43 results about "Pattern search algorithm" patented technology

The invention discloses a method for self-adjusting control parameters of a speed ring of a permanent magnet synchronous motor based on fractional orders. An integer-order proportional integral (PI) controller in an alternating current servo system of the permanent magnet synchronous motor is replaced by a fractional-order PI controller, parameters of the fractional-order PI controller are automatically adjusted, and the alternating current servo system of the permanent magnet synchronous motor can be controlled. The method specifically comprises the following steps of: firstly, acquiring current and speed signals of the alternating current servo system; secondly, identifying a speed ring controlled object model of a permanent magnet synchronous motor servo system according to the acquired signals, and identifying the parameters of the model; and finally, optimally adjusting the control parameters, and thus obtaining optimal control parameters. According to the method, the original integer-order PI controller is replaced by the fractional-order PI controller, and the parameters of the controller are automatically adjusted; the parameters of the controller are optimized by using a pattern search algorithm; and the adjusted parameters of the controller are high in robust, high in anti-disturbance capacity and high in control accuracy.

The invention discloses a PDFF-based AC servo driver control parameter self-tuning method. Firstly, current and speed signals required for identifying an AC servo system model are acquired; secondly, according to the acquired signals, a speed loop controlled object model of a permanent magnet synchronous motor AC servo driving system is identified and parameters of the model are identified; and finally, according to the parameters and performance indexes of the model, searching optimization tuning is performed on control parameters of a servo driver so that the optimal control parameter is acquired. According to the self-tuning method, a PDFF controller is utilized to substitute a conventional PI controller on a speed loop of the servo driver and the parameters of the controller are automatically tuned. The model of the controlled object is identified by the self-tuning method by utilizing a recursion least square method. Then according to the feasible performance indexes, searching optimization is performed on the parameters of the controller by utilizing a mode searching algorithm, and the optimal control parameter is obtained so that the servo driving system can be great in anti-disturbance capability, control precision and robustness.

The invention relates to a photovoltaic string fault diagnosis method based on a kernel function limit learning machine. The method specifically comprises the following steps that 1, voltage-current characteristic scanning is conducted on a photovoltaic string, and curve fitting is conducted to acquire five photovoltaic internal equivalent parameters; 2, the acquired five photovoltaic internal equivalent parameters are integrated and unified; 3, a plurality of coefficients of an optimal KELM algorithm kernel function are calculated by adopting a pattern search algorithm; 4, the calculated coefficients are brought into the KELM, and samples are trained to obtain a training model; 5, fault detection and classification are conducted on the photovoltaic string through the training model. According to the photovoltaic string fault diagnosis method based on the kernel function limit learning machine, the accuracy of fault detection and classification on a photovoltaic power generation array can be effectively improved.

A solution to determining the move set ordering in pattern searching is disclosed that involves driving a pattern search algorithm by a metric other than the step size of the patterns. An instance of this metric is the amount of change in an objective function. Preprocessing algorithms are disclosed which quantify the effect each move has on the objective function. Those moves having a greater effect on the objective function are applied before moves having a lesser effect. We call this effect on the object function the sensitivity of the object function to a particular move and present several methods to quantify it. The sensitivity may be expressed as a function or the moves can be ranked and clustered with the pattern search being driven by the ranked moves or the function. The search may also be driven by an expected change in objective function. Because of the rules governing abstracts, this abstract should not be used to construe the claims.

The invention relates to a quick high-efficiency video coding (HEVC) intra-frame coding unit and pattern decision method which comprises the following steps: (1) the effective length N of a threshold value is set; (2) a first frame of a video sequence is normally coded, and the final coding depth and the discrete total variation (DTV) value of each largest coding unit (LCU) are stored in a cache region; (3) the DTV values in the cache region are counted, and a threshold value T1 and a threshold value T2 are worked out according to the depth range; (4) a quick pattern search algorithm based on direction gradient is adopted on the remaining N-1 frames according to the current DTV values of the LCUs and the threshold values at each depth search, with low-probability depth skipped; and (5) the (N+1)th frame is set as the first frame, and the steps (1), (2) (3) and (4) are repeated until all the frames are coded. The method adopts adaptive DTV threshold decision, the number of searched macro block units is reduced, the process of pattern search is simplified by adopting features based on direction gradient, and the coding speed is increased.

