Variable refrigerant flow system with multi-level model predictive control

Abstract

A model predictive control system is used to optimize energy costs for a variable refrigerant flow (VRF) system. The VRF system includes an outdoor subsystem and a plurality of indoor subsystems. The model predictive control system includes a high-level model predictive controller (MPC) and a plurality of low-level indoor MFCs. The high-level MFC performs a high-level optimization to generate an optimal indoor subsystem load profile curve for each of the plurality of indoor subsystems. The optimal indoor subsystem load profile curve optimizes energy costs. Each of the low-level indoor MFCs performs a low-level optimization to generate optimal indoor setpoints for one or more indoor VRF units of the respective indoor subsystem. The indoor setpoints can include temperature setpoints and / or refrigerant flow setpoints for the indoor VRF units.

Description

[0001] This application is a divisional application of the invention patent application filed on June 28, 2017, with international application number PCT / US2017 / 039798, national application number 201780041246.5, and entitled "Variable Refrigerant Flow System with Multi-Level Model Predictive Control".

[0002] Cross-referencing of related patent applications This application claims the benefit and priority of U.S. Provisional Patent Application No. 62 / 357,338, filed June 30, 2016, the entire disclosure of which is incorporated herein by reference. Background Technology

[0003] This disclosure generally relates to a model predictive control (MPC) system for building heating, ventilation, and air conditioning (HVAC) systems. More specifically, this disclosure relates to an MPC system for optimizing the energy cost of a variable refrigerant flow (VRF) system that includes an outdoor subsystem and multiple indoor subsystems.

[0004] Commercial buildings account for approximately 20% of total energy consumption in the United States, and basic energy expenditures amount to about $200 billion annually. The Energy Information Administration projects that commercial floor space and basic energy consumption will continue to grow in the future. On the other hand, average energy prices are expected to remain relatively stable. Therefore, the amount of energy consumed in commercial buildings will continue to increase significantly. Given the importance of these energy cost values ​​and their projected growth, buildings have become a primary target for control strategies designed to reduce consumption or improve efficiency, particularly in the area of ​​temperature control.

[0005] Many HVAC systems in commercial buildings and educational facilities use simple on / off controllers and proportional-integral-derivative (PID) controllers to control their equipment. These systems rely on temperature controllers whose sole objective is to converge to and maintain a desired temperature setpoint within certain tolerances. However, a better objective is to minimize total energy consumption or total energy costs. In utility markets with time-varying prices, there is potential for cost savings by temporarily shifting heating or cooling loads through some form of energy storage. To achieve these savings, predictive optimization can be used in conjunction with system modeling to forecast future loads. Load shifting alleviates the burden on power plants during peak hours, allowing them to operate more efficiently. Furthermore, coolers operate more efficiently at night when cooling water temperatures are lower.

[0006] MPC is an advanced process control method that has achieved great success over the past two decades. MPC uses a system model that correlates inputs (control actions) with outputs (process measurements). This model is used to predict process variables based on actions taken by the controller over a period of time, known as the range. At each step, MPC uses this model to determine a sequence of control actions that achieves an objective (such as minimizing tracking error or input usage) while adhering to process constraints such as equipment capacity and safety limits to solve an online optimization problem. The first control action in the sequence is implemented, and the optimization problem is solved again at the next step after obtaining new measurements. In economic MPC, the objective of the optimization problem is to minimize the total cost.

[0007] Economically optimal control systems have not yet been widely deployed in the HVAC industry. A fundamental obstacle to the successful deployment of MPC in HVAC systems is the large number of building areas. To implement MPC in an HVAC system, it is desirable to solve the optimization problem within a reasonably short time (e.g., approximately a few minutes). Campus-wide implementations may involve hundreds of buildings and areas. A single combined control system for these applications is impractical and undesirable because the resulting individual optimization problem is too large to be solved in real time. Summary of the Invention

[0008] One embodiment of this disclosure is a model predictive control system for optimizing the energy cost of a variable refrigerant flow (VRF) system. The VRF system includes an outdoor subsystem and multiple indoor subsystems. The model predictive control system includes a high-level model predictive controller (MPC) and multiple low-level indoor MPCs. The high-level MPC is configured to perform high-level optimization to generate an optimal indoor subsystem load distribution curve for each of the multiple indoor subsystems. The optimal indoor subsystem load distribution curve optimizes the energy cost. Each of the low-level indoor MPCs corresponds to one of the indoor subsystems and is configured to perform low-level optimization to generate an optimal indoor setpoint for one or more indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution curve of the corresponding indoor subsystem. Each of the low-level indoor MPCs is configured to operate the indoor VRF unit of the corresponding indoor subsystem using the optimal indoor setpoint.

[0009] In some embodiments, each of the optimal indoor subsystem load distribution curves includes the thermal energy allocation of one of the indoor subsystems at each of a plurality of time steps within the optimization period. In some embodiments, each low-level indoor MPC is configured to use the thermal energy allocation of the corresponding indoor subsystem as an optimization constraint when performing the low-level optimization.

[0010] In some embodiments, the optimal indoor setpoint includes at least one of a refrigerant flow setpoint for each of the indoor VRF units and a temperature setpoint for one or more building areas controlled by the indoor VRF units.

[0011] In some embodiments, each indoor subsystem includes multiple building zones. Each of the lower-level indoor MPCs can be configured to generate an optimal indoor temperature setpoint for each of the multiple building zones in the respective indoor subsystem.

[0012] In some embodiments, each indoor subsystem includes a plurality of indoor VRF units. Each of the lower-level indoor MPCs can be configured to generate an optimal refrigerant flow decision for each of the plurality of indoor VRF units in the respective indoor subsystem and use the optimal refrigerant flow decision to calculate at least one of a flow setpoint, a temperature setpoint, and a valve setpoint for each of the plurality of indoor VRF units.

[0013] In some embodiments, each of the lower-level indoor MPCs is configured to identify the relationship between the thermal load and refrigerant flow rate of each of the indoor VRF units in the corresponding indoor subsystem. Each of the lower-level indoor MPCs can use the identified relationship between the thermal load and refrigerant flow rate to generate the optimal refrigerant flow rate decision for each of the indoor VRF units in the corresponding indoor subsystem.

[0014] In some embodiments, the advanced MPC is configured to generate a predicted temperature distribution curve for each indoor subsystem, derived from the optimal indoor subsystem load distribution curve. In some embodiments, the low-level indoor MPC is configured to generate the optimal indoor setpoint by minimizing the error between the indoor subsystem temperature and the predicted temperature distribution curve.

[0015] In some embodiments, each of the indoor subsystems represents a separate building that is thermally decoupled from each other, such that no direct heat exchange occurs between the indoor subsystems.

[0016] In some embodiments, the high-level MPC is configured to generate an optimal outdoor unit demand distribution curve for the outdoor subsystem. The system may include a low-level outdoor MPC configured to perform low-level optimization to generate optimal outdoor setpoints for one or more outdoor VRF units of the demand-constrained outdoor subsystem based on the optimal outdoor unit demand distribution curve. The low-level outdoor MPC may be configured to operate the outdoor VRF units using the optimal outdoor setpoints.

[0017] In some embodiments, the advanced MPC is configured to determine: the optimal thermal energy allocation for each of the plurality of indoor subsystems at each of a plurality of time steps during the optimization period; and the optimal thermal energy or refrigerant state allocation for the outdoor subsystem at each of the plurality of time steps during the optimization period.

[0018] Another embodiment of this disclosure is a method for optimizing the energy cost of a variable refrigerant flow (VRF) system. The VRF system includes an outdoor subsystem and multiple indoor subsystems. The method includes performing high-level optimization at a high-level model predictive controller (MPC) to generate an optimal indoor subsystem load distribution curve for each of the multiple indoor subsystems. The optimal indoor subsystem load distribution curve optimizes the energy cost. The method includes providing the optimal indoor subsystem load distribution curve from the high-level MPC to multiple low-level indoor MPCs. Each of the low-level indoor MPCs corresponds to one of the multiple indoor subsystems. The method includes performing low-level optimization at each of the low-level indoor MPCs to generate optimal indoor setpoints for one or more indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution curve of the corresponding indoor subsystem. The method includes operating the indoor VRF units in each of the multiple indoor subsystems using the optimal indoor setpoints.

[0019] In some embodiments, each of the optimal indoor subsystem load distribution curves includes the thermal energy allocation of one of the indoor subsystems at each of a plurality of time steps within an optimization period. In some embodiments, performing the low-level optimization includes using the thermal energy allocation of the respective indoor subsystem as an optimization constraint.

[0020] In some embodiments, the optimal indoor setpoint includes at least one of a refrigerant flow setpoint for each of the indoor VRF units and a temperature setpoint for one or more building areas controlled by the indoor VRF units.

[0021] In some embodiments, each indoor subsystem includes multiple building zones. Performing the low-level optimization may include generating an optimal indoor temperature setpoint for each of the multiple building zones in the respective indoor subsystem.

[0022] In some embodiments, each indoor subsystem includes a plurality of indoor VRF units. Performing the low-level optimization may include: generating an optimal refrigerant flow decision for each of the plurality of indoor VRF units in the respective indoor subsystem and using the optimal refrigerant flow decision to calculate at least one of a flow setpoint, a temperature setpoint, and a valve setpoint for each of the plurality of indoor VRF units.

[0023] In some embodiments, performing the low-level optimization includes: identifying the relationship between the thermal load and refrigerant flow rate of each of the indoor VRF units in the respective indoor subsystem and using the identified relationship between the thermal load and refrigerant flow rate to generate the optimal refrigerant flow rate decision for each of the indoor VRF units in the respective indoor subsystem.

[0024] In some embodiments, performing the high-level optimization includes generating a predicted temperature distribution curve for each indoor subsystem, derived from the optimal indoor subsystem load distribution curve. In some embodiments, performing the low-level optimization includes generating the optimal indoor setpoint by minimizing the error between the indoor subsystem temperature and the predicted temperature distribution curve.

[0025] In some embodiments, performing the high-level optimization includes generating an optimal outdoor unit demand distribution curve for the outdoor subsystem. The method may further include performing low-level optimization at a low-level outdoor model predictive controller to generate optimal outdoor setpoints for one or more outdoor VRF units of the demand-constrained outdoor subsystem based on the optimal outdoor unit demand distribution curve. The method may further include operating the outdoor VRF units using the optimal outdoor setpoints.

[0026] In some embodiments, performing the advanced optimization includes determining the following: optimal thermal energy allocation for each of the plurality of indoor subsystems at each of a plurality of time steps during the optimization period; and optimal thermal energy or refrigerant state allocation for the outdoor subsystem at each of the plurality of time steps during the optimization period.

[0027] Another embodiment of this disclosure is a variable refrigerant flow (VRF) system. The VRF system includes multiple indoor subsystems, an outdoor subsystem, an advanced model predictive controller (MPC), and multiple low-level indoor MPCs. Each indoor subsystem includes one or more indoor VRF units. The outdoor subsystem includes one or more outdoor VRF units. The advanced MPC is configured to perform advanced optimization to generate an optimal indoor subsystem load distribution profile for each of the multiple indoor subsystems. The optimal indoor subsystem load distribution profile optimizes energy costs. Each low-level indoor MPC corresponds to one of the indoor subsystems and is configured to perform low-level optimization to generate optimal indoor setpoints for the indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution profile of the corresponding indoor subsystem.

[0028] Those skilled in the art will recognize that the overview is illustrative only and is not intended to be limiting in any way. Other aspects, inventive features, and advantages of the apparatus and / or process described herein, as defined only by the claims, will become clear from the detailed description set forth herein and taken in conjunction with the accompanying drawings. Attached Figure Description

[0029] Figures 1A to 1B The accompanying drawings are of a variable refrigerant flow (VRF) system having one or more outdoor VRF units and multiple indoor VRF units, according to some embodiments.

[0030] Figure 2A These are simplified diagrams illustrating some embodiments. Figures 1A to 1B The VRF system operates in cooling mode.

