435 results about "Lagrange multiplier" patented technology

The invention provides a micro-grid cooperative scheduling method and device, and relates to the technical field of smart grids, and the method comprises the steps: obtaining historical operation data and real-time operation data of a micro-grid system, and data of an external information system; generating load demand and energy equipment output prediction information based on the historical operation data and the data of the external information system; constructing a layered multi-time-scale decision architecture, and performing decision optimization on each layer of agents by adopting a reinforcement learning algorithm; constructing a plurality of heterogeneous agents, and carrying out cooperative scheduling on the plurality of heterogeneous agents by adopting a centralized training and distributed execution multi-agent reinforcement learning algorithm; inputting the prediction information and the real-time operation data into a decision framework, and outputting a real-time control instruction; and setting a security constraint condition, and realizing optimization of the security constraint in combination with a Lyapunov function, a Lagrange multiplier method, a security layer mechanism and a reinforcement learning algorithm. According to the method provided by the invention, the safe, efficient and reliable operation of the micro-grid in the grid-connected / off-grid mode can be realized.

The invention discloses a multi-scale hydrodynamic digital twin model based on a coupling algorithm, which ensures the accuracy and consistency of the model on the overall and local levels through effective information exchange and integration among different scales. The consistency of macroscopic and microscopic variables is ensured on a boundary layer through bridging coupling of different scale models and a Lagrangian multiplier method, and physical behaviors of a complex system are simulated more accurately. Besides, through parameter estimation and data fusion, the model can integrate and utilize real-time data, so that dynamic optimization and prediction are realized, and the practicability and flexibility of the model are greatly enhanced. Therefore, the construction of the multi-scale digital twinborn model not only enhances the description capability of the model to a complex system, but also improves the adaptability and prediction precision of the model in practical application, and provides a more efficient and dynamic method for accurately simulating and predicting behaviors of the complex system.

The invention discloses an electric power system active safety margin assessment method considering new energy output uncertainty. The method comprises the following steps: S1, constructing a first model according to an active power balance relation of an electric power system, phase angle difference constraints at two ends of a transmission line and generator active output constraints; s2, obtaining an active static security domain boundary according to the first model in combination with a maximum volume optimization method; dividing the operation state of the power system into a risk state and a safety state according to the new energy output uncertainty; s3, evaluating the static active safety margin of the system in a risk state according to the active static safety domain boundary in combination with an interval mathematical theory; and evaluating the static active safety margin of the system in a safe state according to the active static safety domain boundary in combination with a safe distance optimization model based on a Lagrange multiplier method. The invention also provides a power system active safety margin evaluation system considering the new energy output uncertainty.

The invention relates to an edge data processing method of an information security monitoring and early warning AI platform and a related device. The method comprises the following steps: carrying out standardization processing on original security alarm data collected by a plurality of edge computing devices to obtain standardized feature data; constructing a target function comprising a processing efficiency item and a resource consumption item based on the standardized feature data, and decoupling the target function into a plurality of sub-optimization problems of edge computing nodes by using a Lagrange multiplier to obtain an initial task allocation scheme; performing local linearization processing and convex optimization decomposition on the initial task allocation scheme to obtain an optimal task decomposition scheme; constructing a state transition probability matrix according to the optimal task decomposition scheme, and generating a dynamic task scheduling strategy; and distributing the dynamic task scheduling strategy to each edge computing node, and performing time sequence analysis and spatial clustering on multi-source data to obtain a security situation assessment result. According to the invention, the utilization efficiency of edge computing resources 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 provides an anomaly detection method and device for image data and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: carrying out the feature extraction of the image data, fusing the spectrum and spatial features to obtain a joint feature matrix, and inputting an anomaly detection model; the model adopts an alternating direction multiplier algorithm to solve a low-rank sparse decomposition problem, and an objective function comprises a data fidelity item, a regularization item and a waveband weight item; the regularization item comprises a low-rank constraint and a sparse constraint, and the wave band weight item acts on the low-rank constraint in a weighting form; updating a background low-rank tensor, an abnormal sparse tensor, a Lagrange multiplier, a sparse constraint weight, a wave band weight item and penalty parameters of an algorithm by adopting an iteration mode in a solving process; and repeating iteration until a preset termination condition is reached, calculating an abnormal score graph pixel by pixel based on the abnormal sparse tensor, and comparing to determine an abnormal target. The problem that an abnormal target is difficult to accurately recognize in a complex scene can be solved, and detection precision and efficiency are improved.

