127results about How to "The output is accurate" patented technology

The invention discloses a numerical simulation determination method of a deep high-stress roadway drilling pressure relief parameter. Before simulation is carried out, a test roadway surrounding rock sample is firstly subjected to a loading and unloading test in a laboratory, an attenuation relationship between a rock sample strength parameter and a damage variable is obtained by fitting and is embedded into a FLAC<3D> (Fast Lagrangian Analysis of Continua) built-in strain softening constitutive model, and an actual measurement mine pressure expression law of a test roadway is taken as a known characteristic value to carry out inversion on the numerical calculation model parameter of a rock mass. A deep roadway drilling pressure relief numerical calculation model is established to simulate and analyze the influence law of a pressure relief drilling orientation, pressure relief opportunity and drilling parameters (diameter, length and inter-row spacing) on the stability of roadway surrounding rock, and a determination method of each factor which affects a pressure relief effect is put forward, and a drilling pressure relief technology system is perfected. The numerical simulation determination method considers the mutual action relationship of a pressure relief degree and the stability of the surrounding rock while pressure relief drilling parameters are determined, and the designed pressure relief drilling can perform a maximum pressure relief effect on a premise that the stability of the surrounding rock is guaranteed. The method is simple and exhibits high application value.

The invention relates to the field of computer vision and discloses a computer vision-based vehicle damage distinguishing method so as to solve a problem that a manual distinguishing approach adopted in conventional technologies is incomplete, inaccurate and inefficient in distinguishing operation. The computer vision-based vehicle damage distinguishing method comprises the following steps: in step a, a binocular image collection system is calibrated; in step b, the binocular image collection system is used for collecting images in a monitored area, and a depth map for the collected images is obtained; in step c, a convolutional neutral network is used for subjecting the depth map to characteristic extraction training operation, and a vehicle damage condition discrimination model is trained; in step d, the vehicle damage condition discrimination model is used for determining a damage degree based on collected vehicle images. The computer vision-based vehicle damage distinguishing method is suitable for distinguishing vehicle damage.

Embodiments of the application disclose a data model training method and apparatus. The data model training method in one embodiment of the application comprises: obtaining original sample data, wherein the original sample data comprises user identifier information; based on the user identifier information, obtaining network operation data of a user; according to the network operation data, determining personalized information of a user; and performing training by taking the original sample data and the personalized information as sample data to obtain a data model. According to the technical scheme provided by the embodiments of the application, the sample training richness can be improved, so that the high-accuracy data model is trained.

The invention discloses a voice-based mouth shape animation synthesis device. The device comprises a memory and a processor, wherein a mouth shape animation synthesis program capable of running on theprocessor is stored on the memory, and the program is executed by the processor through the steps that target text data is acquired, and phonemic characteristics in the target text data are acquiredaccording to a pronunciation dictionary; the phonemic characteristics are input into a pre-trained deep neural network model to output acoustic characteristics, and the acoustic characteristics are input into a voice synthesizer to output voice data; according to the voice data, a pre-trained tensor model and speaker identification information, mouth shape data is acquired; and a mouth shape animation corresponding to the voice data is generated according to the mouth shape data. The invention furthermore provides a voice-based mouth shape animation synthesis method and a computer readable storage medium. Through the voice-based mouth shape animation synthesis device and method and the computer readable storage medium, the technical problem that a mouth shape animation which is matched with synthesized voice data and has a sense of reality cannot be displayed in the prior art is solved.

The invention discloses a pace-making signal detecting method and system and an electrocardio detecting device. The method comprises preprocessing original pace-making signals separated from pace-making electrocardio signals, and then based on the preprocessed signals, computing the basic morphological parameters of the pace-making signals, wherein the basic morphological parameters include the width, the slew rate and the amplitude of the pace-making signals; based on the calculated basic morphological parameters of the pace-making signals, determining whether the pace-making signals are true or false, and recording the position information of the pace-making signals which are determined to be true. The pace-making signal detecting device can detect the pace-making signals more accurately and mark the pace-making signals on electrocardio signals for output.