[0031] Figure 2B It is a directed graph that illustrates the balance of refrigerant states when the VRF system operates in cooling mode, according to some embodiments.

[0032] Figure 3A These are simplified diagrams illustrating some embodiments. Figures 1A to 1B The operation of the VRF system in heating mode.

[0033] Figure 3B It is a directed graph that illustrates the balance of refrigerant states when the VRF system operates in heating mode, according to some embodiments.

[0034] Figure 4A These are simplified diagrams illustrating some embodiments. Figures 1A to 1B The VRF system operates in a combined heating and cooling mode.

[0035] Figure 4BIt is a directed graph that illustrates the balance of refrigerant states when the VRF system operates in a combined heating and cooling mode, according to some embodiments.

[0036] Figure 5 This is a block diagram of a control system for multiple VRF systems according to some embodiments.

[0037] Figure 6 This is a block diagram of a distributed model predictive control system according to some embodiments, the distributed model predictive control system having a high-level model predictive controller, several low-level indoor model predictive controllers, and a low-level outdoor model predictive controller.

[0038] Figure 7 This is a block diagram that illustrates in more detail some embodiments. Figure 6 Advanced model predictive controller.

[0039] Figure 8 This is a block diagram that illustrates in more detail some embodiments. Figure 6 A low-level indoor model predictive controller.

[0040] Figure 9 This is a flowchart of high-level and distributed low-level model predictive control techniques according to some embodiments, which can be used for optimization when the indoor subsystem load is used as a constraint in low-level optimization. Figure 6 The energy cost of the MPC system.

[0041] Figure 10 This is another flowchart of a high-level and distributed low-level model predictive control technique according to some embodiments, which can be used to optimize low-level optimization by tracking a temperature distribution curve provided by high-level optimization. Figure 6 The energy cost of the MPC system. Detailed Implementation

[0042] Referring generally to the accompanying drawings, a model predictive control (MPC) system for a variable refrigerant flow (VRF) system according to some embodiments is illustrated. MPC is a control technique that uses a model of the controlled system to correlate system inputs (e.g., control actions, setpoints, etc.) with system states and system outputs (e.g., measurements, process variables, etc.). The model can be used to predict system states and system outputs based on actions taken by the controller at each time step during an optimization period. At each time step, MPC uses the system model to determine a sequence of control actions to achieve objectives (e.g., minimizing tracking error, minimizing energy costs, etc.) while adhering to process constraints such as equipment capacity and safety limits (e.g., temperature constraints, equipment switching constraints, etc.) to solve an online optimization problem. The first control action in the sequence is implemented, and the optimization problem is solved again at the next time step after obtaining new measurements.

[0043] The MPC system described in this paper can optimize (e.g., minimize) the total cost of energy used to provide heating and / or cooling to a building or campus. Numerous studies have demonstrated that MPC outperforms existing control systems due to its ability to predict the future and anticipate events before they occur. By using passive thermal energy storage (TES) through the building's mass, MPC enables the shift of energy loads from peak to off-peak periods. Energy costs can be reduced by concentrating equipment use during periods of lower resource prices, while maintaining comfort limits within each building.

[0044] In some embodiments, the MPC system includes an MPC layer and a monitoring layer. The MPC layer receives measurement results from the monitoring layer and provides setpoints to the monitoring layer. The MPC layer can generate optimal values ​​for various decision variables, including, for example, zone temperature setpoints, refrigerant flow rates, and equipment on / off decisions. The MPC layer can use system models, such as zone temperature and cooling / heating load models, cooling / heating load and temperature setpoint models, and equipment models, to determine the optimal values ​​of the decision variables. The MPC layer can determine the optimal values ​​of the decision variables by performing an optimization process subject to several constraints. These constraints may include comfort limits for zone air temperature, equipment capacity constraints, refrigerant balance constraints, and rate of change limits for monitoring layer equipment.

[0045] Solving a single MPC problem to determine the optimal values ​​for all decision variables can be difficult for large-scale applications. For example, some VRF systems may include thousands of discrete regions and thousands of unique HVAC units. Discrete decisions (e.g., turning a device on / off) can lead to mixed-integer optimization problems, which further increases the complexity. Due to the difficulty and computational complexity of the MPC problem, an MPC layer can decompose the entire MPC problem into several smaller, more manageable optimization problems.

[0046] A distributed MPC system can decompose the entire MPC problem into high-level and low-level optimization problems. The high-level problem can be solved by a high-level model predictive controller to determine the load distribution curves for each of multiple low-level indoor subsystems and the demand distribution curve for the outdoor subsystem. Indoor subsystems may include indoor VRF units, and outdoor subsystems may include outdoor VRF units. In some embodiments, the high-level controller uses an aggregated low-level model of each indoor subsystem to reduce computational complexity. The high-level controller can determine the load distribution curves that optimize (e.g., minimize) the total operating cost of the MPC system over an optimization period. Each load distribution curve may include the load value at each time step within the optimization period. The high-level controller can provide the load distribution curves to multiple low-level indoor model predictive controllers. The low-level indoor controllers can use the load distribution curves as constraints to define the maximum allowable load value for each indoor subsystem at each time step within the optimization period.

[0047] The low-level optimization problem can be further decomposed into a low-level outdoor optimization problem and one or more low-level indoor optimization problems. Each low-level indoor problem can be solved by one of the low-level indoor controllers to determine the zone temperature setpoint and / or refrigerant flow setpoint for the indoor VRF unit. Each low-level indoor controller can determine the zone temperature setpoint and / or refrigerant flow setpoint that optimizes (e.g., minimizes) the energy consumption of the corresponding indoor subsystem while maintaining the zone temperature within defined temperature limits and not exceeding the load values ​​provided by the higher-level controller. Alternatively, each low-level indoor controller can determine the temperature setpoint and / or refrigerant flow setpoint based on the higher-level optimization problem to track the average indoor subsystem temperature (e.g., predicted subsystem temperature state). These and other components of the MPC system are described in more detail below.

[0048] Variable refrigerant flow system Now refer to Figures 1A to 1BA variable refrigerant flow (VRF) system 100 according to some embodiments is illustrated. The VRF system 100 is shown as including a plurality of outdoor VRF units 102 and a plurality of indoor VRF units 104. The outdoor VRF units 102 may be located outside a building and may operate to heat or cool refrigerant. The outdoor VRF units 102 may consume electricity to switch the refrigerant between a liquid phase, a gas phase, and / or a superheated gas phase. The indoor VRF units 104 may be distributed throughout various building areas within the building and may receive heated or cooled refrigerant from the outdoor VRF units 102. Each indoor VRF unit 104 may provide temperature control for the specific building area where the indoor VRF unit is located.

[0049] The main advantage of the VRF system is that some indoor VRF units 104 can operate in cooling mode, while others operate in heating mode. For example, each of the outdoor VRF unit 102 and indoor VRF units 104 can operate in heating mode, cooling mode, or off mode. Each building area can be controlled independently and can have different temperature setpoints. In some embodiments, each building has up to three outdoor VRF units 102 located outside the building (e.g., on the roof) and up to 128 indoor VRF units 104 distributed throughout the building (e.g., in the various building areas).

[0050] The VRF system 100 has many different configurations. In some embodiments, the VRF system 100 is a two-pipe system, wherein each outdoor VRF unit 102 is connected to a single refrigerant return line and a single refrigerant outlet line. In a two-pipe system, all outdoor VRF units 102 operate in the same manner because only one type of refrigerant, either heating or cooling, can be provided via the single refrigerant outlet line. In other embodiments, the VRF system 100 is a three-pipe system, wherein each outdoor VRF unit 102 is connected to a refrigerant return line, a hot refrigerant outlet line, and a cold refrigerant outlet line. In a three-pipe system, both heating and cooling can be provided simultaneously via dual refrigerant outlet lines. Examples of three-pipe VRF systems that can be used for the VRF system 100 are described in detail below.

[0051] Now refer to Figures 2A to 4B Several simplified diagrams illustrating the operation of a VRF system 100 according to some embodiments in cooling mode, heating mode, and combined heating / cooling mode are shown. Each outdoor VRF unit 102 may include one or more heat exchangers 106 (e.g., Figure 2A , Figure 3A and Figure 4A(As shown in the diagram). When the outdoor VRF unit 102 is operating in cooling mode, the heat exchanger 106 can function as the condenser 128 (as shown in the diagram). Figure 2B and Figure 4B (As shown in the diagram) operates to provide cooling for the refrigerant. When the outdoor VRF unit 102 operates in heating mode, the heat exchanger 106 can function as an evaporator 130 (as shown in the diagram). Figure 3B The condenser 128 and evaporator 130 are designed to provide heating for the refrigerant. It is envisioned that the condenser 128 and evaporator 130 may exist as separate units within the outdoor VRF unit 102 or as a heat exchanger 106, which may operate as both condenser 128 and evaporator 130 depending on the operating mode of the outdoor VRF unit 102. Although only two outdoor VRF units 102 are shown, it should be understood that the VRF system 100 may include any number of... Outdoor VRF unit 102.

[0052] Each indoor VRF unit 104 may include one or more heat exchangers 107 (e.g., ...). Figure 2A , Figure 3A and Figure 4A (As shown in the diagram). When the indoor VRF unit 104 is operating in cooling mode, the heat exchanger 107 can function as the evaporator 105 (e.g., Figure 2B and Figure 4B (As shown in the diagram) operates to provide cooling for the air delivered to the building area. When the indoor VRF unit 104 operates in heating mode, the heat exchanger 107 can function as a condenser 103 (as shown in the diagram). Figure 3B The condenser 103 and evaporator 105 are designed to provide heating for air delivered to the building area. It is envisioned that the condenser 103 and evaporator 105 may exist as separate units within the indoor VRF unit 104 or as a heat exchanger 107, which may operate as both condenser 103 and evaporator 105 depending on the operating mode of the indoor VRF unit 104. Although only three indoor VRF units 104 are shown, it should be understood that the VRF system 100 may include any number of... Indoor VRF unit 104.

[0053] Specific reference Figures 2A to 2BThe diagram illustrates the operation of a VRF system 100 according to some embodiments in a cooling mode. In cooling mode, the heat exchanger 106 of the outdoor VRF unit 102 operates as a condenser 128 to condense superheated gaseous refrigerant 124 into liquid refrigerant 120. The liquid refrigerant 120 from the heat exchanger 106 flows through an expansion valve (EEV) 108 and onto the heat exchanger 107 of the indoor VRF unit 104. In cooling mode, the heat exchanger 107 operates as an evaporator 105 to evaporate the liquid refrigerant 120 into gaseous refrigerant 122, thereby absorbing heat from the air within the building area and providing cooling to the building area. A solenoid valve 110 allows the gaseous refrigerant 122 to return to one or more compressors 112 of the outdoor unit 102. The compressors 112 compress the gaseous refrigerant 122 to produce superheated gaseous refrigerant 124, which is supplied to the condenser 128.

[0054] Now refer to Figures 3A to 3B The diagram illustrates the operation of a VRF system 100 according to some embodiments in a heating mode. In heating mode, the heat exchanger 106 of the outdoor VRF unit 102 operates as a condenser 130 to evaporate liquid refrigerant 120 from the indoor VRF unit 104. The heat exchanger 106 transfers heat to the liquid refrigerant 120, causing the liquid refrigerant 120 to evaporate and form gaseous refrigerant 122. The gaseous refrigerant 122 is provided to a compressor 112, which compresses the gaseous refrigerant 122 to form superheated gaseous refrigerant 124. The superheated gaseous refrigerant 124 is then provided to the heat exchanger 107 of the indoor VRF unit 104. The heat exchanger 107 operates as a condenser 102 to condense the superheated gaseous refrigerant 124 by transferring heat from the superheated gaseous refrigerant 124 to the building area, causing the superheated gaseous refrigerant 124 to lose heat and become liquid refrigerant 120. The liquid refrigerant 120 is then returned to the heat exchanger 106 of the outdoor VRF unit 102.

