28 results about "Linear matrix" patented technology

The invention relates to the technical field of network acceleration, and discloses an edge-end large language model reasoning acceleration method and accelerator, and the method comprises the following steps: reconstructing a calculation process of a decoding stage, and carrying out the deep fusion of a multi-head attention mechanism and the calculation operation of a feedforward network; the weight and key value data are stored in HBM, and the coefficient and the accumulated attention score are stored in DDR; for linear matrix calculation, a unified matrix calculation unit is used for executing multi-precision matrix operation; for nonlinear function calculation, a mathematical transformation and linear fitting method is adopted, a Softmax function is converted into operation with 2 as the bottom through a bottom conversion formula, and truncation and third-order linear fitting are conducted on a Sigmoid function; and constructing a key value screening algorithm based on the accumulated attention score, dynamically adjusting a key value storage position, maintaining a recent key value cache region and an important key value cache region in a limited cache space, and realizing key value efficient cache in long text reasoning.

The invention provides an optical neural network computing system and method based on time domain excitation characteristics, and the system comprises a pulse clock sequence module which is used for generating a reference time sequence; the multi-channel pulse laser array is used for generating semiconductor neuron laser pulses with subnanosecond pulse width based on a reference time sequence; the time domain and wavelength division multiplexing module is used for simultaneously performing time domain coding and wavelength division multiplexing on the laser pulse based on a reference time sequence to generate a first optical signal; the reconfigurable weight matrix module is used for performing all-optical linear matrix operation and weight mapping on the first optical signal to obtain a second optical signal; the neuron activation module is used for performing nonlinear activation processing on the second optical signal based on an LIF theory to obtain an activated optical signal; and the output module is used for carrying out decoding and photoelectric conversion on the activated optical signal to obtain a network reasoning result.

The invention discloses a load frequency coordination optimization control method, and belongs to the technical field of power system load frequency control technologies. The method comprises the following steps: obtaining a frequency response state space model of a hybrid energy storage-containing two-region interconnected power system through mechanism analysis, performing discretization processing, converting the model into a discretization model, and further establishing a discretization prediction model of the system; analyzing constraints existing in the system, and converting the constraints into a linear matrix inequality; and finally, establishing an objective function of the system, performing online rolling optimization on the control quantity, and designing a multi-region controller, thereby realizing a load frequency coordination optimization control effect of the interconnected power system containing hybrid energy storage. The problems of insufficient flexible resources and poor adjusting characteristics of frequency modulation of an electric power system are solved, coordination and cooperation of battery energy storage, pumped storage and a traditional thermal power generating unit are achieved through coordination control, the frequency response performance of the system is improved, the nonlinear constraint problem existing in the system is solved, and potential safety hazards caused by overload work of equipment are avoided.

The application provides a power system transient stability evaluation method and a power system operation trajectory prediction method based on a Koopman operator, the power system transient stability evaluation method comprising the following steps: S1, collecting operation trajectory data of state variables of a power system; S2, using an EDMD method to increase dimensions of the collected state variables to a high-dimensional linear space to approximate an action of the Koopman operator by using a limited-dimensional linear matrix K ; S3, using spectral decomposition of the Koopman operator to locate discrete spectra of stable equilibrium points of the power system, and using characteristics of eigenvalues of the Koopman operator and inverse mapping of dictionary functions to find stable equilibrium points of the power system in an original state space; and S4, based on the obtained stable equilibrium points and analysis of a stable mechanism of a linear system, constructing a power system transient stability criterion, and using the transient stability criterion to judge whether the system is transiently stable after the power system suffers from a large disturbance and collects an arbitrary working point.

