2878 results about "Optimization problem" patented technology

The invention belongs to the technical field of smart home control, and particularly relates to a smart home generation strategy optimization system based on adaptive learning, and the system comprises the steps: collecting the operation history, sensor data and equipment use modes of a user in a smart home system, and determining an evaluation standard and an optimization direction; extracting features according to user behaviors, environment changes and equipment states, and constructing a feature matrix; selecting a reinforcement learning method for modeling according to an optimization problem, designing an adaptive learning mechanism, and enabling the model to be continuously adjusted according to real-time data in an operation process; according to the method, historical data is used for training the model, a reward mechanism in reinforcement learning is used for guiding the implementation of an optimization target, a control strategy adapting to the current environment and demand is generated based on model prediction, and the method has the effects of better meeting the use demand of a user for the smart home and improving the intelligent level of the system and the user experience.

The invention provides a novel power system multi-scale network construction energy storage planning optimization method based on source network load clustering. The method comprises the following steps: firstly, processing source network load data by adopting a multi-dimensional adaptive feature extraction and fusion method, and realizing accurate processing of the data through interval adaptive probability density estimation, adaptive kernel tensor decomposition and marginal information entropy maximization; then, a multi-level data clustering method is adopted, and accurate clustering is achieved through hierarchical distance measurement and boundary self-adaptive adjustment; then, a multi-scale energy storage modeling method is adopted, and accurate modeling of the energy storage system is achieved through the dynamic characteristic matrix and self-adaptive topology embedding; and finally, a multi-objective optimization solution method is adopted, and efficient solution of an optimization problem is realized through adaptive hierarchical decomposition and hybrid evolutionary optimization. According to the invention, the accuracy and reliability of energy storage planning can be obviously improved, and the system operation efficiency is improved.

The invention discloses a multi-agent cooperation method based on a large language model, and relates to the field of agent cooperation control, and the method comprises the steps: constructing a multi-agent system; establishing a multi-agent coordination resource allocation model; a task allocation problem in the resource allocation model is converted into a multi-objective optimization problem, and optimization objectives comprise the minimum task completion time and the optimal resource utilization rate; and solving a multi-objective optimization problem by using an improved particle swarm optimization (PSO) to obtain an optimal cooperation scheme among the master control agent, the task creation agent, the intention arrangement agent and the at least one professional agent. The improved particle swarm optimization (PSO) evaluates the uncertainty of each candidate solution through multiple times of random sampling, and when the uncertainty exceeds a threshold value, cross validation is performed through a multi-agent voting mechanism. For uncertainty accumulation of a multi-agent system under a complex cognitive task, the efficiency and stability of collaborative decision and task scheduling of the multi-agent system are improved.

The invention discloses a reinforcement learning method and system for source network load storage collaborative multi-scene optimization, and the method comprises the steps: constructing a power distribution network optimization model with the minimum cost, and converting a mixed integer nonlinear problem into a solvable mixed integer second-order cone optimization problem; converting a mixed integer second-order cone optimization problem into a reinforcement learning decision model, and performing multi-round assignment on boundary condition parameters by using different operation scene data to form a multi-scene training task set; designing a reinforcement learning decision model multi-scene training loss function, combining the reinforcement learning decision model to construct a strategy neural network, an evaluation network neural network and a scene representation embedded neural network, and completing reinforcement learning decision adaptive to multi-scene optimization; and training a power distribution network multi-scene optimization decision model based on the multi-scene training task set, the multi-scene training reinforcement learning loss function and the neural network structure, and deploying the power distribution network multi-scene optimization decision model to an actual system to complete reinforcement learning decision model application oriented to source network load storage collaborative multi-scene optimization.