The invention discloses a cage asynchronous motor rotor broken-bar fault detection method based on an electronic stability program rotation invariant technology (ESPRIT) and a pattern search algorithm (PSA). The method comprises the following steps of: performing a rotation invariant technology on a stator current instant signal acquired according to a certain frequency so as to obtain accurate frequency values, rough amplitude values and rough initial phase angles of a fundamental wave component and a side frequency component; estimating the accurate amplitude values and the accurate initial phase angles of the fundamental wave component and the side frequency component of the stator current instant signal by applying a pattern search algorithm, and thus obtaining a ratio, which is used as a fault characteristic, of the amplitude value of the current side frequency component to the amplitude value of the fundamental wave component; calculating a ratio of the ratio to a detection threshold value to determine a fault index; and finally determining whether a broken-bar fault exists or not according to the fault index. By the method, the asynchronous motor rotor broken-bar fault can be detected on line with high sensitivity and high reliability by using a small number of stator current signal sampling points; therefore, the influence caused by adverse factors such as load fluctuation, noise and the like can be overcome effectively; and the method is extremely applicable to low slip-ratio running situations of an asynchronous motor.

The invention discloses a dangerous chemical gas leakage isoconcentration line mass center track tracing algorithm. The algorithm includes the following steps: acquiring dangerous chemical gas concentration data collected on site and calculating multiple sets of simultaneous isoconcentration line mass center data; conducting linear fitting on isoconcentration line mass center data sets in time sequence; acquiring gas leakage source position data according to a linear fitting result. First, the mass center of isoconcentration lines at each moment is obtained, and then the mass center of the isoconcentration lines in the time sequence is subjected to linear fitting by utilizing the linear moving track of the isoconcentration line mass center with the time. The location of a leakage source can be found quickly in the reverse wind direction of the fitted line to provide quick guidance for leakage emergency response, and the leakage accident expansion can be effectively controlled. The algorithm also provides a scientific theoretical basis for selection of initial population data and initial values for solving a genetic algorithm and a pattern search algorithm based on the optimized theoretical method, the iterative calculation time for the optimized theoretical method is shortened, and the calculation accuracy is improved.

A weighting function (WF) is previously provided to model the number of search points of a pattern search. However, WF fails to properly describe the behavior of the genetic pattern search algorithms due to some over-simplifications in their models. Therefore, a refined weighting function (RWF) is provided to more accurately describe both genetic and non-genetic pattern searches. Moreover, based on the understanding to RWF, two momentum-directed genetic search algorithms are further provided. These new algorithms check the possible mutations according to their likelihood to the preceding successful mutations and further accelerate the previous genetic pattern searches.

The invention provides a method used for determining the coding type of the macro block of the video frame and the prediction mode. In the implementation modes, the common method is that 1) the coding type selection algorithm (based on the property of the macro block which is easy for calculation) is used to select the coding types (16*16 or 4*4) considered by the macro block initially; if the coding type 16*16 is selected in the procedure 1, based on the distortion threshold, the traditional method or an improved 16*16 prediction mode search algorithm is used to consider the four types of 16*16 prediction modes which can be used for the macro block; 3) if the 4*4 coding type is selected in the procedure 1, based on the position relations of the prediction modes and the traditional method or an improved 4*4 prediction mode search algorithm, the 4*4 prediction mode is selected to be used for each 4*4 module of the 16 modules of the macro block.

