309 results about "Local search (optimization)" patented technology

A computer-implemented method for routing shipments according to criticality includes accessing an initial solution to an optimization problem of routing multiple shipments to multiple locations using multiple vehicles, the initial solution including multiple loads such that each shipment is routed within exactly one load and a global cost across all loads is minimized, the initial solution being generated independent of the criticality of the shipments. Into each of one or more critical loads in a current solution, one or more non-critical shipments are inserted that are within a neighborhood of the critical load, a critical load being a load containing at least one critical shipment. One or more local search operations are executed to improve the initial solution, the operations including at least one of: (a) splitting each of one or more selected critical loads in a current solution into two new critical loads; (b) for each of one or more selected critical load pairs in a current solution, move a sequence of stops from one critical load in the pair to the other critical load in the pair and / or swap two sequences of stops between the critical loads in the pair; and (c) for each of one or more selected critical loads in a current solution that are indirect critical loads having at least one in-transit stop, break up the indirect critical load into a plurality of new direct critical loads having no in-transit stops and execute operation (b) on each of one or more selected critical load pairs, each selected critical load pair including at least one new direct critical load.

The invention discloses a goods location distributing method of a mobile shelf warehousing system for a refrigeration house. The goods location distributing method includes the steps: distributing items with strong correlation to the same sorting roadway, thus reducing the possibility of repeatedly opening the sorting roadway; taking the similarity coefficient of the order item as the basis of thecorrelation; establishing a multi-target goods allocation optimization model by comprehensively considering item sorting frequency and goods shelf gravity center; solving by adopting an improved invasive weed algorithm to obtain an optimal storage position of goods; generating a part of initial population by adopting a greedy algorithm; setting a reasonable spatial diffusion operator; and finallyintroducing an evolutionary reversal operation of a genetic algorithm. According to the goods location distributing method, the global search capacity and the local search capacity are high, and theoptimization effect is remarkable, and the warehouse sorting efficiency and the shelf stability are effectively improved.

The invention discloses a two-stage particle swarm optimization algorithm including independent global search. The two-stage particle swarm optimization algorithm comprises the following steps: species initializing; adopting the chaotization method for initializing the positions X and the speeds V of particles; adopting the fitness function (fitness) to calculate the adaptive values of all current particles, and initializing the record optimal position (pbesti) of each particle and the global optimal position (gbest) of all the particles; carrying out the first stage iterative-global search; carrying out the second stage iterative-local search. The two-stage particle swarm optimization algorithm has the benefits that during each iteration of the first stage iterative-global search, one non-self particle is randomly selected from all the particles for learning, and the random selection guarantees that the species is prevented from tracking the specific particle and that the aggregate phenomenon is avoided; the second stage iterative-local search can quickly converge and can obtain solutions high in accuracy, the accuracy of the optimal solution is increased, and the prematurity defect is remarkably improved.

The invention discloses a multi-target path planning method for an unmanned cruise ship under a dynamic obstacle, and relates to the technical field of water quality sampling and path planning. The method comprises the steps that: an unmanned aerial vehicle collects an image of a lake surface environment, grid segmentation is carried out, and a starting point and a plurality of sampling points areset on a grid map; an improved grey wolf optimization algorithm is adopted to perform sequence optimization on the plurality of sampling points, and the sampling points with the optimal sequence aremarked on a map one by one; an optimal grid path is calculated between every two sampling points marked in the grid map by utilizing a D* Lite algorithm to obtain an optimal path from the starting point to the final sampling point; and finally, the autonomous cruise ship completes cruise along the optimal path. According to the invention, the convergence factor in the grey wolf optimization algorithm is improved, the global search capability and the local search capability of the grey wolf optimization algorithm are balanced, the convergence speed and the stability of the grey wolf optimization algorithm are improved, and the path planning of a plurality of target points in a dynamic unknown environment can be realized.

