76 results about "Competitive algorithm" patented technology

In predictive data mining, a process and tool presents a method to compare given competing algorithms to a derived reference, such as a baseline or benchmark. A result confidence as to the suitability of the competing algorithm to a given task is generated. In an exemplary embodiment, a randomized feature acting, simple, algorithm is used to generate the baseline. In an alternative embodiment, the process and tool is used to determine learnability of the given task. A mechanism to account for overfitting of data is described.

A parallel dynamical system for computing sparse representations of data, i.e., where the data can be fully represented in terms of a small number of non-zero code elements, and for reconstructing compressively sensed images. The system is based on the principles of thresholding and local competition that solves a family of sparse approximation problems corresponding to various sparsity metrics. The system utilizes Locally Competitive Algorithms (LCAs), nodes in a population continually compete with neighboring units using (usually one-way) lateral inhibition to calculate coefficients representing an input in an over complete dictionary.

The invention relates to sparse parallel signal coding using a neural network which parameters are adaptively determined in dependence on a pre-determined signal shaping characteristic. A signal is provides to a neural network encoder implementing a locally competitive algorithm for sparsely representing the signal. A plurality of interconnected nodes receive projections of the input signal, and each node generates an output once an internal potential thereof exceeds a node-dependent threshold value. The node-dependent threshold value for each of the nodes is set based upon the pre-determined shaping characteristic. In one embodiment, the invention enables to incorporate perceptual auditory masking in the sparse parallel coding of audio signals.

The invention discloses a novel multi-objective optimization method for a catalytic cracking separation system. The method comprises the following steps of S1, obtaining production data, collected in actual production, of the catalytic cracking separation system, building a process model of the catalytic cracking separation system by utilizing process simulation software Aspen Plus, and calibrating the process model through the obtained production data, thereby enabling an analog output value to be the same as the production data; S2, comprehensively selecting multiple constraint conditions in a working process of the catalytic cracking separation system, setting optimization variables corresponding to the constraint conditions, and by taking C3 content in a product dry gas and energy consumption of the catalytic cracking separation system as optimization objectives, building a multi-objective optimization model; and S3, establishing an Aspen Plus and Matlab integrated platform, and solving the multi-objective optimization model by applying a fuzzy optimization algorithm and a multi-objective line-up competition algorithm on the integrated platform, thereby obtaining an optimal solution set of the C3 content in the product dry gas and the energy consumption. According to the method, the condition that optimal solutions possibly exist in a transitional area is considered, so that the best optimization effect is achieved.

A system and method for rating the validity of multiple data processing algorithms A₁-A_n on a set of multi-dimensional input data I where P₁-P_n is the output data produced by the competing algorithms. Processing steps include: (1) computing the Fourier Transform or the phase congruence of the input data I, (2) computing the Fourier Transform or the phase congruence of the output data P₁-P_n, (3) computing the phase of the data computed at step 1, (4) computing the phases of the data computed at step 2, and (5) computing a similarity metric, such as the normalized cross correlation of each phase computed at step 4 with the corresponding phase computed at step 3. The similarity metrics computed at step 5 can be ordered to provide a ranking of the algorithm validity.

The invention relates to an improved linear optimal semi-active control method. Since damper has high nonlinearity, control of the damping force by adjusting the control current or voltage is required. Therefore, the control method is composed of two levels of controllers, namely, an improved linear optimal controller (LQR) and an adaptive neuro-fuzzy inference system (ANFIS) inverse model. The control method comprises the following steps: firstly, performing optimal design on the LQR controller by a crow search algorithm (CSA); then calculating desired ideal control force for vibration reduction by using the optimal LQR controller; and finally, improve an ANFIS by using the ideal control force as the input of an ANFIS inverse model with a hybrid training algorithm based on an imperial competition algorithm (ICA) in order to output a control signal of a magnetorheological damper accurately. The method can improve the current prediction accuracy of the MR damper, and make up for the deficiency of the control effect of the existing linear optimal semi-active control, thereby enhancing the vibration reduction effect of the MR damper.