The invention discloses a multi-waveform phase coding method based on a mode search algorithm, belongs to the technical field of radars, and especially relates to a radar waveform coding method. A multi-waveform autocorrelation and cross-correlation function obtained through the method has a low sidelobe level in a specific time delay region, thereby solving a problem that the calculation work is large in the prior art. The implementation process of the method comprises the steps: 1, constructing a target function of a multi-dimensional optimization problem; 2, simplifying the target function, and employing the mode search algorithm to solve the simplified target function; 3, carrying out iteration for many times till a system cannot accept any phase change, i.e., obtaining a waveform design meeting the demands of related characteristics. Compared with a Weighted Cyclic Algorithm New, WeCAN method, the method is high in rate of convergence, is small in calculation work, and can effectively design a phase-coding multiple waveforms with lower autocorrelation and cross-correlation sidelobe in a specific time delay region.

The invention discloses a mode search algorithm and firefly algorithm-based image multi-threshold segmentation method. The method comprises the steps of firstly obtaining a grayscale image needed to be processed, and if the grayscale image is a color image, converting the color image into a grayscale image I(x,y) through rgb2gray() of MATLAB 7.12; and then performing threshold segmentation through an OTSU method based on a mode search algorithm and a firefly algorithm: specifically, the OTSU method based on the mode search algorithm and the firefly algorithm is that thresholds=PS-FA-OTSU(grayImage,num), wherein grayImage is a to-be-segmented grayscale image input in the algorithms, num is a to-be-segmented threshold number, thresholds are segmented thresholds obtained by algorithm-based segmentation, and the output thresholds are the obtained segmented thresholds, so that the threshold segmentation of the grayscale image is finished. The OTSU method-based grayscale image multi-threshold segmentation is quicker, more stable and more efficient.

The invention provides a large-spacing phased-array antenna grating lobe suppression method. The method comprises the following steps: constraining the positions of array elements; constructing a fitness function and a fitness function optimization model; selecting and crossing: selecting a certain proportion of individuals before sorting as parents for generating a new generation of population; then performing crossover operation on the population of the generation to obtain a new generation of population; mutation: for each gene of each population, randomly generating a number r between [0,1], and if r is less than Pm, replacing r with a randomly generated parameter in a value domain, where Pm is a mutation probability; and optimal solution calculation: carrying out position optimization on the large-spacing planar array by utilizing a genetic algorithm, taking an optimization result as an initial solution of a mode search algorithm, and further optimizing to find an optimal solution under the current condition. The array obtained by constraining the positions of the array elements in the optimization process is easy to implement in engineering, the grating lobe suppression effect is good, and the grating lobe can be suppressed to-8dB or below under the condition that the minimum spacing of the array elements is 3 lambda.

An ultra-wideband (UWB)-assisted precise positioning system, which includes a plurality of mobile base stations (MBSs) and a central control base station (CBS). The central CBS and the MBSs are configured to: measure, by the UWB communication module, distances between the MBSs, such that the moving module controls the MBSs to have a predetermined distance from one another and form a predetermined pattern; configure the global satellite positioning (GPS) modules to perform positioning to obtain a plurality of initial positions; execute a pattern search algorithm to generate a plurality of corrected positions; generate a plurality of predicted positions to be input to a cost function for finding a plurality of final predicted positions having relatively smallest errors; measure, by the UWB communication module, distances of a test object relative to the MBSs and calculate a position of the test object.

The invention discloses a meta search method for a gene tissue-specific sequence pattern and a search result assessment method and belongs to the field of biological information science. The search method comprises the following steps: extracting promoter sequences of tissue-specific genes and housekeeping (HK) genes from a bioinformatics database as input initial data; carrying out local a search algorithm and an exhaustive search algorithm on the input initial data respectively; storing result tissue generated from the operation of the two search algorithms into a filter matrix; estimating the pattern probability by utilizing the data in the filter matrix; and merging the motifs. In the assessment method, Bayes factor analysis evaluation statistics are utilized to obtain the significance of the search result of the gene tissue-specific motifs. Compared with prior art, the methods provided by the invention have the advantages that the pattern search frame used in the invention integrates multiple algorithms, the principle of 'average result overcomes single choice' which prevails in bioinformatics is conformed, the robustness and creditability of results are improved, the creditability of search results is increased, and over estimation and low estimation of the pattern are avoided.