The invention belongs to the technical field of communication privacy protection, and particularly relates to a multi-user downlink robust beam forming method based on a differential privacy mechanism. The method comprises the following steps of: estimating a downlink channel vector by a user; the user adds differential privacy noise into the channel vector and feeds back the difference privacy noise to the base station; the base station enables the transmitting power to be minimized under the constraint of a signal to interference plus noise ratio according to the disturbed channel vector modeling optimization problem; using a positive semidefinite relaxation method to convert an optimization problem; using an S-process method to convert the non-convex constraint into a linear matrix inequality; solving a positive semidefinite programming problem by using a CVX optimizer; and restoring the rank-one constraint to obtain an optimal beam forming vector. According to the invention, the location privacy of the user can be effectively protected from being leaked; and on the premise that the requirement of the communication signal to interference plus noise ratio is met, the transmitting power is minimized, excellent system performance is kept, and good robustness and practical value are achieved.

The invention provides an environment adaptive formation control method based on a free parameterization stabilization network, and the method comprises the steps: automatically switching between a compact formation and a standard formation according to the real-time distance from the geometric center of the formation to an obstacle, and achieving the intelligent reconstruction of a formation shape through a preset distance threshold value and a smooth transition mechanism; a free parameterized stable network architecture is adopted to realize parameterized dynamic operator design of each spatial distributed control unit, and distributed information exchange is realized through a sparse interconnection topology isomorphic with a spatial system communication graph; a controller parameter unconstrained scheme suitable for a space environment is constructed by utilizing the inherent stability property of a free parameterized stable network structure, so that the stability of a space closed-loop system can be ensured for any parameter value, the feasibility constraint of a linear matrix inequality is avoided, and unconstrained gradient learning of a three-dimensional distributed controller is realized.

To speed up image processing by reducing tokens to be calculated in ViT (VisionTransformer).SOLUTION: Extracting, from a first input feature amount input at a first time point, a changed portion that has changed from a second output feature amount output at a second time point prior to the first time point; And a sparse MSA (Multi-HeadSelf-Attention) processing unit including an extraction unit that generates a change matrix based on a change portion of a first matrix indicating a first input feature amount, a Linear processing unit that calculates a Linear matrix which is a matrix product of the change matrix and a weight matrix, and a replacement unit that replaces a portion corresponding to a change portion of a second matrix indicating a second output feature amount with the Linear matrix calculated by the Linear processing.SELECTED DRAWING: Figure 1

The invention discloses an anti-DoS attack and actuator saturation power system memory event trigger insurance cost control method and system, and relates to the technical field of network control systems. Establishing a state-space equation based on the dynamic characteristics of the power system; key influence factors in operation of the power system and external influence factors in a network environment are identified; constructing an output feedback controller based on a dynamic memory event triggering mechanism; a convex hull method is adopted to process saturation nonlinear characteristics of the actuator, a closed-loop system is obtained in combination with the state-space equation, and a guaranteed cost function is designed; and judging whether the closed-loop system meets the requirements of local asymptotic stability and a preset weighted disturbance attenuation level, solving a linear matrix inequality, and obtaining an output feedback controller gain and an event trigger weight matrix. The objective of the invention is to solve the problem that the system stability is difficult to guarantee under the complex conditions that a power system suffers from DoS attacks, actuator saturation exists and external disturbance faces.

The application provides an environment adaptive formation control method based on a free parameterization stable network, automatically switches between a compact formation and a standard formation according to real-time distances from a formation geometric center to obstacles, and realizes intelligent reconstruction of a formation shape through a preset distance threshold and a smooth transition mechanism; a free parameterization stable network architecture is used to realize parameterization dynamic operator design of each spatial distributed control unit, distributed information exchange is realized through a sparse interconnection topology isomorphic to a spatial system communication graph; inherent stability properties of the free parameterization stable network structure are used to construct a controller parameter unconstrained scheme suitable for a space environment, so that the stability of a spatial closed loop system can be guaranteed for any parameter value, thereby avoiding feasibility constraints of linear matrix inequalities, and realizing unconstrained gradient learning of a three-dimensional distributed controller.