[0055] Now refer to Figures 4A to 4B The diagram illustrates the operation of a VRF system 100 according to some embodiments in a combined heating and cooling mode. In the combined heating / cooling model, some indoor and outdoor VRF units 102 to 104 operate in heating mode, while others operate in cooling mode. For example, indoor VRF unit-2 is shown operating in heating mode, while indoor VRF units-1 and-m are shown operating in cooling mode. Outdoor VRF units-1 and-n are both shown operating in cooling mode.

[0056] The operation of the outdoor VRF unit 102 in cooling mode is the same as previously referenced. Figures 2A to 2B The description is the same. For example, outdoor VRF unit 102 can receive gaseous refrigerant 122 and condense it into liquid refrigerant 120. The liquid refrigerant 120 can be directed to indoor VRF units-1 and-m to provide cooling for areas-1 and-m. Heat exchangers 107 of indoor VRF units-1 and-m operate as evaporators 105 by absorbing heat from building areas-1 and-m, thereby converting the liquid refrigerant 120 into gaseous refrigerant 122. The gaseous refrigerant 122 is then delivered to compressor 112 of outdoor VRF unit 1022. Compressor 112 compresses the gaseous refrigerant 122 to form superheated gaseous refrigerant 124. The superheated gaseous refrigerant 124 can be provided to heat exchanger 106 of outdoor VRF unit 102, which operates as a condenser 128 to condense the gaseous refrigerant 122 into liquid refrigerant 120. Superheated gaseous refrigerant 124 can also be supplied to indoor VRF unit-2 and used to provide heating to building area-2.

[0057] The operation of the indoor VRF unit-2 in heating mode can be compared with the previous reference. Figures 3A to 3B The description is the same. For example, the heat exchanger 107 of indoor VRF unit-2 can operate as a condenser 103 by discharging heat from superheated gaseous refrigerant 124 to building area-2, thereby turning superheated gaseous refrigerant 124 into liquid refrigerant 120. The liquid refrigerant 120 can be directed to the heat exchangers 107 of indoor VRF units-1 and indoor VRF units-m, which operate as evaporators 105 to absorb heat from building area-1 and building area-m, as previously described.

[0058] In any of the operating modes, the VRF system 100 can operate to ensure that the refrigerant state remains balanced. For example, when operating in cooling mode, the VRF system 100 can operate the outdoor VRF unit 102 and the indoor VRF unit 104 to ensure that the outdoor VRF unit 102 converts the gaseous refrigerant 122 into liquid refrigerant 120 at the same rate as the indoor VRF unit 104 converts the liquid refrigerant 120 into gaseous refrigerant 122. Similarly, when operating in heating mode, the VRF system 100 can operate the outdoor VRF unit 102 and the indoor VRF unit 104 to ensure that the outdoor VRF unit 102 converts the liquid refrigerant 120 into superheated gaseous refrigerant 124 at the same rate as the indoor VRF unit 104 converts the superheated gaseous refrigerant 124 into liquid refrigerant 120.

[0059] In each operating mode, the VRF system 100 can operate the outdoor VRF unit 102 and the indoor VRF unit 104 to ensure that the amount of each refrigerant state (e.g., liquid refrigerant 120, gaseous refrigerant 122, and superheated gaseous refrigerant 124) generated by the outdoor VRF unit 102 and the indoor VRF unit 104 is equal to the amount of each refrigerant state consumed by the outdoor VRF unit 102 and the indoor VRF unit 104. In other words, the VRF system 100 can balance the rate of adding and removing refrigerant from each refrigerant state. In some embodiments, the VRF system 100 imposes a mass balance constraint or a volume balance constraint to ensure that the net amount of refrigerant in each refrigerant state remains balanced at each time step of the optimization period.

[0060] In some embodiments, a predictive energy cost optimization framework is used to control the VRF system 100. For example, the VRF system 100 may include one or more controllers performing high-level and low-level optimizations. High-level optimization seeks to optimize the electricity usage cost plus peak electricity charges (i.e., electricity demand charges) across the entire VRF system 100, subject to several system constraints, by manipulating the requested cooling or heating loads delivered to each zone and the operating modes of indoor and outdoor VRF units 102 to 104. Constraints imposed in high-level optimization may include system constraints, such as refrigerant state balance (as described above) and zone temperature constraints. Zone temperature constraints may require that the temperature in each building zone be maintained within an acceptable temperature range to maintain occupant comfort.

[0061] Low-level optimization can use the requested heating and cooling loads calculated by high-level optimization for each building area as input data. Low-level optimization can manipulate the zone temperature setpoints for each building area so that the zone heating and cooling loads track the requested heating or cooling load distribution curves calculated in high-level optimization.

[0062] In some embodiments, low-level optimization is distributed across several low-level model predictive controllers, each operable to determine a temperature setpoint for a specific building area. For example, the control system may include a high-level model predictive controller (MPC) and several low-level MPCs. The high-level MPC can determine the optimal load distribution curve for each building area and can assign the optimal load distribution curve to the low-level MPCs for that building area. Each low-level MPC can be configured to control a specific building area and can receive the load distribution curve for that building area from the high-level MPC. Each low-level MPC can use the load distribution curve from the high-level MPC to determine the optimal temperature setpoint for that building area. (Refer to...) Figure 6 An example of this distributed implementation is described in more detail.

[0063] Now refer to Figure 5 A block diagram of a control system 500 for multiple VRF systems 510, 520, and 530 according to some embodiments is shown. Each of the VRF systems 510 to 530 may include some or all of the components and / or features of the VRF system 100, as shown in reference to Figures 1A to 4B Described. By introducing an additional control layer (e.g., a supervisory layer) operating on top of the high-level and low-level optimization frameworks, the aforementioned optimization framework can be extended to larger systems including multiple VRF systems 510 to 530. For example, a predictive cost optimization controller can act as a coordinator to coordinate the time-varying power usage of the multiple VRF systems 510 to 530, enabling the multiple VRF systems 510 to 530 to achieve optimal energy cost performance (e.g., the minimum total energy cost of the entire group of VRF systems 510 to 530).

[0064] In various embodiments, cost optimization performed by the predictive cost optimization controller may take into account energy costs (e.g., $ / kWh of electricity consumed), demand charges (e.g., $ / kW of peak power consumption), peak load contribution costs, and / or monetary incentives for participation in an incentive-based demand response (IBDR) program. Several examples of cost optimization that may be performed by the predictive cost optimization controller are described in detail in U.S. Patent Application No. 15 / 405,236, filed January 12, 2017; U.S. Patent Application No. 15 / 405,234, filed January 12, 2017; U.S. Patent Application No. 15 / 426,962, filed February 7, 2017; and U.S. Patent Application No. 15 / 473,496, filed March 29, 2017. The full disclosure of each of these patent applications is incorporated herein by reference.

[0065] In the monitoring layer, each of the individual VRF systems 510 to 530 can be represented as a single asset that converts electricity 502 from 508 into hot air 504 or cold air 506 required for the power utility building area. Hot air 504 and cold air 506 can be delivered to air-side units 512, 522, and 532, which provide heating and / or cooling for the building area served by air-side units 512, 522, and 532. Hot air 504 and cold air 506 can be considered as resources generated by VRF systems 510 to 530, while electricity 502 can be considered as resources consumed by VRF systems 510 to 530. The relationship between resource generation and electricity consumption for each VRF system 510 to 530 can be defined by the system performance curves of each VRF system 510 to 530. The system performance curves can be used in the supervision layer as constraints on the cost optimization performed by the predictive cost optimization controller to ensure that the VRF systems 510 to 530 operate in order to generate sufficient hot air 504 and cold air 506 for the building area.

[0066] At each time step of the optimization period, the amount of hot air 504 and cold air 506 generated by each of the VRF systems 510 to 530 can be determined by the predictive cost optimization controller by executing an asset allocation process. Several examples of asset allocation processes that can be executed by the predictive cost optimization controller are described in detail in U.S. Patent Application Nos. 15 / 405,236, 15 / 405,234, 15 / 426,962, and 15 / 473,496.

[0067] Distributed Model Predictive Control System Now refer to Figure 6 A block diagram of a distributed model predictive control (MPC) system 600 according to some embodiments is shown. The MPC system 600 uses MPC technology to determine optimal setpoints for devices in a VRF system (e.g., VRF system 100) over a time range. The MPC system 600 can be used in conjunction with VRF system 100, as shown in Figures 1 to 100. Figure 5 As described. For example, the MPC system 600 can determine the optimal temperature setting for the indoor VRF unit 104 and the optimal load setting for the outdoor VRF unit 102.

[0068] MPC (Multi-Process Control) is a control technique that uses a model of a controlled system to correlate system inputs (e.g., control actions, setpoints, etc.) with system states and system outputs (e.g., measurements, process variables, etc.). The model can be used to predict system states and outputs based on actions taken by the controller at each time step during the optimization period. At each time step, MPC uses the system model to determine a sequence of control actions to achieve objectives (e.g., minimizing tracking error, minimizing energy costs, etc.) while adhering to process constraints such as equipment capacity and safety limits (e.g., temperature constraints, equipment switching constraints, etc.) to solve an online optimization problem. The first control action in the sequence is implemented, and the optimization problem is solved again at the next time step after obtaining new measurements.

[0069] In economic MPC, the objective of the optimization problem is typically to minimize the total cost, as defined by the cost function. Numerous studies have demonstrated that MPC outperforms existing control systems due to its ability to predict the future and anticipate events before they occur. By using passive thermal energy storage (TES) through the mass of buildings, MPC enables the shifting of energy loads from peak to off-peak periods. Energy costs can be reduced by concentrating equipment use during periods of lower resource prices, while maintaining comfort limits within each building.

[0070] Still refer to Figure 6 The distributed MPC system 600 is shown as including an MPC layer 610 and a supervisory layer 620. The MPC layer 610 is shown as including a high-level model predictive controller 608 and several low-level model predictive controllers 612 to 618. Controllers 612, 614, and 616 are shown as low-level indoor model predictive controllers, while controller 618 is shown as a low-level outdoor model predictive controller. The MPC layer 610 can be configured to determine optimal temperature setpoints and device operating setpoints and provide them to the devices in the supervisory layer 620. In some embodiments, the MPC layer 610 can be retrofitted to any existing VRF system to provide setpoint optimization for the VRF system's devices.

[0071] The monitoring layer 620 is shown as including indoor VRF units 622 to 626 and outdoor VRF unit 628. Indoor VRF units 622 to 626 may include some or all of the indoor VRF units 104, while outdoor VRF unit 628 may include some or all of the outdoor VRF units 102. In some embodiments, the monitoring layer 620 may include a PID controller, operable devices (e.g., outdoor VRF units, indoor VRF units, pumps, fans, valves, etc.), and / or other systems or devices configured to control process variables to setpoints.

[0072] In some embodiments, the distributed MPC system 600 includes a load / rate predictor 602. The load / rate predictor 602 can provide load and rate forecasts to the MPC layer 610, including, for example, disturbance forecasts, electricity prices, demand-based electricity prices, and outside air temperature. The load / rate predictor 602 is shown receiving weather forecasts from a weather service 604. In some embodiments, the load / rate predictor 602 generates disturbance forecasts based on the weather forecasts. In some embodiments, the load / rate predictor 602 uses feedback from a regulatory layer 620 to generate disturbance forecasts. Feedback from the regulatory layer 620 can include various types of sensor inputs (e.g., temperature, flow rate, humidity, enthalpy, etc.) or other data related to the controlled building or campus (e.g., building occupancy data, building power load, etc.). In some embodiments, the load / rate predictor 602 generates disturbance forecasts that include predicted disturbance values ​​for each time step within an optimized time period.

