90results about How to "Improve exploration ability" patented technology

The invention discloses an efficient multi-objective optimization method for a satellite constellation and belongs to the field of constellation system of a spacecraft. The method comprises the stepsof determining an initial condition based on a Walker-delta constellation configuration, and establishing a constellation orbit kinetic equation, an earth coverage analysis module and an earth observation resolution model; optimizing an earth orbit height, an orbital inclination and an ascending node right ascension with a coverage percentage and a ground pixel resolution by use of a sequential radial basis function multi-objective optimization strategy; and constructing an objective function based on l2 weighting and an improved Pareto fitness function, replacing a high time-consuming constellation performance simulation model with an RBF (Radial Basis Function) agent model to optimize design, and updating and managing the RBF agent model through sequence sampling in an interest interval.Therefore, a Pareto non-inferior solution set satisfying engineering requirements is acquired as a satellite constellation design scheme, the coverage percentage of a constellation for a target observation region is realized as high as possible, the pixel resolution of an effective load is realized as low as possible, the calculation cost and the design cost of the satellite constellation are reduced, and the Pareto leading edge search capability is improved.

A CCHP distribute energy station cooperative optimal dispatching method includes such steps as 1, setting up plan operation model considering economic benefit according to operation characteristics ofeach part of equipment of CCHP distributed energy station; 2, synthetically considering that environmental cost and the economic operation cost, expressing the emission pollutant and the environmental carrying capacity by an economic method, and constructing an overall optimal dispatch model for the operation models of each part of the equipment of the CCHP distribute energy station in combination with the constraint conditions; Step 3: using the improved adaptive particle swarm optimization algorithm based on bee swarm optimization operator to solve the overall optimal scheduling model, andthen obtaining the optimal scheduling strategy of each part of equipment in the CCHP distributed energy station. Compared with the prior art, the invention has the advantages of accurate provision ofscheduling plan, easy use by engineers and technicians, high running speed of algorithm, high precision, strong global optimization ability and the like.

The invention discloses an electric power Internet of Things container migration method for edge network load balancing. The method comprises the steps of establishing a container information vector and a container deployment matrix of an edge network; carrying out load balancing detection on the edge network to obtain an overload edge node list; considering the resource utilization balance degree, the residual resource balance degree, the network transmission delay and the container migration downtime, and establishing a container migration model of the load balance joint migration cost; andbased on an improved ant colony system algorithm, performing container migration according to the container deployment matrix, the overload edge node list and the container migration model. The invention can effectively solve the problems of limited edge node resources, obvious business busy degree differentiation among edge nodes caused by obviously unbalanced business request space-time distribution of an edge network, and the like.

The embodiment of the invention provides a business data processing method and device based on a cloud platform. The method comprises the following steps: carrying out security detection of a submitted application object based on the cloud platform; when the application object passes the security detection, searching business data matched with the application object; carrying out security processing of the business data based on the cloud platform; and calling the application object in a designated container, and carrying out business processing of the business data which has been subjected to the security processing. The method and device provided by the invention has the advantages that the cloud platform has been authorized by cloud platform users in advance, so that the diversity of the business data can be greatly improved, and higher value of the business data can be achieved; and the cloud platform achieves big data operation capability through integration, and can provide a big data mining function for a third-party user, so that the capabilities of the third-party user for development, big data processing and big data exploration are improved, and effective products or analysis reports can be further produced.

The invention discloses a method for detecting complex network communities. In order to improve the global convergence performance of a differential evolution algorithm, the method comprises the stepof redesigning three main evolutionary operations: a classification-based self-adaptive mutation strategy, a dynamic self-adaptive parameter adjusting strategy and a history information-based selection operation; on the other hand, in order to make better use of network topology information, proposing an improved neighborhood information-based community adjusting strategy to ensure that sufficientsearch space is provided for global optimal community division while the DE (Differential Evolution) search space is reduced at the same time; and finally, proposing a new modularity optimizing algorithm CDEMO (C Differential Evolution for Multiobjective Optimization) based on the DE algorithm.

