427 results about "Gradient method" patented technology

The invention discloses a gas pipeline leakage detecting and positioning device and a method based on transient model method, which belong to the technical field of failure diagnosis and flow safety security. The detecting device comprises a sensor group, a signal pre-processing device, an I/V plate, a collecting card, a voltage-stabilized power supply, and an operating condition machine; the detecting method includes of establishing a math model for the liquid flowing in the pipe, solving the flow parameters change in the pipe through a computer under a certain boundary condition, and comparing the model calculating values with the true measuring values at the pipe end; when the difference of the true measuring values and calculating values exceeds a certain range, then a leakage can be identified, and the leakage is positioned through a pressure gradient method. The invention provides a leakage detecting method based on the transient model method, and has the advantages of less investment, high sensitivity and good adaptability, and can be used for detecting and positioning gas leakage and leakage point for on-site gas pipeline.

The invention relates to a method and system for establishing a multilayer oil pool inter-well connectivity model. The multilayer oil pool inter-well connectivity model capable of simulating oil and water dynamic states is established oriented to the limitation that by means of a current connectivity model, only the dynamic change of production liquid can be predicted, the dynamic states of the oil phase and the water phase cannot be calculated respectively, and connectivity analysis cannot be carried out layer by layer. According to the model, an oil pool system is layered and scattered into a series of inter-well communication units, matter balance equation calculation is carried out with the inter-well communication units as simulating objects, pressure solving and inter-well flow calculation are achieved, an inter-well saturation degree tracking calculating method is established with the combination of the lead edge boosting theory, and the output dynamic states of oil and water of all layers at well points are obtained; a model parameter inversion method is established in a fitting mode through dynamic states based on a random disturbance approximation method, a projection gradient method and the like. Layered inter-well flow distribution coefficients, single well longitudinal production liquid and production oil splitting coefficients and other information can be obtained in real time, the oil pool plane, the longitudinal oil and water flow relation and the production measure change can be more accurately reflected to guide the practical production of an oil field.

The invention relates to a license plate positioning method integrating multiple methods and adopting the steps as follows: pretreatment is carried out to collected license plate images; a self-adapting matching template is obtained by a self-adapting template matching method which is based on data stream correcting and coarse positioning of the license plate images is realized by a template matching algorithm; edge detection is conducted to the license plate images after the coarse positioning by utilizing morphology gradient method so as to highlight the character areas of the license plates; aiming at distorted license plate images, a method for detecting the main direction of the characters in the license plate images is adopted for correcting; a projection method is adopted for precise positioning so as to achieve precise positioning of the character areas of the license plates, thus realizing the precise positioning of the license plate images. The license plate image positioning method designed by the invention has low requirements on the adaptability of environment and high arithmetic speed, can fulfill the requirements of precision and real-time performance for license plate identification and provides high-quality license plate images for subsequent partitioning and identification of license plate characters.

Concurrent Gradients Analysis (CGA), and two multi-objective optimization methods based on CGA are provided: Concurrent Gradients Method (CGM), and Pareto Navigator Method (PNM). Dimensionally Independent Response Surface Method (DIRSM) for improving computational efficiency of optimization algorithms is also disclosed. CGM and PNM are based on CGA's ability to analyze gradients and determine the Area of Simultaneous Criteria Improvement (ASCI). CGM starts from a given initial point, and approaches the Pareto frontier sequentially stepping into the ASCI area until a Pareto optimal point is obtained. PNM starts from a Pareto-optimal point, and steps along the Pareto surface in the direction that allows improving a subset of objective functions with higher priority. DIRSM creates local approximations based on automatically recognizing the most significant design variables. DIRSM works for optimization tasks with virtually any (small or large) number of design variables, and requires just 2-3 model evaluations per Pareto optimal point for the CGM and PNM algorithms.

The invention discloses a robust controller of a permanent magnet synchronous motor based on a fuzzy-neural network generalized inverse and a construction method thereof. The construction method of the invention comprises the following steps of: combining an internal model controller and a fuzzy-neural network generalized inverse to form a compound controlled object; serially connecting two linear transfer functions and one integrator with the fuzzy-neural network with determined parameters and weight coefficients to form the fuzzy-neural network generalized inverse, serially connecting the fuzzy-neural network generalized inverse and the compound controlled object to form a generalized pseudo-linear system, linearizing a PMSM (permanent magnet synchronous motor), and decoupling and equalizing the linearized PMSM into a second-order speed pseudo-linear subsystem and a first-order current pseudo-linear subsystem; and respectively introducing an internal-model control method in the two pseudo-linear subsystems to construct the internal model controller. The robust controller of the invention has the advantages of overcoming the dependence and local convergence of the optimal gradient method on initial values and solving the problems of randomness and probability caused by using the simple genetic algorithm, obtaining the high performance control, anti-disturbance performance and adaptability of the motor and simplifying the control difficulty, along with simple structure and high system robustness.

