525 results about "Robust control" patented technology

The invention provides an intelligent regulation and control method for a three-stage constructed wetland recirculating aquaculture system, and belongs to the technical field of machine learning. The method comprises the following steps: firstly, continuously collecting water quality data and operation control data of each control unit, and constructing a multi-source heterogeneous data set under a unified time scale; then, constructing a pollution evolution trend prediction model, and capturing a dynamic evolution trend of water quality along with time and control behavior changes; then, under the guidance of a prediction result, analyzing the similarity of historical states and the sensitivity of regulation and control response, automatically identifying key control parameters which influence the water quality change of the system at present, and reasoning the dynamic adjustable boundary of the key control parameters; and finally, constructing a reinforcement learning strategy network fusing state prediction, a parameter boundary and a control target, realizing multi-target tradeoff among pollutant removal efficiency, a water quality standard-reaching rate and operation energy consumption, and outputting an efficient and steady control strategy through continuous interactive training. According to the invention, efficient, accurate and robust operation of the wetland system can be realized.

The invention discloses a renewable energy power generation power prediction and power dispatching method and system, and the method comprises the steps: collecting the historical power generation data and real-time meteorological data of renewable energy power generation, carrying out the linear interpolation of the historical power generation data and the real-time meteorological data, and carrying out the missing value filling and box plot anomaly detection, obtaining a normalized training data set; constructing a hybrid prediction model by using the normalized training data set and adopting a neural symbol acceleration technology with time logic constraints, extracting medium and long term space time features, and generating a renewable energy power generation power prediction result; and according to the renewable energy power generation power prediction result and the system constraint condition, adopting a linear one-dimensional projection constrained distribution robust control method to formulate a scheduling strategy, and utilizing the scheduling strategy to solve an optimal scheduling scheme through mixed integer linear programming. According to the method, the renewable energy power generation power prediction precision and the power dispatching robustness are remarkably improved.

The invention discloses a data-based predefined time heterogeneous multi-agent formation collision avoidance method. The method comprises the following steps: establishing a nonlinear heterogeneous multi-agent system; designing a bimodal artificial potential field function to perform dynamic obstacle avoidance and prevent regional escape; establishing a self-adaptive robust controller used for generating an obstacle avoidance safety motion trail of the root leader; designing a self-adaptive formation zooming mechanism of the leader, and constructing a predefined time affine observer of the follower based on the self-adaptive formation zooming mechanism; designing a unified obstacle function; constructing a virtual control law for processing tracking errors based on the unified obstacle function; designing a neural network estimator; designing controllers of the leader and the follower according to the virtual control law; and forming a collision avoidance decision of the heterogeneous multi-agent formation based on a self-adaptive robust controller, a predefined time affine observer, a neural network estimator and controllers of the leader and the follower. According to the method, safe, efficient and robust cooperative control of the formation in a complex environment is realized, and the safety and task execution efficiency of the heterogeneous multi-agent formation in a limited and unknown environment are remarkably improved.

The invention discloses a four-rotor unmanned aerial vehicle model prediction control method based on a deep Q-network (DQN). The method comprises the following steps: firstly, establishing an inertial coordinate system and a body coordinate system, and establishing a kinematic equation and a kinetic equation of the four-rotor unmanned aerial vehicle based on a Newton second law and an Euler-Lagrange equation; according to the method, an unmanned aerial vehicle robust controller based on model predictive control (MPC) is designed, optimal control input is generated through dynamic optimization, and the trajectory tracking performance of the unmanned aerial vehicle is improved. For various disturbances existing in the whole tracking process, a DQN reinforcement learning algorithm and an MPC method are combined to design a flight control system, control rate errors caused by the disturbances are compensated, the stability of attitude control is enhanced, and the tracking precision and the anti-interference capability are improved. And finally, verifying the robust performance of the flight control system through a simulation experiment.

