134results about How to "Increase the number of samples" patented technology

The invention discloses a method and a system for detecting rice panicle blast based on a deep convolutional neural network. The method comprises the following steps: acquiring a hyperspectral image of outdoor rice panicles and performing panicle blast disease calibration; performing data preprocessing and data enhancement on the hyperspectral image of the rice panicles; establishing a deep convolutional neural network model and optimizing model parameters with a stochastic gradient descent algorithm; detecting the hyperspectral image of the rice panicles to be tested with the trained deep convolutional neural network model, and judging whether the rice panicles are infected by the panicle blast disease or not. The system comprises a hyperspectral camera, a computer, a tripod and a reflection board, wherein the rice panicles are hung on the reflection board; the hyperspectral camera is fixed on the tripod, is connected with the computer, and has a lens aiming at the rice panicles on the reflection board. The method and the system provided by the invention can provide technical support for predicting the outdoor rice panicle blast disease and can also play an instructive role in rational and precise application and management of water and fertilizer, pesticide or other agricultural resources in a production process.

The invention belongs to the technical field of MIMO radar, and particularly provides an MIMO radar multi-target tracking resource management method. Firstly traversal of controllable parameter combinations is performed, and whether the controllable parameter combinations can meet the detection probability and beam direction constraints is judged so as to form an MIMO radar control vector feasible set; then deviation cost and the energy resource consumption cost between the prediction tracking error covariance and the expected covariance of the targets are calculated under each control vector in the feasible set; and the two cost is integrated, and the optimal radar control vectors are acquired according to the criterion of the minimum integrated cost. The advantages that MIMO radar utilizes single beam to irradiate multiple targets can be fully exerted under the multi-target environment so that the sampling frequency of each target can be increased, the time resource utilization rate of radar can be effectively enhanced and the deviation between the actual tracking error covariance and the expected covariance of the targets can be reduced, and the tracking accuracy of the method can be controlled by the method through adjusting the expected tracking error covariance.

The invention relates to a malicious domain name detection method based on SMOTE and BI-LSTM. The method comprises the following steps of carrying out data preprocessing on domain names in a trainingset in order to obtain a domain name character sequence training set D; carrying out equalized data synthesis on the domain name character sequence training set D by utilizing an improved SMOTE algorithm in order to obtain an enhanced and equalized training set D'; constructing and initializing a neural network model comprising a BI-LSTM layer and a full-connection layer; converting the domain names in the training set D' into representation vectors with the fixed length; inputting the representation vectors of the domain names obtained in the step D into the BI-LSTM layer of the neural network model in order to obtain context vectors of the domain names; inputting he context vectors of the domain names into the full-connection layer of the neural network model in order to obtain a trainedneural network model; and converting the domain names to be judged into a character sequence, inputting the trained neural network model, and outputting a judgment result.

The invention discloses an age identification method based on an integrated convolution neural network. The method includes following steps: S1, obtaining and expanding training sub-sets in an age identification training database, obtaining the expanded training sub-sets, and selecting M convolution neural network classifiers obtained by training of the expanded training sub-sets as base classifiers; S2, obtaining a to-be-tested face image; and S3, inputting the to-be-tested face image into M base classifiers obtained in step S1 during a test, fusing age categories output by M base classifiers, and obtaining a final age category. According to the method, the accuracy of age identification is high, the dependency on people by age feature extraction of the face image is reduced, ages of a variety of people can be estimated, and the application is wide.

Disclosed herein is an apparatus for sensing characteristics of an object. In a preferred embodiment, the apparatus comprises an array, wherein the array comprises a plurality of nanoscale hybrid semiconductor/metal devices which are in proximity to an object, each hybrid semiconductor/metal device being configured to produce a voltage in response to a perturbation, wherein the produced voltage is indicative of a characteristic of the object. Any of a variety of nanoscale EXX sensors can be selected as the hybrid semiconductor/metal devices in the array. With such an array, ultra high resolution images of nanoscopic resolution can be generated of objects such as living cells, wherein the images are indicative of a variety of cell biologic processes.

The invention discloses a multi-source data fusion-based system and method for predicting pipeline corrosion defect size. The system comprises a sensor group, a lower computer and an upper computer, wherein the sensor group is arranged at the head end or the tail end of a pipeline and is in contact with a transmission medium; the sensor group is connected with the lower computer connected with the upper computer. The method comprises the steps of firstly building a gray GM (1,1) prediction model and a BP neural network model by using historical data; then using the two models to obtain a prediction value; next building an inducted orderly weighted harmonic average (IOWHA) operator combination model according to the historical data and the prediction value of the two models; for new data, using the three models in sequence to predict the corrosion defect size in three directions of axial direction, circumference and depth respectively, so that a corrosion defect size prediction value of the pipeline is finally acquired. By adopting the system and the method, the prediction accuracy is higher, and the input of data is reduced.