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 relates to the technical field of transient heat conduction, in particular to a composite material transient heat conduction and thermal stress simulation method based on a two-dimensional arbitrary polygon hybrid finite element, which comprises the following steps: S1, determining an analysis model and material attributes, and establishing a transient heat conduction basic equation including boundary and initial conditions; s2, constructing and expanding a generalized variational functional, processing boundary conditions in combination with a Lagrange multiplier method, and discretizing a solution domain into a plurality of independent units; s3, adopting hypothetical temperature field, heat flux field and thermal stress field functions, introducing a discrete finite element format, and deducing a corresponding finite element equation from a functional; and S4, solving a transient temperature field through programming, introducing the transient temperature field into stress analysis, and finally obtaining temperature and stress distribution of the material at different time points. According to the method, temperature distribution and thermal stress of transient heat conduction are solved by hybridizing heat flux finite elements, and a new way is provided for transient heat conduction and thermal stress analysis of the composite material.

The invention discloses a hydrodynamic analysis method for a floating structure with multi-body and multi-coupling characteristics, which comprises the following steps of: acquiring geometric characteristics of each floating body, and dividing boundary element computational grids; wave excitation force, additional mass and radiation damping of each floating body are calculated, and a mass matrix and a rigidity matrix of each floating body are obtained; determining a connection mode and a coupling factor of the floating body; according to a connection mode and coupling characteristics of the system, establishing displacement connection conditions among the floating bodies; constructing a motion constraint matrix according to the displacement continuity condition and the motion constraint relation; determining a frequency domain motion equation of the multi-floating-body system according to a Lagrange multiplier method; and solving the motion equation to obtain the motion response of the multi-floating-body system, and further systematically analyzing the wave energy capture power according to the coupling characteristics of the system. According to the analysis method, the problem that the motion response of each degree of freedom cannot be accurately described when the hydrodynamic response characteristics of the complex multi-body system are obtained through an existing simulation method is solved.

The invention discloses a data processing method based on multi-agent collaborative optimization, and the method comprises the following steps: building a multi-agent collaborative environment, and outputting a constraint threshold value; accessing multiple data sources according to the constraint threshold, and outputting a standardized input data set; distributing the input data set to each agent participating in collaboration, and outputting a processing suggestion set; performing cross evaluation and consensus formation on the processing suggestion set by each agent, and outputting a candidate scheme list; performing feasible region screening on the candidate scheme list, and outputting an optimized candidate scheme set; receiving an optimization candidate scheme set, carrying out joint optimization on candidate schemes by adopting a cooperation bidding rule simultaneously constrained by double Lagrange multipliers, and outputting a bid winning execution scheme; taking the bid winning execution scheme as input, and outputting execution state information; and receiving the execution state information, and outputting a final execution result when the system meets a preset termination condition. According to the invention, a collaborative optimization data processing method is realized.

The invention discloses a digital twinning multi-agent reinforcement learning intelligent decision-making system and method with a secure memory playback mechanism, and the system comprises a digital twinning module which is used for constructing a virtual model and synchronizing the virtual model with a physical entity in real time; the multi-agent reinforcement learning module is used for carrying out strategy learning based on a constrained Markov decision process and balancing performance and safety through a Lagrange multiplier; the safe memory playback module is used for weighting and playing back the experience samples according to the risk and the timeliness so as to improve the learning safety; the reversible grey influence network module is used for causal modeling and reasoning and enhancing decision interpretability; the double-loop self-constraint control module ensures that a control action is always in a physical safety boundary through a barrier function and safety projection; and the convergence and stability criterion module is used for verifying strategy security convergence and system asymptotic stability. According to the method, the problems of strategy border crossing, virtual-real mismatching and the like in the high-risk manufacturing process are solved, and multi-target optimal control under the safety constraint is realized.

