398 results about "Algorithm convergence" patented technology

The invention relates to the field of robot target path planning, and discloses a multi-target path planning method and device based on an improved mantis optimization algorithm, and the method comprises the steps: improving the mantis optimization algorithm, and building a grid environment map and a multi-target fitness function, an obstacle avoidance function and a smooth path function corresponding to path planning; the population position optimization corresponds to path nodes in the grid environment map, iteration is performed on all individuals of the population, and fitness values are formed; calculating comprehensive fitness by using a multi-target fitness function to obtain an updated global optimal position and a corresponding optimal fitness value; completing path planning of a multi-target fitness function; and observing the convergence curve graph and the shortest path graph, and verifying the feasibility of multi-target path planning. According to the method, the convergence speed of the algorithm is improved, the global and local search capabilities are balanced, the shortest path length is effectively reduced, path redundant nodes and path inflection points are reduced, and the path quality and efficiency are improved.

An adaptive variable step-size LMS filter based on a hyperbolic tangent function and its filtering method and computer equipment belong to the field of filter technology, and solve the problem that the LMS algorithm improved based on a nonlinear function requires too many manual adjustments of parameters and that the convergence speed and convergence accuracy of the traditional LMS algorithm cannot be satisfied simultaneously. The method of the present invention includes: determining the filter tap length and the number of iterations; initializing the filter parameters; obtaining the filter input signal vector; determining the instantaneous expected signal of the filter; obtaining the instantaneous output signal; obtaining the instantaneous error signal of the filter and the accumulated amount of the error signal based on the instantaneous output signal; obtaining the instantaneous step size using a step-size update formula based on a hyperbolic tangent function; and continuously iterating the filter tap weight coefficient vector to construct an LMS filter using the filter tap weight coefficient vector obtained after the number of iterations. The present invention is suitable for improving the performance of the LMS filter, and the filter of the present invention can be widely used in various signal processing fields.

The invention relates to a network topology dynamic optimization method and equipment for large-scale power supply and distribution equipment networking. The method comprises the following steps: step S101, real-time data acquisition and state sensing; step S102, carrying out network topology modeling and performance evaluation; step S103, dynamic risk assessment and optimization target generation; s104, carrying out topological optimization decision making based on a feasibility maintenance type genetic algorithm; step S105, carrying out optimal strategy verification and seamless switching; selecting an optimal network topology reconstruction scheme from the Pareto optimal solution set according to a preset decision strategy; and after the reconstruction scheme is verified on a control level, generating an equipment cascade and open circuit control instruction sequence, and guiding related nodes to complete undisturbed switching of the network topology in a preset time window through a distributed cooperative control mechanism. According to the method, generation of invalid solutions can be avoided, and the convergence efficiency of a large-scale network topology optimization algorithm is remarkably improved.

The invention discloses a multi-AGV task scheduling method and device based on a multi-strategy whale optimization algorithm combined with reinforcement learning. The method comprises the following steps: S1, establishing a mathematical model for multi-AGV task scheduling; s2, taking the balance between the path cost and the AGV vacancy rate as an optimization index, and constructing an AGV scheduling optimization model; s3, combining a multi-strategy whale optimization algorithm with a reinforcement learning method framework, and determining a Markov decision process including state, action, reward and value definition; and S4, solving the AGV scheduling optimization model by adopting a multi-strategy whale optimization algorithm combined with reinforcement learning. According to the method, the multi-strategy improved WOA is realized by initializing the population through chaotic mapping, adjusting the non-convergence factor and introducing reinforcement learning to update the whale individual position updating strategy, the calculation precision of the WOA is improved, the algorithm convergence is accelerated, the capacity of the WOA for jumping out of local optimum and searching for global optimum is enhanced, a feasible scheme is provided for multi-AGV task scheduling, and the method is suitable for large-scale popularization and application. And the method has certain advancement and feasibility.