The invention discloses an automatic programming method and system for natural language machine thinking. After a task book is obtained, the following steps are sequentially carried out: S1, converting and expressing a task book text; S2, establishing a corresponding relation library; S3, externally inputting or automatically compiling a task book text, and interpreting, defining and understandingthe task book to be solved; S4, searching an algorithm module library according to the problem type and the problem framework composition mode, and designing an algorithm for solving the problem, a design program and a selected algorithm according to the algorithm module library; S5, entering a natural language generation type system, and outputting an instruction to enter programming software orenter any programming language for programming to obtain a program; And S6, running the program, and checking an output result of the program. Therefore, the machine automatically generates a specific program capable of completing the target after receiving the demand description about the target to be realized by the designed program. The method is a creative artificial intelligence technology,and has a great commercial value.

The invention relates to a data processing method and device based on a table function and a computer storage medium. The method comprises the following steps: performing classification characterization on cognitive contents based on a subject classification table to form different characterization categories and codes; adopting different calculation methods to carry out classification processing according to different representation categories and codes; constructing a data storage module according to the representation category and the coding and classification result; according to the representation category and the coding and classification calculation method, generating output results corresponding to different input information offline, and an input and output truth value mapping relation table based on a preset table function template; and querying in the input and output truth value mapping relation table by adopting a multi-stage mode search algorithm of a self-adaptive resonance network according to input information, and outputting a data processing result based on a mode similarity threshold calculation method. Content required by the user can be quickly and accurately searched, and the requirements of the user are met.

A solution to determining the move set ordering in pattern searching is disclosed that involves driving a pattern search algorithm by a metric other than the step size of the patterns. An instance of this metric is the amount of change in an objective function. Preprocessing algorithms are disclosed which quantify the effect each move has on the objective function. Those moves having a greater effect on the objective function are applied before moves having a lesser effect. We call this effect on the object function the sensitivity of the object function to a particular move and present several methods to quantify it. The sensitivity may be expressed as a function or the moves can be ranked and clustered with the pattern search being driven by the ranked moves or the function.

The invention discloses a method, device and equipment for improving rate distortion optimization calculation efficiency and a storage medium, and the method comprises the steps: searching an intra-frame prediction mode through a preset three-step mode search algorithm, and obtaining a candidate intra-frame prediction mode; in each candidate intra-frame prediction mode, carrying out pixel false reconstruction on the coding tree unit to obtain a false reconstruction pixel value of the coding tree unit; obtaining a reconstruction block according to the false reconstruction pixel value, and calculating a square error sum according to the reconstruction block and the original block; and calculating a rate distortion optimization value of each candidate intra-frame prediction mode according to the square error sum and a pre-calculated code rate. According to the method for optimizing the calculation efficiency through rate distortion provided by the embodiment of the invention, the calculation complexity is lower, the data dependence between the blocks is completely removed, and better hardware time sequence and better layout and wiring can be achieved.

In some embodiments, a method of determining encoding type and predictive mode(s) selections for a macroblock of a video frame is provided. In some embodiments, a general method 1) selects the encoding type (16x16 or 4x4) that is initially considered for a macroblock using an encoding type selection algorithm (based on an attribute of the macroblock that is easy to compute), 2) if the 16x16 encoding type is selected in step 1, consider the four 16x16 prediction modes that may be used on the macroblock using conventional methods or an improved 16x16 predictive mode search algorithm based on distortion thresholds, and 3) if the 4x4 encoding type is selected in step 1, select the 4x4 prediction mode to be used for each of the sixteen 4x4 blocks of the macroblock using conventional methods or an improved 4x4 predictive mode search algorithm based on the positional relationships between predictive modes.

