182results about How to "Meet precision requirements" patented technology

The embodiment of the invention discloses a neural network processor and a convolutional neural network processor. The neural network processor can comprise a first weight preprocessor and a first operation array, wherein the first weight preprocessor is used for receiving a vector V<x> including M elements; the normalized value domain space of the element V(x-i) of the vector V<x> is a real number which is greater than or equal to 0 and is smaller than or equal to 1; an M*P weight vector matrix Q<x> is used for performing weighting operation on the M elements of the vector V<x> to obtain M weighting operation result vectors; the first operation array is used for accumulating the elements with the same positions in the M weighting operation result vectors to obtain P accumulated values; a vector V<y> including P elements is obtained according to the P accumulated values; and the vector V<y> is output. The technical scheme provided by the embodiment has the advantage that the expansion of the application range of neural network operation is facilitated.

The invention discloses a plate-shape parameter online setting method for a dual-UCM temper mill on the basis of mechanism models. The method of the invention is characterized in that: a. the equipment parameters of the dual six-roll UCM type temper mill are collected; b. the key rolling process parameters of the strip to be quickly online set are collected; c. the plate-shape parameters are roughly calculated; d. the plate-shaped parameters are finely calculated; f. the shrinkage movement of the middle rolls of the frame 1model and frame 2model of the temper mill, and the roll bending force of a work roll and a middle roll are calculated correspondingly to the roll shifting coefficient and the roll bending coefficient; furthermore, the online setting of the parameters are realized on the temper mill. When the comprehensive control method provided by the invention is adopted, the plate-shape parameter setting technique on the basis of the mechanism models can be applied online, and can meet the precision requirement on engineering, as well as can effectively ensure the plate-shape quality of the finished product, the service life of the roller, and the service life of the bent roll cylinder, thus bringing obviously economical benefits for the enterprises. The method of the invention is simple and clear and suitable for online application.

The invention discloses an indoor laser positioning system based on a frequency labeling method. A frequency label algorithm module is used for completing demodulation conducted on modulated lasers emitted by a positioning light source and determining the position of a positioning terminal according to the correspondence between the laser frequency and the position coordinates of the area to be positioned. The positioning light source modulates and emits a laser bundle array so that laser light spots with different frequency labels can be formed in positions to be positioned to be detected and processed through a positioning terminal. The positioning terminal receives the lasers which are emitted by the positioning light source and have different frequencies, captures frequency information of the lasers after filtering, photovoltaic conversion and A/D conversion are conducted, and achieves positioning through the frequency label algorithm. Due to the fact that the lasers serve as carrier waves of signals, reliable and stable positioning service can be easily provided; due to the adoption of the frequency labeling method, the position is determined according to the frequency of laser modulation light received by the positioning terminal; the positioning accuracy of the light spots generated by the positioning light source reaches the centimeter level; hardware implementation and conversion are easily achieved so that the product can be popularized and used.

The invention provides a highway mileage stake number determining method based on an electronic map and a GPS coordinate. The method comprises the steps of S1, acquiring a certain highway segment through field acquisition equipment, performing matching with the electronic map according to a proportional relationship, acquiring a certain highway segment of another landform in a same manner, performing matching with the electronic map according to the proportional relationship, performing collection matching on highway segments in different types, and forming a highway segment database; S2, inputting to-be-matched vehicle information data into the database, and setting a GPS coordinate in a manner of calculating a target point, thereby obtaining a mileage stake number; S3, searching a corresponding highway segment through setting the GPS coordinate, and acquiring a matched highway segment name according to the GPS coordinate; and S4, after acquiring the highway segment name, determining a vehicle driving direction in a method of determining the direction according to the GPS coordinate, thereby finally determining the highway segment in which the vehicle exists and the driving direction.

The invention provides a sliding block executing mechanism used for attitude control over a solar sail spacecraft. The sliding block executing mechanism comprises a central box, a pre-tightening regulating unit, a control unit, a driving wheel unit, a driven wheel unit and an adjustable auxiliary driven wheel unit; a telescopic arm penetrates through a hole in the center of a front panel and a hole in the center of a rear panel of the central box; the pre-tightening regulating unit comprises a pre-tightening screw and a pre-tightening nut, and the pre-tightening screw penetrates through an upper panel of the central box; the control motor is driven by obtaining electric energy from a solar battery; the driving wheel unit comprises two protruding idler wheels which are connected with a motor shaft of the control motor through connection shafts, and the outer walls of the two protruding idler wheels make tight contact with the periphery of a stretching arm; two protruding idler wheels of the driven wheel unit penetrates through a rotary shaft and are arranged between two polytetrafluoroethylene sliding rings, the outer walls of the two protruding idler wheels make tight contact with the periphery of the stretching arm, and the rotary shaft is fixed to a support of a lower panel of the central box; the adjustable auxiliary driven wheel unit comprises four protruding idler wheels and four triangular supports, and the triangular supports are regulated so that the outer walls of the protruding idler wheels can make tight contact with the periphery of the stretching arm.

