566results about How to "Avoid local optima" patented technology

The invention discloses a double-layer planning method based on the combination of global path planning and local rolling prediction collision avoidance planning, so as to solve the problem of path planning for a mobile robot in a dynamic environment. The method mainly comprises two parts: the global path planning and the local rolling prediction collision avoidance planning. The path planning method can better realize robot navigation, and improve intelligence of the robot. The double planning method can be utilized to prevent the blindness of planning in the beginning, and searching space of the problem is reduced; based on the uncertainty of the moving direction of a dynamic barrier and by utilizing the two collision prediction strategies and two corresponding collision avoidance strategies, the dynamic barrier can be avoided well; and particularly, in order to adapt to the change of environment better, in the second layer of planning, a Follow_wall behavior based on behavior method is added, so that when the environment changes, the mobile robot can still arrive at the target without touching the barriers safely.

The invention discloses a target following method in a complex interference scene, comprising the steps of: detecting a moving target by confirming the initial position of every moving target which appears in video, detecting with self-adapting Gaussian background models and separating a moving target area with a morphologic treating method, modeling for target feature by acquiring a plurality ofmatching feature messages for each detected moving target area, fusing a plurality of feature messages and building a target feature template, searching the matching area by searching the area which is matched with the target feature template in a specific area in current video as a target following location of current video, updating the target feature template by updating the feature of the target feature template of current video to finish the self-adapting update for the target feature template after locating the target according to the feature in the current video target area and the feature of the target feature template in the former three videos.

The invention discloses an optimization method for solving an omni-channel logistics distribution problem. In a first stage, a Lagrange relaxation technology is used to solve a LAP (Location-allocation Problem) problem. In a second stage, adaptive large-scale neighborhood search is used for solving a multi-vehicle-type vehicle path problem, a feasible solution can be searched in a large range in a solution space through one group of simple destruction and reconstruction algorithms, and a situation of falling into local optimum can be effectively avoided. Meanwhile, a simulated annealing acceptance criteria realized in the adaptive large-scale neighborhood search can guarantee the quality of a solution and the convergence of the algorithm, a result is output after the algorithm is executed for an appointed iteration number, a time constraint requirement can be met, and a good vehicle distribution scheme is solved for enterprises in limiting time. The simpleness of the Lagrange relaxation technology is combined with the efficiency of the adaptive large-scale neighborhood search, so that the integral solving efficiency of a mixed algorithm is high, and the problem of omni-channel logistics distribution can be effectively solved.

The invention discloses an ultra-high-frequency partial discharge signal identification method for gas insulated switchgear (GIS). The method comprises a model training process and a defect identification process, and specifically comprises the following steps of: reprocessing partial discharge signals of the GIS; extracting discharge characteristic parameters such as average discharge amplitude, discharge amplitude standard deviation, discharge phase distribution, discharge polarity, discharge time interval mean, discharge time interval standard deviation; optimizing a weight and a threshold value of a back propagation (BP) neural network by utilizing a genetic simulated annealing tool; training samples by utilizing a BP neural network tool; establishing a corresponding gas statistic algorithm (GSA)-BP model; preprocessing the partial discharge signals to be identified of the GIS; and identifying the samples to be measured in a classified way according to the GSA-BP model after extracting the corresponding characteristic parameters. By the method, the efficiency and the accuracy of partial discharge fault diagnosis of the GIS are improved effectively; and the method is critical to evaluate the insulation state of the GIS and formulate a reasonable maintenance strategy.

The invention relates to a short-term load predicting method of a power grid. The method comprises the steps: step 1, acquiring historical data and pre-treating the data; step2, decomposing the historical load sample data into a plurality of different-frequency sub-sequences by using wavelet decomposition; step 3, performing single-branch reconstruction to each sub-sequence; step 4, dynamically choosing training samples and establishing a neural network predicting model optimized by a vertical and horizontal intersection algorithm; step 5, predicting each sub-sequence 24 hours in advance by using the optimal neural network predicting model; and step 6, superposing the predicted value of each sub-sequence to obtain a whole prediction result. The inherent defects of the neutral network can be overcome by optimizing BP neutral network parameters by a brand-new swarm intelligence algorithm, that is, the vertical and horizontal intersection algorithm instead of the traditional algorithm; the burr problem caused by the impact load processing is solved by the wavelet decomposition, the precision declining resulting from the removal of the effective load in the burr pre-treatment is solved and the predicted value of the hybrid algorithm is more approximate to the actual measured load value.