The invention discloses a method for generating an optimal protection channel of an MIMO radar antenna, and the method comprises the following steps: transmitting a radar signal to a target object ina set range through the MIMO antenna, and generating an initial protection channel through an echo signal of the target object; initializing parameters, wherein the parameters comprise an initial protection channel parameter vector; setting a cost function according to the requirement of the optimal protection channel; optimizing the initial protection channel parameter vector for multiple times by adopting a mode search algorithm; and taking a protection channel parameter vector enabling the cost function to be minimum after multiple times of optimization, and taking the protection channel parameter vector as an optimal protection channel generation vector to generate an optimal protection channel. According to the method, parameter weighting is performed on the antenna output signal to synthesize the protection channel so as to suppress the sidelobe, the method is applied to the MIMO radar, an auxiliary antenna does not need to be independently arranged, and the hardware requirementis low.

The invention discloses a cage asynchronous motor rotor broken-bar fault detection method based on an electronic stability program rotation invariant technology (ESPRIT) and a pattern search algorithm (PSA). The method comprises the following steps of: performing a rotation invariant technology on a stator current instant signal acquired according to a certain frequency so as to obtain accurate frequency values, rough amplitude values and rough initial phase angles of a fundamental wave component and a side frequency component; estimating the accurate amplitude values and the accurate initial phase angles of the fundamental wave component and the side frequency component of the stator current instant signal by applying a pattern search algorithm, and thus obtaining a ratio, which is used as a fault characteristic, of the amplitude value of the current side frequency component to the amplitude value of the fundamental wave component; calculating a ratio of the ratio to a detection threshold value to determine a fault index; and finally determining whether a broken-bar fault exists or not according to the fault index. By the method, the asynchronous motor rotor broken-bar fault can be detected on line with high sensitivity and high reliability by using a small number of stator current signal sampling points; therefore, the influence caused by adverse factors such as load fluctuation, noise and the like can be overcome effectively; and the method is extremely applicable to low slip-ratio running situations of an asynchronous motor.

The invention discloses a method for detecting broken bar fault of a cage type asynchronous motor rotor based on multiple signal classification (MUSIC) and a pattern search algorithm (PSA). The method comprises the following steps of: firstly, applying a multiple signal classification technique to obtain frequency values of a fundamental component and a side frequency component of a stator current momentary signal which is acquired according to a certain frequency; secondly, estimating amplitude values and initial phase angles of the fundamental component and side frequency component of the stator current momentary signal by applying the PSA; thirdly, calculating the ratio of the amplitude value of the current side frequency component to that of the fundamental component and taking the ratio as a fault characteristic; fourthly, calculating the specific value of the ratio to a detection threshold value to determine a fault index; and finally, judging whether a rotor broken bar fault exists according to the fault index. By using the method, the broken bar fault of the asynchronous motor rotor can be detected in an on-line manner with high sensibility and high reliability by using few stator current signal sampling points, so the influence caused by adverse factors such as load fluctuation, noise and the like is effectively overcome, and the method is very suitable for the condition that an asynchronous motor runs at a low slip ratio.

The invention provides a pulse Doppler radar based on a mode search algorithm and a beam forming method, and the method comprises the steps: S1, measuring radar receiving gains of a front end array plane of a radar phased array according to different azimuth angles and pitch angles, and obtaining receiving antenna directional diagrams of different array elements; S2, initializing numbers and wavebeam generation coefficients, designing different search step lengths, determining base vectors in different search directions, and initializing convergence conditions; S3, determining a sum beam generation coefficient objective function according to the sum beam distribution characteristics; S4, solving a target function value corresponding to the beam generation coefficients, and judging whetherthe target function reaches a convergence condition or not; and S5, obtaining different sum beam generation coefficients and corresponding target function values by using the different step lengths and the base vectors, and selecting the sum beam generation coefficient enabling the target function value to be minimum as a final result to be output. According to the invention, an ideal and channelantenna pattern is synthesized through a mode search optimization method.

The invention provides a passenger transportation demand index modeling and computing method based on user query logs. The method comprises the steps of extracting and analyzing user query log data and real ticket booking data from a database; extracting a channel query model classification result based on user query data from the database; allocating weights to different classes of channels, fitting the real ticket booking data, and obtaining the optimum solution of the weight of each class of channels through utilization of a mode search algorithm; and computing a transportation demand indexof each travel route according to query quantity data of different classes of channels for different routes, and the class weights. According to the transportation demand index computing method provided by the embodiment of the invention, passenger travel demands can be reflected relatively visually, and the data support is provided for managers and marketers in the field of transportation.