The invention discloses an angle-controllable power differential omnidirectional wheel, and a control method and an omnidirectional moving platform thereof, and relates to the technical field of machinery manufacturing and assembly and intelligent control. The angle-controllable power differential omnidirectional wheel comprises a wheel frame main body, a rotary support bearing, two wheels, two elastic shaft couplers, two speed reducing motors, a gyroscope and a controller. According to the control method of the omnidirectional wheel, the rotating speed of two wheels is subjected to closed loop control so that the practical azimuthal angle of the omnidirectional wheel tracks the expected value; the flexible steering is realized; the cost is low; the omnidirectional moving requirements of the heavy equipment or cargo transportation can be met. The omnidirectional platform comprises the omnidirectional wheel and a platform bottom plate connected with the omnidirectional wheel through the rotary support bearing; the non-steering flexible movement of the platform bottom plate is realized.

The invention relates to photoetching technology in the semiconductor industry, in particular to an optical proximity correction method. The method comprises the following steps of: dividing a territory into at least two areas according to different critical dimension control accuracies; and respectively performing optical proximity correction on the at least two areas according to the requirement on critical dimension control accuracy, and adopting the same correction accuracy for areas with the same critical dimension control accuracy. By the optical proximity correction method, areas with different critical dimension control accuracies are corrected by using different correction accuracies, the correction accuracy is distributed according to needs, the accuracy requirement of the optical proximity correction is met, the operating time of an optical proximity correction system is effectively shortened, and the production efficiency is improved.

The present invention discloses a pentaprism scanning detection method for an aspheric reflector. The method comprises: 1) arranging a device required for detection; 2) performing scanning detection of a single bus; 3) performing scanning detection of the next bus of the rotation angle a of the aspheric reflector after the scanning of the single bus is completed; and 4) performing detection data processing and fitting. All the angle differences are transited to the slope values of the corresponding positions of the reflective surface of the aspheric reflector, the slope values are fitted with the Zernike gradient polynomial, and the low-order surface-shaped error parameters of the aspheric reflector are obtained. The high-precision low-order surface-shaped error parameters of the aspheric reflector can be obtained.

The disclosed modeling method for equivalent capacitor model comprises: selecting equivalent parameter and desired condition for impedance calculation and decision respectively, which can avoid defect by simple calculation method. This invention can meet different simulation precision request, fit to variable capacitors, and benefit to simulation result analysis.

The invention belongs to the technical field of neural networks, and relates to a convolution kernel cutting method based on an entropy importance criterion model. In order to enable a convolutional neural network model with large parameter quantity, large calculated quantity and excellent performance to meet a real-time requirement in practical application. The invention provides a convolutionalkernel cutting method based on an entropy importance criterion model. The convolution kernel with less information amount is cut by taking a mode of solving image entropy of an activation channel of each convolution layer as a criterion for evaluating the importance of the corresponding convolution kernel. Therefore,, a small model which is excellent in performance and small in parameter amount and calculation amount is obtained, and the small model has the performance advantage and can meet the real-time performance and precision requirements of practical scene application.

The invention provides a bifidobacterium LJM-001 strain. The collection number of the bifidobacterium LJM-001 strain is CGMCC No.4090. The LJM-001 strain is hypersensitive to dimethoate and is a working strain for detecting dimethoate residues in food. The invention also provides a microbiological method for screening dimethoate residues in food. According to the size of antibacterial rings on a screening culture medium, whether the residual quantity of the dimethoate in food samples exceeds the standard or not can be judged. The method for screening the dimethoate residues in food has the characteristics of simpleness and convenience for operation, low cost, high sensitivity and suitability for sample screening, and the screening requirement of a large number of samples in the pesticide residue detection department in China is met.

An anti-fatigue manufacturing method of a car transmission gear comprises the steps that forging blanking, heat pretreatment (isothermal normal fire), gear blank processing, gear hobbing, gear strengthening and toughening treatment (the gear part is subject to two times of induction heating and circulatory quenching), modification of criteria to the drawing demand, gear hobbing, gear shaving and gas nitrizing process are carried out in sequence. By adopting the method, 6-grade or more than 6-grade precise car gears can be efficiently produced in batch at low price, and the produced car gears has the bearing/carrying capacity almost as well as that of alloy cemented steel. The car transmission gear manufactured by adopting the method can replace the carburized and quenched gear in the prior art.