The invention provides a motion object detection method and device. The method comprises the steps that in a compressed domain of video streaming, motion vector prediction is performed on pixel blocks of an intra-frame code in a current frame of image, so that motion vectors of the pixel blocks of the intra-frame code are obtained, wherein each frame of image in the video steaming is divided into a set number of pixel blocks according to setting rules in advance, and the sizes of all frames of images are equal; according to the frequency that motion vectors corresponding to pixel blocks located at the same position are non-vanishing vectors in the current frame of image and the continuous M frames of images before the current frame of image, all non-vanishing vectors in the current frame of image are determined; according to all the determined non-vanishing vectors, a motion area in the current frame of image is determined; the motion area is mapped into a pixel domain, in the motion area, preset targets are detected based on the pixel domain target detection algorithm, and the detected preset targets are indicated and then output. According to the motion object detection method and device, the accuracy of target detection can be improved, and the speed of target detection can be increased.

The invention discloses a rapid area-of-interest detection method based on depth kernelized hashing and mainly aims to solve problems of low detection precision and low efficiency existing in an image area-of-interest detection method in the prior art. The method comprises steps that in a training process, to-be-trained data sets are classified into multiple categories which are inputted to a depth kernelized hashing supervised learning network framework to acquire corresponding hashing codes, fine tuning of network parameters is carried out according to a label information matrix till the optimal learning result is realized; in a test process, inputted test images are pre-processed, and binary codes are acquired through the trained depth kernelized hashing supervised learning network framework; the area-of-interest position of the image is determined according to a decision function and is marked, and area-of-interest detection and identification are accomplished. Through the method, area-of-interest detection analysis performance of the image can be effectively improved, an area-of-interest detection rate and system framework operation efficiency are improved, and the method can be applied to rapid detection and identification of breast image lumps.

The invention relates to a fever disease deep-learning auxiliary diagnosis system based on text medical history for children. The system includes a medical history input module, a text structuralizingmodule, a characteristic extraction module and a result output module, wherein the medical input module is used for receiving the medical history in a free text form; the text structuralizing moduleis linked with the medical history input module and adopts a natural language processing method for processing the medical history in the free text form to obtain structuralized data; the characteristic extraction module is linked with the text structuralizing module and used for extracting diagnosis characteristics from the structuralized data; the result output module is linked with the characteristic extraction module and used for utilizing a neural network to carry out prediction of disease classification on the current medical history records according to the extracted diagnosis characteristics and a diagnosis model learned from the past medical history records, and then outputting diagnosis results. Compared with the prior art, on the basis of a natural language processing and diagnosis model of deep learning, the diagnosis accuracy of the system can exceed a grassroots doctor level, the system can function stably, and therefore a real help can be offered to grassroots medical workers.

The invention discloses an automatic classification system and an automatic classification method for a Web comment viewpoint on the basis of an association rule. The automatic classification system can be divided into four modules, i.e. a frequent feature word extraction module, a frequent feature word optimization module, an association rule extracting and mining module and a viewpoint classification module. According to the invention, the defects that the traditional system (such as systems based on machine learning and sentiment classification) has low precision or needs mass manpower participation or excessively depends on natural language processing and professional knowledge can be overcome. Optimization is carried out in the association rule set extraction process, redundant association rules with low distinction degree are removed, and therefore, the operation efficiency and the result acquisition efficiency of the whole system can be improved. According to the invention, a precise and convenient solution is provided for various situations, such as E-business commodity evaluation, E-government affair feedback and netizen opinion survey.

The invention provides a suction type SF6 gas leakage monitoring device and method. The device comprises an air outlet (1), an air inlet (2), an infrared light supply (3), a window piece I (4), a window piece II (5), a reference spectral filter (6), a measuring spectral filter (7), a dual channel infrared detector (8), a temperature probe(9), a signal conditioning module (10), an A / D (Analog / Digital) conversion module (11), a data processing module (12), a light source driving module (13), a device body (14), an end cover I (15) and an end cover II (16). The invention provides the suction type SF6 gas leakage monitoring device and method have the advantages that the suction type SF6 gas leakage monitoring device has a digital output and thermal compensation function and is simple and reliable, and the sensitivity is high. The method and the device can carry out ration monitoring on SF6 leaking so as to externally output a high-precision SF6 leaking concentration value, and the method and the device have the advantages that the monitoring accuracy is high, the method and the device are stable and reliable, the structure is compact, the volume is small and exquisite and the like.