The invention discloses a multi-aircraft cooperative formation flight path planning method based on a leader and follower model, and relates to the technical field of environment perception and unmanned aerial vehicle cluster cooperation. The method comprises the following steps of: firstly, designing a multi-agent double delay depth deterministic strategy gradient (LFMATD3) based on a leader-follower model, and converting an optimization problem model into a Markov decision process model by introducing an artificial potential field model and a reward function; and secondly, constructing an independent agent for each unmanned aerial vehicle, optimizing a behavior strategy of the unmanned aerial vehicle by combining a reward function based on an algorithm framework of deep reinforcement learning, and performing flight path planning and realizing dynamic formation control. The method provided by the invention can effectively improve the formation stability and collaboration of the unmanned aerial vehicles in a complex environment.

The invention relates to the technical field of simulation model optimization, and discloses an aero-engine model Bayesian optimization method for quantizing uncertainty, and the method comprises the steps: building a probability mapping relation from a component index to an output response through constructing a Bayesian neural network agent model based on a probability weight coefficient; and by taking the difference between the output response and the corresponding complete machine test data as a multi-objective loss function and taking the minimization of the multi-objective loss function as an optimization objective, optimizing the component indexes by adopting a Bayesian optimization method based on a Gaussian process to obtain an optimal component index combination. Not only is a nonlinear relationship between high-dimensional parameters and simulation-test deviation accurately modeled through a neural network, but also efficient search of a parameter space is realized through a Gaussian process. The technical problems that when a traditional optimization method is used for processing the high-dimensional, strong-nonlinearity and multi-parameter coupling complex optimization problem of the aero-engine, the calculation efficiency is low, local optimum is prone to occurring, and result uncertainty cannot be quantified are solved.

The invention discloses a multi-unmanned aerial vehicle cluster cooperative positioning method based on multi-sensor fusion, and the method comprises the following steps: S1, collecting and processing the multi-sensor data of unmanned aerial vehicles, compensating VIO accumulated drift through fusing ultra-wideband ranging information, guaranteeing the time-space consistency of the multi-sensor data through employing a layered time synchronization strategy, and carrying out the collection and processing of the multi-sensor data of the unmanned aerial vehicles; suppressing multipath interference of the UWB signal and eliminating an NLOS error by using adaptive Kalman filtering, and obtaining initial positioning data of the unmanned aerial vehicle; s2, acquiring key frames of multiple unmanned aerial vehicles based on a window sampling method, and performing key frame optimization and outlier processing to balance noise suppression and real-time performance; and S3, constructing a factor graph model of the key frame data based on a graph optimization theory, and obtaining optimized cluster position information by solving a nonlinear least square optimization problem. The method effectively improves the accuracy and stability of multi-unmanned aerial vehicle cluster cooperative positioning, and can be applied to the fields of unmanned aerial vehicle logistics distribution, environment monitoring and the like.

The invention discloses a reservoir multi-target optimization scheduling scheme optimization method and system based on multi-party coupling, and belongs to the field of reservoir scheduling. The system comprises the steps that firstly, boundary conditions of a simulation system are determined, and then a reservoir multi-target optimization scheduling model is constructed; and then solving the reservoir multi-objective optimization scheduling model by using a non-dominated sorting genetic algorithm of the reference points, verifying the effectiveness of the solving method by using a standard multi-objective test function, and evaluating and optimizing a reservoir scheduling scheme by using a method of combining a grey target theory, a cumulative foreground theory and a fuzzy mathematical method. According to the system, the multi-objective optimization problem among flood control, power generation and ecology of the reservoir is coordinated, decision support is provided for formulating a reservoir dispatching scheme of multi-objective optimization dispatching of the reservoir, the contradiction among flood control, power generation and downstream river ecology can be effectively coordinated, and the flood control and power generation benefits are increased on the basis of considering the downstream ecological flow demand as much as possible.