The invention relates to a nonlinear rigidity identification method based on mode search under pulse excitation, which comprises the following steps of: applying a pulse signal to a two-degree-of-freedom local nonlinear vibration system, acquiring actual measurement transient response after the pulse signal acts, defining and extracting characteristic displacement, obtaining instantaneous frequency, and obtaining a frequency-displacement diagram; and establishing a kinetic model corresponding to the obtained frequency-displacement diagram, defining a frequency according to the kinetic model, fitting the frequency with the instantaneous frequency, and determining an unknown nonlinear stiffness parameter of the system based on a mode search algorithm. According to the method, data processing is performed on the signal on the basis of obtaining the actually measured transient response, the nonlinear stiffness parameters of the two-degree-of-freedom local nonlinear vibration system are effectively identified by adopting the mode search algorithm, and the method has practical engineering significance.

The invention discloses a photovoltaic system MPPT (Maximum Power Point Tracking) method based on a leapfrog and pattern search neural network, which comprises the following steps of: (1) obtaining the temperature and irradiance of a photovoltaic module, and obtaining a maximum power point reference voltage by adopting a neural network; (2) enabling the controller to obtain an output quantity according to an error between the reference voltage and the measurement voltage of the photovoltaic module; (3) enabling an enhanced disturbance observer P& Q to obtain the control quantity of the chopper circuit according to the measured voltage, the measured current and the output quantity of the controller, so that the photovoltaic system stably works at the maximum power point along with the illumination change. According to the method, a hybrid recombination leapfrog and pattern search algorithm is adopted to optimize the maximum power point tracking based on the neural network in the photovoltaic system; the method has excellent performance in response speed and precision, and can provide the highest tracking efficiency and the fastest response time under the condition that the steady state and the irradiance are continuously changed, and the response time is 11 seconds.

The invention relates to a Fifteen-phase asynchronous motor rotor broken bar number high-precision diagnosis method based on ESPRIT-PSA and LGBM. According to the method, for an instantaneous reactive power signal sampled in a short time (only 2 seconds), an accurate amplitude Aq of a characteristic component in the instantaneous reactive power signal is obtained by using an ESPRIT (Rotation Invariant Signal Parameter Estimation Technology) and a PSA (Pattern Search Algorithm); corresponding 71 characteristics such as voltage amplitude, current amplitude, average active power P and average reactive power Q are obtained; the method then includes determining Aq and the first five features, namely Aq, P and Q, a first phase voltage amplitude Um and a first phase current amplitude Im, in the 71 features according to an LGBM (Light Gradient Elevator) so as to form a data set, wherein the first five features include Aq, P and Q, and the first phase voltage amplitude Um and the first phase current amplitude Im; and then training and storing an LGBM model and carrying out high-precision diagnosis on the number of broken bars of the rotor (the training precision and the test precision are both 100%, and the five-fold cross validation accuracy is 99.38%).

The present invention proposes a pulse Doppler radar and a beamforming method based on a pattern search algorithm, including: S1: measure the radar receiving gain of the front-end array of the radar phased array according to different azimuth angles and elevation angles, and obtain different array element receiving antennas Direction diagram; S2: Initialize numbers and beamforming coefficients, design different search step sizes, determine basis vectors for different search directions, and initialize convergence conditions; S3: Determine and beamforming coefficient objective function by sum beam distribution characteristics; S4: Solve sum beam The generation coefficient corresponds to the objective function value, and judges whether the objective function meets the convergence condition; S5: Use different step lengths and base vectors to obtain different sum beamforming coefficients and corresponding objective function values, and select the sum beamforming coefficient that minimizes the objective function value as the final The result is output. The invention synthesizes ideal and channel antenna patterns through a mode search optimization method.