Wind turbine control based on optimizing and non-optimizing controller routines is disclosed. A first controller implements a model predictive control (MPC) routine for calculating a predicted first control value. A second controller implements a non-optimizing control routine for calculating a second control value. An actuator controller unit determines an actuator control signal by combining the predicted first control value and the second control value.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for shaping compute load using virtual capacity. In one aspect, a method includes obtaining a load forecast that indicates forecasted future compute load for a cell, obtaining a power model that models a relationship between power usage and computational usage for the cell, obtaining a carbon intensity forecast that indicates a forecast of carbon intensity for a geographic area where the cell is located, determining a virtual capacity for the cell based on the load forecast, the power model, and the carbon intensity forecast, and providing the virtual capacity for the cell to the cell.

A method of controlling building service systems associated with a building for optimizing a plurality of building performance parameters with respect to a region of the building, the building service systems including an air-conditioning and/or heating system, a lighting system and a shading system, is provided. The method includes predicting, based on a shading and lighting prediction model, a visual comfort condition and a lighting condition with respect to the region of the building; optimizing, based on a first multi-component cost function including a plurality of components relating to a plurality of lighting or thermal performance parameters with respect to the region of the building, one or more first control parameters for controlling the lighting system and the shading system based on the predicted visual comfort condition and the predicted lighting condition; predicting, based on a building dynamics model, a plurality of building response parameters based on the predicted visual comfort condition and the predicted lighting condition associated with the region of the building; optimizing, based on a second multi-component cost function including a plurality of components relating to the plurality of building performance parameters, one or more second control parameters for controlling the air-conditioning and/or heating system based on the predicted plurality of building response parameters.

A main computing system maintains and optimizes a predictive control model for an energy system, wherein the main computing system receives state information for the energy system, optimizes the predictive control model, and generates control rules for control of the energy system. The one or more local computing systems, each have a local memory for storing control rules for controlling the associated local state. The main computing system receives local state information and updates control rules, wherein the updated control rules comprise a subset of the control rules generated by the main computing system selected to be appropriate to the local state information received at the main computing system.

A web services platform operates to monitor and control equipment of a building management system. The web services platform includes a data collector and a timeseries service. The data collector is configured to collect feedback samples provided by one or more sensors of a building management system and generate one or more feedback timeseries including a plurality of the feedback samples. The timeseries service is configured to identify a feedback control workflow that uses the feedback timeseries as an input and defines one or more processing operations to be applied to the feedback samples of the feedback timeseries, perform the one or more processing operations defined by the feedback control workflow to generate a control signal timeseries including a set of control signal samples, and provide a control signal including at least one of the control signal samples or the control signal timeseries as an output to controllable building equipment of the building management system that operate using the control signal as an input.

Disclosed herein are related to a system, a method, and a non-transitory computer readable medium for operating an energy plant. In one aspect, the system generates a regression model of a produced thermal energy load produced by a supply device of the plurality of devices. The system predicts the produced thermal energy load produced by the supply device for a first time period based on the regression model. The system determines a heat capacity of gas or liquid in the loop based on the predicted produced thermal energy load. The system generates a model of mass storage based on the heat capacity. The system predicts an induced thermal energy load during a second time period at a consuming device of the plurality of devices based on the model of the mass storage. The system operates the energy plant according to the predicted induced thermal energy load.

A hierarchical model predictive control (HMPC) method based on fuzzy neural network for wastewater treatment process (WWTP) is designed to realize hierarchical control of dissolved oxygen (DO) concentration and nitrate nitrogen concentration. In view of the difference of time scales in WWTP, it is difficult to accurately control the concentration of DO and nitrate nitrogen. The disclosure establishes a HMPC structure according to different time scales. Then, the concentration of DO and nitrate nitrogen is controlled with different frequencies. It not only conforms to the operation characteristics of WWTP, but also solves the problem of poor operation performance of multivariable model predictive control. The experimental results show that the HMPC method can achieve accurate on-line control of DO concentration and nitrate nitrogen concentration with different time scales.

A system includes a plurality of thermostats corresponding to a plurality of HVAC systems that serve a plurality of spaces and a computing system communicable with the plurality of thermostats via a network. The computing system is configured to, for each space of the plurality of spaces, obtain a set of training data relating to thermal behavior of the space, identify a model of thermal behavior of the space based on the set of training data, perform a model predictive control process using the model of thermal behavior of the space to obtain a temperature setpoint for the space, and provide the temperature setpoint to the thermostat corresponding to the HVAC system serving the space. The plurality of thermostats are configured to control the plurality of HVAC systems in accordance with the temperature setpoints.

There is a need for more effective and efficient data modeling and/or data visualization solutions. This need can be addressed by, for example, solutions for performing data modeling and/or data visualization in an effective and efficient manner. In one example, solutions for generating a data model with dynamic relational awareness are disclosed. In another example, solutions for processing data retrieval queries using data models with dynamic relational awareness are disclosed. In yet another example, solutions for generating data visualizations using data models with dynamic relational awareness are disclosed. In a further example, solutions for integrating external data objects into data models with dynamic relational awareness are disclosed.

A matrix converter system operating in current control mode is provided. The matrix converter includes a switching matrix coupled between a low voltage side and a high voltage side, wherein the matrix converter is coupled at its low voltage side to a generator for receiving an input power including an input current and transforming the input power into an output power at its high voltage side, wherein a load is coupled to the high voltage side. The control system includes a load observer and a model predictive controller (MPC). The load observer is configured to estimate a load current that flows to the load from the high voltage side of the matrix converter as a function of a switching state of the switching matrix, an output voltage output at the high voltage side of the matrix converter, and the input current. The MPC is configured to select the switching state of the switching matrix to meet one or more control objectives defined by minimization of a multi-objective function that tracks the output voltage and the input current using the estimated load current.