The invention belongs to the technical field of neural network optimization, and tarticularly relates to an RBF neural network optimization method based on an improved GWO algorithm. A grey wolf population is divided into two sub-populations by setting a threshold value, different search strategies are executed respectively, an improved GWO optimization algorithm is used for searching for optimalinitial parameters of an RBF neural network, a sea clutter prediction model of the RBF neural network is established, and sea clutters are predicted and suppressed. According to the invention, the fitness mean value of each generation of population is calculated; the fitness threshold value is dynamically set, the grey wolf with the fitness higher than the threshold value executes the strategy oflarge-range search, and otherwise, the grey wolf executes the strategy of small-range search, so that each generation of population has global search and local search capabilities, and the convergencerate of the GWO algorithm and the precision of later optimization are improved. The improved GWO optimization algorithm is used for optimizing the initial parameters of the RBF neural network, and the stability and precision of the network are further improved.

The invention discloses a real-time prediction system of a circulating fluidized bed household garbage burning boiler furnace outlet flue gas oxygen content and a method thereof. An integration modeling method of a support vector machine algorithm and a multi-population genetic particle swarm optimization algorithm is used. A rapid, economic and adaptive updating system and a method are constructed so as to predict a boiler furnace outlet flue gas oxygen content in real time, and tedious and complex mechanism modeling work is avoided. A nonlinear dynamical characteristic of a SVM algorithm, a generalization capability and a real-time prediction capability are used to represent a dynamical change characteristic of a flue gas oxygen content. A particle swarm optimization algorithm is used to optimize a SVM algorithm punishment parameter C and a nuclear parameter g so as to increase a generalization capability of a model. A genetic operator and a multi-population migration mechanism are introduced so as to accelerate a convergence speed of a particle swarm algorithm, increase a diversity of a particle swarm optimization algorithm solution, reduce a possibility for particle swarm algorithm optimization calculation to get into local optimum and increase a global search capability and a local search capability of the algorithm.

The invention discloses an aircraft assembly line operation scheduling method based on a genetic variable neighborhood algorithm. The method comprises the steps: firstly building a resource-limited aircraft assembly line operation scheduling model, and converting an operation scheduling problem in actual production into a mathematic model problem for optimization solution; secondly, considering tight-before and tight-after constraints, a resource constraint and a spatial constraint and constructing an aircraft assembly line subsection operation scheduling model by taking the minimization of anassembly operation total construction period as an optimization target; and finally, solving by adopting an improved genetic variable neighborhood algorithm. The invention designs a population initialization method combined with a priority rule so as to reduce a solution space. A variable neighborhood local search mode combined with an acceptance threshold is adopted, and three neighborhood structures considering a tight-before-tight-after relationship are constructed to ensure that a legal solution is generated in a search process, so that the search capability is improved, a traditional genetic algorithm is prevented from falling into local optimum, and an aircraft assembly line operation scheduling scheme obtained by the method can effectively shorten the total construction period of the assembly operation.

The invention relates to an eagle policy and adaptive NM simplex-based photovoltaic model parameter identification method. The method comprises the following steps of S1: obtaining a photovoltaic model by adopting a photovoltaic IV curve test instrument to perform an IV characteristic curve, and setting a range of photovoltaic model parameters according to the number of serial and parallel solar cells of a photovoltaic array; S2: performing a rough global search on the photovoltaic model parameters by adopting a swarm intelligent optimization algorithm, and obtaining a plurality of relatively good photovoltaic model parameter initial value vectors; S3: performing a further rough local search on a group of optimal photovoltaic model parameter vector accessories in the step S2 by adopting a plurality of Nelder-Mead simplexes; and S4: performing a further fine local search on optimal model parameter vectors in the step S3 by adopting a single adaptive Nelder-Mead simplex to obtain optimal photovoltaic model parameter vectors. According to the technical scheme, the precision, speed and reliability of photovoltaic model parameter identification can be remarkably improved.