The invention discloses an electric power tower point cloud registration method and device fused with normal vector clustering, and relates to the technical field of point cloud registration or other related fields, and the method comprises the steps: scanning an electric power tower to obtain point cloud data, dividing the point cloud data into different voxels, and carrying out the point cloud registration; calculating normal vectors of all the point cloud data in each voxel by adopting a principal component analysis algorithm, dividing all the point cloud data in each voxel into different clusters by adopting a clustering algorithm, determining representative point clouds of each cluster, selecting target point clouds and source point clouds from the representative point clouds, and carrying out point cloud clustering on the target point clouds and the source point clouds; the method comprises the steps of determining a matching point cloud of each source point cloud in target point clouds by adopting a binary tree search algorithm to obtain a plurality of matching pairs, calculating transformation parameters of the matching pairs by adopting an iterative nearest point algorithm, calculating a registration error based on the matching pairs and the transformation parameters, and merging the source point clouds and the target point clouds under the condition that the registration error reaches a preset convergence error range. According to the invention, the technical problems of low point cloud registration precision and poor algorithm convergence in the prior art are solved.

The invention discloses an intelligent optimization control method for a wave power generation device based on an improved MPC, and the method comprises the steps: collecting the wave height, period, flow velocity and device motion posture data in real time through a multi-source sensing module, constructing an exciting force prediction model based on a bidirectional long-short-term memory network, and achieving the precise prediction of the wave exciting force of n control periods in the future; prediction data and real-time feedback signals are input into an improved jellyfish search optimization algorithm, through a two-stage chaotic mapping perturbation mechanism, the convergence speed and the global optimization ability of the algorithm are remarkably improved, and finally optimal electric damping control parameters are output. Compared with a traditional MPC method, the method has the advantages that a feed-forward-feedback composite control framework is formed by fusing deep learning prediction and an intelligent optimization algorithm, so that the energy efficiency conversion efficiency of the power generation system is improved.

The invention relates to a ship route planning method, and belongs to the technical field of path planning, and the method comprises the steps: S101, obtaining ship navigation starting point, terminal point and navigation sea area information, and constructing a random tree based on the starting point, the navigation sea area information comprising navigable sea areas, non-navigable sea areas and wind field information; s102, selecting a sampling point from the navigation sea area information, and selecting an adjacent point closest to the sampling point from the random tree; s103, calculating resultant force of the end point, the sampling point, the unnavigable sea area and the wind field to the adjacent points, and expanding the random tree to generate new nodes based on the direction of the resultant force; and S104, judging whether the distance between the new node and the terminal point is smaller than a threshold value or not, if so, outputting an optimal route, and if not, returning to S102. By introducing the artificial potential field method, the expansion direction of the random tree is adjusted and controlled in real time, and the convergence speed of the algorithm and the quality of the optimal solution are improved.

The invention provides an improved particle swarm optimization (PSO) algorithm for mechanical arm trajectory planning, and aims at solving the problems that a traditional PSO algorithm is low in convergence speed and prone to falling into a local optimal solution. Therefore, the inertia weight and the learning factor of the algorithm are dynamically adjusted, so that the inertia weight and the learning factor change along with the increase of the number of iterations, and the global search capability and the convergence speed are enhanced; meanwhile, an elite reverse learning (EL) strategy is introduced, an optimal trajectory planning scheme is selected according to a fitness function, a new search area is explored through reverse particles, and the local search capability is enhanced; and a Gaussian-Cauchy variation (GC) strategy is combined, so that the global search capability is further improved, and a local optimal solution is avoided. In the optimization process, a single-target optimization method is adopted to carry out hierarchical classification on population solutions, and the performance of the algorithm under single-target optimization and constraint conditions is improved. Finally, the trajectory of the mechanical arm is optimized based on the algorithm, and particularly, each time period in a 3-5-3 polynomial interpolation method is finely optimized, so that the time from an initial point to a target point is shortened, a trajectory planning scheme with the optimal time is obtained, and the working efficiency of the mechanical arm is improved.

The invention discloses an intelligent production scheduling method for a long-period operation chemical device, which takes a real-time state of a production device and a predicted future health condition as one of important basis of production scheduling, combines a traditional equipment failure mechanism model with a GBDT algorithm, can predict an equipment health level at a future moment, and improves the production scheduling efficiency. Guiding the dynamic adjustment of the equipment load and the production scheduling; and then iterative optimization of the scheduling scheme is carried out by using an adaptive genetic algorithm, dynamic adaptive crossover probability and mutation probability adjustment is carried out on the scheduling scheme in the intermediate process, and the rate of convergence of the algorithm to a globally optimal solution is improved. From the aspect of application effects, the long-period operation characteristic of the large chemical device and the continuity and stability of the production process can be considered while the production efficiency and the cost are ensured, and a scientific and feasible technical implementation scheme is provided for the production scheduling of the large device of a chemical enterprise.

