Control systems, methods, and algorithms to optimize vapor-compression systems, including refrigeration, heat pump and chiller operations

A computer-readable non-transitory medium storing instructions that cause one or more processors to perform a method, the method comprising:

• receiving sensed condition data from systemic sensors, the condition data comprising one or more of the temperature measurements, pressure measurements, and humidity measurements;
• receiving extrinsic condition data comprising location coordinates, and a weather forecast for a future time period;
• applying, to an optimization model, the sensed condition data, extrinsic condition data, and predictive thermodynamic state data generated from a mathematical model of the operation of the vapor-compression thermal system,
• wherein said mathematical model is not connected to the Internet and comprises ordinary differential equations (ODEs) that forecast the predictive thermodynamic states of the vapor-compression thermal system evolving over time, wherein the predictive thermodynamic states are generated through numerical integration of the ODEs in real-time to estimate thermodynamic states of the vapor-compression thermal system that are not directly measured by sensors in real-time, and
• wherein the optimization model is configured to output simultaneous optimal control commands to operational components of the vapor-compression thermal system;
• applying the optimal commands to a failure machine learning model, wherein the failure machine learning module is trained on input comprising failure cases associated with actuators loaded with the predictive thermodynamic states so as to output modified commands that address potential failures in the operational components of the vapor-compression thermal system.

A vapor-compression thermal system controller, comprising:

• one or more memory devices storing instructions;
• and
• one or more processor devices configured to execute the instruction to perform a method comprising:
• receiving sensed condition data from systemic sensors, the condition data comprising one or more of the temperature measurements, pressure measurements, and humidity measurements;
• receiving extrinsic condition data comprising location coordinates, and a weather forecast for a future time period;
• applying, to an optimization model, the sensed condition data, extrinsic condition data, and predictive thermodynamic state data generated from a mathematical model of the operation of the vapor-compression thermal system,
• wherein said mathematical model is not connected to the Internet and comprises ordinary differential equations (ODEs) that forecast the predictive thermodynamic states of the vapor-compression thermal system evolving over time, wherein the predictive thermodynamic states are generated through numerical integration of the ODEs in real-time to estimate thermodynamic states of the vapor-compression thermal system that are not directly measured by sensors in real-time, and
• wherein the optimization model is configured to output simultaneous optimal control commands to operational components of the vapor-compression thermal system;
• applying the optimal commands to a failure machine learning model, wherein the failure machine learning module is trained on input comprising failure cases associated with actuators loaded with the predictive thermodynamic states so as to output modified commands that address potential failures in the operational components of the vapor-compression thermal system.