Belonging to the field of metallurgical sample detection, the invention relates to a preparation method of a standard spectrum sample. The method includes the processes of: component selection, preparation of a standard sample through a continuous casting middle ladle top pouring method, stability investigation and conclusion. The method is characterized in that the process of preparing the standard sample by the continuous casting middle ladle top pouring method comprises the steps of: model baking, sample model establishment, sample model pouring by the top pouring method, demolding, segmentation and surface treatment, physical property detection, surface treatment, chemical component uniformity test, and fixed value analysis. The standard sample prepared in the invention not only satisfies the standard sample requirements in terms of physical property, texture, component range and uniformity, but also has reduced preparation cost and increased sampling number. At least 200 pieces can be prepared every time, thus being over 20 times as many as a model casting sampling number. And the preparation number can satisfy the current production requirements. Thus, the standard sample preparation method of the invention boasts very high practical value.

A microwave irradiation apparatus proposed for uniformly treating a large number of samples, includes: an irradiation chamber formed in a rectangular resonant cavity of TM (Transverse Magnetic) 110 mode, of which length of X-axis side is “a” (a>0), length of Y-axis side is “b” (b>0) and length of Z-axis side is “c” (c>0); slits respectively formed on Y-Z plane walls of the irradiation chamber; a transfer sheet entering into the irradiation chamber through the slits and moving along a X-Z plane in the irradiation chamber; and a sample holder disposed on the transfer sheet.

The present invention provides an oxide-base scintillator single crystal having an extremely large energy of light emission, adoptable to X-ray CT and radioactive ray transmission inspection apparatus, and more specifically to provide a Pr-containing, garnet-type oxide single crystal, a Pr-containing perovskite-type oxide single crystal, and a Pr-containing silicate oxide single crystal allowing detection therefrom light emission supposedly ascribable to 5d-4f transition of Pr.

The invention belongs to the communication field, to be specific, provides an MIMO radar multi-target adaptive tracking method based on a variable analysis moment. The above mentioned method is used to realize MIMO radar adaptive target tracking in multi-target environment. The method comprises steps that a prediction error covariance threshold method is used to determine the next analyzing moments of various targets, and the minimum value is used as the next analyzing moment of the MIMO radar; a controllable parameter combination constituted by working modes, sub-array quantity, and waveform energy is traversed, and whether the controllable parameter combination is capable of satisfying constraints of a detection probability and wave beam directions is determined, and therefore an MIMO radar control vector feasible set is formed; and finally, in every control vector of the feasible set, an optimal radar control vector is acquired according to a lowest resource consumption principle. In the multi-target environment, the advantage of the MIMO radar using the single wave beam to irradiate a plurality of targets is fully used, and the sampling times of the various targets are increased, and the sampling moments of the targets are determined in an adaptive way according to the motion characteristics of the targets, and therefore the time resource utilization rate of the radar is effectively improved.

The invention belongs to the technical field of plant seedling propagation and discloses a micropropagation method of tilia miqueliana. The method comprises steps as follows: A, culture of seedling of the tilia miqueliana; B, adventitious bud induction; C, high-voltage electrostatic field processing; D, subculture and rooting; E, acclimation and seedling hardening. A large quantity of tilia miqueliana explants easy to disinfect can be obtained, the success rate of primary culture is higher than 98%, the proliferation rate reaches 113 times, the survival rate reaches 95%, the problem of shortage of the tilia miqueliana in the market is effectively solved, and the application prospect is wide.

The invention discloses a preparation method of a high-carbon stainless steel standard sample, belonging to the field of detection of metallurgical samples. The preparation method disclosed by the invention comprises the steps of designing a casting mould, baking the casting mould and a steel ladle, selecting components, controlling the process of a casting process, demoulding, detecting physical performances, processing the sample, detecting the homogeneity of chemical components, carrying out deterministic analysis and determining effective height. The physical performances, the structural state, the component range and the homogeneity of the standard sample prepared by the invention can be satisfied; furthermore, the preparation cost is reduced; the sampling number is increased; at least 100 standard samples can be prepared every time; the number is above 20 times of that of die casting sampling; the preparation number can satisfy current production requirements; and thus, the standard sample preparation process has very high practical value and great popularization value in other matter test rooms.