The invention discloses an adversarial reinforcement learning training method for robust control of a fixed-wing aircraft, and the method comprises the steps: constructing a nonlinear simulation environment and a task set of the fixed-wing aircraft, and constructing a plurality of fixed-wing aircraft agents in parallel in the simulation environment; an adversarial agent module based on a neural network AdvNet is introduced to generate environmental disturbance. A near-end strategy optimized Actor-Critic structure is adopted, an intelligent agent and an adversarial agent module are subjected to joint training through a task set, and a robust control strategy is obtained step by step. In the training process, based on a course learning mechanism, disturbance constraint parameters of the confrontation agency are dynamically adjusted according to the completion condition of a flight task. After training convergence, obtained control strategy parameters are applied to the fixed-wing aircraft, and robust flight control is achieved. According to the method, the robustness and the control precision of the aircraft facing complex environment disturbance can be improved, and the method is suitable for various flight missions with uncertainty.

The invention discloses a data center machine room energy-saving optimization method and system based on thermal environment prediction. The method comprises two stages of offline modeling and online prediction optimization. In the off-line stage, a CFD simulation model is constructed based on a machine room physical structure, equipment layout and thermal load parameters, and high-precision temperature field data is generated; using the thermal environment prediction model to input equipment parameters and load change to output future space temperature distribution; meanwhile, an XGBoost hybrid energy consumption prediction model is constructed based on the wind speed ratio or the fan frequency, the number of running fans and related characteristics. In the online stage, the lowest energy consumption and the minimum temperature deviation serve as targets, and a Pareto optimal solution set is generated through an MOEA / D algorithm; the optimal wind speed ratio / frequency and equipment number combination is selected as required to control operation of the air conditioner; and dynamically updating model parameters by sliding a time window to realize long-term robust control of the system. According to the method, intelligent energy-saving control of the machine room is realized by fusing CFD simulation, time sequence prediction, energy consumption modeling and a multi-objective optimization algorithm.

According to the step motor sensorless control method based on improved robust Kalman filtering, an HBM current state equation is established, and an extended Kalman state equation is established; establishing an inductance deviation model, and deducing a system covariance matrix with errors; parameter uncertainty robust control is introduced, an expansion factor is designed, and a system covariance matrix is reconstructed for limitation compensation; the method comprises the following steps: constructing chi-square test statistics through an innovation sequence, identifying a system gross error, designing a robust weight function based on a system residual vector, selecting whether to carry out variance correction or not according to an identification result, inhibiting the cumulative influence of a model mismatch error on state estimation, updating a state, and adjusting robust intensity. According to the method, parameter uncertainty robust control and an IREKF algorithm based on innovation detection are introduced to carry out dynamic compensation on a model mismatch error, REKF is improved through innovation detection, and the robust intensity is dynamically adjusted.

The invention discloses a power grid broadband oscillation suppression method and device based on adaptive robust control, terminal equipment and a storage medium, and belongs to the field of power systems. The method comprises the following steps: performing fuzzy disturbance classification according to an oscillation mode, an oscillation mode characteristic parameter and a state parameter obtained according to a power grid operation parameter to obtain a disturbance type label and a disturbance degree; the disturbance degree serves as a weighting factor, the initial control parameters are corrected according to the power grid operation parameters, the initial control parameters of the controller and the disturbance type label, and a self-adaptive control strategy is generated after correction; when the disturbance degree is smaller than a disturbance threshold value, the state parameters are regulated and controlled according to a self-adaptive control strategy to update power grid operation parameters; and when the disturbance degree is not less than a disturbance threshold value, taking the minimum sum of the system gain, the frequency deviation and the voltage fluctuation as a target, and regulating and controlling the state parameters under the H infinite robust constraint to update the power grid operation parameters. The problem that in the prior art, broadband oscillation suppression is difficult to adapt to sudden change working conditions is solved.