The invention discloses a manned space and lunar exploration spacecraft system scheme based on a lunar cycle revisiting orbit and an exploration method. A spacecraft system comprises four spacecrafts which are an earth lunar spaceport (ELS), a lunar exploration module (LEM), a crew transfer vehicle (CTV) and a cargo transfer vehicle (CaTV). With the cooperation of an astronaut system, a carrier rocket system and an exploration and control system, earth lunar space and lunar manned exploration are realized. A relatively entire exploration process of manned space and lunar exploration mission cycle is that the ELS is launched to an earth lunar cycle revisiting orbit by a carrier rocket to become an earth lunar spaceport which surrounds the earth, periodically revisits the moon, and chronically and stably moves on a large oval orbit of which the cycle is half of the lunar orbit cycle; then the lunar exploration module, the crew transfer vehicle and the cargo transfer vehicle are orderly launched to a cycle revisiting orbit by the carrier rocket for rendezvous and docking with the earth lunar spaceport; then the lunar exploration module is separated from the earth lunar spaceport to execute lunar manned field exploration; and finally the lunar exploration module returns to the earth lunar spaceport after finishing the lunar mission, and the crew transfer vehicle is separated from the earth lunar spaceport and then returns to the earth.

The invention discloses a reversible remote-control toy car. The reversible remote-control toy car comprises a remote controller, a car body, a front wheel set, a rear wheel set, and a power source part, a control circuit board, a motor and a transmission mechanism which are arranged in the car body, wherein the front wheel set and the rear wheel set are respectively arranged at the front end and the rear end of the car body; the power source part, the control circuit board and the motor form a control circuit; the power source part is used for supplying a working power supply for the whole control circuit; the motor is electrically connected with a corresponding output end of the control circuit board; the motor, the transmission mechanism and the rear wheel set are sequentially connected with one another in a transmission manner; the motor is arranged in the middle in the car body; a counterbalance body is arranged near to the rear end in the car body; and the surface of each wheel of the front wheel set and the rear wheel set is a rough surface with convex grains. The reversible remote-control toy car provided by the invention needs to test the flexible operation of a child on the remote controller and the flexible grasping of the child on inertia force of the car body, so that the exploratory property and interestingness of the car body are enhanced, and the hands-on operating ability of the child is developed.

The invention discloses an effective part extract of valeriana amurensis P.Smirn. and a quality control method and medical use thereof. The content of total lignanoid substances in the extract accounts for over 50 percent of the total weight of the extract, wherein (+) pinoresinol-4,4'-O-beta-D-diglucopyranoside and (+) pinoresinol-8-O-beta-D-glucopyranoside are the main active components in the effective part extract, and the sum of the contents of the two active components accounts for more than 13 percent of the total weight of the extract. The invention also discloses a preparation method and the quality control method of the effective part extract. The effective part extract contains one effective part group of total lignanoid components and can effectively treat central nervous system degenerative diseases including senile dementia, tristimania and the like. The process for preparing the effective part extracts of the valeriana amurensis P.Smirn. for resisting the senile dementia can adapt to industrial production, and provides a new method for treating the central nervous system degenerative diseases by using the valeriana amurensis P.Smirn..

The invention relates to a workshop resource scheduling method based on a heuristic optimization algorithm, which comprises the following production scheduling steps: receiving input data, and inputting the data into the algorithm; setting constraint conditions for screening output results; the method also includes that the algorithm calculates the received data and outputs a result; the algorithm supports input of various constraint conditions, and the constraint conditions comprise minimum waiting time, minimum overdue tasks, priority priority and forced guarantee priority, wherein the constraint condition is a digital quantity, the value range is 1-5, and the default value is 3. According to the method, submitted data are processed by using a genetic algorithm, and a group of locally optimal solutions are obtained under certain constraint conditions; the delivery time accuracy is improved, the consumed resources are reduced, the time required by an enterprise to specify a production plan is reduced, and the production efficiency of the enterprise is improved.

The invention discloses a network node selection method and system based on a whale optimization algorithm, and a storage medium. The method comprises the following steps: setting a binary coded population matrix, setting the maximum number of iterations and the initial number of iterations, and randomly initializing the population matrix; calculating the target function according to the node selection scheme corresponding to each whale individual in the population to obtain the optimal individual position and the optimal fitness function value in the population; calculating a current dynamicconvergence factor and a current dynamic weight according to the current number of iterations; determining a scheme for calculating the whale individual position according to the current dynamic convergence factor and the generated random number, and updating the current whale individual position and the number of iterations; if the number of iterations reaches the maximum number of iterations, returning to the optimal whale individual position, and determining sensor nodes participating in tracking according to the optimal whale individual position; otherwise return computation. According tothe invention, the tracking precision and the real-time performance in the target tracking process in the wireless sensor network can be improved.

