52 results about "Fireworks algorithm" patented technology

The invention discloses a method for image processing based on an improved fireworks algorithm, and relates to the field of image processing. The method comprises the following steps: initializing the positions of N fireworks grains, and recording the fireworks grains as initial grains G(old); calculating the position Sj of explosion spark generated by explosion of each fireworks grain, the explosion radius Aj of each fireworks grain and the position of mutation spark; selecting gains G(new) participating in next iteration from all the grains; determining optimal grains with best fitness values in all the gains included in both the initial grains G(old) and the gains G(new), and optimizing the grains except the optimal grains in the optimal value direction according to a bacterial foraging operator; and taking N grains with highest fitness as initial grains G(old) for next iteration from the optimized grains until an optimal solution is obtained. The improved fireworks algorithm can better overcome the poor result of the traditional fireworks algorithm in classification and regression, and effectively improves the result of optimal value point search.

The invention provides a fireworks algorithm on the basis of simulated annealing and Gauss disturbance. The fireworks algorithm has the advantages that simulated annealing algorithms and the fireworks algorithm are combined with one another, Gauss disturbance is carried out on fireworks with the poorest adaptive values, and accordingly an elitist which is better than the poorest fireworks individuals can be obtained; the probability of acceptance difference solutions is decreased along with gradual decrease of the temperatures, accordingly, the convergence performance of the fireworks algorithm can be improved, and the fireworks algorithm is obviously progressed in the aspects of convergence rates, computational accuracy and stability.

The invention relates to a power distribution network reconstruction method based on a fireworks algorithm. The power distribution network reconstruction method comprises the following steps: independently taking network losses, load balance and voltage decreasing amplitude as bases to define a target function; determining an integrative objective function of the safe operation of a power grid; adding a constraint condition for the integrative objective function; and adopting a firework explosion search mechanism to reconstruct the power distribution network and isolate faults when the load of the power distribution network is changed or the power distribution network fails. The method can quickly improve an optimization direction, effectively avoids acceleration and can easily obtain a global optimal solution.

The invention discloses a distribution network fault recovery method based on an A* algorithm and a fireworks algorithm. The method comprises the steps of (1) reading original data of a distribution network and simplifying the distribution network, (2) converting a contact switch into a virtual distribution power generation device DG and classifying the DG, (3) carrying out first stage algorithm,defining an initial node and a solution node, and using the A* algorithm to carry out power supply recovery on an important load, (4) carrying out second stage algorithm, reading distribution networkdata after first stage optimization, and using the fireworks algorithm to carry out power supply recovery on a rest load, and (5) integrating the optimal programs of the two stages, and determining afinal fault recovery program. According to the method, a staged method is adopted, a globally optimal solution is ensured, and the computing speed is fast.

The invention discloses a grey correlation analysis-based improved fireworks algorithm for carrying out optimization calculation on model parameters, which belongs to the technical field of mid-long term load prediction. The method comprises steps: 1, a GM (1,1) model is built; 2, through one cumulative iteration, a non-negative sequence is processed to generate a modeling sequence; 3, one cumulative sequence is used for building a first-order linear whitening differential equation; 4, a prediction result model is deduced; 5, through inverse accumulated generating operation, a prediction model Y <(0)> for X <(0)> is obtained; and 6, a grey model parameter solution method based on the improved fireworks algorithm is brought forward. The modeling process is practical and simple, the grey algorithm parameters can be quickly and effectively solved, and the mid-long term load prediction precision is improved.

The invention relates to a fireworks depth belief network-based face gender recognition method. The method includes the following steps that: an original image is preprocessed; a depth belief network is trained; the initial parameter space of the depth belief network is optimized through using a fireworks algorithm; and the fireworks algorithm-optimized depth belief network is adopted to carry out face gender recognition. According to the method of the invention, the feature of a face image, which has richer semantic information, can be leaned in an unsupervised stage through adopting the depth belief network; and the initial parameter space of the depth belief network is optimized through using the fireworks algorithm, so that a network model more suitable for a recognition task is obtained. The method of the invention has the advantages of strong anti-interference ability, global optimum, high recognition rate and the like, and can be used for face gender recognition as well as other image recognition and classification.