This application provides a method and apparatus for frequency-tolerant control of a specified performance load in a multi-regional power system. The method includes: establishing a system model of the multi-regional power system; obtaining specified performance indicators for frequency deviation for each control area; converting the specified performance indicators into auxiliary variables through mathematical transformation; constructing a stability function for the multi-regional power system and performing stability analysis to obtain a set of linear matrix inequalities concerning the controller gain matrix of each control area; wherein the stability function is configured to satisfy the convergence of system state variables; and the frequency deviation satisfies the specified performance indicators and the system's H under non-zero initial conditions. ∞ The norm is less than the preset performance index; by solving a system of linear matrix inequalities, the controller gain matrix of each control region is obtained to construct the fault-tolerant controller corresponding to each control region for load frequency control. This application provides a load frequency control scheme for multi-region power systems that balances robustness, fault tolerance, and dynamic quality.

The application discloses an airplane longitudinal fault-tolerant control and reconstruction method under the condition of elevator surface failure and a related device, and belongs to the technical field of aircraft fault-tolerant control.The airplane longitudinal fault-tolerant control and reconstruction method continues to describe the process of airplane elevator sudden failure by using a Markov process.Then, the application establishes a mathematical model of an airplane longitudinal attitude control system with Markov jump characteristic faults.In consideration of the uncertainty of the system model and external disturbance, the application utilizes a sliding mode control technology to propose a fault-tolerant control scheme for the established airplane longitudinal attitude control system with Markov jump characteristic faults.Compared with some existing sliding mode controller design methods, the method proposed by the application needs to solve a smaller number of linear matrix inequalities.Finally, simulation tests show that the fault-tolerant control method designed by the application is effective and advantageous in airplane longitudinal attitude control.

This invention provides a distributed fault detection method and system for a sensor network switching system, belonging to the field of industrial automation control and fault diagnosis technology. The method includes: acquiring the physical structure parameters, historical test data, and sensor network communication topology of the system under various operating modes; establishing a discrete-time state-space equation set; constructing and solving linear matrix inequalities with performance indicators as the optimization objective and bounded reachability sets as constraints to obtain the optimal filter gain matrix corresponding to each operating mode; calculating the theoretical maximum boundary ellipsoid for each operating mode to obtain the safety threshold for each mode; and performing fault detection based on the filter gain matrix and the safety threshold. This invention establishes a strict instantaneous safety boundary for residual signals under bounded disturbances, achieving theoretically zero false alarms; it replaces the sliding window RMS with an instantaneous residual comparison mechanism, eliminating trailing false alarms after fault clearing, and significantly improving the real-time performance and reliability of fault detection.

The invention discloses an unsteady aerodynamic force calculation and optimization method for a wind turbine blade. The method is based on a potential flow theory, a Laplacian control equation is established, corresponding boundary conditions are set, and a wind turbine geometric model composed of a plurality of airfoils is imported; dispersing the surface of the blade into vortex grids formed by vortex lines by adopting a boundary element method, solving a linear matrix to obtain vortex intensity by distributing initial vortex intensity, and iteratively calculating a velocity field and a pressure field according to the Biot-Savart law; and the aerodynamic force and energy output coefficient of the blade is calculated based on the Kutta-like Kovski theorem, and nonlinear correction is carried out through comparison with experimental data, so that the calculation precision is ensured. According to the method, the calculation efficiency and accuracy are considered, the precision is superior to the blade element momentum theory, the calculation resource consumption is lower than that of a computational fluid mechanics method, and the method is suitable for aerodynamic performance evaluation and optimization in the wind turbine design stage.

The application discloses an envelope constraint cooperative filtering method and device for a nonlinear cluster system under communication restriction, comprising the following steps: approximating unknown dynamics of a nonlinear system by using a neural network; for the problem of limited communication bit rate of a cluster topology, designing a double-description coding scheme to cope with possible data packet loss in information interaction; establishing a sufficient condition for ensuring that a filtering error system simultaneously satisfies preset average performance and envelope constraints in probability; converting a filter design problem into a convex optimization problem, and obtaining the filter by solving a set of recursive linear matrix inequalities. The application solves the problem that existing filtering methods are difficult to simultaneously process unknown nonlinear dynamics and guarantee envelope constraint performance under limited communication bit rate, and can be applied to the field of multi-agent cluster cooperative situation awareness and estimation with unknown dynamics and communication constraints.