The invention discloses a multi-mode intelligent tension self-adaptive control method, and belongs to the technical field of tension control of wire rod processing devices. According to the method, an LSTM-Transform hybrid model is utilized to realize the collaborative optimization control of tension prediction, the rotating speed and the pitch; by means of an edge computing architecture and a Modbus-TCP real-time communication protocol, a data processing and control instruction generation period is compressed to be within 50ms, and the dynamic response capability of a system is improved; and meanwhile, a graded safety protection mechanism is designed, a robust control mode is switched to at a millisecond level when vibration energy exceeds a limit or the temperature is abnormal, and the MES system is linked to push a fault code, so that the control precision is remarkably improved, the material loss rate and the operation and maintenance complexity are reduced, and a high-reliability and self-adaptive intelligent solution is provided for high-speed wire rod processing equipment.

The invention provides a power distribution network optimization method based on hierarchical robust control and dynamic decision, which realizes multi-time scale coordination control and real-time adaptive control by constructing a hierarchical architecture of a minute-level equipment layer, an hour-level region layer and a day-preceding-level system layer and combining a dynamic information decision mechanism. The method comprises the steps of preprocessing measurement information in a grading manner, establishing a multi-target robust optimization model, decomposing the model into a local quick response and global coordination problem, constructing a time-varying uncertainty set and embedding mixed integer robust optimization, and realizing the collaboration of millisecond-level correction and hour-level scheduling by adopting a second-order cone relaxation technology. And the weight is dynamically adjusted and optimized through the information decision module, so that the robustness and the operation efficiency of the system are improved.

The invention discloses a reinforcement learning-based quadrotor unmanned aerial vehicle attitude preset performance control method, which comprises the following steps of: establishing an unmanned aerial vehicle attitude kinetic equation by considering a time-varying inertial parameter and external unknown interference; a sub-channel preset performance controller is constructed, and a robust control quantity is generated in combination with preset performance and a non-singular fast terminal sliding mode control technology; designing a reinforcement learning parameter generator, and dynamically optimizing 12 time-varying parameter estimated values through a time sequence feature extraction network and a residual network; and establishing an online reinforcement learning training mechanism, constructing a multi-target reward function, and optimizing the output of the parameter generator through the multi-target reward function. According to the method, a reinforcement learning method is adopted to replace a traditional self-adaptive method to estimate time-varying parameters, multi-degree-of-freedom decoupling optimization is achieved through a sub-channel control architecture, tracking error preset performance constraint is achieved under the working condition of time-varying inertial parameters, and the dynamic adjustment capacity and anti-interference robustness of an attitude system are remarkably improved.

The invention provides a self-adaptive fractional order nonsingular terminal sliding mode control method based on time delay estimation, and relates to the technical field of manipulator control. Designing a non-singular terminal sliding mode surface containing fractional differential and power transformation, and further constructing a reaching law with variable gain; then, the unknown dynamic state and disturbance of the system are compensated in real time through a time delay estimation technology by utilizing historical control input and state information; finally, a control law is generated by integrating the sliding mode surface, the reaching law and the time delay estimation compensation item, high-precision and strong-robustness control over an uncertain nonlinear system is achieved, and the tracking precision and the anti-interference capacity of the manipulator in the uncertain environment are effectively improved.

The invention provides a robust control automatic setting method for a constant-pressure water supply system, and the method comprises the steps: obtaining a system working condition in real time, and extracting multi-dimensional state parameters including real-time flow, pressure deviation, current deviation and power deviation; the current load working condition is judged according to the pressure deviation, disturbance characteristics are analyzed by integrating all parameters, and the disturbance grade and type are determined. On the basis, robust control parameters are automatically set in combination with real-time working conditions, load conditions and disturbance characteristics of the system. And finally, performing multi-stage cooperative control on the system by using the set parameters, so that the constant-pressure water supply system is automatically set. According to the robust control automatic setting method for the constant-pressure water supply system, load working conditions and disturbance characteristics are recognized based on multi-dimensional state parameters, robust control parameters are self-set in a targeted mode, and automatic setting of the constant-pressure water supply system is achieved through multi-stage cooperative control. Therefore, the intelligent level of the constant-pressure water supply system and the adaptability in different operation environments are effectively improved.