The invention relates to a maximum power point tracking method for a photovoltaic power generation system based on a fireworks algorithm. The maximum power point tracking method is characterized by comprising the following steps: (1) finding and tracking a maximum power point: generating a certain number of firework bullets in an output voltage search region based on a unique search mechanism of a fireworks blasting algorithm, and then performing blasting operations in positive and negative directions on each firework bullet, wherein fireworks generated by blasting execute local search within a certain neighborhood range of the original firework bullet (the blasting point), and each blasting includes a voltage size updating phase and a power information updating phase; (2) positioning the maximum power point: repeating the firework blasting process, ending the algorithm if the iterative frequency of actual running of the algorithm is greater than a set parameter T, and determining the last position of the fireworks is the maximum power point. The method cannot be affected by the change of an external environment, and the maximum power point can be found and tracked according to voltage, current and power information output by a photovoltaic array; by the adoption of a selection strategy of partial optimization maintaining and partial random selection, partial optimization in the search process can be effectively avoided, and the position of the maximum power point can be finally accurately positioned; therefore, the search efficiency is improved.

The invention discloses a microgrid short-term load forecasting method based on independent component analysis and support vector machine, which comprises the following steps: adopting Chebyshevv orthogonal polynomial to carry out curve fitting on historical daily load data of the microgrid region, and adding penalty function to carry out optimization; Updating the historical load data and using independent component analysis method to obtain the load independent separation source and mixed matrix of historical load data; For the meteorological data of historical day and forecast day, principal component analysis is used to obtain the principal component vector, and the decision data is combined with the date data. The mixed matrix of historical load data is used as the target data, combined with the historical day decision data, and the support vector machine (SVM) and fireworks algorithm are used to optimize the penalty factor and the kernel parameter, so as to obtain the optimal SVMtraining model. The forecasted daily load curve is obtained by substituting the forecasted daily decision data. The invention can remove noise interference, and can effectively and accurately predictthe load of the microgrid in a short time in the future.

The invention relates to an RFID-based robot indoor locating method. The RFID-based robot indoor locating method is characterized in that secondary development is performed on a signal acquisition system of an original RFID reader-writer, so that position coordinates and RSSI average values, corresponding to electronic tags, can be read, and accordingly, the technical defect that a conventional reader-writer can only read the basic information and signal instant intensity indicated values of the electronic tags is avoided. During design of the ground laying mode of the electronic tags, the strategy that the electronic tags are distributed in a crossed manner is adopted, that is, a longitudinal electronic tag is placed every equal distance in an experimental environment, a transverse electronic tag is placed in the middle of every four longitudinal electronic tags, and accordingly, the problem that the electronic tags are directional for receiving signals of the reader-writer is solved.The conventional fireworks algorithm is utilized for robot position calculation, and through analysis of experimental results, the improvement ensures that the accuracy of positioning by use of a maximum likelihood estimation locating method is greatly improved.

The invention relates to an optimization scheduling method of a tobacco shredding process, belonging to the technical field of intelligent optimization scheduling of a production workshop. The invention determines the tobacco shredding process scheduling model and the optimization target in the factory, and optimizes the target by using the optimization scheduling method based on the improved chaotic fireworks algorithm. The scheduling model is established according to the maximum completion time of tobacco raw materials processed on each piece of processing equipment, and the optimization objective is to minimize the maximum completion time. The invention can obtain an excellent solution to the tobacco silk processing process scheduling problem in a short time, thereby reducing the production cost of the factory, improving the production efficiency of the factory, enhancing the competitiveness of the enterprise, and effectively solving the problems of waste of the factory cost and loweconomic benefit caused by improper processing sequence in the tobacco silk processing process.

The invention discloses an adaptive cuckoo and fireworks hybrid algorithm-based pressure container optimization design method, and belongs to the field of improvement and application of intelligent optimization algorithms. On one hand, the search step length of cuckoo can be adjusted according to a gap between current and optimal environmental adaptation values of a bird nest, and a discovery probability of bird eggs is calculated according to a standard deviation of adaptation values of individuals, so that the search efficiency of a population is improved; and on the other hand, a ''segmented value-taking'' method is adopted for the explosion radius in a fireworks algorithm, the number of sparks is determined according to a search range of an explosion point, and the individuals can perceive a gap between the current explosion point and the optimal value, so that the fireworks algorithm can perform a jumping cross-regional search. Two sub-populations that evolve independently are fused with each other through a fixed number of generations, so that information communication among different individuals can be enhanced. The method combines the advantages of two intelligent algorithms and has a good optimization effect in pressure container optimization design.