The invention discloses a multi-target urban logistics distribution path planning method. The method comprises the following steps: decomposing a three-target vehicle path problem with a time window into a plurality of single-target sub-problems through a group of uniformly distributed weight vectors; initializing the sub-problems by adopting a heuristic strategy; generating a filial generation for the sub-problem by using an evolutionary operator, and designing a target-oriented neighborhood operator to be combined with a variable neighborhood descent algorithm to serve as a local search strategy so as to improve the solving quality of the sub-problem; updating the solution of the sub-problem by adopting a Chebyshev aggregation function; optimizing a non-dominated solution in the archivesby adopting an external archive strategy based on a sorting and congestion degree mechanism; and S3, repeating the steps S3 to S4 until the set maximum number of iterations is reached, and providinga group of feasible vehicle distribution schemes for multi-target urban logistics distribution. Compared with single-target optimization, the method can provide richer decision information for a decision maker, and considers the quality of the solution on the premise of ensuring the convergence and diversity of the algorithm.

The invention discloses a method for scheduling a machine part processing line by adopting the discrete quantum particle swarm optimization, comprising the following steps: reading in the machine part processing process operation time, initializing a particle swarm, calculating the adaptation value of each particle, updating the individual optimal position and the global optimal position of each particle, carrying out global search on the basis of the discrete quantum particle optimization, carrying out local search and drawing a machine part processing sequence Gantt chart according to a global optimal scheduling scheme. The method disclosed by the invention improves the limitation of the traditional quantum particle swarm optimization in the production scheduling field, overcomes the defects that the quantum particle swarm is easy to be subjected to local optimization and has the advantages of high optimizing precision and high optimizing speed. The method is used for scheduling the machine part processing line, can solve to obtain an optimal scheduling scheme in a shorter time and is easy and convenient to operate. The principle has wide range of application and can be popularized to the producing and processing field of the manufacturing industry, the process industry and the like.

The invention relates to a problem of combining a data mining algorithm and a random search algorithm, and proposes an SVM parameter optimization algorithm based on an improved Levy flying particle swarm algorithm. With the advent of the era of big data, how to improve the performance of data mining algorithms has become a research hotspot. When the parameters of SVM are optimized by the traditional PSO algorithm, it is easy to fall into the local extremum, and the later oscillation is serious. Levy flight combines short-range searching with occasional long-range walking, which can fundamentally overcome the problem that PSO algorithm is easy to fall into local extremum. Considering the serious oscillation in the later period of the algorithm, the improved particle position update formulaand the introduction of momentum term can reduce the oscillation of the algorithm. At the same time, considering that the value of the inertia weight trades off the local search and the global search,the adaptive inertia weight based on the distance between particles of the invention can improve the algorithm convergence speed, thereby improving the PSO algorithm searching ability, and further finding the optimal SVM classification model.

The invention discloses an identification method of a power object model in a steam turbine regulating system. The method comprises the following steps of introducing an adaptive factor and a simulated annealing Monte Carlo rule into a basic beetle antennae search algorithm to improve a local searching strategy to obtain an improved beetle antennae search algorithm; collecting actual operation data of the steam turbine regulating system of a thermal power plant, and selecting a section with obvious power change of data sections as sample data; and identifying the power object model through thesample data by using the improved beetle antennae search algorithm to obtain an identification result. Compared with the prior art, according to the method, the simulated annealing Monte Carlo rule is introduced to improve the basic beetle antennae search algorithm, so that the local search speed is higher, the local optimization can be effectively avoided, the global convergence speed is higher,the stability is better, the overall identification effect is better, and the method has important practical significance for improving the automatic control level of a thermal power unit.

