6 results about "Controlling environment" patented technology

Systems and methods are provided for dynamically selecting a control policy from among several available control policies for controlling an energy system having multiple controllable assets. The performance of the selected control policy is monitored and a different control policy may be deployed in its place if the different control policy has a higher chance of providing better performance given the current control environment. Thus, as the control environment changes, the control policy that controls the power system may also be changed in an adaptive manner. In this way, the control policies may be changed as the control environment changes to provide an improved real-time performance compared to the use of a single control policy.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining causal models for controlling environments. One of the methods includes identifying a procedural instance; selecting control settings for the procedural instance, comprising, for a particular one of the controllable elements: assigning the procedural instance to a cluster for the particular controllable element in accordance with current values of a set of clustering parameters for the particular controllable element; and selecting a setting for the particular controllable element for the procedural instances based on a causal model that is specific to the cluster; obtaining environment responses to the selected control settings that define a value of the performance metric for the procedural instance; and updating, for the particular controllable element, the causal model for the cluster for the controllable element to which the procedural instance was assigned based on the value of the performance metric.

This invention relates to the field of industrial control system technology, and discloses an intelligent security integrated management and control method and system for digital operating environments. The method includes: acquiring multimodal raw data and converting it into asynchronous event pulse sequences; using a causal inference kernel to synchronously perform behavioral legality and physical probability inference to determine risks; dynamically generating and modifying control parameters after identifying threats and mapping them to security commands; and injecting the commands into the loop and attaching security posture tags through a symbiotic actuator. The system includes a perception encoding module, a causal inference kernel, a command arbitration unit, and a symbiotic actuator. This invention, through the deep integration of security and control logic, can accurately distinguish between attacks and fluctuations, maintaining production continuity while ensuring security, and constructing an intrinsic security system in industrial control environments.

The invention discloses a PID parameter adjustment method and device based on PILCO reinforcement learning migration PID control, and relates to the technical field of control architecture migration, and the method comprises the steps: obtaining target interaction data based on a PILCO reinforcement learning algorithm when an intelligent agent interacts with a control environment; a state error, an error integral and an error derivative in target interaction data are calculated according to an expected state of a controlled object, and a PID control parameter of a PID controller is identified by applying a recursive least square method combined with a forgetting factor. According to the method, parameter estimation is carried out by using a recursive least square method combined with a forgetting factor, a PILCO offline learning result is migrated and converted into PID control, a PILCO reinforcement learning algorithm and a PID controller architecture do not need to be changed, and the problems that the PILCO reinforcement learning algorithm is high in calculation complexity, poor in high-dimensional state space expansibility and difficult to deploy online are solved.

The application discloses a kind of gate real-time opening and closing control strategy optimization methods based on reinforcement learning, comprising the following steps: gate operation data are collected and preprocessed, to obtain state data set;Gate opening and closing control environment is constructed;Control strategy model based on reinforcement learning is constructed, the control strategy model uses Actor-Critic architecture, and is initialized;Gate opening adjustment action is obtained, and input gate opening and closing control environment, to obtain the next time state and reward value;MO-DDPG algorithm is used to train control strategy model, to obtain the control strategy model of training completion;Real-time operation data are collected, and input the control strategy model of training completion, to output gate opening control instruction;Gate opening is controlled according to gate opening control instruction, and the running data after control are fed back to gate control system, the application improves the real-time performance, accuracy and stability of gate opening and closing control, reduces the degree of artificial intervention.

This invention discloses an artificial intelligence scheduling and operation system and method based on external control and unidirectional interaction constraints. The system combines a data interaction channel with a security management terminal to enable task execution and data interaction of AI computing units in a controlled environment, ensuring full-process controllability, auditability, and traceability. The system includes a data access unit, isolated AI computing units, a data interaction channel, and a security management terminal. Data is transmitted from the data access unit to the isolated AI computing units through the data interaction channel, which is either a physically unidirectional structure or a logically unidirectional transmission structure constructed by the security management terminal based on access control policies. The security management terminal, as the sole control hub of the system, provides unified control over task scheduling, data interaction, result acquisition, and anomaly handling, ensuring that the isolated AI computing units operate only as passive execution units, lacking task initiation and communication decision-making capabilities. This invention is applicable to environments without or with limited networks, achieving secure, controllable, and auditable operation of the AI ​​system through external control and unidirectional interaction constraints.