The invention discloses a spraying equipment automatic control system based on the industrial internet, and belongs to the technical field of industrial automatic control. The system comprises a state sensing unit, a feature extraction unit, a risk prediction unit and a closed loop correction unit. The state sensing unit obtains a real-time working condition data set; the feature extraction unit generates a core process feature set; the risk prediction unit generates a quality risk prediction set; the closed-loop correction unit calculates a target process set point. The system senses multi-source data such as environment temperature and humidity and coating flow in real time, quantifies the influence of environment and process fluctuation on spraying quality, predicts indexes such as coating film thickness and defect probability in a prospective mode, and adjusts process parameters in a self-adaptive mode on the basis of working condition risk levels. Compared with the prior art, complex working condition insight, quality risk look-ahead prediction, dynamic intelligent optimization and extreme working condition robust control are achieved, the contradiction between high quality and high efficiency in automatic spraying is relieved, and the spraying quality stability and the production efficiency are improved.

The invention belongs to the technical field of aviation hybrid power, and discloses a self-adaptive intelligent power distribution control method for an aviation hybrid power system, which comprises the following steps: constructing a full-dimensional state model through multi-dimensional perception fusion, predicting working condition transition in combination with deep reinforcement learning, and dynamically adjusting a power distribution strategy; the response speed is guaranteed preferentially during emergency takeoff, energy recovery is emphasized during conventional landing, the problem that a traditional distribution mode is disjointed with actual requirements is solved, power output of an engine and a battery better fits the flight working condition, and stable operation can still be achieved in complex environments such as sudden airflow; by introducing digital twinborn rehearsal verification and adaptive robust control, risks of engine overtemperature, battery overcharge and the like are exposed and corrected in advance through virtual simulation, disturbance is dealt with in combination with a'basic + compensation 'power structure, part faults caused by unreasonable power distribution in traditional control are avoided, and the reliability of the system is improved. And the reliability of the hybrid power system in key stages of takeoff and landing is improved.

The invention discloses a tunnel pipe roofing construction dynamic deviation correction method based on laser ranging and optical fiber sensing, and relates to the technical field of measurement and sensing, and the method comprises the steps: S1, collecting laser and optical fiber data, obtaining real strain through temperature coupling compensation, and calculating global pose parameters; s2, generating an eigenmode component based on the real strain and the global pose parameter, and screening and constructing an intelligent feature vector according to the contribution degree of the physical coupling feature; s3, constructing a physical information driving state estimation model in combination with pipe joint mechanical constraints, and outputting a pipe joint state estimation value in real time; s4, performing construction deviation prediction by using the pipe joint state estimation value, and generating an optimal control sequence and instruction; and S5, applying the control instruction to the construction device, and adaptively updating the key parameters of the system through a meta-optimization algorithm based on feedback. And through multi-modal data fusion and physical constraint intelligent prediction, high-precision deviation correction, closed-loop robust control and system self-adaptive optimization of the pipe joints are realized, and the intelligence and long-term precision of tunnel pipe curtain construction are improved.

The invention discloses a fuzzy neural network PID photovoltaic series welding temperature control method based on improved PSO optimization. The method comprises the steps that a target photovoltaic series welding equipment heating transfer function model is acquired; setting an initial PID parameter; adjusting a PID increment parameter of the PID control module in real time according to the temperature error, the temperature error change rate and a preset fuzzy rule base; the neural network is combined with fuzzy control, and parameters and rules of fuzzy control are automatically modified through training data; according to the photovoltaic series welding temperature condition, the fitness function of the PSO algorithm is improved, the adjustment mode of the inertia weight is improved, and the improved PSO algorithm is used for optimizing the initial PID parameters of fuzzy neural network PID photovoltaic series welding temperature control. Self-adaptive precise control and robust control of the welding temperature control system are achieved, the temperature fluctuation phenomenon in the photovoltaic series welding process is effectively improved, and the robustness and control reliability of the system are enhanced.

