Described herein are systems and methods for inverse reinforcement learning to leverage the benefits of model-based optimization method and model-free learning method. Embodiments of a framework combining human behavior model with model predictive control are presented. The framework takes advantage of feature identification capability of a neural network to determine the reward function of model predictive control. Furthermore, embodiments of the present approach are implemented to solve the practical autonomous driving longitudinal control problem with simultaneous preference on safe execution and passenger comfort.

A system, method, and apparatus is provided for production processes to coordinate the utilization of the operational zone for process variables and automate the execution of the target settings to maximize the production goals without requiring human intervention.

This disclosure includes various systems and methods for determining an operating stage based on electrical conditions in electric power delivery systems and identifying a control strategy based upon the operating stage. The control strategy may be selected and customized to avoid or to ameliorate stresses in an electric power delivery system while maintaining the stability of electric power delivery systems. Various embodiments consistent with the present disclosure may include a distributed controller configured receive a plurality of indications of electrical conditions from a plurality of control devices in electrical communication with the electrical power delivery system. The distributed controller may determine an operating stage from among a plurality of operating stages based upon the plurality of indications of electrical conditions. The distributed controller may further identify a control strategy based upon the operating stage. The control strategy may be communicated to and implemented by the plurality of control devices.

Distributed analytics system used to control the operation of at least one monitored system; the system includes an architect subsystem and an edge subsystem, wherein the edge subsystem comprises at least one edge processing device associated with at least one monitored system. The architect subsystem deploys at least one analytic model to an edge processing device based on characteristics of a monitored system associated with the edge processing device, the analytic model to be used by the edge processing device to provide control signals to a monitored system; and, receives information related to the monitored system from the edge processing device, the information utilized by the architect subsystem to modify the analytic model deployed to the at least one edge processing device to improve system performance of the monitored system. An edge processing device receives an analytic model from the architect subsystem; provides control signals to the monitored system according to the analytic model; and, sends information related to the monitored system to the architect subsystem, the information to be used by the architect subsystem to modify the analytic model to improve system performance of the monitored system.

The present invention relates to a control system for use with industrial systems. More particularly, the present invention relates to an automated supervisory control system for providing decisions in relation to operation of one or more industrial plants or factories. Aspects and/or embodiments seek to provide an automated supervisory control system for providing and/or automating decisions in relation to operation of one or more industrial plants and factories.

It is disclosed a weather-predictive apparatus for controlling a climatization plant, comprising a weather-climate data sensor associated with a building, a processing unit and a signal transceiver. The signal transceiver is configured to transmit a current measured value of the weather-climate data associated with the building to a weather forecast device and it is configured to receive from the weather forecast device a plurality of weather forecast data associated with the building in a forecast time interval. The processing unit is configured to: calculate a change in a nominal activation instant of the climatization plant of the building, calculate a modified activation instant; check whether the current instant is equal to the modified activation instant. In case wherein the current instant is equal to the modified activation instant, the processing unit is configured to generate a command signal having a value representative of the activation of the climatization plant.

A building energy system includes HVAC equipment, green energy generation, a battery, and a predictive controller. The HVAC equipment provide heating or cooling for a building. The green energy generation collect green energy from a green energy source. The battery stores electric energy including at least a portion of the green energy provided by the green energy generation and grid energy purchased from an energy grid and discharges the stored electric energy for use in powering the HVAC equipment. The predictive controller generates a constraint that defines a total energy consumption of the HVAC equipment at each time step of an optimization period as a summation of multiple source-specific energy components and optimizes the predictive cost function subject to the constraint to determine values for each of the source-specific energy components at each time step of the optimization period.

A control device that outputs a command voltage value includes: a feedback controller to perform feedback control based on a current value, and output a first voltage value; a feedforward controller established by using an inverse model of a plant to be controlled; a feedforward voltage corrector to correct a voltage disturbance that occurs due to a modeling error between a plant model and the actual plant; a repetitive controller to learn periodic current disturbances; and a switch to control input/output to the repetitive controller. The switch is turned ON when the current response is in a steady state, and the repetitive controller learns the current disturbances and corrects a command current value. The feedback controller outputs the first voltage value by performing the feedback control based on a current value that is acquired from the corrected command current value. The feedforward controller inputs the command current value to the inverse model to generate a second voltage value. The control device outputs a sum of the first voltage value and the second voltage value as a command voltage value.

A method for exchanging electrical energy between a plurality of private electricity networks each comprising: a coordination unit, groups of electrical devices, and a plurality of relays arranged on the power supply lines of the devices. The method comprises: a) transmitting device status data from the devices to the coordination unit, b) establishing a set of electrical energy resources and requirements, c) comparing the set of resources and requirements with those of other private networks, d) allocating the requirements and resources to one another, e) ensuring at least part of the routing of electrical energy by means of digital certificates and instructions for drawing electrical energy that are sent to the relays, f) keeping record of each energy exchange in order to define a transaction between two private networks.

A method of determining one or more direct voltage reduction (DVR) factors at a point of load (POL) in a power grid voltage environment. An energy processing unit (EPU) is installed at each individual POL; each EPU includes at least one AC-AC series voltage regulator. An EPU-regulated output voltage is increased under specific controlled conditions. One or more independent DVR factors for each individual POL are determined during the increased and decreased EPU-regulated output voltage.