[0073] In some embodiments, load / rate predictor 602 uses a deterministic plus stochastic model trained from historical load data to generate disturbance predictions. Load / rate predictor 602 can use any of a variety of prediction methods to generate disturbance predictions (e.g., a linear regression model with a deterministic component and an autoregressive model with a stochastic component). Load / rate predictor 602 can predict one or more different types of disturbances to a building or campus. For example, load / rate predictor 602 can predict heat loads generated by heat transfer between air inside the building and outside air through the building walls. Load / rate predictor 602 can predict heat loads generated by internal heating within the building (e.g., heat generated by electronics in the building, heat generated by building occupants, etc.). In some embodiments, load / rate forecaster 602 uses the techniques described in U.S. Patent Application No. 14 / 717,593, filed May 20, 2015, entitled “Building Management System for Forecasting Time Series Values ​​of Building Variables,” the entire disclosure of which is incorporated herein by reference.

[0074] Load / rate predictor 602 is shown receiving utility rates from utility 606. Utility rates can indicate the cost or price per unit of resource (e.g., electricity, gas, water, etc.) provided by utility 606 at each time step within an optimization period. In some embodiments, utility rates are time-varying rates. For example, the price of electricity is higher at certain times of day or on certain days of week (e.g., during high-demand periods) and lower at other times of day or on other days of week (e.g., during low-demand periods). Utility rates can define various time periods and the cost per unit of resource during each time period. Utility rates can be the actual rate received from utility 606 or the predicted utility rate estimated by load / rate predictor 602.

[0075] In some embodiments, utility rates include demand charges for one or more resources provided by utility 606. Demand charges may define a separate cost imposed by utility 606 based on maximum usage (e.g., maximum energy consumption) of a specific resource during a demand charge period. Utility rates may define various demand charge periods and one or more demand charges associated with each demand charge period. In some instances, demand charge periods may partially or completely overlap with each other and / or with optimization periods. Advantageously, MPC layer 610 may consider demand charges in an advanced optimization process performed by advanced model prediction controller 608. Utility 606 may be defined by time-varying (e.g., hourly) prices, maximum service levels (e.g., by physical infrastructure or by the maximum consumption rate allowed by the contract), and (in the case of electricity) demand charges or costs for peak consumption rates over a period of time. Load / rate predictor 602 may store predicted loads and utility rates in memory and / or provide predicted loads and utility rates to advanced MPC 608.

[0076] MPC layer 610 can receive measurement results from supervisory layer 620 and provide setpoints to supervisory layer 620. MPC layer 610 can generate optimal values ​​for various decision variables, including, for example, zone temperature setpoints, variable refrigerant flow setpoints, operating mode setpoints (e.g., heating or cooling), and / or equipment on / off decisions. MPC layer 610 can use system models, such as zone temperature and cooling / heating load models, cooling / heating load and temperature setpoint models, and equipment models, to determine the optimal values ​​of the decision variables. MPC layer 610 can determine the optimal values ​​of the decision variables by performing an optimization process subject to several constraints. These constraints may include comfort limits for zone air temperature, equipment capacity constraints, refrigerant state balance constraints, and equipment rate of change limits for supervisory layer 620.

[0077] As discussed above, solving a single MPC problem to determine the optimal values ​​of decision variables can be difficult for large-scale applications. For example, a building or building system may include thousands of discrete areas and thousands of unique HVAC units. Discrete decisions (e.g., turning a device on / off) can lead to mixed-integer optimization problems, which further increases the complexity. Due to the difficulty and computational complexity of the MPC problem, MPC layer 610 can decompose the entire MPC problem into several smaller, more manageable optimization problems.

[0078] like Figure 6 As shown, the distributed MPC system 600 can decompose the entire MPC problem into high-level optimization problems and low-level optimization problems. The high-level problem can be solved by a high-level controller 608 to determine the load distribution curves for each low-level indoor subsystem 632-636 and the demand distribution curve for the outdoor subsystem 638. In some embodiments, the high-level controller 608 uses an aggregated low-level model of each indoor subsystem 632-636 to reduce computational complexity. The high-level controller 608 can determine the load distribution curves that optimize (e.g., minimize) the total operating cost of the MPC system 600 during the optimization period. Each load distribution curve can include the load value at each time step in the optimization period. Low-level indoor controllers 612-616 can use the load distribution curves as constraints to define the maximum allowable load value for each indoor subsystem 632-636 at each time step in the optimization period. The high-level controller 608 can provide the load distribution curves to each of the low-level indoor controllers 612-616. (Refer to...) Figure 7 The advanced optimizations performed by the advanced controller 608 are described in more detail.

[0079] The low-level optimization problem can be further decomposed into a low-level outdoor optimization problem and one or more low-level indoor optimization problems. Each low-level indoor problem can be solved by one of the low-level indoor controllers 612 to 616 to determine a zone temperature setpoint for the indoor VRF units 622 to 626 of each indoor subsystem 632 to 636. Each low-level indoor controller 612 to 616 can determine a zone temperature setpoint that optimizes (e.g., minimizes) the energy consumption of the corresponding indoor subsystem 632 to 636 while maintaining the zone temperature within defined temperature limits and not exceeding the load value provided by the high-level controller 608. Alternatively, each low-level indoor controller 612 to 616 can determine a temperature setpoint that tracks the average indoor subsystem temperature (e.g., the predicted subsystem temperature state) based on the high-level optimization problem. (Refer to...) Figure 8 The low-level optimizations performed by the low-level controllers 612 to 616 are described in more detail.

[0080] Low-level outdoor problems can be solved by a low-level outdoor controller 618. In some embodiments, the low-level outdoor problem is a mixed-integer linear procedure. The low-level outdoor controller 618 can determine optimal setpoints for the outdoor VRF unit 628 that minimize operating costs while satisfying the demand distribution curve from the high-level controller 608. Decision variables optimized by the low-level outdoor controller 618 may include, for example, equipment on / off states, cooler heat load, compressor setpoints, pump flow rates, and setpoints for other auxiliary outdoor equipment. The low-level outdoor controller 618 can use the demand distribution curve from the high-level controller 608 as input to the total demand satisfied by the outdoor VRF unit 628 at each time step of the optimization period.

[0081] In some embodiments, the low-level outdoor controller 618 decomposes the low-level outdoor optimization problem into a first outdoor optimization problem and a second outdoor optimization problem. The first outdoor optimization problem can allocate the demand specified by the high-level controller 608 across multiple outdoor VRF units of the outdoor subsystem. The second outdoor optimization problem can be decomposed into a mixed-integer optimization problem so that each sub-facility determines the optimal device on / off state and device settings for the outdoor VRF unit 628. Examples of optimization techniques that can be used by the low-level outdoor controller 618 are described in detail in U.S. Patent Application No. 14 / 634,609, filed February 27, 2015, the entire disclosure of which is incorporated herein by reference.

[0082] like Figure 6 As shown, each low-level indoor model predictive controller 612 to 616 can control subsystems 632 to 636 of the entire VRF system 100. Each low-level indoor controller 612 to 616 can perform an individual indoor optimization process to determine the optimal temperature setpoint for the indoor VRF units 622 to 626 of the corresponding indoor subsystem 632 to 636. Each indoor subsystem 632 to 636 may include one or more indoor VRF units, each of which can be configured to deliver air to one or more building areas.

[0083] In some embodiments, the high-level controller 608 uses an aggregated model of each indoor subsystem 632 to 636 and distributes the thermal load to each indoor subsystem 632 to 636. Low-level indoor controllers 612 to 616 can use a more detailed region-level model during the low-level optimization process to determine the optimal temperature setpoint for each building area of ​​the respective indoor subsystem. Decomposing the VRF system 100 into individual indoor subsystems 632 to 636 can improve computational performance and significantly reduce the amount of time required to solve low-level MPC problems. For example, all low-level MPC problems can be solved within minutes.

[0084] In some embodiments, each indoor subsystem 632 to 636 represents a separate building. Significant coupling between indoor subsystems 632 to 636 (e.g., heat exchange between subsystems 632 to 636) can impact performance because low-level controllers 612 to 616 are not required to coordinate their solutions. One approach to decomposing the VRF system 100 to ensure no coupling exists between subsystems 632 to 636 is to decompose it by buildings, since separate buildings cannot exchange heat with each other. For this reason, it may be desirable to select indoor subsystems 632 to 636 such that each indoor subsystem 632 to 636 represents a separate building. Each indoor subsystem 632 to 636 may include multiple zones (e.g., rooms or spaces within a building), each of which may be controlled by a separate indoor VRF unit. In other embodiments, each indoor subsystem 632 to 636 may represent a single building zone, a set of zones within a building, or even multiple buildings.

[0085] In the MPC system 600, the advanced model predictive controller 608 determines the thermal load to be allocated to each indoor subsystem 632 to 636 (e.g., each building or each zone) and the demand distribution curve for the outdoor subsystem 638. Each indoor subsystem 632 to 636 may include a separate low-level indoor controller 612 to 616 that calculates the temperature setpoint or variable refrigerant flow rate (e.g., the setpoint for each indoor VRF unit in the building) for each zone within that indoor subsystem 632 to 636. Low-level indoor problems can be solved in a distributed manner. Low-level indoor problems can be easily scaled up to handle implementations across large industrial and campus scales without increasing computational complexity.

[0086] The distributed MPC system 600 offers several advantages over alternative control strategies. For example, the higher-level controller 608 can coordinate the operation of lower-level indoor subsystems 632 to 636 via load distribution curves provided to each lower-level controller 612 to 616. By including demand charges in the higher-level objective function, the higher-level controller 608 can generate load distribution curves that represent the operation of lower-level indoor subsystems 632 to 636. In other words, the higher-level controller 608 can generate load distribution curves that ensure lower-level indoor subsystems 632 to 636 do not all consume power simultaneously. This allows the higher-level controller 608 to coordinate the operation of lower-level indoor subsystems 632 to 636 and consider demand charges without requiring communication between lower-level indoor controllers 612 to 616. The coupling to all indoor subsystems 632 to 636 caused by the presence of a single outdoor subsystem 638 is also resolved by the higher-level controller 608. Therefore, the lower-level control problems are completely independent, eliminating the need for iteration and communication between lower-level controllers 612 to 616.

[0087] Data communication between the MPC layer 610 and the regulatory layer 620 can also be greatly reduced. For example, as Figure 6 As shown, data communication between the MPC layer 610 and the regulatory layer 620 can be limited to measurement results and setpoints. This allows the MPC system 600 to integrate with any existing BMS. The advanced controller 608 can use the aggregation model of each indoor subsystem 632 to 636 and the outdoor subsystem 638 to reduce computational complexity during advanced optimization. Each low-level indoor controller 612 to 616 can provide the advanced controller 608 with the aggregation perturbation estimate, aggregation temperature, and aggregation system parameters (e.g., heat capacity, heat transfer coefficient, etc.) for the corresponding indoor subsystem 632 to 636.

[0088] The distributed MPC system 600 can be used to control a variety of different systems, including, for example, cooler facilities, air handling units, rooftop units, variable refrigerant flow systems, air-side systems, water-side systems, building management systems, and / or other types of systems (for which power consumption or thermal load can be distributed to different subsystems). Most building temperature regulation methods do not consider detailed models of VRF devices or integer decision variables, which reduces the accuracy of energy cost calculations. However, the MPC system 600 can include integer variables in the optimization problem to determine when to turn devices on and off, rather than relying on heuristics.

[0089] Advanced Model Predictive Controller Now refer to Figure 7 A more detailed block diagram of an Advanced Model Predictive Controller (MPC) 608 according to some embodiments is shown. The Advanced MPC 608 is shown as including a communication interface 702 and processing circuitry 704. The communication interface 702 may include wired or wireless interfaces (e.g., receptacles, antennas, transmitters, receivers, transceivers, wire terminals, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 702 may include an Ethernet card and port for sending and receiving data via an Ethernet-based communication network and / or a WiFi transceiver for communication via a wireless communication network. The communication interface 702 can be configured to communicate via a local area network (LAN) or a wide area network (WAN) (e.g., the Internet, building WAN, etc.) and can use various communication protocols (e.g., BACnet, IP, LON, etc.).