The invention discloses a machine tool manufacturing system processing and energy saving optimization method based on a particle swarm algorithm, and solves the problems high energy consumption and long processing time caused by improper selection of processing parameters in component processing of a machine tool. The adopted method comprises the steps that firstly: a machine tool manufacturing system processing energy consumption and processing time mathematical model is constructed; secondly: corresponding constraint conditions are set according to the actual processing situation; thirdly: a multi-target model is solved by adopting the improved particle swarm algorithm; and fourthly: a solution set is processed by adopting an analytic hierarchy process in one solved solution set so that one optimal solution is objectively selected. Cutting speed, cutting depth, feed amount and other parameters of processing are optimized through the swarm intelligent algorithm so that the parameters are enabled to be optimal and the objective of energy saving can be realized.

The invention discloses a recurrent neural network short-term power load prediction method for improving a whale algorithm, and relates to the technical field of short-term power load prediction. A recurrent neural network is used for short-term power load prediction, similar daily load data of a day to be predicted is used as input data of the recurrent neural network, and the number of input neurons, the number of output neurons, the number of hidden layers, the learning rate and the gradient descent algorithm of the recurrent neural network are determined. And a prediction model of the recurrent neural network is constructed. And the whale optimization algorithm is improved by using a differential evolution algorithm, so that the high-dimensional global optimization capability of a common whale algorithm is improved. An improved whale algorithm is adopted to pre-train the weight in the recurrent neural network, after pre-training is finished, the trained weight is put into a recurrent neural network model, then a gradient descent algorithm is adopted to train the recurrent neural network model, and after training is finished, a neural network model with the fixed weight is obtained, and then load prediction is carried out.

The invention discloses a fiber optic gyroscope temperature drift error compensation method based on an optimized least square-support vector machine (LS-SVM). The method comprises the following specific steps: (1) initializing an artificial fish swarm algorithm (AFSA) optimization algorithm and LS-SVM model parameters; (2) determining sample data for model training and testing according to fiber optic gyroscope output data, and preprocessing the model input data; (3) training a LS-SVM model by using the training data, and continuously iterating the optimized model parameters through the AFSA algorithm; (4) predicting output of the fiber optic gyroscope according to the trained model, and performing temperature drift error compensation. According to the method, the optimized parameters under set conditions can be acquired according to the training sample by adopting the AFSA algorithm with high global optimization capacity, the condition that local optimization is possibly caused in the optimization process is avoided, and the prediction accuracy of the model can be improved to a certain degree.

The invention belongs to the technical field of unmanned driving, and discloses a target following and dynamic obstacle avoidance control method for a speed difference slip steering vehicle, and the method comprises the steps: building four neural networks through employing a depth determinacy strategy in reinforcement learning; constructing a cost range of the obstacle so as to determine a single-step reward function of the action; determining continuous action output through an actor-critic strategy, and updating network parameters continuously through gradient transmission; and training a network model for following and obstacle avoidance according to the current state. According to the method, the intelligence of vehicle following and obstacle avoidance is improved, and the method canbetter adapt to an unknown environment and well cope with other emergencies. the complexity of establishing a simulation environment in the reinforcement learning training process is reduced. By utilizing a neural network prediction model trained in advance, the position and posture of each step of the target vehicle and the obstacle can be obtained according to the initial position and posture ofthe target and the obstacle and the action value of each step, so that the simulation accuracy and efficiency are improved.

The invention relates to a temperature compensation method and system for a silicon micro-accelerometer based on an improved PSO (Particle Swarm Optimization) optimized neural network, and the temperature compensation method and the temperature compensation system are designed for improving temperature compensation precision. The method comprises the following steps: acquiring a training sample ofPSO optimization and a BP (Back Propagation) neutral network; constructing the BP neutral network on the basis of the training sample; using an optimal extreme point optimized by an adaptive weight PSO as an initial weight value and a threshold value of a BP neutral network model; introducing mutation operation into a PSO algorithm, updating the particles, then reinitializing the particles at a certain probability and expanding a population search space which is continuously reduced in iteration by the mutation operation; establishing the BP neutral network by calling the parameters, realizing real-time temperature compensation of the silicon micro-accelerometer and outputting a compensation result. The temperature compensation method and the temperature compensation system disclosed by the invention have the advantages that the problems of solving of an optimal compensation result and temperature globality are solved, and finally improved compensation precision and global improvementof the silicon micro-accelerometer are realized.