The present invention provides a method for preprocessing vibration signals of a hydropower unit. The present invention proposes five strategies to improve the optimization accuracy of a dung beetle algorithm so that it has good robustness. The improved dung beetle algorithm MSIDBO jumps out of the local optimum through PWLCM chaotic mapping, nonlinear boundary convergence factor, rolling ball behavior integrating dynamic golden sine, sparrow alert mechanism and parameter adaptive adjustment to improve the accuracy of the optimization problem. With the minimum envelope entropy as the fitness function, MSIDBO is used to optimize parameters such as the white noise amplitude weight Nstd and the number of noise additions NE of ICEEMDAN, and the optimized ICEEMDAN is used to perform modal decomposition of the vibration signal of the hydropower unit. The MSIDBO algorithm has a faster convergence speed, higher solution accuracy and better applicability than the standard dung beetle algorithm. Through simulation signal analysis, it is concluded that the decomposition accuracy of the optimized ICEEMDAN is higher than that of the conventional decomposition algorithm, which improves the accuracy of the preprocessing of the vibration signal of the hydropower unit and provides a basis for the intelligent vibration monitoring and diagnosis of the hydropower unit.

The invention discloses a deep neural network model parallel reasoning method for a heterogeneous GPU cluster, and relates to the field of distributed machine learning, and the method comprises the steps: obtaining current information, remaining selectable DNN models, deployed DNN models on each GPU server, and GPU servers which do not meet the number constraint of the DNN models; the scheduler selects a DNN model and deploys the DNN model on the selected GPU server, and calculates throughput for executing parallel reasoning at the moment; the combination of the DNN model and the GPU server with the maximum throughput is found, and related information is updated; judging whether the DNN models deployed on the GPU meet the number constraint or not, and updating GPU cluster information until all GPUs meet the specific DNN model number constraint; and repeating the steps until the algorithm converges. According to the method, limited heterogeneous GPU resources are fully utilized, and the DNN model with high compatibility is selected to deploy and execute parallel reasoning, so that the throughput is maximized.

The invention discloses a multi-agent path planning algorithm based on an adaptive memetic algorithm, and aims to solve the problems that the multi-agent path planning algorithm is not mature enough and the convergence speed of the algorithm is low. According to the optimization method, through an algorithm structure in which a global optimization algorithm is matched with local search, chromosomes are used as generation paths in a two-dimensional grid map, strategies such as'selection 'and'crossing' are carried out on the generation paths to optimize fitness values of the generation paths, and a hill climbing algorithm is used as a sub-algorithm to carry out local optimization on the generation paths, so that a generated solution is prevented from falling into local optimum, and the generation efficiency of the two-dimensional grid map is improved. Therefore, the multi-agent motion control precision is improved, and the convergence speed and optimization efficiency of the algorithm are enhanced. The problems that the algorithm convergence speed is low and the path optimization precision is low in the current multi-agent-based path planning field are solved. Experimental results show that the method can accurately realize multi-target-point path optimization for multiple agents, each path is converged to the global optimal path length along with the population iteration average length, and the algorithm test achieves an expected effect.

The invention discloses a machine room group control multi-objective optimization digital twin platform and an optimization method, and relates to the technical field of machine room group control. Constructing a multi-objective optimization model, wherein the multi-objective optimization model comprises a three-dimensional objective function and constraint conditions; the three-dimensional objective functions comprise an energy consumption minimization function, a temperature and humidity control precision maximization function and an equipment life loss minimization function. Compared with a traditional static twinborn model, the method has the advantages that the deviation is greatly reduced, and a precise virtual-real mapping basis is provided for optimization decision making. According to the whale optimization algorithm, the global search capability and the NSGA-IIPareto solution screening capability are fused, the convergence speed of the algorithm is improved through fuzzy membership degree fitness calculation and crowding degree screening, when deviation exceeds a threshold value, optimization can be shortened to 15 seconds, compared with an existing algorithm, the response speed is greatly improved, and the real-time control requirement under the dynamic load of a machine room is met.