The invention provides an input error correction method and apparatus. The method comprises the steps of judging whether an input word character string is full pinyin or not; if yes, performing pinyin error correction processing on the full pinyin of the word character string, performing a reverse query on the full pinyin subjected to the error correction to obtain Chinese characters, and obtaining a first error correction result; or otherwise, performing word segmentation processing on the word character string, converting the word character string into the full pinyin, performing the pinyin error correction processing on the full pinyin obtained by conversion, performing a reverse query on the full pinyin subjected to the error correction to obtain the Chinese characters, and obtaining a second error correction result. According to the method and the apparatus, a similarity calculation method is skillfully applied to similarity calculation of pinyin characters and similarity calculation of Chinese characters; and by applying the method and the apparatus to Chinese search engines and intelligent question-answer systems, the accuracy of query and question-answer of information input for words in the Chinese search engines and the intelligent question-answer systems can be remarkably improved.

The invention discloses a vehicle-mounted integrated navigation system based on information filtering. The system comprises a GPS navigation unit coupled to a motion object, a dead reckoning (DR) navigation unit coupled to the motion object, a turn judgment unit and an information filter (IF) coupled to the GPS navigation unit, the DR navigation unit and the turn judgment unit; the GPS navigation unit periodically calculates GPS navigation information and judges the output result of a GPS turn judgment unit coupled to a GPS according to historical course angle information; DR navigation information is periodically reckoned according to the reference position and direction stored in the DR navigation unit, and a turn judgment unit coupled to the DR navigation unit outputs a judgment result according to the course angle change and curvature; the turn judgment unit integrates the GPS turn judgment unit and the DR turn judgment unit to obtain the state of the motion object and correct a DR formula; the IF is used for periodically calculating the navigation information of the motion object. The IF is used for integrating the GPS navigation information and the DR navigation information, and therefore a navigation function with higher precision and higher reliability is achieved.

The invention discloses a scene analysis method based on binocular vision, and the method comprises the steps: obtaining a first original image and a second original image corresponding to the binocular vision, and extracting the multi-scale features of the first original image and the multi-scale features of the second original image; combining the multi-scale feature of the first original imagewith the multi-scale feature of the second original image, performing depth estimation operation, and outputting primary depth estimation information of the first original image based on the combinedmulti-scale feature; performing semantic segmentation calculation by using the multi-scale features of the first original image, and outputting primary semantic segmentation information correspondingto the first original image; and fusing the primary depth estimation information, the primary semantic segmentation information and the multi-scale features of the first original image, and outputtingfinal depth estimation information and semantic segmentation information. The calculation efficiency can be improved, the calculation speed is increased, and the calculation precision is improved.

The invention discloses a CO2 gas concentration monitoring device and method with the automatic temperature and air pressure compensation. The CO2 gas concentration can be accurately detected in a relatively wide temperature range and a relatively wide air pressure range, and the device and the method are applied to monitoring the CO2 gas concentration in a low-temperature low-pressure environment. The CO2 gas concentration is monitored by adopting a dual-wavelength infrared monitoring principle and a suction structure, and the device has the advantages of adaptation to a low-temperature low-pressure environment, high precision, high reliability and the like. The interference caused by factors such as light source instability is eliminated by adopting a dual-channel infrared detector, the absorption length of infrared light is increased by adopting a reflector, the environment suitability of the device is enhanced by adopting a heating element and a heat preservation material, the receiving light intensity is enhanced and the automatic temperature and air pressure compensation and correction is performed by adopting a conical optical collector, and the CO2 gas monitoring precision is effectively improved.

The invention discloses a roller kiln temperature prediction integrated modeling method capable of combining a mechanism with data. The method comprises the following steps that: through the analysis of factors which affect temperature change, from a perspective of the temperature change and energy, establishing a mechanism model; then, considering situations that roller kiln sintering is a very complex process, a whole sintering process can not be described through a single mechanism model and the mechanism model has model errors through simplification, establishing a data model to predict model errors so as to make up mechanism output, i.e., utilizing errors to serve as a training sample to establish an error prediction model of a nonlinear time-varying process based on local weighted kernel principal component regression; and finally, combining the mechanism model with the data model to establish a roller kiln temperature prediction integration model. By use of the model, the state change of a process can be better tracked, and a good guidance function is provided for roller kiln temperature control so as to improve the production quality and the percent of pass of a product.