The invention provides an anaglyph coding method based on three-dimensional self-organizing mapping. The method comprises the steps that a three-dimensional SOM network structure for processing a three-dimensional image is established, a three-dimensional SOM algorithm is utilized to code an anaglyph, and a three-dimensional image compressing code is obtained. The method comprises an initializing algorithm, a competition algorithm, a neighbourhood algorithm and a learning algorithm. The method achieve mapping from two dimensions to three dimensions, a designed code book has good performance in the aspect of the peak signal to the noise ratio of a three-dimensional reconstruction image, three-dimensional image coding is achieved, and a right image reconstructed by decoding the anaglyph and a left image has good subjective and objective evaluation. The anaglyph coding method has important principle and engineering practical significance and wide application prospects and can be widely applied to three-dimensional image processing, remote sensing image processing, medical image processing, target identification, three-dimensional video coding and the like.

The invention relates to an optical character recognition method based on an LVQ (Learning Vector Quantization) neural network. The method includes the following steps: performing preprocessing including image denoising, character segmentation, binaryzation and characteristic extraction on a to-be-recognized character image; sending a single character characteristic into an input layer of the LVQ neural network and acquiring a winning competition-layer nerve cell according to a competition algorithm and performing a recognition rejection test; and inputting a character characteristic which complies with a recognition condition to the competition layer of the LVQ neural network for recognition and outputting the recognition result. The optical character recognition method based on the LVQ neural network is applicable to the field of automatic recognition of a small character set which has strict demands on false rate.

The invention discloses a regional power grid carbon emission management method suitable for a new energy connection background. The method comprises the steps that S1, a regional carbon emission management model is established; S2, probabilistic flow based on a stochastic response surface method (SRSM) is adopted to calculate power grid flow distribution containing multiple associated new energy power stations; S3, an imperialistic competitive algorithm is adopted to solve the multi-target regional carbon emission management model to obtain a Pareto non-inferior solution set; S4, a penalty function is introduced to check and adjust obtained Pareto solutions, and node voltage limit exceeding or branch overloading is avoided; S5, an interactive satisfaction decision making method is utilized to select an ''optimal solution'' of the model from the Pareto solution set according to different demands of system operating personnel; and S6, comprehensive benefits of different schemes are evaluated by changing settings of new energy power station connection locations and all objective function reference memberships. Through the method, comprehensive management of practical power system carbon emission can be realized, and a new thought is provided for evaluation of address selection schemes of new energy stations.

The invention relates to a multi-harmonic source identification method for a power distribution network. The method comprises the following steps of (1) placing a phasor measurement unit (PMU) in thepower distribution network; (2) performing optimization configuration of the PMU on the power distribution network: determining a mathematic model of PMU optimization configuration; solving the mathematic model of the PMU optimization configuration by utilizing a binary imperialist competitive algorithm to obtain an optimization configuration scheme of the PMU; (3) obtaining harmonic voltage Vh data in a period of time at a node for configuring the PMU in the power distribution network; (4) obtaining a fast change component Vh.fast in a harmonic voltage Vh by using a mobile filter; (5) preprocessing Vh, namely, performing centralization processing and whitening processing; and (6) applying the fast change component Vh.fast to an improved independent component analysis algorithm, determining a separation matrix, and enabling the separation matrix to act on the harmonic voltage Vh for separation to obtain harmonic currents Ih injected into the power distribution network. The method has the characteristics of systematization and accurate estimation result, and can be used for the multi-harmonic source identification process under unknown harmonic impedance in the actual power distribution network.