The invention relates to a preparation method of a core-shell-structure nickel cobalt lithium aluminate and belongs to the technical field of lithium ion battery positive material application. The core-shell-structure material is in a two-layer structure, wherein the inner layer core part is LiNiaCo[1-a-b]AlbO2, a is more than 0.7, b is less than or equal to 0.05 and more than or equal to 0, and a+b is less than 1; the molecular formula of the shell part is LiNicCo[1-c-d]AldO2, c is more 0.5, d is more than 0 and less than 0.5, and c+d is less than 1. In the method, two kinds of aqueous alkali are used as precipitators, so that the materials of the core part and the shell part are consistent in crystal from and compact in structure in the crystallization process, and a solubility gradient method is adopted in the alternating process of the two kinds of aqueous alkali so that the core-shell-structure material is consistent in crystal from and high in crystallization degree, and an obvious core-shell interface does not exist. Compared with a common homogeneous phase material, the core-shell-structure material disclosed by the invention has the advantages that the high capacity is kept, the cycling stability and the heat stability are improved, the air expanding rate is obviously reduced, and the cost performance is higher, so that the core-shell-structure material is more suitable for being used in power batteries.

An image processing apparatus, a feature point tracking method and a feature point tracking program, which enable efficient feature point tracking by taking the easiness of convergence of a displacement amount according to the image pattern into account in a hierarchical gradient method, are provided. A displacement calculating unit reads a hierarchical tier image with the smallest image size from each of a reference pyramid py1 and a tracking pyramid py2 in an image memory, and performs repetitive detection of a tracking point using a gradient method. Then, the displacement calculating unit repeatedly calculates the displacement amount (dx_i, dy_i) between the feature points obtained as a result of the repetitive detection, and outputs the value of the repetition count inum_last of tracking point detection until convergence of the displacement amount (dx_i, dy_i) and the value of the displacement amount (dx_last, dy_last) for the point of time of convergence to a repetition criterion setting unit. The repetition criterion setting unit changes the criterion for determination of displacement amount convergence based on the repetition count inum_last and displacement amount (dx_last, dy_last) input from the displacement calculating unit. The displacement calculating unit reads upper hierarchical tier images and detects a tracking point according to the convergence determination criterion set by the repetition criterion setting unit. Then, the above-described repetitive tracking point detection and convergence determination criterion setting change are repeated. Consequently, the repetition count of repetitive tracking point detection is suppressed.

A visualization system and method for visualization of network data, which is data that represents elements and links between elements. The network data is converted into a data structure, which represents a grid arrangement of the elements, where each element is placed on an individual grid position of a lattice. The data structure is suitable for use by a graphics display. The visualization tool comprises a processing unit that generates an initial data structure which represents an initial grid arrangement of the elements. It then assigns a global value to this initial grid arrangement and employs a gradient method for converting the initial grid arrangement into another grid arrangement which has a decreased or an increased global value.

The invention discloses a power system harmonious wave compressed signal reconstruction and detection method based on compressed sensing. The power system harmonious wave compressed signal reconstruction and detection method based on compressed sensing comprises the steps that harmonious wave original signals of a power system and a measurement matrix with a binary sparse random measurement matrix serving as a power system harmonious wave data compressed sample are sent to a frequency mixer for compressed sampling in a simulation domain, then A/D conversion is carried out on the simulation signals, and a compressed sampling value is obtained; it is determined that a compressed sensing sparse base is a discrete Fourier transform base; initialization of fundamental wave filtering is carried out; fundamental wave filtering is carried out; the frequency, amplitude and phase of a fundamental component are detected; the fundamental wave constituent in the compressed sampling value is filtered away; parameter initialization is carried out on a spectrum projection gradient method; a sparse vector estimation value of a harmonious wave component is reconstructed through the spectrum projection gradient method; the frequency, amplitude and phase of the harmonious wave component are detected; reconstruction of the harmonious wave original signals of the power system is finished. The power system harmonious wave compressed signal reconstruction and detection method based on compressed sensing overcomes the defect that all existing recovery algorithms do not take the influence of the fundamental wave component in the harmonious wave signals on signal reconstruction into account, so that the recovery effect is not ideal.

The invention discloses a multimodal feature extraction and matching method based on ASIFT (affine scale invariant feature transform), and the method is mainly used for realizing the point feature extraction and matching of the multimodal image which cannot be solved in the prior art. The method can be realized through the following steps: carrying out sampling on the ASIFT affine transformational model tilting value parameters and longitude parameters, thus obtaining two groups of views of two input images; adopting a difference of Gauss (DoG) feature detection method to detect the position and size of the feature point on the two groups of views; using an average squared-gradient method to set the principle directions of the features and setting the feature vector amplitude by a counting method; calculating the symmetric ASIFT descriptor of the features; and adopting a nearest neighborhood method to carry out coarse matching on the symmetric ASIFT descriptor, and using an optimized random sampling method to remove mis-matching features. In the invention, features can be extracted and matched in the images sensed by various sensors, and the method provided by the invention has the characteristic of invariability after complete affine, and can be applied to the fields of object recognition and tracking, image registration and the like.