The embodiment of the invention discloses a text data processing method, a text data processing device and text data processing equipment. The method comprises the following steps: acquiring target text data aiming at an appointed service; performing word segmentation processing on the target text data, and determining word segmentation positions in the target text data; inserting a predeterminedseparator at a word segmentation position in the target text data; and inputting the target text data into which the separator is inserted into a predetermined language model to obtain result data related to the specified service.

The invention provides a touch method, a touch screen and a writing pen. The touch method comprises the steps that the touch screen receives a position signal sent by the writing pen and converts the position signal into first coordinates, wherein the position signal is obtained in the mode that after identifying an identification label of the touch screen, the writing pen converts obtained position information; the touch screen determines second coordinates according to the signal receiving and transmitting conditions of a receiving module and a transmitting module of the touch screen; when the writing pen is switched on and used for writing, the first coordinates and the second coordinates are converted into output coordinates, and otherwise, the second coordinates are determined as output coordinates. By the adoption of the touch method, the touch screen and the writing pen, seamless switching between handwriting and pen writing can be achieved; when the writing pen is switched on and used for writing, a signal transmitted by a signal sending module and a signal transmitted by the signal transmitting module will not disturb each other due to different transmitting times; due to the scheme that the first coordinates and the second coordinates are converted into the output coordinates, an output result is more accurate; to sum up, user experience is good, and the handwriting application environment and the pen writing application environment are compatible with each other.

The invention provides a smart city ground parking space image processing method based on artificial intelligence and CIM, and the method comprises the steps: carrying out the projection transformation and image splicing of a city road image collected by a camera, and projecting the city road image to a ground plane of the city road CIM to obtain a city road panorama; cutting the urban road panorama to obtain a plurality of sub-images; sending the sub-images into a semantic segmentation network, processing the sub-images to obtain a vehicle mask, a parking space area mask and a parking space line mask, calculating an IoU between the vehicle mask and the parking space area mask and an IoU between the vehicle mask and the parking space line mask, and judging ground parking space informationbased on the IoU and a set threshold value; transmitting the ground parking space information to the CIM, carrying out the visual processing of the data through employing the Web GIS technology, and displaying the ground parking space information while carrying out the real-time imaging. The method not only can detect whether the vehicle is parked in the parking space in real time, but also can detect whether one vehicle occupies two parking spaces, thereby improving the utilization rate of the parking spaces.

The invention discloses a generative abstract extraction method based on an encoder-decoder, and the method is based on an encoder-decoder framework containing an attention mechanism. The idea of a variational self-coding network and replication is introduced. An encoder part of the method is the same as a conventional encoder. A basic unit used is a GRU. The decoder part has three layers in total, the first layer and the second layer are GRU decoding layers and are used for determining the output of the part, the third layer is a variational self-encoding layer and is used for outputting a potential structure variable part, and an attention mechanism is introduced into the second layer of GRU. In an output part of the network, the hidden layer states of a copying part and a generating part are combined into a whole and then mapped into the output of the network. Meanwhile, a historical dependency is added to the variational self-coding layer, so that the variational self-coding layercan adapt to a time sequence network. According to the method, the state information of the hidden layer of the encoder is fully utilized, the output accuracy is improved, and better performance is achieved.

The invention discloses a data sampling method. The data sampling method includes the following steps that: S1, data are sampled through a sensor, so that sampling data can be obtained; S2, the sampling data are sent to a first processing unit, so that the first processing unit can carry out calculation to obtain a first result, at the same time, the sampling data is sent to a second processing unit, so that the second processing unit can carry out calculation to obtain a second result, the first processing unit and the second processing unit carry out the same calculation processing on the sampling data; and S3, whether the first result and the second result are identical is verified, if the first result and the second result are identical, the first result is correctly outputted to a communication bus. The invention also discloses a corresponding sampling system. The sampling system comprises a sensor, an A / D conversion unit and a calibration output module which are connected with one another sequentially. The data sampling method and the sampling system of the invention can verify results, automatically identify wrong data and ensure the safety and correctness of output data. The data sampling method and the sampling system of the invention have the advantages of multi-redundancy performance, high stability, flexibility in expansion, convenience and high integration.