The invention discloses a fault prediction method based on characteristic quantity optimization and a wavelet kernel function LSSVM, and relates to the technical field of fault prediction of power transformers. The method comprises the following steps: firstly, acquiring a ratio of DGA characteristic quantity, secondly, establishing a support vector machine model, and selecting a radial basis function as a kernel function of a classification model; Then, encoding the candidate characteristic quantity and the penalty factor c kernel parameter of the SVM classification model to the same chromosome, optimizing the chromosome by adopting a genetic algorithm, wherein the optimal characteristic quantity is the optimal chromosome selected from the genetic algorithm; taking the optimal feature combination as the input of the next fault prediction and diagnosis function; establishing a wavelet kernel function-least squares support vector machine prediction model; And optimizing the penalty factor of the prediction model and the kernel parameters of the wavelet kernel function by using an imperialism competition algorithm to obtain an optimal parameter combination, and constructing an optimal prediction model based on the parameter combination. Operation state analysis and fault prediction of the transformer at a future moment are realized.

The invention provides a multi-objective real-time optimization method and system in a delay coking process. The method comprises the steps that S1 the production data of the delay coking process are acquired in real time; S2 a delay coking process model is established and calibrated; S3 based on the optimized delay coking process model, a hybrid algorithm based on a multi-objective queuing competition algorithm and a non-dominated genetic algorithm is realized; S4 a number of optimization variables in the delay coking process are comprehensively selected, and raw material heat transfer final temperature and product yield are used as double optimization objectives for analog computation to acquire the variation trend chart of double optimization objectives; and S5 a GUI interactive interface is written, the results of data reading, model establishing, algorithm optimization and the variation trend chart are displayed in real time to select an optimization variable corresponding to the optimum optimization objective in the variation trend chart. The algorithm provided by the invention has the advantages of good optimization effect, strong searching ability, energy saving, environmental protection, can effectively help a petrochemical plant to improve the production efficiency, and has significant economic benefits.

The invention relates to a magneto-rheological damping fuzzy controller design method based on the imperialist competition algorithm. The design method comprises the steps that the structural response of buildings under the effect of earthquakes serves as the input variable of a fuzzy controller, and the control current output by the fuzzy controller serves as input of a magneto-rheological damper; the magneto-rheological damper outputs damping force to the buildings, and the buildings generate new excitation and output new structural responses to the fuzzy controller; and after the earthquakes stop, the objective function value is calculated, whether the objective function value is convergent or not is judged, and further the optimal parameter of the fuzzy controller is output. According to the invention, an obtained fuzzy control system is optimized through the imperialist competition algorithm, the performance of the magneto-rheological damper is given full play, and the structural response of the buildings under the effect of the earthquakes is effectively weakened.

The invention provides an alternating current/direct current hybrid microgrid optimized operation method considering micro-source surplus Internet access. The method comprises the steps that an alternating current/direct current hybrid microgrid optimized operation mathematical model including wind power generation, photovoltaic power generation, a battery, a diesel generator and a fuel cell is established; the alternating current/direct current hybrid microgrid mathematical model comprises a multi-objective, multi-constraint and non-linear objective function constructed with consideration ofpower grid purchase cost, micro-source fuel cost, environmental benefit cost, network loss, operation and maintenance cost, and micro-source surplus Internet access income, and the mathematical modelobeys the constraints of internal microgrid power balance, controllable micro-source climbing rate, upper and lower limits of battery charge and discharge per unit time, upper and lower limits of battery charge state, battery power balance before and after a time period, and unchanged battery charge state; a basic queue competition algorithm is improved to acquire an improved queue competition algorithm; and the improved queue competition algorithm is used to solve the alternating current/direct current hybrid microgrid optimized operation mathematical model. According to the invention, the microgrid operation cost can be effectively reduced, and the new energy consumption rate is promoted.