The invention discloses a building load forecasting method and device based on an improved IHCMAC (Hyperball Cerebellar Model Articulation Controller) neural network model. The method comprises the steps of: simulating the actual operation of a building to obtain building cold/heat load data and influencing factor data; determining input variables of the model according to the degree of correlation between the influencing factors and the building cold/heat load; clustering the input variables according to a particle swarm-K mean clustering algorithm to obtain values of L clustering centers, i.e., model node values, and defining a Gaussian kernel function for each node; and updating the weights of the nodes via a weight training algorithm to obtain a building load forecasting value of the model. The method has the advantages of fast convergence, high learning precision and strong generalization ability, and can provide a decision basis for energy-saving optimization control of a building system.

The invention discloses a pressure guide wire temperature compensation method of an improved Particle Swarm Optimization neural network. The method includes the following main steps: collecting pressure guide wire output voltage and parameters related to the environment where a pressure guide wire is, and performing normalization processing on data; building a three-layer front feedback neural network model having an error back propagation capability; utilizing improved Particle Swarm Optimization (PSO) to optimize the weight and threshold value of the built neural network; training the neural network after the weight and threshold value are optimized; and utilizing the neural network model obtained by training to perform temperature compensation on pressure guide wire measured data. The pressure guide wire temperature compensation method of the improved PSO neural network utilizes the improved PSO neural network algorithm to build a pressure guide wire measurement inverse model, the trained model is high in compensation precision, generalization ability and stability, and the defects that a Back Propagation (BP) neural network is easy to fall into local optimum and a standard PSO BP neural network is easy to skip global optimum are overcome.

The invention discloses a hierarchical planning method for a power distribution network containing a distributed power supply, and belongs to the field of power distribution network planning of a power system. The implementation method comprises the following steps of: obtaining a target object; establishing a multi-objective optimization model by taking the annual minimum comprehensive investmentoperation cost of the power distribution network as an objective, converting the multi-objective optimization model into a hierarchical planning model, establishing an objective function by taking the annual minimum comprehensive investment operation cost of a line as an objective in upper layer planning, solving a line decision variable, obtaining an optimal grid structure, and transmitting theoptimal grid structure to a lower layer; On the basis of the upper-layer net rack, the lower-layer planning establishes an objective function with the minimum sum of the annual average investment construction and operation maintenance cost of the distributed power supply DG, the line network loss cost, the power purchase cost from the upper-level network and the environmental pollution treatment cost avoided by accessing the DG; And solving the upper and lower layer models by using a PSCO optimization algorithm to obtain a final grid structure and DG access position and capacity configuration.The method has lower annual comprehensive economic cost and more stable system voltage level, and the power supply reliability can be improved.

The invention discloses a method for selecting the optimal path of a power distribution and utilization communication network based on the genetic algorithm. The method is characterized by including the following steps that first, a unidirectional topology network is established; second, the population is initialized through binary coding; third, the number of iterations is set; fourth, selecting operation is performed; fifth, crossover operation is performed; sixth, mutation operation is performed; seventh, judgment is performed; eighth, the optimal path is output. On the basis of meeting the constraint conditions, the optimal path is efficiently found so that the search efficiency can be improved and the search complexity can be reduced.

The invention relates to an intelligent ship path planning method based on a fast search genetic algorithm. The method comprises the steps of (S1) performing rasterization on a test site electronic chart, obtaining obstacle points for the rasterized chart, obtaining starting point coordinates and target port coordinates, presetting a maximum number G of iterations, an initial temperature T0, an ending temperature Tf and an attenuation value a, and obtaining an initial path set pop0 and an inflection point of an unmanned ship, (S2) obtaining the inflection point spacing sum D of each path in the initial path set pop0, and (S3) obtaining an updated path set popm through G times of iterations by using crossover, variation, proportional selection and annealing optimization operations accordingto the inflection point spacing sum D, a target temperature T and a preset attenuation value a, performing temperature updating according to an attenuation coefficient a, and selecting a shortest path in the updated path set popm as an optimal path when T is smaller than Tf. The planned path of the invention has a small steering angle, the trajectory is smooth, an obstacle is actively avoided, the method is close to actual navigation application, the convergence speed is fast, and the problem that a traditional genetic algorithm is easy to fall into a local extremum is overcome.