For each pixel in an operation block, an effective pixel determining unit sets a predetermined pixel difference (threshold value) on a time-base pixel difference. When the time-base pixel difference is less than the set predetermined pixel difference, it is determined that the pixel contributes to (or are effective in) motion vector detection, and uses the pixel in the gradient method in the case of performing motion vector detection. When the time-base pixel difference is not less than the set predetermined pixel difference, it is determined that the pixel is regarded as having a possibility to generate irregularity in movement, and the pixel is excluded from the gradient operation in the case of performing motion vector detection. The present invention is applicable to a signal processing apparatus for performing frame frequency conversion from a 24 P signal to a 60 P signal.

The invention belongs to the field of development of oil and gas fields and particularly relates to a well pattern optimization method in a large-scale oil field development process. The well pattern optimization method comprises the following steps of: firstly, acquiring oil reservoir data, establishing a three-dimensional digital oil reservoir geological model and simulating an oil reservoir value; secondly, constructing a self-adaptive rule well pattern unit and generating a well pattern; thirdly, calculating economic benefits and evaluating an initial well pattern; fourthly, setting parameters to make preparations for well pattern optimization; fifthly, optimizing the unit and constructing parameters to generate a new well pattern by using a gradient method; and sixthly, calculating and evaluating economic benefits of the new well pattern and carrying out iterative operation until an optimal well pattern is generated. According to the well pattern optimization method disclosed by the invention, the well pattern is optimized by using the optimization theory, so that an optimal well pattern form can be found; the well pattern optimization is combined with injection-production optimization to optimize the optimal injection-production amount under different well pattern forms; the corresponding optimal total injection-production amount is obtained while the well pattern form is optimized; and well patterns with various rules can be optimized.

A document classification system updates the logistic regression parameters of the logistic regression algorithm used to classify the documents, using a coherent gradient method based upon the coherency between the update data regarding the classifications of the documents and the combined data including both the update data and the original data used to generate the original logistic regression parameters. The classification system can be applied to an email filtering system for filtering spam email.

The invention discloses a novel epitaxy method for growing an ultraviolet LED. The method can obviously improve the crystal quality of ultraviolet LED epitaxial growth materials and improve the light intensity of the ultraviolet LED. According to the method, an Al component gradient method is adopted to carry out doping in AlN, stress of crystal lattice adaption from a substrate layer can be released gradually, gradient component growth can reduce defects of crystal lattice adaption of an interface, the crystal quality of a material interface can be improved, slip dislocation can be carried out in gradient growth, penetration dislocation segregation is achieved, penetration dislocation is well stopped from entering a quantum well area, a quite good substrate is provided for growth of the quantum well, influences on quantum well growth caused by stress generated due to the crystal lattice adaption are reduced greatly, the integral crystal quality of materials is improved, the electron hole wave function spatial overlapping of a quantum well layer is enhanced, and optical and electrical properties of the whole ultraviolet LED are improved.

The invention discloses an irinotecan or irinotecan hydrochloride lipidosome and a preparation method thereof. The lipidosome contains irinotecan or irinotecan hydrochloride, neutral phospholipid and cholesterol, wherein the weight ratio of the cholesterol to the neutral phospholipid is 1:(3-5), and the irinotecan or irinotecan hydrochloride lipidosome is prepared by an ion gradient method.

The invention relates to a method for controlling a flexible satellite based on a feature model, which is characterized in that the time dimension, the sampling time and the parameters M and m are determined according to the kinetic equation of the flexible satellite; the coefficient range of the feature model is determined according to each obtained variable; the parameters of the feature model are identified by utilizing a gradient method; and a control law is designed according to the coefficients of the feature model obtained through the identification, and the attitude angle of the flexible satellite is controlled through the kinetic equation that the control law is fed back to the flexible satellite. The method has the advantages that the time dimension and the sampling period of the flexible satellite are introduced; the change rate of the flexible satellite is depicted; the bottleneck problem of the feature modeling of the flexible satellite is solved; the expression of the parameter range of the feature model of the flexible satellite is provided; the parameter property of the feature model is qualitatively researched; the boundary of the parameters of the feature model is relative to the sampling period, the modeling error, the system order and the change rate of the system from the given parameter range; and the theoretical foundation of the self-adaptive control based on the feature model is laid for the flexible satellite. The method is suitable for the feature model of the attitude kinetics of an aircraft so as to lay the foundation for the aircraft based on the attitude control of the feature model.