The invention discloses a multi-target task scheduling method and system under a cloud computing system. The multi-target task scheduling method comprises the steps: taking the minimization of the maximum completion time, the minimization of the maximum equipment workload and the minimization of the total workload of all equipment as targets, and constructing task scheduling under the cloud computing system into a mixed workshop scheduling model; solving the hybrid workshop scheduling model by adopting a hybrid discrete artificial bee colony algorithm embedded with a disturbance structure to obtain a scheduling optimization scheme; and scheduling the tasks under the cloud computing system by utilizing the obtained scheduling optimization scheme. A hybrid discrete artificial bee colony algorithm is adopted, and the flexible task scheduling problem under a cloud computing system is optimized, and an HFS model is modeled; eight disturbance structures are embedded, so that the developmentcapability of the algorithm is enhanced; the adaptive disturbance structure balances the development and exploration capabilities, and the improved following bee mechanism has a deep mining function,so that the local search capability can be further enhanced; and the convergence capability of the algorithm can be improved by designing a good reconnaissance bee algorithm.

The invention discloses an orthogonal wavelet constant modulus blind equalization method based on an IWPA (Improved Wolf Pack optimization Algorithm). According to the invention, A CM (Complex Method) with high local searching capacity is embedded into a WPA with high global optimizing capacity; an update mechanism of a wolf pack is improved to obtain an excellent IWPA; the new method promotes optimizing capacity of the WPA; a reciprocal of a cost function in the constant modulus blind equalization method CMA is used as a fitness function of the IWPA; an input signal of the CMA is used as an input of the IWPA; a leader wolf position captured by utilizing the IWPA is used as an initial weight vector of the CMA; then signal correlation is reduced by a wavelet; signals are output in an equalizing mode by the CMA; and an excellent equalizing effect can be obtained. Compared with the prior art, the orthogonal wavelet constant modulus blind equalization method has the advantages that correlations between the signals and between the signals and noise can be reduced; an algorithm convergence speed is improved; a steady state error is reduced; balancing quality is improved; and the orthogonal wavelet constant modulus blind equalization method has a certain practical value.

The invention relates to the related technical field of tobacco processing process control, in particular to a performance evaluating method oriented to a PID control loop in the tobacco processing process. The performance evaluating method is characterized in that oriented to the PID control loop in the tobacco processing process, a blended genetic algorithm combining a genetic algorithm and a correction Newton method is proposed to optimize and calculate the minimum variance capable of being achieved by a PID controller, the situation that a traditional optimization algorithm is likely to fall into local optimization is avoided, and the local searching optimization capability of the algorithm is also improved; in addition, punishment of cost control is taken into consideration in a target function, and the changes of control action are also considered in calculation of the minimum variance, so that more objective and more efficient performance evaluating indexes are obtained, and the process control performance monitoring level is increased. By means of the method, equivalence analysis can be carried out on the series PID loop in the tobacco processing process, so that the application range of the method is effectively widened, the process control performance evaluating capability is improved, and the more objective and more efficient performance evaluating indexes are obtained.

The invention is applicable to the technical field of route planning, and provides a dynamic vehicle pickup-and-delivery route planning method and device based on meme calculation. The method comprises the steps of firstly performing code initialization, then performing route correction on N chromosomes, carrying out crossover and mutation operations in a genetic algorithm, and performing local search optimization according to a mode of the sorting priority on the basis of a pre-established sorting library; then screening out N chromosomes from the N new optimized chromosomes and the N original chromosomes to serve as N route planning schemes to be executed; selecting one route planning scheme from the N route planning schemes to serve one of the service points, extracting the service point from the N chromosomes after the service to serve as a served service point, judging whether all service points are served or not, and if not, returning the correction step until all of the servicepoints are served. The method provided by the invention can shorten the distance and improve the working capacity, and improves the local search ability.

The invention discloses multiple scent seeking robots and a dangerous gas leakage source positioning system and method for an underground comprehensive pipe gallery. Through a gas sensor array and a camera, the scent seeking robots timely monitor concentration of leaked gas, and timely collect the wind direction and wind speed through a wind speed sensor, a host controller automatically finds an optimal path by judging the direction of a gas leakage source, a camera survey method, obstacle avoidance information detected by an ultrasonic sensor, and finally, a main control plate drives a continuous current dynamo to make the scent seeking robots move autonomously. Positioning of a dangerous gas leakage source is realized by a cooperative method of multiple scent seeking robots based on a swarm intelligence optimization algorithm, in order to avoid the scent seeking robots leaving a smoke plume area which is searched and reduce the possibility of the scent seeking robots falling into thelocal optimum, a group of the scent seeking robots is constructed, through searching by a smoke plume concentration gradient method and a robot anti-collision mechanism, the problem of local search in the process of robot search is solved, and detection and positioning of the dangerous gas leakage source are realized.