The invention discloses a Gaussian representation SLAM method based on dense matching priori and factor graph constraint, which comprises the steps of inputting a current image and a key frame image, outputting a point graph corresponding to the image through a pre-trained model, returning a matching condition of two frame image points and respective point cloud information, and obtaining a point-level matching result based on a point-level matching result. The method comprises the following steps: constructing a joint optimization problem of a current frame and a key frame by taking a luminosity consistency error and a geometric projection error as targets, performing joint estimation on a camera pose and a point cloud of the current frame, realizing high-precision pose solution, generating point diagram data after Gaussian scene representation and rasterized rendering processing, and transmitting the point diagram data to a rear end for global optimization. And the rear end receives the pose and point cloud data, executes loopback detection to identify repeated key frames, and performs Gaussian rendering through an optimized key frame image to complete global dense three-dimensional reconstruction. The method effectively solves the problem of track drift and scene inconsistency caused by lack of pose priori and global geometric constraints in an existing system.

The invention provides a self-adaptive constraint model active mismatch prediction control method based on voltage and current ripple resistance of a power converter in a hybrid energy storage system, and the method comprises the steps: building an island DC micro-grid system model, collecting the voltage and current information of a system, and designing a model prediction control frame of a power loop and a cost function of the model prediction control frame; designing a model predictive control framework of the current loop and a cost function of the model predictive control framework; designing an active model mismatch mechanism to enhance the response capability to dynamic change; determining an optimization problem with a normal value, and designing an adaptive constraint optimization algorithm to dynamically adjust constraint parameters and optimize control performance; and designing a steady-state and dynamic-mode trigger for adapting to the running state of the system. According to the invention, the trajectory tracking precision of the system under dynamic and steady-state working conditions can be significantly improved, current ripples caused by control delay can be effectively eliminated, damage to electrical equipment due to overlarge ripples can be avoided, and the service life of the equipment can be prolonged.

The invention discloses an industrial chain multi-objective optimization solving method and a computer program product, and belongs to the technical field of data processing method.The industrial chain multi-objective optimization solving method comprises the steps that a mathematical model is constructed, and multiple objective functions of an industrial chain are quantized; constructing a swarm intelligence bat optimization algorithm; parameters in the swarm intelligence bat optimization algorithm are initialized; executing a search process, updating the speed and position of each bat individual, and comprehensively evaluating the fitness value of each bat individual by using a plurality of objective functions; a Pareto dominating strategy is adopted to optimize the multiple objective functions, and a Pareto non-dominated solution set is screened out; using a congestion distance algorithm to calculate a congestion distance between solutions in a Pareto non-dominated solution set, and rejecting over-concentrated solutions to form an optimal solution; carrying out back propagation on the optimal solution and the fitness value, and optimizing a next iteration result; and outputting an optimal solution when an iteration termination condition is satisfied. According to the method, the solving precision, the convergence speed and the global search capability of the industrial chain optimization problem can be improved.

The invention provides a multi-agent system initialization method and device for a question and answer large model, and relates to the technical field of question and answer large language model multi-agent systems. The method comprises the following steps: disassembling a user question and answer query task into a plurality of sub-tasks by adopting a pre-constructed question and answer planning agent, generating an agent corresponding to each sub-task, carrying out standardization processing through a formatting agent to obtain a standardized agent corresponding to each sub-task, and carrying out evaluation through observing the agents to obtain feedback information. Generating a candidate agent set based on multiple iterations; a pre-training text encoder is adopted for encoding, and semantic representation and task representation of each agent are generated; task correlation and agent diversity are calculated, and a multi-objective optimization problem is constructed; and obtaining a Pareto optimal frontier team set by adopting a non-dominated sorting algorithm, and selecting an optimal agent team set by selecting agents. According to the invention, the cooperation efficiency of the multi-agent system can be improved.

The invention relates to the technical field of information, and discloses an agricultural sewage treatment control method and system based on the Internet of Things, and the method comprises the steps: obtaining sewage water quality data in different time periods, and carrying out the preprocessing of the data, and obtaining standard water quality data; extracting time sequence data of the water quality indexes from the water quality indexes, and obtaining smoothed time sequence data by using a time sequence analysis method; according to the smoothed time sequence data, through a trained long-short-term memory neural network model, obtaining prediction data of a future time sequence of the water quality index; according to the prediction data, constructing an objective function of an optimization problem so as to establish a multi-objective optimization model; solving the multi-objective optimization model by adopting a particle swarm optimization algorithm to obtain an optimal combination of process parameters; and converting the optimal combination into a control instruction, issuing the control instruction to execution equipment of the sewage treatment system, and controlling the operation state of the execution equipment. According to the method, the process parameters can be adjusted in time, and the efficiency and the effect of agricultural pollution treatment are improved.