The invention discloses a secondary forest development stage division method based on forest stand state characteristics, and belongs to the technical field of forest development stage division. Comprising the following steps: selecting an initial index set, performing index forward and standardization processing on sample plot survey data, and forming a data set; classifying the data by adopting a system clustering analysis method, checking and analyzing a state index influencing a classification result, and selecting plt; the index of 0.05 constitutes a secondary forest development stage division index system; carrying out weight assignment, then correcting the weight by adopting an entropy evaluation method, and determining the comprehensive weight of the index by utilizing a Lagrange multiplier method; the sum of the dimension-removed value of the division index and the comprehensive weight value product is a development stage value; according to an equidistant method, the development stage of the secondary forest is divided into a forest gap stage, a renewal stage, a differentiation stage, a built-up stage and a stable stage. Depending on forest age is avoided, and the method is suitable for multi-tree mixed secondary forests; the indexes are easy to obtain, and the division result can accurately guide forest management.

The invention relates to the technical field of electric power system dispatching, in particular to a multi-microgrid collaborative optimization dispatching method based on an improved ADMM, and the method comprises the steps: constructing a microgrid operation cost minimization objective function based on a microgrid mathematical model, and carrying out the operation constraint of each microgrid; iteratively updating a local variable, a global variable and a Lagrange multiplier by using an ADMM model to solve the operation cost of the micro-grid; and income distribution is carried out based on the asymmetric Nash negotiation model. According to the method, the problem that the convergence speed and the calculation complexity need to be further improved when an existing model is used for multi-microgrid dispatching optimization is solved.

The invention relates to a reservoir parameter intelligent inversion method and system, a processing device and a storage medium, and the method is characterized in that the method comprises the steps: obtaining the time-space information of a to-be-explained point in a reservoir, and the time-space information comprises a time-space coordinate and a code of the time-space coordinate; inputting the spatio-temporal information of the to-be-explained point into a pre-trained parameter prediction model, and performing inversion to obtain reservoir parameters of the reservoir; the parameter prediction model is a neural network model, and a Lagrange multiplier method is introduced into the neural network model to guide the residual network, so that high-precision parameter inversion can be realized, and the method can be widely applied to the technical field of petroleum engineering.

The invention relates to a low-dimensional subspace clustering method based on projection matrix guidance, and the method comprises the steps: extracting a light response non-uniformity PRNU noise residual error from input image data through employing a denoising filter, and constructing a PRNU feature data set of an image; performing feature dimension reduction on the feature data set by adopting a projection matrix method, constructing a projection matrix maintaining a geometric structure, mapping the projection matrix to a low-dimensional potential subspace, and further constructing a model for the subspace by utilizing a sparse self-representation method; constraint is applied to sparse self-representation in the low-dimensional potential subspace, and joint optimization is carried out through an augmented Lagrange multiplier method ALM and an alternating direction minimization ADM strategy to be used for efficient clustering of data in the low-dimensional potential subspace. According to the method, the projection matrix maintaining the geometric structure is constructed, the high-dimensional PRNU features are mapped to the low-dimensional potential subspace, the local neighborhood relation and the global distribution structure are reserved in the dimension reduction process, the calculation cost is reduced, and the clustering robustness and performance are effectively improved.

The invention relates to an improved constraint weighted least square TDOA / FDOA passive positioning method based on a Lagrange multiplier, and belongs to the field of passive radar positioning. The method comprises a two-step method and comprises the following steps: firstly, converting a time difference and frequency difference nonlinear positioning equation about a target position and speed in a passive positioning problem into a pseudo-linear equation, on this basis, converting into a quadratic programming problem, simplifying constraint conditions, and solving a position and speed initial solution of target positioning by applying a Lagrangian multiplier method; and then a nonlinear equation set about the target position and the speed is constructed according to the original constraint condition, and the initial solution obtained in the first step is substituted into the constructed nonlinear equation set to further carry out optimization solution on the target position and the speed value. The method has good positioning precision and real-time performance, an estimation RMSE curve of a target positioning parameter can still be close to CRLB in a large-noise environment, and the method has good robustness in a high-noise environment and has good advancement.