The invention provides an underwater robot dynamic positioning method for coping with complex disturbance, and the method comprises the steps: collecting the attitude angle, angular velocity, linear velocity and linear acceleration information of an underwater robot, and generating fused high-frequency and high-precision pose information; designing an extended state disturbance observer, and estimating a disturbance value in real time through a feedback function and an auxiliary variable equation; adding the disturbance estimation value and the output force of the dynamic model, constructing a constraint condition of model prediction control, and solving the optimal control input of N prediction time domains in the future through rolling time domain optimization; and based on the optimal control input, constructing a quadratic programming problem of thrust distribution in combination with the space pose distribution matrix of the thrusters, solving the minimum energy solution of the thrusters, and outputting the minimum energy solution to a thruster execution mechanism. According to the technical scheme of the invention, the robust control of the underwater robot in a strong disturbance environment can be realized. The method mainly solves the problem of fixed point and attitude determination when the underwater unmanned robot faces underwater turbulence or other complex disturbances.

The invention discloses a new energy electric power robustness control method, system and device for preventing chain overload of a power transmission section and a medium. The method comprises the following steps: obtaining the active power flow size and the operation state of each branch in the power transmission section; calculating the comprehensive sensitivity of all the controllable nodes to each branch in the power transmission section, and generating a composite priority index based on the calculated comprehensive sensitivity, the classification of the nodes and the new energy fluctuation risk; performing priority ranking on the generator nodes to generate a node sequence; in combination with a reverse equivalent pairing principle, selecting an adjustment node pair, and calculating an adjustment amount of node output by considering an adjustment amount dynamic margin constraint of new energy output uncertainty; recalculating the active power flow and the operation state of each branch in the section; and judging whether the section power flow and the margin meet a termination condition or not to realize optimal control. According to the method, the operation robustness of the new energy power system is improved, and the overload of the transmission section can be quickly and efficiently eliminated.

The invention discloses an under-actuated unmanned surface vehicle safe formation tracking control method based on robust obstacle control. The method comprises the following steps: establishing a multi-unmanned ship group and performing real-time communication, and enabling each follower to follow a leader according to a formation configuration and to avoid an obstacle; a safe formation tracking control framework of the underactuated unmanned surface vehicle is established, during formation tracking, followers input state information into the control framework, processing is carried out according to a reference trajectory of a leader and real-time information of a current follower and an obstacle, and forward and steering control torque instructions are output to control the current follower. Obstacles are avoided and formation configuration is restored to continue following the leader. According to the method, the underactuation characteristic problem of the unmanned surface vehicle is solved, the worst case error generated when the neural network approaches the unknown dynamic state can be effectively compensated, it is ensured that the unmanned surface vehicle can strictly avoid collision with the obstacle in the complex environment, and meanwhile the formation tracking performance is optimally kept.

The invention belongs to the technical field of robot motion control, and particularly relates to a quadruped robot control method based on error symbol robust integral feedback, which combines an RISE control mechanism with a constraint-based optimization method, and realizes high-precision trajectory tracking and strong-robustness control under the condition that a system model has relatively large uncertainty. According to the method, RISE controllers are designed in a position subsystem and a posture subsystem respectively, quadratic programming (QP) optimization distribution of ground contact force is fused, control input is dynamically adjusted while the physical feasibility of foot end force is ensured, and closed-loop stable control over the whole-body movement of the quadruped robot is achieved. The control strategy has an asymptotic error convergence characteristic, and can be widely applied to quadruped robot motion tasks with severe load change and frequent interference.

