak value, and finally analyzing the acceleration signal at the position of a rotating cycle of a wheel in a shaft space by the method. With the method, the impact feature of the rail local defects can be detected, the line modification is free, and a detection system is simple. Compared with the prior art, the method has high reliability, low cost, fewer channels, a simple system, high anti-interference performance, a wider application space and a higher application value.