The invention discloses a multi-view migration interpretable method based on soft variable embedding and discriminant structure preserving, and particularly relates to the technical field of machine learning, and the method comprises the steps: obtaining a plurality of multi-view features for EEG samples of a tagged source domain and an untagged target domain through the extraction of a plurality of features; after TSK-FS antecedent network mapping, constructing inter-domain connection by adopting migration soft variable embedded consequent learning; a data structure is retained through a local-global structure retention item, so that a discriminant neighborhood relationship of original data is retained to the greatest extent in a migration process; and a target function is constructed in combination with a multi-view learning strategy, iterative optimization is performed through an enhanced Lagrange multiplier algorithm, and finally a result is output by using a given classifier.

The invention provides a method for accurately positioning and installing electromechanical equipment in a complex area, and belongs to the technical field of electromechanical equipment installing.The method comprises the steps that three-dimensional space point cloud data is obtained through a multi-dimensional locator to construct an initial positioning vector database, and an installation coordinate matrix is determined in combination with the technical specification of the electromechanical equipment; a coordinate measuring arm and a laser tracking interferometer are used for measuring and calculating a space compensation variation vector and a thermal expansion variation vector to form a comprehensive variation vector. Performing computer simulation on an installation environment based on the acquired data, determining an optimal installation path, constructing a spatial region overlapping matrix and a spatial region idle matrix, performing calculation to generate a selectable region, determining an optimal installation point location by applying a Lagrange multiplier constraint optimization equation set, performing installation and adjustment, monitoring a deviation value in real time, and determining the optimal installation point location. When the deviation value is smaller than a preset threshold value, fixed installation is completed, and finally the installation precision is verified to meet the design requirement through operation testing.

The invention discloses a dual robust principal component analysis image denoising method based on a graph, and belongs to the technical field of image processing. The method comprises the following steps: flattening an image to be processed to construct a data matrix in a column vector form; constructing a graph structure by using a K nearest neighbor method, and generating a graph Laplacian matrix based on the graph structure; jointly considering an image reconstruction error, a sparse noise item, a linear mapping error item and a graph structure regular item, and constructing an optimization model; and carrying out variable alternating optimization by adopting an augmented Lagrange multiplier method and an alternating direction solution method, and obtaining an image denoising result according to the low-rank principal component. According to the method, image structure information and a double constraint mechanism are introduced into a robust principal component analysis framework, so that the detail retention capability and the structure consistency of the image are effectively enhanced, the robustness and the visual quality of image denoising are improved, and the method is suitable for application scenes such as image processing, video monitoring and target detection under complex backgrounds.

The invention provides an active power distribution network reactive voltage optimization method and related equipment. The method comprises the following steps: constructing a centralized optimization model; decomposing the boundary constraint conditions according to sub-partitions, and determining a sub-partition optimization problem; solving each sub-partition optimization model based on the sub-partition optimization problem, converting the sub-partition optimization model into a sequential quadratic programming sub-partition optimization model based on the solved first power flow solution, and determining each sub-partition matrix; extracting coupling information in adjacent sub-partition matrixes to obtain an optimization sub-problem about a Lagrange multiplier coefficient; determining the optimal search direction of each sub-partition; and correcting the first search direction by utilizing the optimal search direction, carrying out iterative updating until a preset condition is met, and ending optimization processing on the reactive voltage of the active power distribution network. According to the embodiment of the invention, in the calculation process, the partitions are mutually independent, only the coupling information of the adjacent nodes of each partition is exchanged, and the information privacy among the sub-partitions is protected.