The invention provides a wavelet transform compression and/or decompression-based neural network processing method. The neural network processing method comprises the following steps of: compressing off-chip data and sending the off-chip data onto a chip; decompressing the off-chip data which is compressed and sent onto the chip; receiving the decompressed data and executing a neural network operation; compressing the data obtained via the neural network operation and sending the data off to the chip; and decompressing the data which is compressed and sent off to the chip and storing the dataas off-chip data, wherein the compression and/or compression carried out on the data is based on wavelet transform. According to the wavelet transform compression and/or decompression-based neural network processing method, data compression is carried out during data loading and storage, so that the IO quantity is decreased and the time and energy overheads are reduced.

The invention relates to an oil well layered yield splitting method based on dynamic and static coupling, which can solve the technical problems of poor adaptability and large calculated amount of thetraditional splitting method. The method comprises the following steps: S100, determining a static splitting coefficient; resetting single well perforation data according to a layering result, and generating a static splitting coefficient according to a single well logging interpretation result and the perforation thickness; S200, determining a single well profile splitting coefficient; generating a profile splitting coefficient after the liquid production profile data is reset layer by layer; and S300, determining a dynamic splitting coefficient according to the static splitting coefficientand the profile splitting coefficient. The oil well layered yield splitting method based on dynamic and static coupling has the following beneficial effects that (1) coupling splitting of a single well production and suction profile is achieved, the method is reasonable and effective, and the precision requirement of residual oil research can be met; (2) the method is simple, easy to operate andhigh in generalization performance; and (3), compared with a numerical simulation research method, the working time can be effectively saved, and the working efficiency is greatly improved.

The invention provides a calibration device and calibration method for a compound-eye type spherical vision system. The calibration device comprises a data acquisition system and an angle adjustment device. The angle adjustment device consists of a sensor assembly and a mechanical adjustment mechanism. The sensor assembly includes an angle sensor A, a linear displacement sensor B and an angle sensor C, wherein the sensors are connected with an input terminal of a multi-channel data acquisition card of the data acquisition system by leads; the angle sensor A collects a rotation angle theta2 of a rocker arm around an Y axis; the linear displacement sensor B collects a distance d between a light source and a spherical vision system; and the angle sensor C collects a rotation angle theta 1 of the spherical vision system around a Z axis. Compared with the traditional calibration method, the provided calibration method has the following advantages: one-time integrated mounting and calibration of the large-range spherical surface can be realized. While the precision requirement is met, the cost is reduced, the efficiency is improved, and the operation becomes simple and convenient.

The invention provides a testing device and method for measuring soil body fracture toughness by the aid of soil body self-weight. The testing device comprises a testing platform and a soil beam mold, wherein the testing platform comprises a base platform, a fixed support and a sliding support, the base platform adopts a flat-plate structure, the fixed support is a long strip with the rectangular cross section and is fixed on one side of the upper plane of the base platform, the sliding support is placed on the upper plane of the base platform and can slide on the plane, the cross section of the sliding support is rectangular, and rolling bearings or rolling balls are arranged in upper and lower planes of therectangle; an inner cavity of the soil beam mold is rectangular and adopts a detachable rigid structure. With the adoption of the testing method for measuring the soil body fracture toughness by the aid of the soil body self-weight, soil body fracture conditions can be reflected relatively truly and objectively, the soil body fracture toughness parameter can be determined by recording the initial fracture extension length, operations are simple, the device is simple, themeasurement accuracy is higher, and the accuracy requirement for the engineering soil body fracture parameter can be met.

The invention provides a bifidobacterium longum LJM002 strain, the preservation No. of which is CGMCC No. 4900. The LJM002 strain is ultra-sensitive to acephate and is a work strain for detecting the residue of the acephate in foods. The invention also provides a microorganism method for screening the residue of the acephate in the foods, which comprises the step of judging whether the residue limit of acephate in food samples to be detected exceeds the standard according to the size of a bacteriostatic ring on a screening medium. The method for screening the residue of the acephate in the foods has the characteristics that the method is simple to operate, has low cost and high sensitivity and is suitable for sample screening. The method conforms to the screening requirements of China's pesticide residue detection department for a large quantity of samples.

The invention relates to a remote-sensing image target recognition method based on feature tensors and a support tensor machine, and aims to solve the problem that existing vector structure features cannot maintain spatial structure information which targets in remote-sensing images have, and system models trained and obtained thereby are very difficult to directly satisfy precision requirements.A process of the method includes: first, selecting a training image set on remote-sensing images, and selecting slices of to-be-recognized target and background samples on training images according tothe size of to-be-recognized targets on the training images; second, extracting SIFT or SURF features of the slices, and constructing the feature tensors; third, deriving a classification decision hyperplane; fourth, utilizing the support tensor machine to carry out target recognition on a to-be-recognized remote-sensing image to obtain slices containing objects labeled as the targets; and fifth,extracting the slices containing the objects labeled as the targets, and carrying out slice merging to obtain a final recognition result. The method is used in the field of remote-sensing image target recognition.