A global navigation satellite system (GNSS) spoofing detection and classification technique is provided. An optimization problem is formulated at the baseband correlator domain by using an optimization algorithm such as the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm, for example. A model of correlator tap outputs of the intended received signal is created to form a dictionary of pre-computed waveform functions (e.g., triangle-like-shaped functions). Sparse signal processing can be leveraged to choose a decomposition of pre-computed waveform functions from the dictionary. The optimal solution of this minimization problem can discriminate the presence of a potential spoofing attack peak by observing the decomposition of two different code-phase values (authentic and spoofed) in a sparse vector output. A threshold can be used to mitigate false alarms. Furthermore, a variation of the minimization problem can be provided that enhances the dictionary to a higher resolution.

The invention discloses an optimal design method for a floating wind power-wave energy multi-energy complementary power generation platform, which comprises the following steps of: constructing an integrated and parameterized system model which is a fully-coupled and parameterized numerical model comprising all key components of the floating wind power-wave energy multi-energy complementary power generation platform; all key design parameters influencing the system performance are set as parameterized variables; establishing a multidisciplinary coupling dynamic simulation model; defining a multi-objective optimization problem including decision variables, objective functions and constraint conditions; the decision variable selects a group of core variables from the parameterized variables as optimization input; and combining the multi-objective optimization problem with a multidisciplinary coupling dynamic simulation model, executing a multi-objective optimization cycle, and generating and deciding a Pareto optimal solution set to obtain typical design schemes with different characteristics. According to the method, the global optimization design of the floating wind power-wave energy multi-energy complementary power generation platform can be realized.

A profit-aware Offloading Framework towards Prediction-assisted MEC Network Slicing is provided. Towards MEC network slicing, formulate the optimization problem of maximizing long-term ESP profits and decouple it into the sub-problems of Edge network Slicing (EnS) and Computation offloading Access (CoA). For the slicing sub-problem, use a gated recurrent neural network (GRNN) to accurately predict user requests in different regions, and then use the optimal partitioning of network slices with the predicted requests and expected demands; For the offloading sub-problem, incorporating results from slice partitioning, use an improved deep reinforcement learning with twin ritic-networks and delay mechanism, solving the Q-value overestimation and high variance for approximating the optimal offloading and resource allocation.

The invention discloses a single-bit direction of arrival estimation method and system with signal amplitude recovery, a terminal and a storage medium, and the method comprises the steps: obtaining each snapshot time signal from a signal source of a uniform linear array receiver, introducing a non-zero threshold value to carry out the single-bit quantization of each snapshot time signal, and obtaining a signal model; reconstructing the signal model, replacing a sign function of the reconstructed signal model with a smooth hyperbolic tangent function, and performing reconstruction to obtain a target signal model; based on the target signal model, constructing a single-bit DOA estimation optimization problem by using a truncated Frobenius norm and an L2, 0 norm, and converting the optimization problem into a target optimization problem by using a semi-quadratic technology; and solving a target optimization problem by using a near-end alternating minimization algorithm and a forward-backward splitting algorithm to obtain a signal source estimation direction, a signal source estimation number and a signal estimation amplitude. According to the method, the amplitude of the signal source is accurately estimated while the prior dependence on the number of the signal source is eliminated.