The embodiment of the invention discloses a data classification prediction method and system based on an improved grey wolf optimizer. The method comprises the steps that historical data is acquired,and the acquired historical data is subjected to normalization processing and classification; the historical data obtained after normalization processing is used as training samples of a support vector machine, and the preset improved grey wolf optimizer is utilized to optimize a penalty coefficient and kernel width of the support vector machine; a prediction model is constructed according to theoptimized penalty coefficient and kernel width of the support vector machine; and to-be-predicted data is acquired, the to-be-predicted data is used as to-be-predicted samples to be imported into theprediction model, and classifications of the to-be-predicted data and a predicted value corresponding to each classification are obtained. Through the data classification prediction method and system,the problems that the grey wolf optimizer is trapped in a local optimal solution and is low in convergence speed can be solved, classification and prediction on problems in specific domains are realized, and decision precision is improved.

The invention belongs to the technical field of program control, and discloses a system and a method for efficiently controlling a robot. The system for efficiently controlling the robot specificallycomprises a binocular stereoscopic vision module, a path planning module, a remote operation module, an upper computer control module and a communication module, wherein the binocular stereoscopic vision module is used for acquiring an image by using a camera, and a three-dimensional coordinate of an object point is calculated; the path planning module is used for finding an optimal path which isfree from collision and arrives at the target object position for the robot in a specific environment; the remote operation module is used for improving the operation efficiency and operability of therobot by using a Teleoperation and Remote Control operation mode; the upper computer control module is used for packaging each module in a program through MFC programming; and the communication module is used for completing the two-way information exchange between an upper computer and a rear lower computer. According to the system for efficiently controlling the robot, the robustness and the fault-tolerant processing of the system are effectively improved, and the operability, the operation efficiency and the operation precision degree of the robot can be improved.

The invention discloses an examination system using intelligent test paper generation based on an improved ant colony algorithm. The examination system comprises the components of a user management module which is used for managing account number information and attribute information; a question database management module which is used for dividing the question database into a plurality of sub-question-databases according to the question type; a test paper generation module which is used for extracting questions according to a test paper generation strategy and generating the test paper; a test paper management module which is used for storing and managing the test paper; an online testing module which is used for calling the test paper, receiving a user input and performing real-time storage of question answering data; an online monitoring module which is used for searching examination arrangements and examinee examination state recording information; and a score analyzing module which is used for marking answering paper, outputting examination scores and performing statistics on examination scores. The examination system according to the invention has advantages of efficient and reasonable test paper generation, flexible examination form, and remarkable examination cost reduction.

The invention discloses a service request distribution method oriented to an edge computing environment. The method comprises the steps of (1) monitoring an available resource and a service request execution situation of each edge apparatus in a system in real time; (2) collecting service requests collected by all edge apparatuses in the system; (3) selecting an optimal edge server or a cloud server for all service requests by use of a GASD (Globally Aligned Spatial Distribution) algorithm; and (4) performing service request scheduling on all edge servers. Compared with the prior art, from twoaspects of distribution and scheduling of the service requests, the service response time is optimized to the utmost extent; meanwhile, a heuristic method is adopted, a temperature control mechanismin a simulated annealing method is introduced into a genetic algorithm, the convergence rate of the algorithm can be slowed down at the initial phase of the algorithm, thereby effectively avoiding local optimization; and the convergence rate is accelerated at the termination phase of the algorithm, thereby improving the efficiency of the algorithm.

The invention provides a remote sensing image change detection method and system based on mask classification, and the method comprises the steps: 1, the preprocessing of a multiband remote sensing image; 2, the single-band separation of the multiband remote sensing image after preprocessing; 3, the constructing of a single-band difference image of two-phase single-band remote sensing images; 4, the construction of a change mask difference image, and cutting the change mask difference image based on an Otsu method; 5, the multiplication fusion of a change mask with one of the two-phase single-band remote sensing images, thereby obtaining a change region; 6, the clustering of the change region through a fuzzy C mean value method; and 7, determining the change type of the change region through combination of the prior knowledge of the ground object type of the other one of the two-phase single-band remote sensing images according to the clustering results of the change region. The method and system are resistant to nose interference, can alleviate local optimum effectively, can consider the qualitative and quantitative description of an image change process, and is high in change detection precision and reliability.