The invention discloses an engineering machinery industry green factory evaluation method and system. The method comprises the steps of obtaining a pre-formulated green factory evaluation index system; calculating the subjective weight of each secondary index relative to the target layer; introducing a Monte Carlo simulation algorithm, and calculating an objective weight and an evaluation matrix of each secondary index corresponding to each Monte Carlo simulation relative to the target layer; based on a minimum relative information entropy principle and a Lagrange multiplier method, combined with the subjective weight and the objective weight of each secondary index relative to the target layer, calculating the comprehensive weight of each secondary index relative to the target layer corresponding to each Monte Carlo simulation; on the basis of the comprehensive weight and evaluation matrix of each secondary index relative to the target layer, constructing a fuzzy comprehensive evaluation matrix corresponding to each Monte Carlo simulation; and obtaining a green factory grade evaluation result based on the fuzzy comprehensive evaluation matrix corresponding to each Monte Carlo simulation and a preset score interval corresponding to each comment grade. According to the method, the accuracy and objectivity of green factory evaluation can be remarkably improved.

The invention relates to the field of engineering project management, and discloses an engineering project full-process management system construction method, which comprises the following steps that project optimization objectives and constraint conditions are defined, the optimization objectives comprise project progress, cost and resource utilization rate, and the constraint conditions comprise task dependence, resource availability, construction period and budget limitation; based on the targets and the constraint conditions, a multi-target optimization model is constructed, and the multi-target optimization model adopts a weighted sum method to merge a plurality of target functions; and solving the optimization model by using a Lagrange multiplier method to obtain an optimal resource allocation and task scheduling scheme. By combining the dynamic control theory, the Internet of Things technology and the fluid mechanics model, real-time dynamic adjustment of project resource scheduling is achieved, resource allocation is optimized, project progress and cost control efficiency is improved, and it is ensured that the project efficiently and flexibly responds to changes in the execution process.

The invention discloses an artificial wetland process design system and method based on a neural network multi-boundary condition. The system comprises a data preprocessing module, a neural network model construction module, a multi-boundary constraint processing module and an optimal solution output module. Wherein the multi-boundary constraint processing module uses a Lagrange multiplier method and a penalty function method to manage constraint conditions of the constructed wetland involved in the optimization process; the optimal solution output module is used for randomly generating an initial constructed wetland process combination, inputting the initial constructed wetland process combination into the trained deep feed-forward neural network, and calculating a fitness value of the initial constructed wetland process combination after key performance indexes of operation of the constructed wetland are obtained; and on the basis of the fitness values and constraint conditions of the constructed wetland, correcting the process combination which does not meet the constraint conditions by adopting an evolutionary algorithm, and outputting the process combination with the highest fitness as an optimal solution after the optimization process is ended. The result obtained by the method is high in accuracy.

The invention provides a large model training data desensitization protection method and system based on differential privacy, and relates to the technical field of artificial intelligence security, and the method comprises the steps: calculating a differential privacy parameter, mapping training data to a feature subspace, and constructing a dimension reduction manifold projection operator; solving an optimal disturbance vector by using a Lagrange multiplier method to form a de-sensitization feature set; probability distribution is constructed based on kernel density estimation, and Gaussian noise is injected; and dynamically adjusting a parameter group updating strategy in model training. According to the method, data utility and privacy protection can be effectively balanced, the large model training safety is enhanced, and the model convergence efficiency is improved.

A production green quality comprehensive evaluation method for a distribution transformer comprises the steps of collecting data, forming a quality detection index matrix, a production energy consumption index matrix and a carbon footprint characteristic index matrix, screening procedures which have great influence on energy consumption by using a K-means clustering algorithm, constructing a production energy consumption index set, updating the production energy consumption index matrix, and evaluating the production green quality of the distribution transformer. And performing normalization processing on various index matrixes to form a normalized index data matrix. On the basis, respectively calculating the weight of each index by adopting a CRITIC method and an entropy weight method to obtain an overall CRITIC weight vector and an overall entropy weight vector; according to a minimum identification information principle, solving an optimal combination weight through a Lagrange multiplier method; and calculating a comprehensive evaluation score and a corresponding grade. The multi-dimensional data fusion mode can reflect the green production quality of the distribution transformer more comprehensively and accurately.