The invention provides an improved method for solving a problem of flexible job shop scheduling through an empire competition algorithm, and irons out the application defects of a conventional empire competition algorithm in a discrete flexible job shop scheduling problem. Each country individual employs two-segment type integer coding for expressing one scheme, and the coding mode can adapt to multiple flexible job shop scheduling problems, and reduces the computing complexity of an algorithm. Three initialization methods are mixed to generate an initial country group method, thereby improving the quality of a group overall solution, improving the operation efficiency of the algorithm, and obtaining an optimal solution more quickly. The assimilation operation in an empire is reconstructed, and a mode of excellent information exchange between colonial countries and imperialist countries enables excellent information to be shared, thereby enabling the colonial countries to be close to the imperialist countries. The probability changing of the revolution operation in the empires is carried through the iteration times of the empire competition algorithm, thereby improving the diversity of country individuals, and preventing the premature convergence of the algorithm.

The invention belongs to the technical field of power system power distribution network planning and operation, and particularly relates to a reconstruction method for a power distribution network with a distributed power supply after serious fault. The method for reconstructing the power distribution network with the distributed power supply after serious fault is suitable for rapid reconstruction of the power distribution network with the distributed power supply after the serious fault, and is convenient to model, high in simulation precision, high in calculation efficiency and strong in expansibility. The method comprises the following steps: 1, establishing an objective function and constraint condition model of a reconstruction problem after the power distribution network has the fault; 2, performing modeling analysis on the distributed power supply when power distribution network fault occurs; 3, adopting a spanning tree model based on the graph theory to solve an optimal methodof island division; 4, completing the weight dynamic division of a multi-objective function through a membership function, and normalizing a power distribution network reconstruction multi-objectiveoptimization problem into a single-objective optimization problem; and 5, solving the reconstruction optimization problem based on an empire competition algorithm to obtain an optimal reconstruction strategy after the serious fault of the power distribution network.

The invention provides an improved method for solving an FJSP (flexible job shop scheduling problem) with an ICA (imperialist competitive algorithm) and overcomes the defects of the ICA in applicationto the discrete type FJSP. Two-segment integer encoding is adopted for each individual country to express one scheme, the encoding mode can solve multiple FJSPs, and calculation complexity of the algorithm is reduced. With the adoption of the method for producing an initial country group by mixing of three initialization methods, quality of the entire solution of the group is improved, operatingefficiency of the algorithm is improved, and the optimal solution is obtained more quickly. Assimilation operation in an empire is improved, good information is shared between colonial countries and imperialist countries by exchanging the good information, so that the colonial countries are enabled to approach the imperialist countries; by means of revolutionary operation in the empire, probability change is performed according to iterations of the ICA, diversity of the individual country is improved, and the algorithm is prevented from premature convergence.

The invention relates to a fault recovery method for an active power distribution network with the consideration to a charging, discharging and storage integrated station. The method comprises the following steps: 1) establishing an integrated station load emergency support model and a scheduling strategy; 2) enabling the integrated station to receive a load emergency support instruction of a power grid dispatching center and converting the load emergency support instruction into a power supply state from a load state, wherein CS, BSS and CESS in the station serve as energy storage unit for ordered discharge support of partial power-losing loads of the power grid; 3) establishing an active power distribution network fault recovery model and strategy with the consideration to the integratedstation; 4) optimizing the island power supply ranges of the integrated station and the DG and the output of the two sides; 5) performing fault reconstruction on the residual network of the power grid; 6) optimizing and solving the active power distribution network fault recovery model with the consideration to the integrated station through using a chaotic simulated annealing empire competitionalgorithm. Compared with the prior art, the method overcomes the technical difficulty in optimizing the output of each part of the energy storage subsystem of the dispatching integrated station, improves the operation state of the fault recovery stage of the active power distribution network, and improves the self-healing capability of the active power distribution network.

The invention discloses an ultrasonic signal feature acquisition method based on improved sparse decomposition, and relates to the technical field of signal processing. The evolutionary idea of whaleoptimization is combined in the empire competition algorithm for aiming at the problem of sparse decomposition of the ultrasonic detection signal, and the algorithm is improved by reference to chaoticthinking and entropy analysis and other theories to form a new improved empire competition algorithm, and the new improved empire competition algorithm is combined with the matching tracking class sparse decomposition method to perform deep analysis and feature extraction of the ultrasonic signal. The method is wide in range of application and high in practicality and has important meaning for the research and application field of nondestructive detection, vibration analysis, seismic waves and medical signals so as to provide the effective way for further enhancement of the industrial production efficiency.