The invention discloses a reactive power optimization control method of distribution network. The traditional reactive power optimization model is based on definite system conditions without considering the uncertain factors. The invention utilizes moth flame optimization algorithm to solve reactive power optimization control problem of distribution network, maps control variable from each moth toload flow date, and then calculates load flow to obtain transmission loss through matpower software; at each iteration, the update position of each moth is relative to the flame, and after the updateposition, the transmission loss of the corresponding moth is obtained; updated control variables are checked to see if they exceed the limit, and if they do, they are marked at the lower and upper limits for accurate results; 3) The evaluation process is repeated until it is terminated by the maximum iterative algebra restriction. The invention does not need many control parameters when solving the reactive power optimization dispatching problem, and the optimization method is simple and easy to implement.

The invention discloses a flexible workshop scheduling optimization method and system considering a crane transportation process. The method comprises the following steps: obtaining the parameters ofa flexible workshop, wherein the parameters comprise the number of machines in the target factory, the number of workpieces, the machining process corresponding to each workpiece, the machining machine corresponding to each process, the machining time of the workpieces and the position coordinates of the crane; constructing a flexible workshop scheduling model based on the parameters of the flexible workshop, wherein the flexible workshop scheduling model aims at minimizing the maximum completion time and the total energy consumption; and solving the flexible workshop scheduling model based ona mixed algorithm of distribution estimation and variable neighborhood search, and outputting a flexible workshop scheduling scheme after solving, wherein all individual solutions in the output solutions of the flexible workshop scheduling model are arranged according to the increasing sequence of fitness values. According to the invention, a hybrid algorithm of distribution estimation and variable neighborhood search is provided to solve the flexible workshop scheduling problem, so that the factory production efficiency is improved.

A protein conformation space optimization method based on differential evolution local disturbance is disclosed. Under the framework of a differential evolution algorithm, information exchange betweenindividuals in a population is used to enhance the exploration ability of the algorithm; and simultaneously, the differential evolution algorithm is used to achieve the fine tuning of a loop area andincrease the diversity of a loop area structure so that the exploration of the loop area is further enhanced based on an existing structure and then the exploration efficiency and the prediction precision of an integral body are increased. By using the protein conformation space optimization method based on the differential evolution local disturbance, prediction precision is high.

The embodiment of the invention discloses a battery SOH prediction method and device. The method comprises the following steps: extracting characteristic factors related to battery capacity decline from an original data set to construct a characteristic vector; obtaining a training sample set and a test sample set corresponding to the feature vectors according to the original data set; performing parameter optimization on the support vector regression model through an improved ant lion optimization algorithm to obtain an optimal parameter combination; taking the optimal parameter combination as a parameter of the support vector regression model, and training the support vector regression model through the training sample set; and predicting the test sample set by adopting the trained support vector regression model, and outputting an SOH prediction result. By adopting the technical scheme provided by the embodiment of the invention, the prediction precision of the SOH of the battery can be improved.

The invention discloses a Bayesian optimization method based on a sequential Monte Carlo method, and the method comprises the following steps: S1, building an objective function, and obtaining the posterior distribution of the objective function; S2, adding a noise signal, and calculating an edge Stuent-t distribution of the objective function; and S3, optimizing the objective function through a sequential Monte Carlo approximation method. Weight distribution is calculated through a sequential Monte Carlo method; the improved sequential Monte Carlo method is expanded to global optimization through Student-t process regression instead of Gaussian process regression, maximum value distribution can be more effectively obtained under the condition that the number of samples is small, and higher exploration capacity and abnormal value adaptability are achieved.

The invention discloses a three-dimensional group exploration method based on multi-head attention asynchronous reinforcement learning. The method comprises the following steps that 1, a command center host process sets a shared sample multiplexing cache and initializes a reference exploration strategy; 2, the command center starts a sub-process; 3, the command center adopts a pixel control algorithm to optimize an unmanned aerial vehicle exploration strategy based on the shared sample multiplexing cache; 4, the command center obtains the flight path of the unmanned aerial vehicle group based on the shared sample multiplexing cache by adopting a trust domain strategy algorithm; 5, the steps 2, 3 and 4 are repeatedly executed until the action track of the unmanned aerial vehicle group does not change any more; and 6, the command center sends an optimal trajectory transfer instruction to the unmanned aerial vehicle group. According to the method, the problem of low sample sampling efficiency of a reinforcement learning algorithm is solved, a better data acquisition effect is achieved by the algorithm when the same number of samples are used for learning, and an optimal track for maximizing data acquisition is further obtained.