The invention discloses a cloud platform energy consumption optimization management method based on a fireworks algorithm, and on the basis of the fireworks algorithm, load balance, performance and energy consumption of each device in a cloud computing platform are combined, so that the overall energy consumption of the cloud computing platform is reduced, and meanwhile, the computing capability of the cloud computing platform is ensured.

The invention discloses a fireworks-algorithm-based wireless sensor node deployment method. Sensor deployment is carried out based on a fireworks algorithm for solving an optimization problem by using simulation of firework explosion. The fireworks algorithm not only carries forward many advantages of the existing meta-heuristic algorithm but also has own explosive, transient, simple, local coverage, and distributed parallel characteristics. Moreover, individuals of the fireworks algorithm are independent of each other and each operator is capable of carrying out searching locally and independently, so that parallel processing is realized well and a problem of slow convergence on the condition of large node scale is solved. Meanwhile, compared with random distribution or other heuristic algorithms, the method enables the high coverage rate of the three-dimensional area to be realized. Because of full consideration of the landform and terrain, connectivity among all nodes of the network is enhanced on the premise that network effectiveness is guaranteed.

The invention discloses an improved substation engineering cost prediction method belonging to the technical field of substation engineering cost prediction. The method includes the steps of 1, selecting input index of substation engineering cost prediction and data preprocessing; 2, establishing a prediction model of optimized LSSVM base on that fireworks algorithm; 3, inputting the input index data set obtained in step 1 into the prediction model obtained in step 2 to obtain the prediction result. The invention optimizes the coefficient of the LSSVM model through the fireworks algorithm, andthe prediction result obtained is higher in prediction precision and stronger in robustness than the basic LSSVM model and the support vector machine model. The model provided by the invention can beapplied to the cost level of substation engineering projects of different regions, different voltage grades and different scales, and can obtain ideal prediction effect, and the model constructed bythe invention has strong adaptability and stability.

The invention discloses a multi-label classification method and a multi-label classification system based on a fireworks algorithm. The method specifically comprises the steps of acquiring a label ofa known training sample, wherein the training sample is provided with a plurality of training sample nodes; performing multi-label classification on a to-be-predicted sample according to the label ofthe training sample, and calculating an optimal characteristic weight value between the to-be-predicted sample and the training sample by use of the fireworks algorithm; calculating a weighted Euclidean distance between the to-be-predicted sample and the training sample according to the optimal characteristic weight value; acquiring K-nearest nodes nearest to the to-be-predicted sample from the plurality of training sample nodes according to the weighted Euclidean distance; and acquiring the label of the to-be-predicted sample according to the label of the k-nearest nodes. The optimal characteristic value in a classification algorithm is calculated by use of the fireworks algorithm, and thus the precision of the multi-label classification is improved.

The invention provides a visible light communication distributed light source position optimization method based on a fireworks algorithm. Through building a calculation model for a signal-to-noise ratio of receiving optical signals and a signal-to-noise ratio factor of a receiving plane with multiple points of LED light sources as transmission points in the visible light communication space, a certain number of LED light sources with the same power are placed in the communication space with a fixed length, a width and a height, a fireworks algorithm is used for finding an LED light source three-dimensional position with the most uniform signal-to-noise ratio distribution for optical signals received by the receiving plane, SNR fluctuation can be reduced effectively, an SNR mean value is improved, the signal-to-noise ratio distribution for the optical signals on the receiving plane is the most uniform, the optimal communication effects and the optimal lighting effects can be achieved, and users at different positions can be ensured to receive signals with the consistent communication quality in a maximal area mode.

The invention discloses a backbone network frame search method considering element comprehensive importance and network survivability. The method comprises the steps of introducing an economic factor based on conventional electric betweenness, and establishing an index for evaluating branch importance; according to node neighbor information and a clustering coefficient, establishing an index for evaluating node importance; based on the branch and node importance indexes, and by introducing a whole network survivability index, building a backbone network frame mathematic model considering element importance and network survivability; and based on a guide fireworks algorithm, implementing the model. According to the method, a line importance index, the node importance index and the network survivability index are combined; the backbone network frame mathematic model considering the element comprehensive importance and the network survivability is built; the model can assess the importance of lines and nodes in a power system more comprehensively, so that a most reasonable backbone network frame is obtained and the capability of a power grid in resisting a power failure accident is improved; and the method is suitable for construction of the backbone network frame of the power grid and reconstruction calculation of the power grid.