The invention discloses a GIS fault diagnosis method based on a spectrum energy analysis and a self-organizing competition algorithm, and is mainly used for diagnosing mechanical failures such as insulator looseness fault and insulation faults such as spike discharge and surface flashover for a GIS. The method comprises steps of collecting vibration signals in multiple periods, adaptively filtering the collected signals, and performing multi-level reconstruction on the filtered signals by using wavelet transform to obtain reconstructed signals in different frequency bands, calculating ratio ofenergy of the reconstructed signals in different frequency bands to total frequency band, extracting a energy ratio of frequency band with obvious change on characteristics as an input sample, training the input sample by the self-organized competition algorithm to obtain a weight center, thereby achieving fault classification and diagnosis of GIS faults. The method is easy to implement, and hasobvious results and high diagnostic accuracy.

The invention discloses a microgrid optimization scheduling method based on an improved imperialist competitive algorithm. The method comprises the steps of: determining features of energy storage devices and energy supply devices in a selected microgrid, predicting the generated energy of a photovoltaic cell, the generated energy of a fan and a load capacity; for the load features of different seasons, establishing the total cost of operation of a target function--microgrid for optimization of a scheduling scheme, determining the constraint conditions of the energy storage and energy supply devices, and forming an optimal scheduling model by constraint conditions and the target function; taking the predicted electricity output quantity of the photovoltaic cell and the fan as input, and atthe same time inputting the predicted load capacity and the constraint conditions to take the target function as the fitness of each country, and solving the optimal scheduling scheme of the optimalscheduling model through adoption of the improved imperialist competitive algorithm. According to the load features of different seasons, a seasonal scheduling scheme is provided, an optimal scheduling scheme of an optimal scheduling model is obtained through adoption of an improved imperialist competitive algorithm, and the convergence speed and the calculation precision are effectively improved.

The invention discloses an intelligent fault diagnosis method for a magnetic suspension rotor system. The intelligent fault diagnosis method comprises the following steps: acquiring an axis trajectorydiagram of the magnetic suspension rotor system under different fault conditions; carrying out purification and denoising processing on each axis trajectory diagram; converting each axis trajectory diagram into seven corresponding invariant moments through a Hu invariant moment theory, wherein the seven invariant moments form a feature vector; screening and simplifying the feature vectors of theaxis track graphs through an empire competition algorithm, and removing unqualified axis track graphs; carrying out classification training on the directed acyclic graph support vector machine model by using the simplified feature vectors of the axis trajectory graph under multiple different fault conditions; and monitoring the track of the magnetic suspension rotor system in real time, extractingan axis track diagram of the magnetic suspension rotor system in a fault, inputting the extracted axis track diagram into the trained support vector machine model for classification and identification, and judging a corresponding fault type according to a classification and identification result. The identification of the axis track of the magnetic suspension rotor system in the fault is more accurate.

The invention relates to an imperialistic competitive algorithm-based carbon-energy combined flow solving method. Carbon-energy combined flow of a power system is optimized by adopting a multicultural transfer-based imperialistic competitive reinforcement learning algorithm; each empire adopts a value function matrix in the reinforcement learning algorithm as a cultural matrix of the empire; each optimization nation updates the cultural matrix through interaction with the environment and performs action selection according to the cultural matrix to improve the globality of a policy; and knowledge in the cultural matrix is stored in real time and multicultural transfer is performed to improve the optimization efficiency of subsequent new tasks. According to the adopted method, the cultural transfer-based imperialistic competitive reinforcement learning algorithm is a relatively new intelligent algorithm and has the advantages of high convergence speed, better global convergence, higher stability and the like, so that quick carbon-energy combined flow optimization of the large-scale power system is realized.