The invention discloses a virtual power plant decentration AGC method and system based on a precise diffusion algorithm. According to the method, an AGC instruction decomposition model with optimal operation cost and highest frequency modulation performance as double optimization targets is designed according to the adjustment upper and lower limits of each resource and the actual line capacity constraint; a Lagrange multiplier of the model is decoupled, and an accurate diffusion algorithm is adopted to realize rapid and accurate solution of an instruction; and finally, an intelligent agent communication model containing communication noise is obtained. A precise diffusion algorithm is rewritten and a gain function is introduced, so that the precise diffusion algorithm has the capability of translating noise, and the robustness of the algorithm is improved.

The invention discloses a mixed regular variational mode decomposition method, device and equipment suitable for an unsteady flow field of an aircraft and a storage medium, and relates to the technical field of flow analysis, and the method comprises the steps: determining a bandwidth estimation value and a frequency separation weight coefficient based on the unsteady flow field, an expected mode number, a spatial mode and a time evolution coefficient, constructing a first regular term and a second regular term, weighting to obtain a mixed regular term, introducing a flow field reconstruction term to establish a flow field modal decomposition variational optimization problem, converting the problem into a frequency domain optimization problem through Fourier transform, and simplifying the problem into a sub-convex optimization problem by adopting an alternating direction multiplier method; an analytic solution is obtained through a variational method and iteratively updated to obtain a flow field space mode, a time evolution coefficient frequency domain representation, a center frequency and a Lagrange multiplier, the time evolution coefficient is obtained through inverse Fourier transform, a mode decomposition result is obtained, and therefore high-precision mode separation of the unsteady flow field is achieved. And the calculation efficiency and the iteration convergence speed of the variational optimization problem are improved.

The invention discloses a manifold optimization transmitting directional diagram synthesis method for an analog digital hybrid array. According to the manifold optimization transmitting directional diagram synthesis method for the analog digital hybrid array, the HAD array transmitting directional diagram synthesis problem with the criterion of minimizing the matching error between an expected beam directional diagram and an actual beam directional diagram is researched. According to the method, the solution is carried out by utilizing block coordinate descent (BCD) and Riemannian manifold optimization. Firstly, a constant modulus auxiliary variable is introduced to equivalently represent that an objective function is a smooth biquadratic function; then, alternately optimizing a scaling factor, a digital weight, a simulation weight and a constant modulus auxiliary variable under a BCD framework, reexpressing the simulation weight and the auxiliary variable as a quadratic programming problem under constant modulus constraint, and solving through a Riemannian gradient descent method; the globally optimal solution of the digital weight is obtained through a Lagrange multiplier method. A numerical result shows that compared with other representative algorithms, the algorithm provided by the invention has better beam matching performance.

The invention discloses a marginal device intelligent sensing method and system based on information freshness, and a Markov decision process model is constructed by analyzing key factors in the marginal device intelligent sensing system. And secondly, the system state is updated in real time through dynamic adjustment of the sensing state, and parameters of equipment are adjusted according to new environment changes, so that the system can quickly respond in a complex environment. In the aspect of joint optimization, joint optimization is carried out on information age and energy consumption by utilizing a Lagrange multiplier method, and the freshness of information is maximized while energy constraints are met. And finally, through the deep Q network, the system can continuously optimize a decision strategy through an experience playback mechanism, so that the edge equipment can minimize the information age under the condition of limited energy in a continuously changing environment. According to the invention, the timeliness of sensing information is improved.

The invention relates to a secure communication optimization method for a double-IRS-assisted simultaneous wireless information and power transfer system based on a graph neural network, and belongs to the field of wireless communication security. According to the method, reflected beam forming of two IRSs and transmitted beam forming of a base station are jointly optimized to maximize the total information transmission rate of a system, the transmitted power constraint, the maximum eavesdropping rate constraint, the energy collection constraint and the modulo-one constraint of IRS reflection coefficients of the base station are met at the same time, and a machine learning method combining a Lagrangian multiplier method and GNN is adopted. Pilot frequency information used for estimating a channel is parameterized into a GNN representation vector, the representation vector is updated layer by layer according to a GNN connection graph, and beam forming at a BS position and a reflection coefficient at an IRS position are optimized according to a system target. Compared with a traditional method using channel estimation and non-convex optimization, the method provided by the invention obtains a higher information transmission rate on the premise of meeting the requirements of energy collection users and information collection users.