The invention provides a servo motor control method and device. The servo motor control method comprises: acquiring historical data of a running speed of a servo motor; carrying out offline identification on the historical data based on a particle swarm algorithm and constructing an object model of a servo drive system; designing an discrete time differential flat controller according to the object model to obtain a differential flat feedback control law; optimizing control parameters of the feedback control law by using a chaotic dynamic fireworks algorithm so as to obtain an optimized feedback control law; and according to the optimized feedback control law, controlling the running speed of the servo motor. Therefore, the rotating speed control of the servo motor is realized; the operation becomes simple; the system can quickly return to the stable state when being affected by the strong external disturbance; the adjustment time is reduced and the robustness and stability of the whole system are improved; and thus the control performance of the entire system is improved. The servo motor control method and device have the great significance in the actual industrial production.

The invention relates to an improved wolf pack algorithm based on firework explosion behaviors, belongs to the technical field of robot path planning, and aims to reduce the exploration pace of exploded fireworks close to a prey and improve the local exploration capability by introducing the exploration rule of the exploded fireworks of a fireworks algorithm, and increase the search radius of theprey far away from the prey and improve the global search capability. The moving step length of the running behavior is improved, the step length is automatically adjusted according to the position ofeach wolf, and the running behavior efficiency is improved. And the updating rule of the wolf pack algorithm is improved, and wolf pack individuals are selected and eliminated according to probabilities, so that the global optimization capability of the algorithm is enhanced. The method has the advantages of high convergence speed and high convergence precision.

The invention discloses a multi-target multi-robot task allocation method based on an improved fireworks algorithm. The method is used for solving the multi-target multi-robot task allocation problem.The improved firework algorithm is applied to the multi-target multi-robot task allocation problem. According to a Pareto superior theory and a specific solving target of multi-target multi-robot task allocation, a selection mechanism of a firework algorithm is improved, and before a multi-robot system starts to execute tasks, a plurality of task allocation schemes which meet a certain constraintcondition, have the best quality and are large enough in difference are found. The problems that an existing algorithm is small in search range and poor in solving quality can be effectively solved.Under the constraint condition that the task completion degree is met, the time and consumption can be minimized, and the task distribution quality is improved.

The invention discloses an internet spammer account identification method and system. The method includes the steps of obtaining relevant indicators of each user; determining a position vector of eachuser according to the relevant indicators of each user; assigning weights to the relevant indicators of each user to obtain a plurality of weight vectors; determining a fitness function according tothe dot product of the position vector and the weight vectors; using the fitness function to calculate an explosion radius; according to the explosion radius, using a fireworks algorithm to determinea spark which is an explosion spark or a Gaussian spark; using the fitness function to calculate the adaptive degree of the spark; determining whether or not the adaptive degree is greater than an adaptive degree threshold value, and if yes, determining that the user corresponding to the spark is a spammer account. The identification method and system can improve the recognition efficiency and search efficiency of network spammer accounts.

The invention discloses a parallel computing-based logistics transportation scheduling method, device and equipment and a readable storage medium. For logistics transportation scheduling, the fireworks algorithm is adopted to search the optimal path, on one hand, the fireworks algorithm searching capability is improved through the parallel operation strategy, on the other hand, the fireworks algorithm is optimized through the multi-group cooperative strategy and the novel evolution strategy, local convergence of the algorithm is effectively prevented from occurring too early, and the robustness and stability of the algorithm are improved. The scheme has the characteristics of high operation speed, high convergence capability and high optimization efficiency in the process of realizing logistics transportation scheduling.

The invention discloses a fireworks algorithm-based multi-target software and hardware division method. The method comprises the following steps of: randomly initializing N solutions satisfying a hardware area constraint condition, wherein the N solutions correspond to N fireworks in a fireworks algorithm; putting the N solutions into a set P, finding out non-dominated solutions in the set P in sequence according to a Pareto optimization theory, and setting non-dominated grades of the found non-dominated solutions to be 1, 2, and the like in sequence; calculating dispersibility of non-nominated solutions in a same non-dominated grade according to a congestion calculation method; sorting all the obtained non-dominated grades and the dispersible non-dominated solutions; calculating a numberof sparks generated by firework explosion and a firework explosion amplitude according to the sorting information; generating all the Gaussian sparks; sorting all the fireworks, explosion sparks and Gaussian sparks, and taking the obtained first N fireworks or explosion sparks or Gaussian sparks as the next generation of fireworks; and achieving a set iteration frequency. The method is capable ofimproving the operation speeds of systems and reducing the power consumption of the systems.