The invention provides a thermally driven micro-nano compliant mechanism topological optimization method considering a scale effect. The method comprises the following steps: defining design conditions of a thermally driven micro-nano compliant mechanism, and setting material attribute indexes; establishing a finite element analysis model of the thermally driven micro-nano compliant mechanism; obtaining a structure displacement response; establishing a topological optimization model of the thermally driven micro-nano compliant mechanism; calculating and optimizing an objective function and constrained sensitivity information; performing smoothing processing on the sensitivity information; and solving the optimization problem of the thermally driven micro-nano compliant mechanism by adopting a moving progressive optimization algorithm, judging whether the convergence condition of the moving progressive algorithm is met or not, and if so, outputting the optimal topological configuration of the thermally driven micro-nano compliant mechanism considering the scale effect. The thermally-driven micro-nano compliant mechanism obtained through topological optimization has an obvious scale effect, the topological configuration of the thermally-driven micro-nano compliant mechanism is changed along with increasing of scale parameters related to the feature length, and it is indicated that the scale effect becomes more obvious.

The invention provides a multi-objective optimization intelligent stereoscopic warehouse scheduling method and system, a medium and computer equipment. The method comprises the steps that S1, attribute data of goods in a warehouse are collected in real time; s2, based on the attribute data, weights of a plurality of optimization targets are dynamically determined through a self-adaptive weight adjustment mechanism, and the optimization targets at least comprise the space utilization rate, the access efficiency and the goods shelf gravity center; s3, carrying out multi-objective optimization on the shelf layout by adopting a particle swarm optimization algorithm, and carrying out weighted calculation on an optimization objective by the particle swarm optimization algorithm according to the dynamically adjusted weight in the step S2 to generate an optimal shelf layout scheme; s4, feeding back the optimal shelf layout scheme to a warehouse management system for execution, continuously adjusting weights and optimizing strategies according to cargo attribute data collected in real time, and dynamically balancing the space utilization rate, the access efficiency, the shelf stability and other targets through the combination of an adaptive weight adjustment mechanism and a particle swarm optimization algorithm, so as to achieve the optimal shelf layout scheme. And meanwhile, the algorithm convergence speed and the system real-time response capability are improved.

The invention proposes a robot path planning algorithm based on a particle swarm optimization algorithm and a dynamic window method, and relates to the field of control theories and electronic information, and the method comprises the steps: introducing an inertia weight updating strategy based on a state factor, setting adaptive parameters and crossover and mutation operators to improve the global search capability, increase the population diversity, and improve the robot path planning precision. On-line self-adaptive updating of inertia weight and learning factors is realized on line in combination with Q-learning, gene combination modes are enriched through crossover operators, convergence and exploratory performance of the algorithm are improved, an obstacle avoidance strategy of a traditional DWA algorithm is improved, weight parameters of a dynamic window evaluation function are dynamically adjusted according to real-time information of a target and an obstacle, and an obstacle avoidance algorithm is established. According to the method, the global planning is adopted, the path points generated through global planning are adopted as temporary targets, fusion of MOQLCOPSO and the improved DWA algorithm is achieved, local optimum is effectively avoided, the planning efficiency and path safety are improved, and the method is suitable for mobile robot navigation under the complex three-dimensional terrain.

The invention discloses a super-multi-target vehicle path planning method based on deep reinforcement learning assisted evolution, and aims to dynamically guide a search direction through deep reinforcement learning for multi-target optimization and path planning problems. The core of the method comprises the following steps: 1) constructing a super-multi-target vehicle path problem mathematical model; 2) designing a deep Q network as a decision model, and adaptively selecting a high-potential search direction in iteration; 3) designing a global search strategy, and balancing global search and local optimization; 4) generating a filial generation and evaluating the fitness, and executing environment selection to update the population; and 5) calculating a reward value according to a population evolution effect, and optimizing DQN network parameters. According to the method, through a collaborative mechanism of reinforcement learning and an evolutionary algorithm, a search strategy is dynamically adjusted, the convergence speed and stability of the algorithm are effectively improved, the adaptability to a complex path planning problem is enhanced, and the method has remarkable advantages in solving efficiency, solution set quality and dynamic environment response.