[0090] Communication interface 702 may be a network interface configured to facilitate electronic data communication between the advanced MPC 608 and various external systems or devices, such as weather service 604, utilities 606, low-level controllers 612 to 618, BMS equipment, etc. For example, the advanced MPC 608 may receive weather forecasts from weather service 604, utilities rates from utilities 606, and / or load and rate forecasts from load / rate forecaster 602. The advanced MPC 608 may receive measurement results from the BMS indicating one or more measurement states of the controlled building or campus (e.g., temperature, humidity, electrical load, etc.) and one or more states of outdoor subsystem 638 (e.g., equipment status, power consumption, equipment availability, etc.).

[0091] The advanced MPC 608 can receive indoor subsystem disturbance estimates from each low-level indoor controller 612 to 616. These indoor subsystem disturbance estimates can indicate the estimated thermal load for each indoor subsystem 632 to 636. The advanced MPC 608 can receive aggregated system curves, aggregated subsystem parameters, and / or performance coefficients from each low-level controller 612 to 618. The advanced MPC 608 can use the information received at communication interface 702 to generate load distribution curves for each indoor subsystem 632 to 636 and demand distribution curves for the outdoor subsystem 638. The advanced MPC 608 can provide the load distribution curves and demand distribution curves to the low-level controllers 612 to 618.

[0092] Processing circuitry 704 is shown as including processor 706 and memory 708. Processor 706 may be a general-purpose or special-purpose processor, application-specific integrated circuit (ASIC), one or more field-programmable gate arrays (FPGAs), a set of processing units, or other suitable processing units. Processor 706 may be configured to execute computer code or instructions stored in memory 708 or received from other computer-readable media (e.g., CD-ROM, network storage device, remote server, etc.).

[0093] Memory 708 may include one or more means (e.g., memory cells, memory devices, storage devices, etc.) for storing data and / or computer code used to perform and / or facilitate the various processes described in this disclosure. Memory 708 may include random access memory (RAM), read-only memory (ROM), hard disk drive storage devices, temporary storage devices, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. Memory 708 may include database components, object code components, scripting components, or any other type of information structure for supporting the various activities and information structures described in this disclosure. Memory 708 may be communicatively connected to processor 706 via processing circuitry 704 and may include computer code for (e.g., by processor 706) performing one or more processes described herein. When processor 706 executes instructions stored in memory 708, processor 706 is typically configured with advanced MPC 608 (and more specifically processing circuitry 704) to perform this activity.

[0094] Still refer to Figure 7 The advanced MPC 608 is shown as including a temperature modeler 714. The subsystem temperature modeler 714 can generate a temperature model for each indoor subsystem 632 to 636. The temperature model generated by the subsystem temperature modeler 714 can be referred to as a building temperature model under the assumption that each indoor subsystem 632 to 636 represents a separate building. However, if indoor subsystems 632 to 636 represent other types of spaces, the temperature model generated by the subsystem temperature modeler 714 can be a temperature model for other types of subsystems. The subsystem temperature modeler 714 can generate a temperature model for each indoor subsystem. Although... Figure 6 Only three indoor subsystems 632 to 636 are shown, but it should be understood that any number of buildings and indoor subsystems 632 to 636 can exist. Typically, the subsystem temperature modeler 714 can generate... Indoor subsystem temperature model, where, It is the total number of buildings and / or interior subsystems 632 to 636.

[0095] In some embodiments, the subsystem temperature modeler 714 uses a building heat transfer model to model the temperature of each indoor subsystem. The dynamics of heating or cooling a single building area are described by energy balance: in, It is the heat capacity of the building area. It is the environmental heat transfer coefficient of the building area. It is the temperature of the building area. It is the ambient temperature outside the building area (e.g., outside air temperature). It is the amount of cooling applied to the building area (i.e., the cooling load), and This refers to the external loads, radiation, or other disturbances experienced by the building area. In the previous equation, This indicates the removal of heat (i.e., cooling) from a building area by the HVAC system and is therefore negative. However, if heating is applied to a building area instead of cooling, then... The symbol above can be switched to a positive sign, making This indicates the amount of heat applied to a building area by the HVAC system (i.e., the heating load).

[0096] The previous equation combines all mass and air properties of the building area into a single zone temperature. Other heat transfer models that can be used by the subsystem temperature modeler 714 include the following air and mass zone models: in, and It refers to the heat capacity and temperature of the air in the building area. It is the ambient air temperature. It is the heat transfer coefficient between the air in a building area and the ambient air outside the building area (e.g., through the exterior walls of the building area). and It refers to the heat capacity and temperature of non-air mass within the building area, and It is the heat transfer coefficient between air and non-air mass in a building area.

[0097] The previous equation combines all the mass properties of the building area into a single area mass. Other heat transfer models that can be used by the subsystem temperature modeler 714 include the following air, shallow mass, and deep mass area models: in, and It refers to the heat capacity and temperature of the air in the building area. It is the ambient air temperature. It is the heat transfer coefficient between the air in a building area and the ambient air outside the building area (e.g., through the exterior walls of the building area). and It refers to the heat capacity and temperature of the shallow mass within the building area. It is the heat transfer coefficient between the air and the shallow mass in the building area. and It refers to the heat capacity and temperature of the deep mass within the building area, and It is the heat transfer coefficient between shallow and deep layers.

[0098] In some embodiments, the subsystem temperature modeler 714 uses the following subsystem temperature model to model the temperature of each indoor subsystem: in, and It is an index The specified heat capacity and temperature of the indoor subsystem, It is the indoor subsystem External ambient air temperature (e.g., outside air temperature). It is the indoor subsystem The heat transfer coefficient between the air and the ambient air. This is the amount of cooling applied to the indoor subsystem by the MPC system 600 (i.e., the amount of heat removed from the indoor subsystem), and It is the indoor subsystem The external loads, radiation, or disturbances experienced. If heating is supplied to the indoor subsystem instead of cooling, then... The sign on the symbol can be switched from negative to positive.

[0099] Subsystem temperature modeler 714 can use indoor subsystem disturbance estimates received from low-level indoor controllers 612 to 616 to identify each indoor subsystem at each time step during the optimization period. External interference The appropriate value. In some embodiments, the subsystem temperature modeler 714 uses weather forecasts from weather service 604 and / or load and rate forecasts provided by load / rate predictor 602 to determine the appropriate value for each indoor subsystem at each time step during the optimization period. Ambient air temperature and / or external interference The appropriate value. and The value can be specified for the indoor subsystem. Parameters from the indoor subsystem The low-level indoor controller receives, receives from the user, retrieves from memory 708, or is otherwise provided to the subsystem temperature modeler 714. The subsystem temperature modeler 714 can be configured for each indoor subsystem. Generate the indoor subsystem temperature model, where, ,and This represents the total number of indoor subsystems.

[0100] Advanced MPC 608 is shown as including an outdoor unit demand modeler 722. The outdoor unit demand modeler 722 can generate a model representing the demand for the outdoor subsystem 638 as allocated to each indoor subsystem at each time step of the optimization period. thermal load The function. In some embodiments, the outdoor unit demand modeler 722 uses the following equation to model the outdoor unit demand: in, In time step Outdoor unit requirements at the location (e.g., at a time step) (Thermal or refrigerant status distribution of the outdoor subsystem 638), and In time step Assigned to indoor subsystem The thermal load. The previous equation indicates the total demand for outdoor subsystem 638. It is allocated to each indoor subsystem thermal load The sum of these equations. This equation can be used by the advanced MPC 608 as an energy balance constraint to ensure that the outdoor subsystem 638 generates sufficient thermal energy at each time step. The indoor subsystem is covered.

[0101] Advanced MPC 608 is shown as including a constraint modeler 710. The constraint modeler 710 can generate optimization constraints and impose these constraints on an optimization procedure executed by the advanced optimizer 712. The constraints imposed by the constraint modeler 710 may include, for example, equipment capacity constraints and indoor subsystem temperature constraints. In some embodiments, the constraint modeler 710 imposes the following constraints: Ensure each time step Outdoor unit requirements Less than or equal to the maximum capacity of outdoor subsystem 638 .

[0102] In some embodiments, the constraint modeler 710 constrains the indoor subsystem temperature. Apply constraints. For example, constraint modeler 710 can constrain a minimum temperature as shown in the following equation. With maximum temperature Indoor subsystem temperature : in, and The value can be adjusted based on the temperature setpoint of the indoor subsystem. In some embodiments, the constraint modeler 710 automatically adjusts based on information received from the BMS of the lower-level indoor controller and / or the indoor subsystem. and The value. For example, the constraint modeler 710 can automatically adjust each time step using the temperature setpoint schedule and / or occupancy schedule of the indoor subsystem. superior and The value of . This allows the constraint modeler 710 to use temperature limits based on the time-varying setpoint temperature range of the indoor subsystem, such that the indoor subsystem temperature Maintain at the appropriate temperature limit for the time and Inside.

[0103] Still refer to Figure 7 The advanced MPC 608 is shown as including an energy cost modeler 720 and a demand charge modeler 718. The energy cost modeler 720 can generate an energy cost model representing the cost of energy consumed by the MPC system 600. The energy cost can include the cost per unit of energy resource (e.g., electricity, water, gas, etc.) consumed by the MPC system 600 during the optimization period, and a demand charge based on maximum power consumption. The demand charge component of the energy cost model can be modeled by the demand charge modeler 718 and enforced via demand charge constraints. In some embodiments, the energy cost model only considers the energy resources consumed by the outdoor subsystem 638. In other embodiments, the energy cost model also considers the power consumption of indoor subsystems 622 to 626, which can be modeled by the indoor power consumption modeler 716. Examples of these two scenarios are described below.

[0104] Example 1: Energy cost model without indoor power consumption In some embodiments, the energy cost modeler 720 generates an energy cost model that considers the energy consumption of the outdoor subsystem 638 but excludes indoor power consumption. For example, the energy cost modeler 720 can model the total energy cost during the optimization period using the following equation: The first term of the energy cost model considers each time step during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). In some embodiments, In time step Consumption at the point to meet the time step Total outdoor unit requirements The cost per unit of energy, parameters It is the reciprocal of the performance coefficient of the aggregated VRF subsystem (e.g., 0.1). 0.25), and It is the time step The duration. Therefore, the item Indicates the time step Consumption during the period to meet the needs of outdoor units The total amount of energy consumed (e.g., kWh). Multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) is generated at the time step. The total cost of energy consumed during the period ($). An energy cost model can include the energy cost at each time step. The sum of energy costs during the period is used to determine the total cost of energy consumption during the optimization period.

[0105] The second term in the energy cost model considers demand charges. In some embodiments, This is the demand-based electricity rate (e.g., $ / kW). It is the peak outdoor unit demand during the demand electricity pricing period (e.g., (the maximum value), and It is the reciprocal of the performance coefficient of the aggregated VRF subsystem. Therefore, the term This indicates that the peak outdoor unit demand is met. Peak power consumption. Multiplied by the demand rate. The total cost of generating demand electricity (e.g., $).

[0106] In some embodiments, the demand charge modeler 718 generates demand charge constraints to ensure that It has an appropriate value. If the demand period is fully included in the optimization period (e.g., within the time step). and (between), then The appropriate value is only the maximum value during the optimization period. The demand pricing modeler 718 can implement the following demand pricing constraints: Ensure peak outdoor unit demand The outdoor unit demand is always greater than or equal to the outdoor unit demand at each time step. This forces peak outdoor unit demand. It is at least as large as the maximum outdoor unit demand during the optimization period.

[0107] If the demand period starts before the optimization period, then the demand period will be... The maximum value may occur before the start of the optimization period. The demand cost modeler 718 can enforce the following demand cost constraints: Ensure: Even if the peak outdoor unit demand occurs before the start of the current optimization period, the peak outdoor unit demand will be [not specified]. It is also always greater than or equal to the maximum outdoor unit demand occurring during the same demand period. In some embodiments, the demand charge modeler 718 is updated whenever a new maximum outdoor unit demand is set. This is to ensure that the energy cost model accurately represents the demand charges imposed by the electricity utility.