The invention discloses an effective community division method based on fireworks algorithm and local double rings. The method includes the steps of initializing node labels and algorithm parameters in a network; calculating the importance of nodes in the network and sorting according to the importance; initializing fireworks and calculating the number of sparks generated by the explosion of the fireworks and the scope of the explosion; generating sparks, and choosing fireworks or sparks to enter the next iteration; determining whether the termination condition of the algorithm is reached, andif an iterative termination condition is reached, outputting a community division result represented by the firework with the highest degree of modularity. Compared with the prior art, the method provides a concept of local double rings to preprocess the network for community division, can achieve non-overlapping community division of unweighted and undirected graphs, and has higher accuracy andbetter community structure division result.

The invention discloses a keyword recommendation method and system based on a search engine. The method includes the steps: acquiring input information inputted to the search engine by a user; extracting data information related to the input information from an engine database according to the input information of the user; acquiring the explosion frequency M of a fireworks algorithm; acquiring the explosion point of the K explosion of the fireworks algorithm according to N pieces of data information and the explosion frequency M; acquiring the explosion radius and the spark number of theK explosion of the fireworks algorithm according to the N pieces of data information and the explosion frequency M; performing explosion by the fireworks algorithm to obtain keywords related to the input information of the user; recommending the keywords to the user through the search engine for selection by the user. The data information related to the input information comprises the N piecesof data information. By the method and the system, keyword recommendation accuracy and efficiency are improved.

The invention provides a molecular conformation searching method based on a hybrid fireworks algorithm. The molecular conformation searching method is characterized by comprising the following steps of: S1, setting a docking area of receptor molecules, and expressing the docking area by using a docking box which is used for storing ligand conformations; S2, initializing a plurality of initial fireworks, wherein each firework represents a ligand conformation, expressing the ligand conformations as solution vectors, and setting a receptor-ligand binding affinity scoring function as a fitness function; S3, constructing a solution space according to the solution vectors, wherein the solution space comprises a plurality of layers; S4, constructing an operator of a fireworks algorithm; and S5, constructing the hybrid fireworks algorithm by combining the fireworks algorithm and a local search algorithm, and searching for an approximately optimal ligand conformation in the docking box by usingthe hybrid fireworks algorithm. According to the invention, the average time for docking on a test compound in molecular docking is reduced, and a molecular docking speed is increased; and meanwhile,the approximate optimal value of the fitness function can be found, and the precision of molecular docking is improved.

The invention discloses a low-carbon vaccine cold chain optimization and distribution method based on a multi-target firework algorithm. The method comprises the following steps of (1) reading the problem information; (2) initializing the algorithm parameters; (3) randomly generating an individual, decoding according to a decoding mode for eliminating vehicle constraints, calculating a target vector of the individual, and distinguishing a feasible solution and a non-feasible solution; (4) partially mapping an explosion operator; (5) operating a mutation operator; (6) realizing a co-evolution mechanism of double external archives; (7) updating the population; (8) after selecting the next generation of fireworks, randomly selecting an individual from the fireworks, and calculating the similarity between the individual and other fireworks; and (9) judging whether a termination condition is met or not, if yes, terminating the iteration, and outputting a feasible solution set. The method has better convergence and diversity, is accurate and stable in solution, has good expandability, and is suitable for solving the constraint multi-objective optimization problems, such as a low-carbon vaccine cold chain distribution problem, etc.

The invention discloses a multi-target scheduling method based on a fireworks algorithm and a genetic algorithm. The multi-target scheduling method is characterized by comprising the steps of setting initial parameters; generating an initial population, and starting iteration by taking the initial population as a current population; calculating a non-dominated solution of the current population, judging whether a new non-dominated solution is generated or not; and if yes, inputting the new non-dominated solution into the optimal solution set; if not, judging whether a set total number of iterations is reached, and if yes, outputting an optimal solution set; if not, calculating the firework scale to obtain a firework group; performing firework explosion operation and Gaussian mutation operation on the firework group; carrying out genetic selection; selecting fireworks; performing population crossover operation; performing population variation operation, and continuing iteration by taking the population subjected to the population variation operation as a current population; the method has the advantages that the genetic algorithm and the fireworks algorithm are combined, the convergence speed is high, the solving precision is high, and therefore the scheduling efficiency and the scheduling precision under the multi-variety small-batch production mode are effectively improved.