The invention discloses a resource optimization method for safe calculation unloading in a mobile edge computing network, which comprises the following steps: S1, initializing parameters and configurations of a user and a mobile edge server; S2, obtaining an initial communication and computing resource allocation scheme and a corresponding initial objective function value; S3, according to that initial communication and computing resource allocation scheme, a projection gradient method is adopted to obtain a new communication and computing resource allocation scheme and correspond new objective function values; S4, judging whether that difference between the new objective function value and the last objective function value is within the threshold value range, and if so, outputting a current communication and calculation resource allocation scheme and terminating optimization; Otherwise proceeding to step S5; S5, taking the current communication and computing resource allocation schemeas the initial communication and computing resource allocation scheme, and returning to step S3. The method considers the scene of eavesdropper in the system, improves the security of the system, realizes the minimum delay, and improves the service quality.

The invention discloses an anti-collision control method based on depth reinforcement learning. A depth deterministic policy gradient method (DDPG algorithm) is adopted for depth reinforcement learning, and the anti-collision control method includes the steps that first, vehicle parameters and environmental vehicle parameters are extracted; second, a virtual environment model is constructed through the vehicle parameters and the environmental vehicle parameters; third, according to the vehicle parameters, the environmental vehicle parameters and the virtual environment model, basic parametersof the depth deterministic policy gradient method are defined; fourth, according to the basic parameters defined in the third step, a neural network in depth reinforcement learning is used for constructing an anti-collision control decision making system, and the anti-collision control decision making system comprises a strategy network and an evaluation network; and fifth, the strategy network and the evaluation network are trained, and the anti-collision control decision making system is obtained. According to the anti-collision control method based on depth reinforcement learning, the anti-collision control decision making system based on the depth neural network and the anti-collision control decision making system for constantly optimizing network control results based on a time difference reinforcement learning method are constructed, and the control performance of the anti-collision control decision making system is improved effectively.

The invention provides a pretreatment process for a CVD diamond film deposited on a hard alloy surface by adopting a gradient method in view of the problems of low intensity of the seamed edges and sharp corners, affecting the product performance and even the corrosion of seamed edges and sharp corners and so on caused by the faster removal of Co on the seamed edges and the sharp corners than Co on a face in the pretreatment process of the existing CVD diamond films deposited on a hard alloy surface; the main process procedures are provided in sequence as follows: polishing (1) arrow cleaning(2) arrow etching of WC (3) arrow cleaning(4) arrow first etching of Co(5) arrow cleaning(6) arrow drying (7) arrow glue spreading (8) arrow second etching of Co(9) arrow cleaning(10) arrow sol(11) and seeding with diamond powder (12), particularly, the gluing technique of the procedures ensures that the hard alloy surface meets the deposit demands of the CVD diamond films after being pretreated and at the same time, the Co content of the seamed edges and the sharp corners of the hard alloy product and the removal Co content on the face are caused to be basically consistent, which ensures the intensity of the seamed edges and the sharp corners and improves the performance of the CVD diamond films.

The invention discloses a form-finding method of a cable-strut tensile structure, belonging to the field of building design and structure design of a space structure. The method provided in the invention comprises the following steps: setting a topological relation of the cable-strut tensile structure and rendering constraint conditions thereof and lengths of struts; determining an objective function in form-finding analysis of the cable-strut tensile structure; setting weight coefficients of the lengths of each cable and strut as well as weight coefficients of each boundary condition, and rendering an initial step length in a gradient optimization method; and calculating a gradient vector of a coordinate variable of each node in XK. The method provided in the present invention is based on viewpoint of energy, constraints such as boundary conditions of the cable-strut tensile structure and the like are written into the performance function of the gradient optimization method, thereby a form-finding problem of the tensile structure having the constraint conditions being easily solved. Too many constraint conditions and geometric constraints are added in traditional form-finding methods, while, few constraint conditions exist in the method provided in the present invention, therefore, a large number of irregular systems can be found.

An ocean sectional diagram drawing method based on a GIS platform comprises the following steps of (1) data loading, wherein loaded data comprises ocean environmental element data and seabed landform data used for drawing; (2) data frame template loading; (3) layer symbolization, wherein symbolization of a sectional distribution diagram is a gradient blank diagram represented by isopleths and a hypsometric tending method, and graded chroma for filling follows the gradient method and the color system is maintained fixed; (4) note generation and edition; (5) layer masking, namely, isopleth covering, which is used for covering noted characters with isopleths; (6) diagram finishing; and (7) diagram outputting. The invention firstly integrates all flows of ocean sectional diagram drawing on the GIS platform, changes the present situations that drawing formats and platforms of existing ocean sectional diagram drawings are different, typesetting and editing are difficult, and standards are not uniform, and automatic drawing of a traditional sectional diagram on the GIS platform is achieved.