Lowest cost usage scheduling of an energy system during a time interval of interest is achieved by utilizing two components of learning and optimization. First, a number of learning approaches including Linear Least Square Regression (LLSR), Auto Regressive Integrated Moving Average (ARIMA), and Multi-Layer Perceptron Deep Learning (MLPDL) are used to forecast energy production and storage of the energy producing components of a given group of energy components as a function of historical production data and weather data collected from weather and geo reports. Then, an optimization problem is formulated to create the optimal schedule of energy usage during time intervals of interests for the given group of energy components subject to scheduling and equipment constraints. Two alternative solutions, namely Simplex Branch and Bound (SBB) and Interior Point Branch and Bound (IPBB) are presented to solve the optimization problem. Accordingly, integrated iterative methods, programs, and systems are described aiming at minimizing the cost of energy consumption for the given group of energy components within time intervals of interest.

The invention discloses a stability analysis method for a hybrid grid-connected system of a grid-following converter, and the method comprises the steps: constructing precise impedance models of a grid-following type (GFL) converter and a grid-constructing type (GFM) converter respectively based on a phase-locked loop (PLL) mechanism and a virtual synchronous generator (VSG) control strategy, and the details are shown in an abstract figure 1; deducing the total equivalent impedance of the system at a point of common coupling (PCC) according to a circuit equivalence principle; performing stability analysis and sensitivity evaluation on different converter combination proportions, topological structures and key control parameters of the hybrid grid-connected system based on an impedance ratio criterion; and verifying the accuracy of the analysis result through frequency domain simulation. The method can be widely applied to a new energy power system, solves the problem of system stability evaluation and optimization under the high-proportion renewable energy access background, and has the characteristics of high adaptability, accurate model and comprehensive analysis.

The invention provides a coal seam hydraulic fracturing parameter dynamic optimization method based on deep learning, and belongs to the technical field of coal seam parameter optimization based on computer data processing. The method comprises the following steps: firstly, acquiring a coal seam slice CT image with spatial continuity; secondly, designing a coal seam structure-physical parameter intelligent sensing model, and performing joint prediction on coal seam image semantic recognition and physical property estimation; then, performing coal seam three-dimensional structure reconstruction and regional layering modeling by utilizing a prediction output result, and performing finite element mesh division and material attribute accurate mapping on the basis to construct a coal seam three-dimensional finite element model; and finally, a coal seam hydraulic fracturing parameter multi-objective optimization problem is constructed, a coal seam physical information embedded agent model is designed, the optimization problem is dynamically solved in combination with a multi-objective evolutionary algorithm, and optimal hydraulic fracturing parameters are obtained. On the basis of deep learning, multi-source information, structural parameters and material attributes are fused, and intelligent development of coal seam hydraulic fracturing design is promoted.

The invention relates to a heterogeneous GPU cluster scheduling optimization method and system based on resource affinity, and belongs to the field of computer resource scheduling. Comprising the steps of 1, according to an affinity historical library of historical jobs, matching the historical jobs with relatively high similarity through job characteristics, and estimating resource requirements; 2, for a new job, receiving the job, performing probe testing on each type of GPU, and obtaining and recording job performance data; 3, establishing an affinity evaluation model for each job by using historical information or test data, evaluating affinity values of the jobs under different resource configurations, and providing data support for resource allocation decisions; and 4, entering a scheduling optimization period cycle, constructing an allocation matrix according to a target function of an optimization problem, and allocating corresponding resources to each job. According to the method, accurate resource matching and data support are achieved, the dynamic adaptive capacity is guaranteed, and the system expandability is enhanced.

The invention relates to a quantitative control method for a nonlinear system, computer equipment and a storage medium. The method comprises the following steps of: performing sector bounded processing on the output of a static logarithmic quantizer to establish an expression for quantizing control input; constructing a nonlinear system with disturbance, time lag and parameter uncertainty, and constructing a T-S fuzzy linear model of the nonlinear system according to the expression; constructing a target function according to the T-S fuzzy linear model, and constructing an optimization problem on an infinite time domain by taking a system state constraint, a control input constraint and a disturbance constraint as conditions so as to realize minimum quantization control input under maximum disturbance; and performing stability analysis on the T-S fuzzy linear model by constructing an LKF, and solving an optimization problem under the condition of meeting the stability analysis to obtain quantitative control input. By adopting the method, stable control on a nonlinear system can be effectively realized.