The invention discloses a power distribution network harmonic suppression method and system, and the method comprises the steps: building an objective function with the minimum total harmonic distortion rate of nodes of a whole network, building a power flow constraint, a harmonic distortion rate constraint and an active filter device related constraint, and obtaining a power distribution network harmonic suppression model according to the objective function and constraint conditions. A distributed partially observable Markov decision process is established based on the MADDPG algorithm, and harmonic suppression is performed on the power distribution network based on the distributed partially observable Markov decision process by using the PRE-combined MADDPG algorithm, so that the problem of low sample efficiency of a reinforcement learning algorithm is effectively solved, the stability of a training process and the convergence speed of the algorithm are promoted, and the robustness of the power distribution network is improved. According to the method, the effectiveness of the finally obtained harmonic suppression strategy is ensured, based on a centralized training-distributed execution architecture, it can be ensured that a plurality of active filter devices can make a cooperative control strategy giving consideration to the harmonic problem of the whole power distribution network in real time only by means of distributed observation, and therefore efficient whole network harmonic cooperative governance is achieved.

The invention discloses an energy storage frequency modulation control method and system for a super capacitor coupled lithium battery, and belongs to the technical field of energy storage and frequency modulation control, and the method comprises the steps: collecting data from an original frequency modulation instruction sequence, generating a substitution sequence through a search method, and generating a regular optimal substitution sequence through positive, negative and comprehensive optimization; training a GRU model by using the optimal substitution sequence, and optimizing hyper-parameters of a GRU neural network through an improved whale optimization algorithm; and inputting the optimal substitution sequence into the optimized GRU model to generate a preliminary prediction result, and correcting the prediction result in combination with the deviation between the prediction residual error and the original data to obtain a final prediction result. According to the method, prediction errors are remarkably reduced, the prediction precision of the frequency modulation control system is improved, the local optimum problem is effectively avoided, the globality of the optimization process is improved, and it is ensured that the generated optimal sequence has higher quality; the whale optimization algorithm is improved, the algorithm convergence efficiency is remarkably improved, and falling into a local optimal solution is avoided.

The invention provides an improved salmons swarm optimization algorithm, and aims to solve the problems that the global exploration capability of the algorithm is weak, and the algorithm is easy to fall into local optimum in the later stage, and the method comprises the following steps: firstly, solving a population with more uniform distribution as an initial population through adding Cubic chaotic mapping, increasing the diversity of salmons population, and improving the convergence speed of the algorithm; secondly, a spiral search strategy is added in the prey search stage, so that the salmons have multiple search paths to better adjust the positions of the salmons, and the global search performance of the algorithm is improved; and finally, a sparrow early warning mechanism is introduced, so that the rate of convergence of the Karat swarm algorithm is increased more quickly. According to the method, 12 basic test functions of a CEC2017 test set are used for testing the improved algorithm, and compared with path planning experiments of PSO, GWO, AWOA, GA and SCSO algorithms under a complex map, the optimal values of the shortest paths of the improved algorithm are reduced by 9.46%, 14.83%, 14.32%, 5.76% and 1.06% respectively. It is verified that the search efficiency of the algorithm is improved, the capability of avoiding falling into local optimum is high, and the convergence speed, the convergence precision and the optimization time are all superior.

In some implementations, a transmitter may determine that a difference between a total age of incorrect information (AoII) obtained in a current iteration of an iterative algorithm and a total AoII in a previous iteration of the iterative algorithm satisfies a tolerance factor for algorithm convergence. The transmitter may use the iterative algorithm to solve a convex optimization problem based on a plurality of iterations that continue until an objective function associated with AoII converges to within the tolerance factor for algorithm convergence. The transmitter may determine a precoder matrix and a vector of scheduling indicators from a solving of the convex optimization problem. The transmitter may transmit, using a downlink multi-user communication framework based on rate-splitting multiple access (RSMA) in a semantic-aware network, one or more updates to one or more receivers based on the precoder matrix and the vector of scheduling indicators.