The invention discloses a wind power plant frequency modulation capability prediction method considering wind speed uncertainty. The method comprises the steps that a dimension raising correlation model among the wind speed of a wind power plant, the rotating speed of a draught fan and a wind power plant droop coefficient is built based on the Koopman theory; acquiring historical operation data information of the target wind power plant; according to the obtained data information, a Koopman linear matrix of the target wind power plant is obtained through calculation based on the constructed model; calculating a probability density expression of the maximum droop coefficient of the target wind power plant based on the relationship between the system frequency of the wind power plant and the droop coefficient of the wind power plant; and calculating to obtain frequency modulation capability prediction data of the target wind power plant so as to complete prediction of the frequency modulation capability of the target wind power plant considering the wind speed uncertainty. The invention also discloses a system for realizing the wind power plant frequency modulation capability prediction method considering the wind speed uncertainty. The wind power plant frequency modulation capability prediction considering the wind speed uncertainty can be realized, the reliability is higher, the accuracy is better, and the efficiency is higher.

This invention discloses a dynamic neural network event-triggered control method for motor load frequency under external disturbances and internal coupling, relating to the field of motor control technology. The method includes the following steps: obtaining the system matrix parameters of the motor system; establishing a motor state-space model to obtain discrete states; based on the discrete states, representing external disturbances and internal coupling in the motor system as a common disturbance term, and estimating the common disturbance term online using a neural network to obtain disturbance estimates; obtaining the trigger update time of the control signal based on the state deviation between the discrete states and their most recent trigger state; constructing a dynamic neural network state feedback controller at the trigger update time; substituting the dynamic neural network state feedback controller into the motor system model, establishing linear matrix inequalities, and implementing dynamic event-triggered control by solving the linear matrix inequalities, which can better cope with unknown disturbances and reduce communication burden.

The disclosed embodiments include systems and methods to perform automated drilling for both surface and downhole control. The method includes receiving a curvature reference set point, and calculating a curvature error with respect to the curvature reference set point. The method also includes providing the curvature error to a linear matrix inequalities based robust linear quadratic regulator controller. The method further includes determining one or more steering commands based on an output of the linear matrix inequalities based robust linear quadratic regulator controller. The method further includes providing the one or more steering commands to be implemented by a drilling tool during a drilling operation.

This application belongs to the field of power system load frequency control, specifically disclosing a plug-and-play power system load frequency control method based on invariant sets. The method includes: constructing a load frequency control model for a multi-region interconnected power system; predefining fixed-scale alliance subsystems within each control region, and constructing a two-layer control architecture of centralized control within the alliance and decentralized control between alliances; constructing a state feedback control law, using invariant set theory to constrain the state of the closed-loop system within an ellipsoidal invariant set using linear matrix inequalities; and using the absolute error integral of the frequency deviation and the time-weighted absolute error integral as optimization objectives, employing the Grey Wolf optimization algorithm to globally optimize the invariant set parameters and control gain, and substituting these into the state feedback control law to perform load frequency control on the system. This application can effectively improve the robustness and adaptability of power system load frequency control in plug-and-play operation scenarios with frequent unit switching.

The application provides a single-target tracking method based on sparse optical flow motion enhancement, belongs to the field of target tracking, and aims to improve the accuracy of motion modeling through a sparse optical flow algorithm, and to reduce the parameter amount of a tracking model by using a linear matrix when the tracking model is constructed, so as to avoid the problem that the tracking efficiency is reduced due to excessive model parameter amount and motion modeling error propagation; the method comprises the following steps: based on a current frame image, a previous frame image adjacent to the current frame image and the current frame image, a sparse optical flow algorithm is used to perform motion modeling on a tracking target, so as to determine a target candidate region image in which the tracking target is located in the current frame image; image feature extraction is performed on the target candidate region image to obtain multi-level resolution features; and the multi-level resolution features are input into a preset tracking model to estimate the position of the tracking target in the current frame image.