The invention discloses a power system economic load distribution method based on an improved whale algorithm. The power system economic load distribution method comprises the steps of establishing a corresponding constraint condition expression according to the requirement of a power system economic load distribution problem in practice; establishing an objective function of the power system economic load distribution problem according to the constraint condition, and converting an actual application problem into a solution of a nonlinear programming problem; proposing an improved whale optimization algorithm, and performing optimization solution on the power system economic load distribution problem by utilizing the algorithm; and updating the position by adopting a self-adaptive weight strategy through the improved whale optimization algorithm, updating the position again by an individual through a random differential variation strategy, obtaining the final position before and after the change, and further obtaining the optimal result of economic load distribution of the power system. In order to improve the search capability of the whale optimization algorithm, an adaptive weight and differential variation strategy is introduced, so that the algorithm can perform global search in the early stage and perform accurate local search in the later stage.

The invention relates to the technical field of neural network optimization, in particular to an RBF neural network optimization method based on an improved Harlisia eagle algorithm, which optimizes RBF initial parameters through the improved Harlisia eagle algorithm and realizes accurate prediction and suppression of sea clutters. According to the invention, the coefficient r3 in the position updating formula in the exploration stage is correspondingly improved, and the balance between the global search capability and the local search capability of the algorithm is fully considered. A part of individuals with poor fitness is selected to carry out non-uniform variation and greedy selection. Besides, the E is segmented to ensure that global search is executed in the early stage of iteration of the algorithm, the global search capability can be kept under a certain probability in the later stage of iteration, and the possibility that a global optimal solution cannot be found due to search stagnation caused by falling into local optimum is reduced. The global search capability of the improved Harris eagle algorithm is enhanced, and the RBF network optimized by the improved Harris eagle algorithm provides more rising space for the sea clutter prediction precision.

The invention relates to the field of quantum communication, and provides a quantum key distribution parameter optimization method based on a random forest algorithm. Under the condition of limited data length, globally optimized parameters can significantly improve security code rate of an actual decoy state QKD system. A traditional local search algorithm can be used for searching optimal parameters, but generally consumes more time resources and computing resources, and cannot meet the requirements of real-time parameter optimization of a high-speed QKD system and optimal configuration of large quantum communication network resources. According to the method, the random forest model is trained in advance by utilizing the original data, and then the optimal parameter is directly predicted by utilizing the random forest model, so that the to-be-optimized transmission parameter can be rapidly and accurately predicted according to the current system configuration, and the parameter optimization process is greatly accelerated. Numerical simulation proves that compared with a traditional search algorithm, the method is lower in time cost, higher in prediction precision and good in application prospect in a high-speed QKD system and a future large-scale quantum communication network.

The invention provides a vehicle tailgate structure optimization method based on a two-step improved PSO (particle swarm optimization) algorithm. The method comprises the following steps: determiningdesign variables from a finite element model of a vehicle tailgate structure and calculating response of the model at each sample point, and establishing a high-precision surrogate model between eachworking condition response and the design variables; establishing a target function for the optimization of the vehicle tailgate structure, and using the two-step improved PSO algorithm, namely, a constrained PSO algorithm to perform global search and local search at a constrained boundary for optimization to obtain the optimized vehicle tailgate structure. The method has the advantages that the problem of dependence of the PSO algorithm optimizing capacity on adjustment of control parameters is solved, the optimizing capacity of the PSO algorithm in optimization of the vehicle tailgate structure with complex constraints can be improved effectively, and an optimized design result of the vehicle tailgate structure is improved remarkably.