The invention discloses a robot control optimization method based on deep reinforcement learning, and the method comprises the steps: firstly initializing a control system, constructing an experience playback buffer pool, and setting an Actor network strategy function, a Critic network weight, a hyper-parameter, and a target network parameter; then, the robot generates an action according to the Actor network, collects environment feedback after executing the action, and stores the state transition tuple into a buffer pool; next, a small batch of data is sampled, a time difference error is calculated, and a network parameter is updated accordingly to minimize a value estimation deviation, synchronously optimize the network parameter, maximize an expected value of a state-action value function, and improve control performance. According to the method, a multi-thread architecture and a deep reinforcement learning technology are fused, the real-time performance, stability and generalization ability of robot control are improved, the problem of algorithm convergence in a non-stationary environment is effectively solved, the cost of adjustment and calculation is reduced, and the robustness of long-term stable operation and deployment feasibility in a complex scene are enhanced.

The invention relates to a differential privacy distributed gradient tracking optimization method and system based on a state decomposition technology, and belongs to the technical field of information communication. According to the method, the state, related to privacy, of each node is decomposed into two sub-states, only one sub-state is exchanged with a neighbor, the other sub-state is not shared, and therefore potential attackers are prevented from inferring sensitive information. In addition, Laplacian noise is introduced into the exchanged sub-states and decision variables to achieve differential privacy. Through a state decomposition mechanism, the method can reduce the influence of Laplacian noise on algorithm convergence precision, and shows higher precision under the same privacy protection level. The algorithm is specially designed for a complex network scene, is suitable for applications needing privacy protection and complex network topology, such as a smart power grid and energy management, and shows wide applicability and innovation in distributed optimization.

The invention discloses an edge data transmission optimization method based on energy consumption minimization and load balancing, which belongs to the technical field of edge computing, is realized through a system consisting of a central control server, an edge node, a transmission link and an SDN (Software Defined Network) controller, and comprises the following steps: S1, network topology modeling and data collection; s2, link selection optimization is carried out; s3, topological optimization and dynamic adjustment are carried out; s4, executing and controlling data transmission; s5, performing adaptive dynamic adjustment; and S6, the steps S1 to S5 are executed in a circulating mode. According to the method, a multi-target reinforcement learning algorithm is adopted, and joint optimization of energy consumption minimization and load balancing is achieved in edge data transmission; transmission links are dynamically adjusted and redundant nodes and links are closed, so that the overall energy consumption of the system is reduced; through link classification and load balancing optimization, the data transmission efficiency and the network stability are improved; and a heuristic factor and a Chebyshev scaling function are introduced, so that the convergence speed of the algorithm is accelerated.

The invention discloses a satellite edge computing task scheduling method and system based on multi-agent deep reinforcement learning, and the method comprises the steps: S1, constructing a system model which comprises a user equipment model and a satellite edge computing server model; s2, constructing a target function, wherein the target is to minimize the total delay and total energy consumption of task processing of the user equipment and the satellite edge computing server; s3, modeling an original optimization problem into a decentralized partial observation Markov decision process, proposing an efficient cross-plot Transform multi-agent near-end strategy optimization algorithm, and training a task unloading and resource allocation model until the algorithm converges; and S4, the user equipment makes an appropriate task unloading and resource allocation decision based on the trained task unloading and resource allocation model according to the current satellite edge computing environment state and task requirements. According to the method, satellite movement and calculation and communication resource constraints of equipment in a satellite edge calculation environment are fully considered, joint optimization of task unloading and resource allocation strategies is realized through a multi-agent deep reinforcement learning algorithm, the total delay and total energy consumption of the system are effectively reduced, and meanwhile, the task completion rate is improved.