[0108] The advanced optimizer 712 can use energy cost models, demand-to-electricity models, indoor subsystem temperature models, outdoor unit demand models, and optimization constraints to formulate the optimization problem. In some embodiments, the advanced optimizer 712 seeks to minimize the total energy cost (i.e., energy cost and demand-to-electricity cost) consumed by the outdoor subsystem 638, which is subject to indoor subsystem temperature constraints and other constraints provided by the advanced models described herein. For example, the advanced optimizer 712 can formulate the advanced optimization problem as: Subject to the following constraints: And ensure the cooling dose for each state (e.g., liquid, gas, superheated gas, etc.) at each time step. Any refrigerant balance constraints required to maintain equilibrium.

[0109] In some embodiments, the advanced optimizer 712 transforms one or more of the identified models and / or constraints into a state-space form for use in advanced optimization problems. For example, the advanced optimizer 712 can transform the aforementioned equations into a discretized state-space model of the following form: in, In time step The system state vector at that point, In time step The system input vector at that location, In time step The measurement result or system output vector at that location, In time step The interference vector at that location, and In time step The (predicted) system state vector at the location. Table 1 shows the variables that can be included in each of these vectors: As shown in Table 1, the system state vector Including indoor subsystem temperature In some embodiments, the system state vector include The indoor subsystem temperature of each indoor subsystem This makes the system state vector Total number of variables equal Input vector This can include allocation to each indoor subsystem thermal load In some embodiments, the input vector include The thermal load of each of the indoor subsystems , so that the input vector Total number of variables equal Interference vector This can include the ambient air temperature of each indoor subsystem. and estimated interference In some embodiments, the interference vector include Estimated disturbance for each of the indoor subsystems and individual ambient air temperature This makes the interference vector Total number of variables equal .

[0110] In some embodiments, measurement vector With system state vector Same. This indicates that all system states are directly measurable (i.e., ), and in the state-space model and The value of the metric is and In other embodiments, the measurement results can be used as a basis. To construct or predict system state For example, the advanced MPC 608 can use a Kalman filter or other predictive techniques to adjust the prediction based on the measurement results. To build system state Therefore, it is possible to utilize To replace system state In this context, the hat symbol indicates that the state is predicted. System recognition techniques can be used to identify the state-space representation. , , ,and The value of the metric. Examples of state prediction and system identification techniques that can be used by the advanced MPC 608 are described in detail in U.S. Patent No. 9,235,657 entitled “System Identification and Model Development”, filed March 13, 2013, the entire disclosure of which is incorporated herein by reference.

[0111] Example 2: Energy cost model with indoor power consumption In some embodiments, the energy cost modeler 720 generates an energy cost model that considers both the energy consumption of the outdoor subsystem 638 and the energy consumption of the indoor subsystems 622 to 626. For example, the energy cost model may consider the energy consumed by fans and other types of devices 622 to 626 in the indoor VRF units to deliver the allocated heat load to the indoor subsystems. power In some embodiments, the power consumption of each indoor subsystem 632 to 636 It is the thermal load allocated to this indoor subsystem. The function.

[0112] The indoor power consumption modeler 716 can generate indoor power consumption models. With thermal load The relevant indoor power consumption model. In some embodiments, the indoor power consumption modeler 716 models indoor power consumption using the following equation: in, It is composed of indoor subsystem The indoor VRF unit 622 consumes heat loads to deliver heat. Power quantity. Conversion factor. It can be a function of the performance coefficient of the indoor VRF unit 622 (e.g., the reciprocal of the performance coefficient). In some embodiments, It is a constant, which indicates indoor power consumption. With thermal load The relationship is linear. In other embodiments, the indoor power consumption modeler 716 can use operational data to calculate... It is used as a nonlinear function of load and other parameters.

[0113] In some embodiments, the indoor power consumption modeler 716 calculates the conversion factor. The power consumption modeler 716 can be used as a function of various system parameters, such as the type of indoor VRF unit in the indoor subsystem, time of day, comfort limits, environmental conditions, cooling water supply temperature, cooling water supply flow rate, and / or other parameters characterizing indoor subsystems 632 to 636 or any of the indoor subsystems 632 to 636. For example, the indoor power consumption modeler 716 can collect operational data from indoor VRF unit 622 and / or low-level indoor controller 612 and use the operational data to determine... The appropriate value.

[0114] In some embodiments, the indoor power consumption modeler 716 calculates... As a heat load And functions of various fan power models. For example, air at 20°C can have a density as shown in the following equation. and heat capacity : Thermal load provided by airflow The following model can be used to represent this: in, It is the volumetric velocity of the supply air entering the building area. It is the temperature of the building area, and This refers to the temperature of the supplied air. Assume the temperature of the supplied air is... Approximately 55°F and room air temperature At approximately 72°F, the indoor power consumption modeler 716 can calculate the thermal load per unit volume of airflow (e.g., the air's cooling capacity) as follows: The 716 indoor power consumption modeler can use this cooling capacity value. Estimating power consumption per unit volume of a typical indoor VRF unit fan. The value. For example, a typical HVAC fan consumes approximately 1 horsepower (hp) to provide airflow between 1000 cubic feet (CFM) and 1500 CFM per minute. These values ​​can be converted to metrics as follows: Substituting these values ​​into the indoor power consumption model, we get: This indicates the indoor power consumption of each indoor subsystem 632 to 636. Approximately the heat load delivered by the indoor subsystem 10%.

[0115] Assuming that indoor power consumption can be... Modeling as The energy cost modeler 720 can then use the following equation to model the total energy cost during the optimization period: The first part of the energy cost model considers the outdoor subsystem 638 at each time step during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). In some embodiments, In time step The cost of each unit of energy consumed. It is the time step The duration, and It is the reciprocal of the performance coefficient of the outdoor subsystem 638 (e.g., 0.2). Item Indicates the time step During this period, the outdoor subsystem 638 meets the requirements of the outdoor unit. The amount of power consumed (e.g., kW). Multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) and duration (For example, hours) generated by the outdoor subsystem 638 in time steps The total cost of energy consumed during the period ($). This can be used across all time steps of the optimization period. The first part of the energy cost model is summed to determine the total energy consumed by the outdoor subsystem 638 over the duration of the optimization period.

[0116] The first part of the energy cost model considers the time step of each indoor subsystem during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). As mentioned above, It is the reciprocal of the performance coefficient of the indoor subsystem (e.g., 0.1), and It is composed of indoor subsystem time step The heat load delivered at the point. (Item) Representation Subsystem Power consumption of indoor VRF units It can be used across all indoor subsystems. And across all time steps The second part of the energy cost model is summed to determine the total power consumption of all indoor subsystems over the duration of the optimization period. This is then multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) and duration (e.g., hours) generates the total cost ($) of energy consumed by the indoor subsystem over the duration of the time step.

[0117] The third part of the energy cost model considers demand charges. In some embodiments, It is the demand-based electricity rate (e.g., $ / kW), and This is the peak aggregated indoor and outdoor power consumption during the applicable demand-rate period, multiplied by the demand-rate rate. The total cost of generating demand charges ($). In some embodiments, the demand charge modeler 718 generates demand charge constraints to ensure It has an appropriate value. If the demand period is fully included in the optimization period (e.g., within the time step). and (between), then An appropriate value is any time step during the optimization period. The maximum combined indoor / outdoor power consumption at the specified location. The demand cost modeler 718 can implement the following demand cost constraints: Ensure peak power consumption The outdoor power consumption is always greater than or equal to the outdoor power consumption at each time step. With indoor power consumption The sum of these. This forces peak power consumption. It should be at least as large as the maximum combined indoor / outdoor power consumption during the optimization period.

[0118] If the demand charge period begins before the optimization period, the maximum peak power consumption during the demand charge period may occur before the start of the optimization period. The demand charge modeler 718 can enforce the following demand charge constraints: Ensure: Even if the maximum power consumption occurs before the start of the current optimization period, the peak power consumption... It is also always greater than or equal to the maximum power consumption that occurs during the same demand electricity billing period. In some embodiments, the demand cost modeler 718 is updated whenever a new maximum power consumption is set. This is to ensure that the energy cost model accurately represents the demand charges imposed by the electricity utility.

[0119] The advanced optimizer 712 can use energy cost models, indoor power consumption models, demand cost models, indoor subsystem temperature models, outdoor unit demand models, and optimization constraints to formulate the optimization problem. In some embodiments, the advanced optimizer 712 seeks to minimize the total energy cost consumed by the aggregated VRF system, which is subject to indoor subsystem temperature constraints and other constraints provided by the advanced models described herein. For example, the advanced optimizer 712 can formulate the advanced optimization problem as: Subject to the following constraints: And any refrigerant balance constraints as previously mentioned.

[0120] In some embodiments, the advanced optimizer 712 transforms one or more of the identified models and / or constraints into a state-space form for use in advanced optimization problems. For example, the advanced optimizer 712 can transform the aforementioned equations into a discretized state-space model of the following form: in, In time step The system state vector at that point, In time step The system input vector at that location, In time step The measurement result or system output vector at that location, In time step The interference vector at that location, and In time step The (predicted) system state vector at each location. The variables included in each vector can be the same as those shown in Table 1 above.

[0121] The advanced optimizer 712 can perform optimization procedures to determine the time step during the optimization period. Vector at position The optimal value for each input variable in the input variables. For example, the advanced optimizer 712 can determine the optimal value for each input variable at each time step. Assigned to each indoor subsystem thermal load The optimal value for each heat load within the system. With the same indoor subsystem index. Each group of thermal loads The load distribution curves for specific indoor subsystems are generated and included at each time step during the optimization period. The load value is displayed on the screen. The advanced optimization 712 can provide indoor subsystem load distribution curves to the lower-level indoor controllers 612 to 616.

[0122] In some embodiments, the advanced optimizer 712 generates a predicted temperature state vector for each low-level indoor subsystem 632 to 636. Each predicted temperature state vector This can include each time step during the optimization period. Predicted temperature state Temperature conditions can be predicted using any of a variety of forecasting techniques. The prediction technique includes, for example, a Kalman filter as described in U.S. Patent No. 9,235,657. The advanced optimizer 712 can provide predicted temperature states to the low-level indoor controllers 612-616 of the corresponding low-level indoor subsystems 632-636. Each vector. In some embodiments, the low-level indoor controllers 612 to 616 use predicted temperature states. Generate at each time step Tracking and predicting temperature status The area temperature setting value.

[0123] In some embodiments, the advanced optimizer 712 calculates at each time step Total requirements for outdoor subsystem 638 As a time step Thermal load of indoor subsystem The sum of these outdoor unit demand values ​​forms the demand distribution curve for outdoor subsystem 638 and includes each time step within the optimization period. The demand value at a specific time step. Demand value This indicates that the outdoor subsystem 638 is at this time step. The total demand that must be met. The advanced optimizer 712 can provide the outdoor unit demand distribution curve to the low-level outdoor controller 618.

[0124] Low-level indoor model predictive controller Now refer to Figure 8 A more detailed block diagram of a low-level indoor model predictive controller (MPC) 612 according to some embodiments is shown. Although only one low-level indoor MPC 612 is shown in detail, it should be understood that any other low-level indoor MPC in the control system 600 (e.g., low-level indoor MPCs 614 to 616) may include some or all of the same components as low-level indoor MPC 612. The control system 600 may include any number of low-level indoor MPCs, wherein each low-level indoor MPC can operate independently to monitor and control separate low-level indoor subsystems (e.g., indoor subsystems 632 to 636).

[0125] A low-level indoor MPC 612 is shown including a communication interface 802 and processing circuitry 804. The communication interface 802 may include wired or wireless interfaces (e.g., receptacles, antennas, transmitters, receivers, transceivers, wire terminals, etc.) for data communication with various systems, devices, or networks. For example, the communication interface 802 may include an Ethernet card and port for sending and receiving data via an Ethernet-based communication network and / or a WiFi transceiver for communication via a wireless communication network. The communication interface 802 can be configured to communicate via a local area network (LAN) or a wide area network (WAN) (e.g., the Internet, building WAN, etc.) and can use various communication protocols (e.g., BACnet, IP, LON, etc.).