The invention discloses an optimal dispatching method for a power station containing a coal-fired combined heat and power generation unit, and the method comprises the steps: taking the minimum coal consumption of a power station level as an optimization target, adjusting the coal consumption of each unit through optimizing the electric heating load of each unit, enabling the coal consumption of a power station to be the minimum, and achieving the optimization target; the optimization method comprises the following optimization steps: reading scheduling related information, establishing an objective function by adopting a variable working condition theory, establishing an equality constraint condition according to an external load of the power station, establishing an inequality constraint condition according to safe operation domain data of each coal-fired combined heat and power generation unit in the power station, and establishing an optimization scheduling model according to the objective function and the constraint condition; and solving the model by adopting a multi-starting-point local search algorithm to obtain a scheduling scheme of each coal-fired combined heat and power generation unit in the power station. The coal consumption of the power station containing the coal-fired combined heat and power generation unit can be remarkably reduced, the operation economy of the power station is improved, and pollutant emission is reduced. The optimal scheduling method is simple to execute and easy to implement.

The invention relates to a mixed firework particle swarm synergic algorithm for solving unmanned aerial vehicle constraint route planning. The mixed firework particle swarm synergic algorithm is usedfor solving the problem of unmanned aerial vehicle route planning. The method employs a mode of searching an optimal path by two groups in a parallel and independent way, one of the groups employs animproved firework algorithm to search, and the other group employs a particle swarm optimization algorithm. A rough region of an optimal solution is sought in the whole search space by explosion of firework and is provided for particles, a direction is guided for subsequent search of the particles, the particles are used for performing detailed local searching on the region during the iteration process, therefore, the two groups are combined to search, the optimal solution of the planning problem is further obtained. During the whole searching process, a safety path and an unsafety path are divided by setting safety class, the safety class is gradually improved during the iteration process again and again, the search range of the optimal path is further limited in the safety path, and thesafety of the planned path is ensured.

Provided is a traffic signal timing optimization method based on principal component analysis and a local search improvement orthogonality genetic algorithm. The algorithm is provided by analyzing internal relation among the genetic algorithm, image processing and mode recognition and can be used for solving various function optimization problems. By means of the algorithm, an improvement orthogonality cross operator based on principal component analysis is provided. The operator first conducts PCA projection on the population before cross, individual length is reduced during cross, orthogonal cross operation is implemented on the projection area, the projection is projected to the original space after cross, redundant individual number and calculation expenses caused by redundancy are reduced, algorithm convergence speed is further improved, and the local search strategy is further introduced. The algorithm is applied to single-crossing signal timing optimization. By means of testing comparison with the existing algorithm, the method improves algorithm generality and efficiency, effective timing time is acquired, and the number of the queuing vehicles in front of a crossing is reduced.

The invention discloses an iterative compilation optimization method based on program characteristic analysis. The method comprises the following steps that: S1: classifying compilation options in a compiler according to functions; S2: constructing a relational tree diagram; S3: carrying out dataset testing on a target program, and collecting and arranging testing results; S4: analyzing the testing results, and extracting the program characteristic and the calling relationship of the target program; S5: according to the program characteristics of the target program, finding a corresponding compilation option category, and forming an iterative search space; S6: in the iterative search space, taking the category as a basis to carry out local optimal search to obtain a global optimal solution; and S7: saving, and ending. By use of the method, program characteristic analysis is combined with a local search algorithm to fully consider an influence among all compilation options while the search space is greatly reduced. Compared with an existing method, the method disclosed by the invention can find an optimal compilation option sequence suitable for the target program.