The invention provides a breast medical image processing method and device, an AEC exposure parameter acquisition method, a computer device and a storage medium. The method comprises steps that firstly, an acquired breast medical image is processed to obtain a projection curve of a human body area; secondly, the peak of the projection curve is determined according to a first trough and a second trough in the projection curve; thirdly, the first trough position, the second trough position and the peak point position in the breast medical image are determined according to the first trough, the second trough and the peak in the projection curve; and lastly, the breast area in the breast medical image is determined according to the acquired first valley point position, the second valley pointposition and the peak point position. The method is advantaged in that accuracy of the trough of the human projection curve of the breast medical image is higher compared with the output result of thepeak of the acquired curve, the peak position in the breast medical image can be more accurately determined based on the trough of the projection curve, and thereby the result of the breast area is more accurately obtained according to the exact peak position of the method.

The invention discloses a video dense event description method based on a generative adversarial network, and belongs to the field of deep learning and image recognition. The method comprises the steps of constructing a video dense event description network; the network comprises a video feature extraction module for performing feature extraction on video frames to obtain video frame features; thetime sequence action feature extraction module is used for performing video frame feature learning by utilizing the characteristics of forward propagation and reverse propagation of the video to obtain each time sequence action feature; the natural language encoder fuses the video frame features and the time sequence action features by utilizing an attention mechanism to obtain natural statements; the discriminator is used for enhancing the accuracy of natural statements; and performing video dense event description by using the trained video dense event description model. According to the method, the characteristic of video two-way propagation is fully considered, video characteristics and time sequence action characteristics are fully utilized when the natural sentences are learned to be generated, and the grammar discriminator and the content discriminator are constructed, so that the accuracy of the natural sentences is effectively enhanced.

The invention discloses a roller kiln temperature soft measurement modeling method based on local twice-weighted kernel principal component regression. The method includes establishing a roller kiln temperature soft measurement model based on the local weighted kernel principal component regression through utilizing local sample data with high similarity, combining characteristics of the existinghigh dimensionality, nonlinearity and process time-varying of the roller kiln and introducing techniques of kernel tricks and instant learning separately; and performing twice-weighing on local modeling variables, and building the roller kiln temperature soft measurement model based on the local twice-weighted kernel principal component regression to achieve the precise prediction of the roller kiln temperature in consideration of the different influence degree of the input variables of local modeling sample data on the output variables. The obtained model can better track the state changes ofa process and provide good guidance effects for roller kiln temperature control, and therefore, product production quality and qualified rates can be enhanced.

The invention discloses a street lamp intelligent energy-saving method based on random forest regression prediction algorithm. An intelligent street lamp system obtains original data information, likevisitors flow rate, vehicle flow rate, sound, visible light, infrared rays, latitude and longitude, altitude, etc. Through a large amount of data analysis, the optimum illumination (measured by power) meeting the demand of illumination under the condition can be obtained. The random forest regression prediction algorithm is made, wherein the series of data obtained by the system can be used as independent variable X, the corresponding power values can be set as dependent variable Y, and the random forest training conducts interpretation on Y through X, obtaining the random forest regression prediction model. The model is applied in the intelligent street lamp system, and the street lamp power can be intelligently controlled based on the circumferential environment condition of each streetlamp. The invention is advantageous in that with the advantages of the random forest algorithm, the method not only greatly increases the power prediction accuracy of the system, but also has the energy-saving effect under the condition that illumination is met.

The invention discloses a high-accuracy phase frequency detector, which comprises a basic phase frequency detection unit and a rapid locking unit, wherein the basic phase frequency detection unit can be used for realizing detection of two input waveform frequencies and phases; and the rapid locking unit can be used for increasing the accuracy of the phase frequency detector. When two input waveform frequencies and phases of the phase frequency detector are consistent, the rapid locking unit does not work, and two paths of outputs of the phase frequency detector are at low levels; when two input waveform frequencies or phases of the phase frequency detector are inconsistent, the rising edge of one input waveform comes to turn corresponding output to a high level immediately, and the other output is kept at a low level; and when the rising edge of the other input waveform comes, the rapid locking unit starts working, and two outputs turn to low levels once again. The high-accuracy phase frequency detector is compact in structure, and has small phase frequency detection errors and low resource consumption.