The invention discloses a high-frequency motor drive multi-rate control method for time sequence disturbance compensation, and the core principle of the method is that low-rate disturbance compensation and a reference tracking trajectory are improved and converted into a time sequence processing process matched with a high-rate control instruction; specifically, a single disturbance compensation value which can be known only at a sampling moment Ts is expanded into time sequence prediction of disturbance changes on N control points in the future through a generalized proportional-integral observer; meanwhile, a low-frequency single reference instruction is constructed into a smooth high-frequency reference trajectory covering N control points through a Lagrange interpolation method, and closed-loop correction of disturbance compensation and reference tracking is achieved on each high-frequency control point Tc. The problem that open-loop prediction errors are continuously accumulated and spread due to mismatching of information updating rates in traditional multi-rate control is fundamentally solved, and therefore high-precision robust control over the high-frequency motor driving system is achieved.

The invention relates to the technical field of closed-loop drift control, and particularly discloses a closed-loop drift control method and system for an autonomous vehicle in an uncertain environment, and the method comprises the steps: building a vehicle kinematic model based on a Frenet coordinate system, and constructing an MPC path tracking controller with a preview mechanism; establishing a three-degree-of-freedom vehicle dynamics model with additional yawing moment; the method comprises the following steps: establishing a UniTile-Ctrl tire dynamic model; designing a robust drift controller by combining LQR and integral sliding mode control on the basis of the whole vehicle dynamic model and the tire dynamic model, designing a torque distribution module by taking minimization of distribution error and tire load utilization rate as targets on the basis of ideal rear wheel rotating speed and additional yawing moment output by the robust controller, and outputting a torque distribution model; according to the drifting vehicle tracking control method, closed-loop drifting control can be achieved, the drifting vehicle tracking control performance and robustness are improved, and safety guarantee is provided for limit control over the uncertain road surface of the vehicle.

The invention discloses a graphical interface automatic control code generation method based on a large model agent, which comprises the following steps of: firstly, generating a task execution track through interaction between a ReAct agent based on a large language model and a graphical interface environment, and translating a hard coding action in the track into a soft coding action based on a semantic positioning page element; constructing a track library of a tree structure; then, on the basis of information in the trajectory library, a retrieval enhancement mechanism is adopted to generate an RPA function capable of being independently executed, and the RPA function is synthesized on the basis of soft coding actions and is automatically injected into robustness control logic; then executing the RPA function, if execution fails, triggering breakpoint analysis, and optimizing the RPA function through a mixed trajectory; and finally, testing the generated RPA function, and selecting to execute the RPA code or back to the ReAct agent for execution according to a verification result. According to the method, an automatic generation process from an intelligent agent interaction track to a robust and reusable RPA function is realized.

The invention relates to an automatic printing calibration method and system combining vision and AI and a storage medium, a multi-mode sensing unit integrated on a printing head assembly synchronously collects two-dimensional images, three-dimensional shapes and working spacing data of a printing area in real time, and data fusion is carried out by using a model based on a neural radiation field; generating a high-fidelity multi-dimensional digital mapping; carrying out defect identification and positioning by adopting a pre-trained convolutional neural network CNN model, and predicting an optimal adjustment parameter according to the defect information and the historical sequence through a long short-term memory network LSTM model; the edge calculation and control unit is combined with a material characteristic database to generate a hierarchical calibration instruction in a millisecond level, adopts fuzzy PID self-optimization control to dynamically set PID control parameters, and is combined with feed-forward compensation of the inertial measurement unit to perform accurate, smooth and robust control on an execution mechanism, so that the accuracy of calibration is improved. And full-automatic, high-precision and real-time online calibration of the printing process in a complex industrial environment is realized.