The invention discloses an efficient particle swarm optimization method based on an RBF proxy model. The method comprises the following steps: initializing a population, sampling in a search space to generate NP individuals as an initial population, evaluating the individuals, and adding the individuals into a sample data set; pre-selection: constructing a global agent model, and performing pre-selection by using a particle swarm algorithm as an optimizer; performing local search: constructing a local agent model for neighborhoods of population individuals, and selecting by using a particle swarm algorithm as an optimizer to obtain better individuals of local search; performing updating: using a better individual guide population obtained by local search to perform speed updating and position updating, and selecting a part of individuals to update the population and a sample data set after sorting; judging whether a termination condition is met or not; extracting the guiding information through local search, so the convergence capability of the population is ensured under the limited fitness evaluation times, and the optimization efficiency of the particle swarm algorithm is improved; the method has the characteristic of accelerated population convergence speed.

The invention provides a three-dimensional flight path planning method for an unmanned aerial vehicle. The method comprises the following steps: constructing a three-dimensional terrain environment model, a flight path evaluation model and a flight path optimization constraint model; constructing a cell membrane structure; performing track search space discrete reduction and parameter initialization setting; searching an optimal track point; and performing track smoothing processing. According to the method, the global search capability of FA and the local search capability of ABC are combined, the overall search performance is enhanced, and the search probability and real-time performance of a three-dimensional flight path constructed by the method are remarkably improved; meanwhile, theinvention further provides a three-dimensional flight path planning device for the unmanned aerial vehicle, and the device comprises a cell membrane unit, a power supply unit, a clock unit, an EEPROMand an SRAM.

The invention discloses an electric vehicle distribution path optimization method supporting a charging and discharging strategy, and the method comprises the steps: firstly building a target functionwhich takes the fixed cost, driving cost, punishment cost and charging and discharging cost of a vehicle as optimization, and then according to the coordinates of a customer point and a charging station, performing path planning by adopting a hybrid algorithm of a genetic algorithm and local search to obtain an optimal path scheme; and finally, according to the optimal path scheme of the electricvehicle distribution path obtained in the step 2, in combination with the time-of-use electricity price, guiding the electricity market to make a charging and discharging decision for profit maximization in a power supply mode of battery exchange of the battery swap station. According to the distribution path optimization method, an improved genetic algorithm (GA-LS) combined with local search isdesigned, the influence of charging and discharging of the electric vehicle on a power grid is considered, and the distribution path optimization method has the advantages of being high in robust stability, high in iteration efficiency, high in solving quality and the like.

The invention discloses a short-term wind speed prediction method and system based on an improved whale algorithm optimized ELM. The method comprises the steps: (1) obtaining the time series of various historical meteorological data of a wind power plant within a preset time range, and carrying out the preprocessing of the data; (2) analyzing the influence of each collected meteorological factor on the wind speed, calculating the weight of the characteristic quantity through the correlation degree obtained by grey correlation analysis, and taking the characteristic quantity with high correlation degree as input; (3) determining a network structure of the extreme learning machine and selecting an activation function; (4) adding chaos initialization and hill-climbing local search into the basic whale optimization algorithm, and adding inertia weight for improvement; and (5) establishing an extreme learning machine algorithm model based on improved whale algorithm optimization. According to the method, the technical problem that the wind driven generator cannot generate power according to the ideal wind power curve due to the uncertainty of the wind speed is solved, so that the technical effect of improving the short-term wind speed accurate prediction precision is achieved, and the utilization of wind energy resources by a wind power plant is improved.

The invention discloses a software micro-service combination optimization method for an edge cloud system. The method resets a step length factor and an attraction degree formula, gives full play to the detection capability of the step length factor at the early stage of an algorithm, and balances the local search and global search; micro-services are classified by utilizing the characteristics ofthe micro-services, and the services with the same input and output are classified into the same class, so that the number of combinations is reduced, and all combination schemes can be enumerated completely; in a combination algorithm, residual tangent initialization and residual tangent disturbance are adopted to replace random initialization and random disturbance, initialization and updatingof a micro-service combination scheme are improved, and the diversity of a final combination scheme is greatly improved; meanwhile, service attribute values stored in a service registration center arehashed and stored as a hash root through the concept of a hash tree, and therefore it is guaranteed that all the service quality attribute values stored in the service registration center are safe and reliable.