The invention discloses a near-field flux-sensing integrated Bayesian Cramer-Rao bound beamforming optimization method, which aims at minimizing the Bayesian Cramer-Rao bound of an extended target response matrix, and constraint conditions comprise the maximum transmitting power of a base station, the minimum signal to interference plus noise ratio requirement of a communication user and the minimum energy efficiency of a system. Auxiliary variables are introduced, an original optimization problem is converted into an equivalent full-digital beam forming matrix optimization problem, and iteration is carried out by adopting a sequential convex approximation method until convergence is carried out to obtain a local optimal solution; aiming at the local optimal solution, constructing an optimization problem by using a method of approaching an equivalent full-digital beam forming matrix by using a hybrid beam forming matrix; and an analog beam forming matrix and a digital beam forming matrix are obtained through an alternate optimization method. According to the method, the Bayesian Cramer-Rao bound minimum transmitting beam design problem of the near-field communication and sensing integrated system is researched, and sensing of an extended target can be realized under the condition that the communication energy efficiency of the system is met.

The invention discloses a dynamic optimization system and method for a compound essential oil ratio, and aims to solve the problems that the traditional technology depends on artificial experience, the optimization efficiency is low and the quality tracing is insufficient. According to the method, raw material molecular fingerprint data are collected in real time through a high-precision spectrum sensor array and a micro-fluidic chip, a production environment field domain is constructed in combination with a computational fluid mechanics model, a formula optimization problem is encoded into an Ising model by using a quantum annealing algorithm, and global search is realized through a quantum tunneling effect. The system integrates digital twin bodies, covers molecular dynamics simulation, phase equilibrium prediction and olfactory receptor activation probability simulation, and accurately maps a physical system state; a multi-modal data fusion network and a reinforcement learning framework are adopted to dynamically adjust formula parameters, and an NSGA-III algorithm is combined to optimize efficacy collaboration degree, cost and stability indexes. And the execution control module realizes nanoscale flow regulation through the piezoelectric micro-injection valve array, and realizes non-tampering evidence storage of formula parameters and quality data based on a block chain technology. Dynamic closed-loop optimization of the compound essential oil ratio is achieved, the production efficiency, the product consistency and the supply chain transparency are remarkably improved, and the method is suitable for the field of intelligent manufacturing of high-end essential oil.

The invention discloses a dynamic equipment management service system and method for deep learning. The method comprises the following steps: S1, collecting real-time operation data of equipment and preprocessing the real-time operation data; s2, constructing a long-short-term memory network model, and outputting comprehensive early warning indexes of the equipment; s3, constructing an optimization problem based on the comprehensive early warning index of the equipment, and setting an objective function and constraint conditions; s4, performing global search on the optimization problem by adopting an improved dragon fly algorithm, and generating a plurality of resource scheduling and maintenance candidate schemes; s5, performing local optimization on the candidate scheme by using a simulated annealing algorithm, and determining an optimal resource scheduling and maintenance decision scheme; and S6, feeding back the real-time prediction result and the optimal decision result to form closed-loop control. Through fusion of deep learning prediction, the improved dragon fly algorithm and the simulated annealing algorithm, real-time closed-loop control of accurate early warning of the equipment state and dynamic resource scheduling decision is realized, so that the failure rate is effectively reduced, and the equipment management efficiency is improved.

The invention belongs to the technical field of communication, and discloses a positioning method and system based on differential time delay and an alternating direction multiplier method, and the method comprises the steps: obtaining the position information and differential time delay of a receiver; performing joint estimation on the position of the target object and the position of the transmitter based on an alternating direction multiplier method; wherein during joint estimation, a first variable and a second variable are introduced, an original optimization problem is decomposed into two sub-problems related to the first variable, the second variable and a Lagrangian multiplier coefficient during iterative calculation in the joint estimation process, and in each iterative process, analytical solutions of the two sub-problems are solved respectively; and updating the first variable, the second variable and the Lagrange multiplier coefficient. According to the method, the accuracy of target object position estimation can be improved while the calculation complexity can be remarkably reduced.