This invention provides a finite-time control method for a pulse-switched positive system based on an event-triggered mechanism. The method includes constructing a pulse-switched positive system with nonlinear and disturbance terms, applying sector constraints to the nonlinear terms and boundedness constraints to the disturbance terms; defining the switching signal in the pulse-switched positive system based on the average dwell time; defining the event-triggered error of the pulse-switched positive system to design the event-triggered mechanism and event-triggered controller; solving for the construction parameters given linear matrix inequalities and reconstructing the switching signal and event-triggered controller; and using the reconstructed switching signal and event-triggered controller to perform closed-loop control of the pulse-switched positive system to achieve finite-time bounded control. Implementing this invention solves the problems of overly complex event control mechanisms and performance degradation caused by the combined effects of pulses and external environmental disturbances.

The invention relates to a quad-rotor unmanned aerial vehicle trajectory tracking method based on grid Kalman filtering and model prediction control, and belongs to the technical field of unmanned aerial vehicle control. According to the method, firstly, a sample point based on Cranley-Patterson random shift is generated by adopting a Korov lattice rule, and high-precision estimation of the pose and the speed of the unmanned aerial vehicle is realized by combining a Kalman filtering framework constrained by a linear matrix inequality, so that the accuracy and the real-time performance of state estimation are improved. A robust predictive controller is designed for bounded interference, actuator saturation and other actual constraints, and compared with a conventional model predictive control method, the problems of system model uncertainty, external bounded interference and actuator saturation can be more effectively solved. And finally, trajectory tracking experimental verification under simulation and physical environment is carried out through an MATLAB simulation platform and an indoor optical localization unmanned aerial vehicle experimental platform. Through collaborative design of high-precision state estimation and robust control, the method can be applied to various unmanned aerial vehicle autonomous operation scenes such as air reconnaissance, material delivery and the like.

This invention discloses a design method and system for an event-triggered controller based on a Markov model, belonging to the field of controller design technology. The method includes the following steps: establishing a hidden Markov model, including system state, control input, output, external disturbances, and time-varying state delay; designing event triggering conditions based on the hidden Markov model to dynamically trigger state information transmission; designing an asynchronous controller to generate control input based on the state information transmitted under the asynchronous event triggering conditions and handling actuator saturation nonlinearity; and analyzing the stochastic stability and H0 of the closed-loop system using Lyapunov functions and linear matrix inequalities. ∞ Performance; a matrix augmentation method is used to separate the product of the controller gain matrix and the Lyapunov variables, obtaining a result that simultaneously ensures the stochastic stability of the system and satisfies H ∞ The performance metrics are consistent with the gain matrix of all asynchronous and event-triggered controllers. The event-triggered controller designed in this invention can guarantee H... ∞ While ensuring anti-interference capabilities, the number of communication triggers is minimized to achieve on-demand control.

The invention relates to the technical field of multi-agent cooperative control, and particularly discloses a second-order multi-agent system periodic event triggering approximate optimal formation control method and system under a DoS attack. Comprising the following steps: establishing a second-order multi-agent system model with a bimodal Markov switching topology, and embedding DoS attack frequency and duration constraints through a transition probability matrix; a distributed periodic event triggering mechanism based on a fixed sampling period is designed, and an intelligent agent is synchronously awakened only at a discrete moment and detects a triggering condition; event triggering and self-adaptive dynamic planning are fused, the Critic network weight is updated at the triggering moment so as to approach a solution of a Hamilton-Jacobi-Bellman equation, and collaborative optimization of formation tracking errors and control energy consumption is achieved; based on a multi-Lyapunov function and a linear matrix inequality, controller gain, trigger parameters and a sampling period are designed, and it is ensured that the system is semi-globally consistent and finally bounded under DoS attacks. According to the method, the random DoS attack is effectively resisted, and meanwhile, the consumption of communication and computing resources is remarkably reduced.