The invention provides a training method and device for a reinforcement learning model in a battle game, belongs to the technical field of computers, and relates to artificial intelligence and computer vision technologies. The method comprises the steps that a target battle model and a similar opponent model of the target battle model are acquired, the similar opponent model is a historical battlemodel with the grade score difference between the similar opponent model and the target battle model smaller than a score threshold value, and the grade score is used for evaluating the battle ability of the model; based on the battle state characteristics of the two battle parties, the prediction operation of the target battle model and the prediction operation of the similar opponent model aredetermined respectively; the target battle model and the similar opponent model are used for controlling the two battle parties to execute prediction operation so as to conduct battle; an operation value of the target battle model in battle is determined; and the target battle model is trained based on the battle state characteristics, the prediction operation and the operation value.

The invention relates to a simulated annealing genetic algorithm based reactive power optimization method of an AC/DC system. The reactive power optimization method comprises the following steps of (1) building an AC/DC system model, and calculating the power flow of the AC/DC system, wherein the AC/DC system comprises an AC side and a DC side of which powers are transferred through a converter; and (2) building a reactive power optimization model according to a target function expressed by a usage cost value of network loss, and figuring out the reactive power optimization model according to a set constraint condition. The reactive power optimization method has the advantages that a simulated annealing genetic algorithm is introduced into reactive optimization and voltage control analysis of the AC/DC system; and through the combination of the characteristics of two algorithms, the respective advantage of the two algorithms is absorbed, the global optimal solution can be found with large probability, and meanwhile, the convergence rate is higher.

According to the MPPT control method for the photovoltaic power generation system, different algorithms are adopted to track the maximum power point of the photovoltaic power generation system according to different shielded conditions of a photovoltaic array; when the sunlight intensity changes slowly, local optimization is performed by adopting a particle swarm optimization algorithm; and when the sunlight intensity changes drastically, global search is carried out by adopting a self-adaptive radial motion optimization algorithm. When global search is carried out by adopting a self-adaptiveradial motion optimization algorithm, an optimization process is started by dispersing a plurality of particles in a predefined search space, the discrete particles are taken as an optimization solution, the particles move along the radius around the center at different speeds, the fitness value of each particle in the optimization process is calculated by utilizing a target function, and the position of the optimal particle is determined to obtain the maximum power point of the photovoltaic power generation system. According to the technical scheme, the method has higher accuracy and optimization capability, the memory required by the algorithm is smaller, the calculation speed is higher, and the method is suitable for large and complex search space.

The invention relates to a particle swarm positioning algorithm based on adaptive differential. The particle swarm positioning algorithm mainly comprises relation establishment between signal intensities and distances, least squares algorithm position computation, and a new search method, and includes receiving the distance information of an anchor node by using a unknown node, computing a position by using a least squares algorithm, generating a new swarm by using an improved adaptive differential algorithm, performing local search by using a particle swarm algorithm and a new mutation strategy, comparing with adaptive values and repeatedly performing iteration in order to achieve asymptotic convergence, and finally obtaining the position of the unknown node. The diversity of the swarms generated by the new differential algorithm is guaranteed and it increases rate of convergence and positioning precision to perform the local search by the combination of the particle swarm algorithm and the differential algorithm.

The invention discloses a manipulator autonomous obstacle avoidance planning method and system in a dynamic environment, and the method comprises the steps: carrying out the simplified modeling of a manipulator and an obstacle, and building a collision detection model of the manipulator and the obstacle; establishing a model-free deep learning algorithm model and carrying out collision training; according to a model-free deep learning algorithm model, an autonomous obstacle avoidance path of the manipulator is obtained, and an obstacle avoidance object is described as a reinforcement learning object by setting a collision negative reward, that is, an obstacle avoidance grabbing problem of the manipulator in a dynamic environment is described as a problem of searching a strategy for maximizing a total reward. In addition, a collision detection algorithm is designed for the two irregular entities of the manipulator and the complex obstacle, and the problem of collision detection between the two irregular entities is solved.

The invention discloses an artificial bee colony-based Monte Carlo localization method. In the Monte Carlo localization method, a behavior of simulating honey collection of bees is added for realizingthe localization. The method comprises the steps of firstly, initializing an initial particle sample set S1 of a certain quantity of particles in a given space; secondly, building a robot motion model, and forming a primary new particle sample set S2 according to the motion model based on all the particles in the initial particle sample set S1; thirdly, building an observation model, and taking the observation model as a fitness function of an artificial bee colony algorithm; fourthly, taking the primary new particle sample set S2 as an initial honey source position of an artificial bee colony, and simulating the honey collection behavior of the bees for performing global optimization; and finally, updating particle weights, and calculating out a robot pose. The exploration capability ofthe particles is effectively improved to avoid localization failure due to the fact that the particles fall into local optimum in complex environments or environments with similar structures; and while the particle diversity is ensured, the real position of a robot is quickly converged.