The invention provides a method, a device, a program and a readable storage medium for identifying characters. The method comprises the steps of: identifying the input character samples to generate the identification result; generating degree of confidence of the identification result; in accordance with the degree of confidence, judging whether the input character samples belong to deformed samples, if not, regarding the identification result as the final identification result; if so, identifying the deformed samples and generate the final identification result. The technical proposal of the invention can effectively process the deformed samples by directly regarding the samples which are incorrectly identified at the first level as the basic training samples at the second level, thus improving the identification accuracy of the system.

The invention discloses a smoke and fire identification model establishing method based on deep learning and a smoke and fire identification method. According to the smoke and fire identification model establishing method, a can network is used for synthesizing a smoke and fire picture and a normal picture of a to-be-detected area, the number of samples of a data set in an actual application scenecan be increased, and then the to-be-established smoke and fire identification model is obtained through yolov3 labeling, training and verification. According to the method, the problems of abstractsmoke and fire characteristics, diverse shapes and unobvious and unspecific characteristics in a traditional smoke and fire identification method can be solved, and meanwhile, the situation that smokeand fire image data is small or does not exist in an actual application scene is solved. The smoke and fire identification method is based on the established smoke and fire identification model, andthe misjudgment problem generated in large-scale scene long-distance image identification is solved.

The invention relates to a UAV (unmanned aerial vehicle) water sampling device. The device comprises a UAV, wherein a control system is arranged in the UAV, and a sampling tank is connected to the lower part of the UAV and provided with a conversion acquisition device, and the conversion acquisition device comprises rotating devices arranged bilaterally symmetrically in the sampling tank, water collecting bottles vertically arranged on the rotating devices in a sleeving manner, hoppers vertically placed on the water collecting bottles, balls placed at the center of the hoppers and electromagnets I tightly attached to the upper right ends of the rotating devices; a reel I is arranged right above the center of each rotating device on the UAV, an electromagnet II is arranged below the reel I,and a connecting wire winds around the reel I and is connected to a sampling bucket; the sampling bucket comprises a lifting ring arranged above the sampling bucket, a sampling bucket cover right below the lifting ring, a water outlet pipe arranged at the bottom of the right end of the sampling bucket cover, a valve arranged on the water outlet pipe, an iron block vertically arranged at the upperend of the valve and a magnet vertically arranged at the lower end of the valve; a reel II is arranged in the middle of the device and wound with a wire of an ultrasonic probe.

The invention discloses a method of analyzing structural reliability of an aeroengine turbine disc; a random variable mathematical model is established according to non-uniformity of turbine disc material, and a stress analysis model is established according to stress condition; PDS (probabilistic design system) in ANSYS environment is used to perform Monte Carlo sampling on random variables of a turbine disc three-dimensional model to obtain maximum stress MAXSTRESS distribution of the turbine disc under different conditions; Monte Carlo simulation tests of different levels are performed on the random variables on MATLAB (matrix laboratory) platform, and reliability of the turbine disc is calculated according to structural functions to obtain the structural reliability of the turbine disc under different conditions; sensitivity of the random variables influencing the structural reliability of the turbine disc is calculated according to sensitivity equation. The method has the advantages of high convergence speed, low cost and high precision, and theoretical basis is provided for the optimal design and reliability management for a turbine disc.

The invention is suitable for the technical field of data processing, and provides a speech recognition method based on artificial intelligence (AI), a device, a terminal and a storage medium. The method comprises the steps of obtaining a to-be-recognized target speech signal; determining a target language type of the target speech signal; and outputting text information of the target speech signal through a real-time speech recognition model corresponding to the target language type, wherein the real-time speech recognition model is obtained by training a training set containing original speech signals and extended speech signals, and the extended speech signal is obtained by converting an existing text based on a basic language type. According to the technical scheme provided by the invention, the number of samples required for training the real-time speech recognition model of the non-basic language can be increased, so that the accuracy and applicability of speech recognition are improved.

A delay line system for high frequency signal transmission comprises a first delay line and a second delay line that are each tapped. The first delay line comprises a first terminal and a second terminal, and the second delay line comprises a first terminal and a second terminal. The first delay line and the second delay line are configured in a manner whereby an analog input signal applied to the first terminal of the first delay line propagates through the first delay line in a first direction, a clock signal applied to the first terminal of the second delay line propagates through the second delay line in a second direction, and the first direction is opposite to the second direction. Further, an oscilloscope for measuring high frequency signals comprises an ADC, which comprises a high frequency sampler, which comprises such a delay line system.