The invention discloses a motor imagery EEG pattern recognition method based on the time-frequency parameter optimization of an artificial bee colony. The method comprises the steps of conducting the leads selection based on the linear decision rule, selecting time-domain and frequency-domain optimal parameters based on the artificial bee colony algorithm, extracting features based on the common spacial pattern algorithm, and finally classifying features based on the linear discriminant analysis algorithm. The result of the method shows that, a lead channel of larger inter-class distinction degree can be effectively selected based on the lead selection algorithm. At the same time, based on the time-frequency parameter optimization algorithm of the artificial bee colony, a time window and a frequency band of larger inter-class distinction degree can be automatically selected, so that a better classification effect is obtained. The method is capable of effectively recognizing different motor imagery modes. Compared with the traditional parameter manual selection method and the frequency-domain parameter automatic selection algorithm, global optimal parameters can be automatically searched in both time domain and frequency domain at the same time based on the above method. Therefore, the feature extraction and feature classification effect for motor imagery EEG signals is improved.

The present invention discloses a quick image registration method based on a visual remarkable area, which belongs to the field of image processing. The quick image registration method is used for matching an image A to be matched with a reference image B and respectively acquiring remarkable areas of A and B. The mass centers of three remarkable areas of A are vertexes of a triangle, and the triangle is used as a first characteristic triangle. Similarly a second characteristic triangle is acquired. The first characteristic triangle and the second characteristic triangle which are similar form a similar triangle pair, wherein the similarity of (a,b) is maximal. An affine transformation model for transforming from A to B is established based on (a,b), wherein the related initial matching parameters comprise horizontal translation amount, vertical translation amount, rotation angle and scale parameter. From the initial matching parameter, Powell searching for setting step length is performed for acquiring searching values. Each searching value is used for performing multiple registration tests on the A and B, and the searching value which corresponds with an optical registration test is the optimal matching parameter as a result. The optimal matching parameter is used for performing affine transformation on A for acquiring a final registration result.

The invention belongs to the radar communication technology field and particularly relates to a radar LFM composite waveform design method which has properties of Doppler tolerance, low interception probability characteristic and low relevant sidelobe characteristic and is for motion target information acquisition. The radar LFM composite waveform design method is in combination with a low relevant sidelobe waveform design method and an LFM noise waveform design concept, an LFM composite waveform mathematics model is constructed in a phase weight mode, and relevant sidelobe template vectors are introduced to construct a corresponding target function; phase constraint and constant modulus constraint conditions are analyzed, an iteration spectral approximation relaxed projection phase correction algorithm framework is constructed, LFM composite waveform optimization output programmed steps are provided, iteration spectral approximation relaxed projection phase correction constant modulus LFM composite waveform coding design concepts as described are employed, and the low relevant sidelobe characteristic and the low interception probability characteristic of waveforms can be relatively improved. The method further has properties of high efficiency, small time consumption and good robustness, and the method is more suitable for LFM composite waveform online design.

The invention provides a wave band correction change detection method of remote sensing image fuzzy clustering and a system thereof. The method comprises the following steps: step1, carrying out pretreatment of a multi-band remote sensing image, which means filtering and enhancement processing of the multi-band remote sensing image; step 2, carrying out single band separation on the multi-band remote sensing image after the pretreatment and acquiring a single-band remote sensing image; step 3, constructing a single band difference image of two time-phase single-band remote sensing images; step 4, constructing a multi-band combined image based on two time-phase multi-band remote sensing images after the pretreatment and acquiring a mutual neighborhood information content of each pixel of the multi-band combined image; step 5, correcting the mutual neighborhood information content of the pixels of the multi-band combined image and acquiring a correction image; step 6, using a fuzzy C-means method to carry out change detection on the correction image. By using the method and the system of the invention, detection precision is high, an anti-noise interference ability is good and an automation degree is high.