A finite time dynamic mechanical arm control method based on a speed reconstruction strategy comprises the steps that firstly, a mechanical arm system physical model containing joint torque is constructed, joint tracking errors are defined, and a mechanical arm system state space model is obtained; 2, constructing a finite-time high-bandwidth gain speed estimator, defining a state estimation error, and obtaining an estimation error dynamic equation; 3, defining an auxiliary state to obtain an auxiliary state dynamic equation; 4, constructing a finite time non-recursive adaptive controller; and 5, obtaining a final input torque of the controlled system, and outputting the final input torque as a joint torque. According to the method, through the processes of error modeling, state estimation, auxiliary conversion, control law design and closed-loop feedback, the dynamic tracking problem of the mechanical arm is converted into the finite time convergence problem of the error state, meanwhile, dynamic gain and robust compensation are introduced, the adaptability of a system to parameter uncertainty and external disturbance is improved, and the system reliability is improved. And a high-precision and high-robustness control target is realized.

The invention discloses an order-lifting sliding mode robust control method for an active-passive vibration isolation system, and belongs to the technical field of vibration isolation system control, and the method comprises the steps: introducing a feed-forward control path, predicting the vibration frequency and amplitude at a future moment based on the historical state information of a controlled system, building an overall vibration isolation target model based on the prediction, and carrying out the control of the overall vibration isolation target model. Resolving a target transfer function of an active execution mechanism through kinetic analysis of the active-passive vibration isolation system, so as to calculate a feedforward control signal; the method comprises the following steps: modeling an active-passive vibration isolation system, constructing a nonsingular terminal sliding mode surface, collecting system state error information, and generating a feedback control signal of an actual controlled system through integral controller design; a feedforward control signal and a feedback control signal are reasonably coupled and superposed to serve as control input of a controlled system. According to the order-lifting sliding mode robust control method for the active-passive vibration isolation system provided by the invention, the anti-interference capability and the stability of the active-passive vibration isolation system are improved.

The invention relates to the field of robot control, in particular to an air-ground cooperative multi-task constraint following control method based on a Udwadiia-Kalaaba equation, and the method comprises the following steps: building a kinetic model of an air-ground cooperative system composed of unmanned control equipment, and building system constraints of the air-ground cooperative system according to a U-K method; based on the analysis of the control problem, establishing a multi-domain behavior control constraint between the unmanned control devices; the constraint following error is expressed as a control following object, and a control problem is converted into a solvable constraint following problem; multi-domain behavior control tasks among unmanned control devices are integrated into a problem framework, constraint force in the problem framework is obtained, and an adaptive robust controller is designed. By adopting the multi-task processing framework, a plurality of tasks can be integrated into a unified constraint, the complexity of the system is effectively reduced, and the conciseness and reliability are improved.

The invention relates to the field of electric power automation, and discloses an AI-based remote control method and system for a high-low voltage power distribution cabinet, and the method comprises the steps: sensing the data of a power grid, and constructing a dynamic graph; aI state and risk prediction; carrying out robust optimization control decision making; issuing a control instruction and feeding back a state; performing rolling optimization and continuous learning; the system comprises a data perception and feature engineering module, an AI state and risk prediction module, a robust control decision module, a control instruction execution and feedback module and a system cooperation and continuous learning module. According to the method, the dynamic power grid attribute graph is constructed, the space-time diagram neural network is applied to quantify and predict uncertainty, the coordination instruction is generated in combination with robust control, and a closed loop is formed through rolling optimization and continuous learning, so that high-precision state prediction, robust control decision, equipment coordination and system evolution are realized; and the safety, the real-time performance and the self-adaptability of remote control are effectively improved.

The invention discloses a self-adaptive optical system parameter adjusting method and device combining offline pre-training and online reinforcement learning, and belongs to the field of intelligent control combining self-adaptive optics and reinforcement learning. Multi-modal state modeling, off-line strategy pre-training and an online strategy enhancement mechanism guided by behavior advantages are introduced, and key control parameters in the AO system are dynamically optimized by using a reinforcement learning algorithm. According to the method, multi-modal heterogeneous data can be effectively integrated, the dynamic modeling and prediction capability of the system is improved, and stable, efficient and robust control strategy updating is realized.