[0126] Communication interface 802 may be a network interface configured to facilitate electronic data communication between the low-level indoor MPC 612 and various external systems or devices (e.g., weather service 604, advanced MPC 608, indoor VRF unit 622, etc.). For example, the low-level indoor MPC 612 may receive weather forecasts from weather service 604 and / or load forecasts from load / rate predictor 602. The low-level indoor MPC 612 may receive measurement results from BMS indicating one or more measurement states of the controlled building or campus (e.g., temperature, humidity, electrical load, etc.) and one or more states of indoor subsystem 632 (e.g., equipment status, power consumption, equipment availability, etc.). The low-level indoor MPC 612 may receive predicted temperature states and / or load distribution curves from advanced MPC 608. The low-level indoor MPC 612 can use information received at the communication interface 802 to generate zone temperature setpoints for each zone of the low-level indoor subsystem 632 and / or refrigerant flow setpoints for each indoor VRF unit of the low-level indoor subsystem 632. The low-level indoor MPC 612 can provide the zone temperature setpoints and / or refrigerant flow setpoints to the indoor VRF unit 622.

[0127] Processing circuitry 804 is shown as including processor 806 and memory 808. Processor 806 may be a general-purpose or special-purpose processor, application-specific integrated circuit (ASIC), one or more field-programmable gate arrays (FPGAs), a set of processing units, or other suitable processing units. Processor 806 may be configured to execute computer code or instructions stored in memory 808 or received from other computer-readable media (e.g., CD-ROM, network storage device, remote server, etc.).

[0128] Memory 808 may include one or more means (e.g., memory cells, memory devices, storage devices, etc.) for storing data and / or computer code used to perform and / or facilitate the various processes described in this disclosure. Memory 808 may include random access memory (RAM), read-only memory (ROM), hard disk drive storage devices, temporary storage devices, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. Memory 808 may include database components, object code components, scripting components, or any other type of information structure for supporting the various activities and information structures described in this disclosure. Memory 808 may be communicatively connected to processor 806 via processing circuitry 804 and may include computer code for (e.g., by processor 806) performing one or more processes described herein. When processor 806 executes instructions stored in memory 808, processor 806 typically configures low-level in-system MPC 612 (and more specifically processing circuitry 804) to perform this activity.

[0129] Still refer to Figure 8 The low-level MPC 612 is shown as including a zone interference predictor 824. The zone interference predictor 824 can be configured for each zone of the indoor subsystem 632. Generate interference predictions. Throughout this disclosure, the index... Used to represent various regions, among which... ,and This represents the total number of zones within a given indoor subsystem. In some embodiments, each zone... Uniquely configured for controlling the area The temperature of a specific indoor VRF unit is associated with this. Therefore, the index It can also represent a single indoor VRF unit. Area Interference prediction can include interference value vectors , where the vector Each element includes a specific time step for the optimization period. Predicted interference value .

[0130] A regional interference predictor 824 is shown receiving a weather forecast from weather service 604. In some embodiments, the regional interference predictor 824 generates an interference forecast based on the weather forecast. In some embodiments, the regional interference predictor 824 uses feedback from regulatory layer 620 to generate the interference forecast. Feedback from regulatory layer 620 may include various types of sensor inputs (e.g., temperature, flow rate, humidity, enthalpy, etc.) or other data related to the controlled building or campus (e.g., building occupancy data, building power load, etc.). In some embodiments, the regional interference predictor 824 uses a deterministic plus stochastic model trained from historical load data to generate the interference forecast. The regional interference predictor 824 may use any of a variety of forecasting methods to generate the interference forecast (e.g., a linear regression model with a deterministic component and an autoregressive model with a stochastic component).

[0131] The 824 area interference predictor can predict interference in each building area. One or more different types of interference. For example, the area interference predictor 824 can predict interference caused by building areas. The heat load generated by heat transfer between the interior air and the exterior air through the building walls. The area interference predictor 824 can predict the heat load generated by internal heating within the building area (e.g., heat generated by electronic devices within the building area, heat generated by area occupants, etc.). In some embodiments, the area interference predictor 824 uses the prediction technique described in U.S. Patent Application No. 14 / 717,593 to perform interference prediction.

[0132] Still refer to Figure 8 The low-level MPC 612 is shown as including a zone temperature modeler 814. The zone temperature modeler 814 can be configured for each building zone of the indoor subsystem 632. Generate a temperature model. The indoor subsystem 632 can have any number of zones. In some embodiments, the temperature of each zone can be controlled and / or regulated independently. Some building zones can exchange heat with each other (e.g., if building zones are adjacent to each other), while other building zones cannot directly exchange energy. Typically, a zone temperature modeler 814 can generate... A regional temperature model, in which... This is the total number of areas in the indoor subsystem 632.

[0133] In some embodiments, the zone temperature modeler 814 uses a zone heat transfer model to model the temperature of each building zone. The dynamics of heating or cooling a single building zone are described by energy balance: in, It is the heat capacity of the building area. It is the environmental heat transfer coefficient of the building area. It is the temperature of the building area. It is the ambient temperature outside the building area (e.g., outside air temperature). It is the amount of cooling applied to the building area (i.e., the cooling load), and This refers to the external loads, radiation, or other disturbances experienced by the building area. In the previous equation, This indicates the removal of heat (i.e., cooling) from a building area by the HVAC system and is therefore negative. However, if heating is applied to a building area instead of cooling, then... The symbol above can be switched to a positive sign, making This indicates the amount of heat applied to a building area by the HVAC system (i.e., the heating load).

[0134] The previous equation combines all mass and air properties of the building area into a single zone temperature. Other heat transfer models that can be used by the zone temperature modeler 814 include the following air and mass zone models: in, and It refers to the heat capacity and temperature of the air in the building area. It is the ambient air temperature. It is the heat transfer coefficient between the air in a building area and the ambient air outside the building area (e.g., through the exterior walls of the building area). and It refers to the heat capacity and temperature of non-air mass within the building area, and It is the heat transfer coefficient between air and non-air mass in a building area.

[0135] The previous equation combines all the mass properties of the building area into a single area mass. Other heat transfer models that can be used by the area temperature modeler 814 include the following air, shallow mass, and deep mass area models: in, and It refers to the heat capacity and temperature of the air in the building area. It is the ambient air temperature. It is the heat transfer coefficient between the air in a building area and the ambient air outside the building area (e.g., through the exterior walls of the building area). and It refers to the heat capacity and temperature of the shallow mass within the building area. It is the heat transfer coefficient between the air and the shallow mass in the building area. and It refers to the heat capacity and temperature of the deep mass within the building area, and It is the heat transfer coefficient between shallow and deep layers.

[0136] In some embodiments, the zone temperature modeler 814 uses the following zone temperature model to model the temperature of each building zone: in, and It is a regional index The heat capacity and temperature of the specified building area, It is a region External ambient air temperature (e.g., outside air temperature). It is a region The heat transfer coefficient between the air and the ambient air. The MPC system 600 is applied to the building area. The amount of cooling (i.e., the amount of heat removed from the area), and It is a region The external loads, radiation, or disturbances experienced. If heating is provided to the area instead of cooling, then... The sign on the symbol can be switched from negative to positive.

[0137] parameter Characterization region With another region (For example, with the region) The degree of coupling between adjacent building areas. If the area and If the regions are not adjacent and / or cannot directly exchange heat with each other, the region temperature modeler 814 can be set. The value is zero. The regional temperature model can include building areas. With each other building area The sum of heat transfer between them is used as the regional temperature. and With coupling coefficient The function. In other embodiments, external disturbance estimates can be used. Let's consider heat transfer between regions.

[0138] The region temperature modeler 814 can use the region disturbance estimate received from the region disturbance predictor 824 to identify each region at each time step during the optimization period. External interference The appropriate value. In some embodiments, the area temperature modeler 814 uses weather forecasts from weather service 604 and / or load and rate forecasts provided by load / rate predictor 602 to determine the appropriate value for each area at each time step during the optimization period. Ambient air temperature and / or external interference The appropriate value. and The value can be specified as a parameter of the region. From the management of building areas The BMS is received from the user, retrieved from memory 808, or otherwise provided to the area temperature modeler 814. The area temperature modeler 814 can provide temperature data for each area. Generate a regional temperature model, where... ,and It represents the total number of buildings.

[0139] Still refer to Figure 8 The low-level MPC 612 is shown as including a subsystem load modeler 816. The subsystem load modeler 816 can generate the total amount of heat energy delivered to the indoor subsystems. (For example, the total amount of heating or cooling delivered to the subsystem) as a function of the load of a single area. The model. In some embodiments, the subsystem load modeler 816 models the total subsystem load using the following equation: in, It is the total amount of heat energy (e.g., heating or cooling) delivered to the indoor subsystem, and It is a specific area The rate at which heat energy is delivered (in units of power). Indoor subsystem load models can be used to calculate the heat load for each building area. Summation is performed to calculate the total indoor subsystem thermal load. The total indoor subsystem thermal load is the total amount of thermal energy delivered to the indoor subsystem. The derivative of .

[0140] The low-level indoor MPC 612 is shown as including a cooling / heating load modeler 820. The cooling / heating load modeler 820 can generate one or more models that represent the thermal load of each building area. Defined as the regional temperature shown in the following equation and area settings Functions: The model generated by the cooling / heating load modeler 820 can be used as an optimization constraint to ensure thermal load. It will not drop to the point where the regional temperature will be lower. Values ​​that deviate from the acceptable or comfortable temperature range.

[0141] In some embodiments, the region setting value Including operating mode settings (e.g., heating or cooling) and refrigerant flow setpoint For example, the area setting value. It can be represented as follows: Among them, variables Indicates that the indoor VRF unit is for a building area. The operation mode, and variables Indicates that the indoor VRF unit is for a building area. The refrigerant flow rate setpoint. In some embodiments, the operating mode It is a binary variable, as shown in the following equation: Among them, the value The indicator shows the operating mode as cooling, while the value... The indicator shows whether the operating mode is heating or non-cooling.

[0142] In some embodiments, the cooling / heating load modeler 820 uses the following model to represent the regional thermal load. (i.e., building area) Cooling load or heating load ) and refrigerant flow Related: in, It is the area cooling load (i.e., the area thermal load when the operating mode is cooling). ) and refrigerant flow The conversion factor between them, and It is the area heating load (i.e., the area thermal energy load when the operating mode is heating). ) and refrigerant flow The conversion factor between them. In some embodiments, the conversion factor... and Both are regional temperatures. A function of regional temperature. With conversion factor and The relationship between them can be predefined, allowing the cooling / heating load modeler 820 to model based on measured or estimated zone temperatures. To calculate and The value of . In other embodiments, the conversion factor and One or both of these can have constant values, in which case the cooling / heating load model can be formulated as a mixed integer linear program (MILP).

[0143] In some embodiments, the region setting value It is the zone temperature setpoint. The Cooling / Heating Load Modeler 820 can use multiple models to represent regional thermal loads. With regional temperature and zone temperature setpoint Related. For example, the cooling / heating load modeler 820 can use a model of the zone supervisory controller to determine the control actions performed by the controller as a reference for zone temperature. and zone temperature setpoint The function is shown in the following equation, which illustrates an example of such a regional regulatory controller model: in, It is the airflow to the building area The rate (i.e., the control action). Function This can be identified from the data. For example, the cooling / heating load modeler 820 can collect... and The measurement results and identification The corresponding values. The cooling / heating load modeler 820 can use the collected data. , and The values ​​are used as training data to perform the system recognition process to determine a function that defines the relationship between these variables. .

[0144] The cooling / heating load modeler 820 can be used to control actions. With regional thermal load The relevant energy balance model is shown in the following equations: Among them, the function This can be identified from the training data. The cooling / heating load modeler 820 can use the collected data. and The values ​​are used to perform the system identification process to determine the function that defines the relationship between these variables. .