The invention provides a double-target medical waste transportation method based on an improved RSA-VNS, and relates to the technical field of path optimization. The method comprises the following steps: firstly, acquiring a medical waste transportation task and vehicle scheduling resources; secondly, on the basis of medical waste transportation tasks and vehicle scheduling resources, constructing a dual-target medical waste transportation model by taking minimization of cost and storage risk as optimization targets; and finally, solving the dual-target medical waste transportation model by adopting an improved RSA-VNS algorithm, and outputting a Pareto external file so as to decode and obtain an optimal medical waste transportation scheme. The storage risk of the medical waste is effectively quantified by integrating the service starting time of the medical institution, the type and strength of the medical waste and the possibility of disease transmission, and modeling is more in line with the characteristics of recovery and transportation of the medical waste. In addition, the proposed hybrid algorithm effectively improves the convergence speed of the algorithm and the diversity of the solution, and also enhances the local convergence capability at the end of the algorithm.

The invention relates to a cascade reservoir water storage period combined power generation optimization scheduling method and system and a storage medium. The method comprises the following steps: collecting basin hydrological data information; basic calculation parameters are determined; determining a cascade reservoir water storage period power generation optimization scheduling target, setting a cascade reservoir water storage scheduling constraint condition, and establishing a reservoir water storage period scheduling model; obtaining an initial feasible solution; based on the reservoir number M and the discrete number F, constructing all representative possible state combinations with a small number by using a key sampling method, and enabling the state of each reservoir to meet a feasible region range; and calculating a target optimization function sub-problem by using a parallel dynamic programming algorithm until an optimization algorithm convergence condition is reached, and obtaining a final optimization water storage scheduling scheme. According to the method, an optimization result can be ensured to be close to a globally optimal solution, and compared with a traditional dynamic programming algorithm, the state variable search space can be effectively reduced, so that the storage amount and the calculation amount required by the algorithm are reduced.

The invention relates to an integrated energy optimization scheduling method of a hybrid adaptive quantum particle swarm. The invention provides an optimal scheduling scheme for a comprehensive energy system containing carbon capture and power-to-gas conversion. The method comprises the following steps: improving a quantum particle swarm algorithm: combining a contraction and expansion coefficient and a fitness value in the quantum particle swarm algorithm to improve the search capability of the algorithm; using a difference strategy to update the random position of the particle, so that the particle approaches the optimal position of the population; the Levy flight strategy is added into the particle position updating process, so that the population diversity in the later convergence period of the algorithm is improved. And scheduling optimization is performed on the integrated energy system by using the improved algorithm, so that the flexibility and the economical efficiency of the operation of the integrated energy system are improved.

The invention provides an intelligent stowage method in a container wharf ship cabin. The intelligent stowage method comprises the following steps that S1, information of a container to be assembled and pre-stowage information of a target ship are obtained; s2, based on the information of the container to be assembled and the pre-assembly information of the target ship, establishing a multi-target optimization function including a dynamic weight adjustment mechanism; s3, carrying out iterative solution through an improved Monte Carlo tree search algorithm; and S4, when a preset number of iterations is reached, outputting an optimal stowage scheme. Through the Monte Carlo search algorithm for dynamic weight adjustment, the defect of parameter solidification in multi-objective optimization of a traditional method is effectively overcome, the convergence speed of the algorithm is increased, and meanwhile, a local optimal trap is avoided. According to the multi-target cooperation mechanism, core indexes such as box turnover operation, box area conflicts and field bridge movement are brought into dynamic optimization, the effects of reducing the box turnover amount, reducing the box area conflicts and reducing invalid field bridge movement are achieved, and the operation cost can be obviously reduced.

The invention relates to the technical field of robot path planning, in particular to a navigation method and device of a mobile robot based on Transform-DQN, and can solve the problems that an existing algorithm is slow in algorithm convergence time, poor in path planning strategy performance and the like in a complex dynamic environment to a certain extent. According to the method, environment information and state information are obtained through multi-sensor fusion sensing environment; obtaining an expected action of the current mobile robot by using an optimal strategy obtained by a DQN algorithm; and controlling the movement of the mobile robot according to the current expected action. According to the technical scheme, the reward function considering multiple factors is set to interact with the mobile robot, so that the accuracy of the algorithm is improved; in the training process, an attenuation mode of an adjustable greedy factor is set, exploration and learning of the mobile robot in environments with different complexity degrees are balanced, a Transform model is introduced into an experience playback mechanism, the long-term dependency relationship between experiences is captured, the learning effect of the robot is enhanced, and the training efficiency is improved.