The invention discloses an information processing method and system for improving the health level of children. The method is applied to a community health system. The method comprises the steps thata first activity image of a first child in a classroom is obtained through a first camera in the classroom at the first time; first behavior information of the first child is obtained according to thefirst activity image; the first behavior information is input into a first training model to obtain a first output result of the first training model; a second activity image of the first child in the family is obtained through a second camera in the family according to the second time; second behavior information of the first child is obtained according to the second activity image; the second behavior information is input into a second training model to obtain a second output result of the second training model; a third output result is obtained according to the first output result and thesecond output result, the third output result represents whether the first child is healthy or not. The technical effects that the health level of the child is monitored in real time and accurately judged, and the health level of the child is improved are achieved.

The invention discloses a heating furnace temperature control method and system based on improved particle swarm optimization, and belongs to the technical field of heating furnace automatic control. A heating furnace temperature model is established based on an MATLAB / Simulink simulation platform, a fuzzy control rule is established for a temperature change curve, the temperature error and the error change rate of the heating furnace are obtained through temperature monitoring equipment at a heated material inlet and a heated material outlet and serve as input parameters of the fuzzy PID control module, parameter optimization of output KP, Ki and Kd of the PID control module is directly fuzzy through a particle swarm, an optimal value is selected and assigned to the gain module, and the three initially calculated PID parameters are combined with the output parameters of the fuzzy control module passing through the gain module so that the temperature curve of the heating furnace can be adjusted. In order to solve the problem of low furnace temperature control precision of the heating furnace in the prior art, chaotic mapping is utilized to enable the particle swarm to overcome the premature phenomenon, local optimization is avoided, and the method has the advantages of high response speed, high stability and good anti-interference performance.

The present invention discloses a super-resolution identification method of medical CT images. The method comprises the steps of creating a training set in advance to obtain a high resolution dictionary, a low resolution dictionary and an optimization dictionary, and establishing a local characteristic base of a plurality of classification target images; obtaining a to-be-processed first CT image; carrying out the up-sampling on the first CT image, and obtaining a first low resolution image corresponding to the first CT image; extracting a low resolution characteristic of the first low resolution image, and calculating a sparse coefficient of the low resolution characteristic under a dual sparse model; according to the low resolution characteristic, the sparse coefficient and the high resolution dictionary, recovering a corresponding high resolution characteristic; according to the high resolution characteristic and the first low resolution image, obtaining a corresponding first high resolution image; carrying out the pathological local characteristic comparison on the local characteristic of the first high resolution image and the local characteristics of the classification target images, and outputting a comparison result. According to the super-resolution identification method of the medical CT images, the personnel not having the professional medical knowledge can know the pathological information in the CT images.

The invention provides a plant leaf moisture content detecting method and device, and relates to the technical field of detection. The plant leaf moisture content detecting method comprises the stepsof: obtaining hyperspectral data of a plant leaf to be detected; and processing the hyperspectral data of the plant leaf to be detected according to a preset diagnosis model to obtain a diagnosis result output by the diagnosis model, wherein the diagnosis model comprises a preset corresponding relationship between the hyperspectral data of the plant leaves and moisture contents, and the diagnosisresult is the moisture content of the plant leaf to be detected. According to the plant leaf moisture content detecting method and device provided by the invention, the measured hyperspectral data ofthe plant leaf are input to the established diagnosis model, so that the moisture content of the plant leaf can be directly output, the output result is accurate, and the timeliness is improved.

The invention discloses a prediction method for harmonic current of a permanent magnet motor. A motor magnetic field distribution model considering the ferromagnetic material saturation effect is established, the input of the motor magnetic field distribution model is the harmonic current of the permanent magnet motor, and the output is the terminal voltage of the permanent magnet motor; establishing a double Fourier series model of inverter output voltage harmonics based on space vector modulation of a modulation function, and inputting the modulation depth to obtain the output voltage of theinverter; establishing a matching degree between the terminal voltage of the permanent magnet motor and the output voltage of the inverter and taking the matching degree as an optimization target; and proposing a harmonic current matching method, i.e., carrying out global optimization on each order of harmonic current of the motor magnetic field distribution model through an optimization algorithm, and solving to obtain the amplitude and phase of each order of harmonic current so as to obtain the harmonic current. The time for processing the harmonic current is shorter than that of a finite element processing mode, and an output result is consistent and accurate with a result calculated by a finite element.