The invention relates to a multi-driving-style high-risk automatic driving cut-in scene test method and system, and the method comprises the steps: collecting original driving track data, carrying out the clustering of driving styles, and generating a risk cut-in scene track cluster with consistent driving styles through employing a Cutin-TimeGAN network model in combination with physical feasibility constraints; constructing a cost function, calculating a cost function value for each risk cut-in scene trajectory cluster, and selecting a target cut-in reference trajectory as a target reference state vector; constructing a game framework, and combining a target reference state vector to construct a utility function according to the state vectors of the VUT and the test confrontation vehicle in the framework; constructing a risk confrontation utility function based on the predicted collision time; a game optimization problem is constructed and solved, an optimal interaction track of the test confrontation vehicle is obtained, and a closed-loop high-risk automatic driving scene switching test is carried out based on the optimal interaction track; the system is used for implementing the method. Compared with the prior art, the method has the advantages that testing of authenticity and high-risk confrontation of multiple driving styles is considered.

The invention is suitable for the technical field of mobile edge computing, and provides a hybrid energy WPT-MEC system-based task unloading and resource allocation method, which comprises the following steps of: establishing a system model; an optimization problem with the goal of minimizing the long-term average power grid energy consumption is drawn up and converted into a Markov decision process; implementing an action space reduction strategy, and executing data normalization processing; constructing an enhanced depth deterministic policy gradient (EDDPG) algorithm; and generating and outputting an optimal strategy according to a real-time system state by using the trained network. According to the method, the wireless power transmission (WPT) technology, the green energy harvesting technology and the mobile edge computing technology are combined, so that the energy use duration and the computing capacity of the wireless equipment are improved. The online decision-making mechanism based on the EDDPG algorithm can dynamically optimize task unloading and resource allocation strategies, adapts to a dynamic environment, reduces energy cost and environmental pollution, and provides reliable technical support for large-scale deployment of the Internet of Things.

The embodiment of the invention provides a hyper-parameter adjustment method based on safety reinforcement learning, and the method comprises the steps: firstly converting an optimization problem of a control strategy into a Lagrange problem with a constraint condition, defining a primary loss function and a secondary loss function, and setting a safety threshold value; secondly, combining with a Lagrange multiplier to construct an initial Lagrange function so as to minimize main loss and meet security constraints; secondly, setting an initial value of a Lagrange multiplier by solving a dual problem, and then introducing a damping factor to construct an enhanced function; and finally, after model parameters are initialized, updating the model parameters by using a gradient descent method and updating the multiplier by using a gradient ascent method in iteration, and substituting into an enhanced function to evaluate whether the strategy is optimal, namely, the main loss is minimum and the security constraint is met, until a global optimal solution is found. Therefore, the optimization of the control strategy is realized, the system performance can be improved, the system safety is guaranteed, the oscillation of the learned strategy when the constraint is satisfied is prevented, and the stable proceeding of the reinforcement learning process is ensured.

The invention belongs to the technical field of vehicle-mounted edge computing (VEC), and particularly relates to an unloading and resource allocation method for a DAG task in a vehicle-mounted edge computing environment. The method comprises the following steps: firstly, constructing a VEC unloading system in a two-way lane scene, and collecting task and equipment information and establishing an optimization model in combination with a vehicle moving model, a communication model and a calculation model; and secondly, aiming at a task in a directed acyclic graph (DAG), a task priority scheduling method based on a DAG topological structure is designed, so that a task scheduling strategy is optimized. On the basis, the problem is modeled as a Markov decision process by taking minimization of task completion time delay and system energy consumption as optimization objectives. The invention relates to the field of resource allocation, in particular to a task-dependent computing unloading and resource allocation algorithm based on a depth deterministic policy gradient (DDPG), and further provides a task-dependent computing unloading and resource allocation algorithm based on the depth deterministic policy gradient (DDPG) so as to solve the optimization problem.