The invention provides a distributed robust load frequency control method for a multi-region power system, and relates to the field of power control, and the method comprises the steps: building a multi-region LFC system model; converting the multi-region LFC system model into a low-dimensional error system through tree transformation; performing robust consistency analysis based on a low-dimensional error system, and proposing a stability criterion meeting the performance; designing a distributed robust controller based on the stability criterion, solving the gain of the controller through a linear matrix inequality, and achieving the control of the distributed robust load frequency through a solving result. The technical scheme of the invention is suitable for a power system with three or more regions, and the communication topology is a switching structure and comprises full communication, partial communication and chained topology.

The application provides a motor imagery electroencephalogram signal classification method and system based on convex optimization, the method comprising: collecting electroencephalogram signals by a visual prompting method, extracting motor imagery electroencephalogram signals as a data set, and calculating a covariance matrix thereof; preprocessing the covariance matrix, mapping the preprocessed covariance matrix to a latent space through a spatial filter, generating a potential signal, and extracting a spectral power feature thereof; using a linear matrix regression model to map the spectral power feature to a binary classification task for classification training, and parameterizing the model using a weight vector. The model is optimized by minimizing a loss function and introducing a regularization term to prevent overfitting; the optimized model is used to predict test samples, calculate a predicted value, and perform binary classification determination, and finally the classification accuracy is calculated by comparing the predicted category with the actual category. The method realizes effective classification of electroencephalogram signals through the combination of spatial filtering and linear matrix regression.

The invention relates to a distributed energy storage cooperative control method and system in a source load uncertain environment. The method comprises the following steps: switching a positive state space model based on non-negative constraint of physical characteristics of an energy storage battery and a dynamic switching behavior of a micro-grid communication network; designing a switching observer based on the switching positive state space model; constructing a self-adaptive power distribution controller; constructing a global closed-loop augmentation system; designing a group of constraint conditions and average residence time conditions based on a linear matrix inequality, and verifying the non-negativity of the closed-loop augmented system matrix; and selecting a common positive Lyapunov function, verifying that the closed-loop augmentation system is globally exponentially stable in a source load uncertain environment, and realizing cooperative control of the distributed energy storage system. The problem of power distribution under communication topology frequent switching and source load random fluctuation is effectively solved, collaborative stable operation of the system on the premise of meeting positive constraint is guaranteed, and the robustness of the system is improved.

The invention discloses a heterogeneous power system fault diagnosis method and device based on multilateral interaction, a medium and computer equipment. The method comprises the following steps: establishing a state space model of a heterogeneous multi-subsystem of the distributed power system; designing a distributed state and fault estimation observer based on an unknown input method, and constructing a dynamic equation for simultaneously estimating a system state and a fault signal; defining a local state estimation error vector and a fault estimation error vector, and deducing a global error system model based on the dynamic equation and the state space model of the observer; performing stability analysis on the global error system model based on the Lyapunov stability theory, and converting a stability condition into an inequality equation set of a linear matrix to obtain a matrix parameter of the observer; and configuring an observer through matrix parameters, and outputting estimation on a system state and a fault signal in real time through the observer so as to complete fault diagnosis. According to the method, the fault diagnosis accuracy of the heterogeneous power system can be improved.

This invention discloses a spacecraft thermal control method and system based on distributed model predictive control, belonging to the field of spacecraft thermal control and automatic control technology. It addresses the problems of excessive computational complexity, poor robustness, and low anti-interference capability in complex space thermal environments found in existing spacecraft thermal control technologies. The method includes: establishing a space-interconnected T-S fuzzy heat transfer model to address the nonlinear characteristics of the thermal properties of spacecraft insulation materials as a function of temperature; introducing model predictive control theory based on tubular constraints to design distributed robust constraint conditions including disturbance boundary estimation; offline solving of the common Lyapunov matrix and feedback gain of each fuzzy subsystem using linear matrix inequalities; and online distributed collaborative optimization calculation of the thermal control input of each subsystem to achieve high-precision and high-stability collaborative control of the spacecraft insulation structure temperature field under multi-source disturbances. This invention is applicable to spacecraft thermal control.