[0145] In some embodiments, and There exists a linear relationship between them. Assume an ideal proportional-integral (PI) controller and... and The linear relationship between them can be used to calculate the thermal load of each building area using a simplified linear controller model. Defined as regional temperature and zone temperature setpoint The function is shown in the following equation as an example of this model: in, Is it the steady-state heating rate or cooling rate? It is the proportional gain of the scaling region PI controller. It is the integral time of the regional PI controller, and It is the setpoint error (i.e., the zone temperature setpoint). With regional temperature (The difference between them). Saturation can be determined by... The constraints are used to represent this. If the linear model is insufficient to accurately represent the PI controller and the region... Modeling the heat transfer in the indoor VRF unit can replace the use of a nonlinear heating / cooling load model.

[0146] Advantageously, modeling the supervisory controller (e.g., a zone PI controller) in a low-level optimization problem allows the low-level indoor MPC 612 to utilize the dynamics of the supervisory controller when determining the optimal temperature setpoint. In some embodiments, the supervisory controller's response may be slow. For example, some zones may require up to an hour to reach a new temperature setpoint. Using the dynamics of the supervisory controller in a low-level MPC problem allows the low-level indoor MPC 612 to consider the time between control actions and effects, enabling the selection of the optimal temperature setpoint based on time-varying energy prices.

[0147] Still refer to Figure 8 The low-level MPC 612 is shown as including a constraint modeler 810. The constraint modeler 810 can generate optimization constraints and impose these constraints on the optimization procedure executed by the low-level optimizer 812. The constraints imposed by the constraint modeler 810 may include, for example, equipment capacity constraints and area temperature constraints. In some embodiments, the constraint modeler 810 imposes area temperature constraints. Apply constraints. For example, constraint modeler 810 can constrain a minimum temperature as shown in the following equation. With maximum temperature Temperature in the region between : in, and The value can be adjusted based on the temperature setpoint of the indoor subsystem.

[0148] In some embodiments, the constraint modeler 810 automatically adjusts based on information received from the BMS of the building area. and The value. For example, the constraint modeler 810 can automatically adjust each time step using a temperature setpoint schedule and / or occupancy schedule for the building area. Place and The value. This allows the constraint modeler 810 to use the temperature limit of the region-based time-varying setpoint temperature range, making the region temperature... Maintain at the appropriate temperature limit for the time and Inside.

[0149] In some embodiments, the constraint modeler 810 applies constraints to ensure that at any time step... The total indoor subsystem load during this period shall not exceed the thermal load allocated to the indoor subsystem by the advanced MPC 608. For example, the constraint modeler 810 may impose the following constraints: in, In time step Total energy consumed by indoor subsystems In time step Total energy consumed by indoor subsystems It is assigned to the indoor subsystem by the advanced MPC 608. The heat load, and This represents the duration of each time step. The left side of the equation represents the duration at each time step. The thermal load of the indoor subsystem during the period (i.e., the change in total thermal energy delivered between consecutive time steps divided by the duration of the time step), while the right side of the equation represents the change in thermal energy delivered between time steps. During this period, the indoor subsystem is assigned by the advanced MPC 608. Thermal load.

[0150] In some embodiments, the constraint modeler 810 applies additional constraints to ensure that the total amount of delivered thermal energy does not decrease between consecutive time steps, as illustrated in the following equation: because It is the sum of the amounts of thermal energy delivered over a long time step. Therefore, this constraint prevents the low-level optimizer 812 from choosing the time step. Thermal load The negative value of . In other words, the rate of heat delivery (i.e., It can be added to the total amount of heat energy delivered during the optimization period, but cannot be subtracted from the total amount of heat energy delivered.

[0151] The low-level optimizer 812 can use a zone temperature model, an indoor subsystem load model, a cooling / heating load model, and optimization constraints to formulate the optimization problem. In some embodiments, the high-level optimization problem seeks to optimize the thermal energy used by the indoor subsystem 632, which is subject to zone temperature constraints and other constraints provided by the low-level model described herein, during the optimization period. Minimize the total amount. For example, the low-level optimizer 812 can formulate a low-level optimization problem as: Subject to the following constraints: Among them, as mentioned above, the function It is based on the regional thermal load With area setting value (For example, zone temperature setpoint) or operating mode With refrigerant flow rate It is defined by the relationship between them.

[0152] In some embodiments, the low-level optimizer 812 transforms one or more of the identified models and / or constraints into a state-space form for use in low-level optimization problems. For example, the low-level optimizer 812 can transform the aforementioned equations into a discretized state-space model of the following form: in, In time step The system state vector at that point, In time step The system input vector at that location, In time step The measurement result or system output vector at that location, In time step The interference vector at that location, and In time step The (predicted) system state vector at the location. Table 2 shows the variables that can be included in each of these vectors: As shown in Table 2, the system state vector Including regional temperature Integral of regional tracking error and the total heat energy delivered to the indoor subsystem. In some embodiments, the system state vector include The temperature of each region in each area and integration region tracking error And a single total heat energy value, which makes the system state vector Total number of variables equal Input vector This can include region settings. (For example, zone temperature setpoint) or operating mode and refrigerant flow rate In some embodiments, the input vector include Regional settings for each region in each region , so that the input vector Total number of variables equal .

[0153] In some embodiments, measurement vector With system state vector Same, but without the tracking error in the integral region. This indicates the regional temperature. and the total amount of heat energy delivered It is measured directly. It can be based on... and The difference between them is used to calculate the tracking error in the integral region. The value of the interference vector. This can include ambient air temperature Estimated interference for each region and the steady-state heating / cooling rate of each region. In some embodiments, the interference vector include Estimated disturbance for each indoor subsystem in the indoor subsystem and steady-state heating / cooling rate and individual ambient air temperature This makes the interference vector Total number of variables equal .

[0154] In some embodiments, the measurement results can be used as a basis. To construct or predict system state For example, the low-level indoor MPC 612 can use a Kalman filter or other predictive techniques to adjust the measurement results. To build system state System identification techniques can be used to identify the state-space expressions. , , ,and The value of the metric. Examples of state prediction and system identification techniques that can be used by the low-level indoor MPC 612 are described in detail in U.S. Patent No. 9,235,657.

[0155] Still refer to Figure 8 The low-level MPC 612 is shown as including a model aggregator 818. The model aggregator 818 can generate aggregated values ​​for various indoor subsystem parameters and / or variables used in the low-level optimization. For example, the model aggregator 818 can generate aggregated values ​​for individual zone temperatures of each zone in the subsystem. Aggregation is performed to generate aggregated subsystem temperature for the lower-level indoor subsystem. In some embodiments, the model aggregator 818 uses the following equation to generate the temperature of the aggregation subsystem. : in, It is a region The heat capacity, and It is a region The temperature. The numerator of the previous equation represents the total heat in the subsystem, while the denominator represents the total heat capacity of the subsystem. Summing these two quantities across all subsystem regions. The model aggregator 818 can estimate the average subsystem temperature by dividing the total heat by the total heat capacity. The model aggregator 818 can calculate each time step of the optimization period. Temperature of the polymer subsystem on .

[0156] Model aggregator 818 can calculate, for example, the heat capacity of a subsystem. Subsystem heat transfer coefficient and estimation of subsystem disturbances Aggregated values ​​of other subsystem parameters or variables. In some embodiments, the model aggregator 818 uses the following equation to calculate the aggregated values ​​of these variables and parameters: Among them, the heat capacity of the subsystem It is the regional heat capacity of each subsystem region. The sum of values, the heat transfer coefficient of the subsystem It is the regional heat transfer coefficient of each subsystem region. The sum of values, and estimation of subsystem disturbances. It is the estimated subsystem disturbance for each subsystem region. The sum. Model aggregator 818 can calculate the sum of each time step in the optimization period. On , and The aggregate value.

[0157] In some embodiments, the model aggregator 818 aggregates subsystem parameters and variables. , , and This information is provided to the high-level MPC 608. The high-level MPC 608 can use these values ​​as input to the high-level models, constraints, and optimization functions used in advanced optimization. Advantageously, model aggregation performed by the model aggregator 818 helps reduce the amount of information exchanged between each low-level indoor MPC 612 to 616 and the high-level MPC 608. For example, each low-level indoor MPC 612 to 616 can provide the aforementioned aggregated values ​​to the high-level MPC 608, instead of individual values ​​for these variables and parameters for each building area.

[0158] Still refer to Figure 8 The low-level indoor MPC 612 is shown as including a temperature tracker 822. In some embodiments, the temperature tracker 822 performs alternative optimizations that may supplement or replace the low-level optimizations performed by the low-level optimizer 812. The temperature tracker 822 does not measure the total thermal energy used by the low-level indoor subsystem. Instead of minimizing, it can generate a predictive subsystem temperature state that tracks the advanced subsystem temperature state generated by the advanced MPC 608 based on advanced optimization. Area temperature setpoint For example, the advanced MPC 608 can calculate the predicted subsystem temperature state for each low-level indoor subsystem 632 to 636 to generate the load distribution curve generated by the advanced MPC 608. As mentioned above, the predicted temperature state can be calculated using a Kalman filter or any other type of state prediction technique. The advanced MPC 608 can predict temperature conditions. Provided to each low-level indoor MPC 612 to 616 for use in the temperature tracking process.

[0159] Temperature tracker 822 is shown receiving predicted temperature status via communication interface 802. Temperature tracker 822 can also receive the temperature of the aggregated subsystem generated by model aggregator 818. The temperature tracker 822 can be used to seek the temperature of the polymer subsystem. With predicted temperature state The objective function for minimizing the error between them is formulated as shown in the following equation: in, It refers to the total amount of heat energy used by the indoor subsystem during the optimization period. A smaller penalty factor. This can be adjusted. The value is adjusted to increase or decrease the total amount of heat energy allocated relative to the temperature tracking error. The weight.

[0160] Temperature tracker 822 can use the previous equation as the objective function in the optimization process to determine the regional temperature setpoint. The optimal value. The optimization performed by the temperature tracker 822 can be similar to the optimization performed by the low-level optimizer 812, except that the temperature tracker 822 is not constrained by the load distribution curve provided by the high-level MPC 608 and seeks to optimize different objective functions. For example, the temperature tracker 822 can minimize the objective function: Subject to the following constraints: Among them, as mentioned above, the function It is based on the regional thermal load With area setting value (For example, zone temperature setpoint) or operating mode With refrigerant flow rate It is defined by the relationship between them.

[0161] In some embodiments, the temperature tracker 822 converts one or more of the identified models and / or constraints into a state-space form for use in low-level optimization problems. For example, the temperature tracker 822 can convert the aforementioned equations into a discretized state-space model of the following form: in, In time step The system state vector at that point, In time step The system input vector at that location, In time step The measurement result or system output vector at that location, In time step The interference vector at that location, and In time step The (predicted) system state vector at each location. The variables included in each vector can be the same as those shown in Table 2 above.

[0162] flow chart Now refer to Figure 9 A flowchart 900 illustrating model predictive control techniques that can be performed by an MPC system 600 according to some embodiments is shown. Flowchart 900 is shown as including: performing advanced optimization at an advanced model predictive controller (MPC) to generate optimal load distribution curves for each of multiple indoor subsystems in a variable refrigerant flow (VRF) system (box 902). In some embodiments, the advanced optimization is performed by an advanced MPC 608, as referenced... Figures 6 to 7 As described. For example, advanced optimization may include generating an advanced energy cost function that defines energy cost as the outdoor unit demand distribution curve. The function. In some embodiments, the outdoor unit demand distribution curve Indicates multiple time steps within the optimization period. The thermal or refrigerant state distribution of each of the outdoor subsystems.

[0163] In some embodiments, the advanced MPC 608 generates an energy cost function that considers the cost per unit of energy consumed by the outdoor subsystem 638 and the demand charge, but does not consider the energy consumption of the indoor subsystems 632 to 636. An example of such an energy cost function is shown in the following equation: The first term of the energy cost function considers each time step during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). In some embodiments, In time step Consumption at the point to meet the time step Total outdoor unit requirements The cost per unit of energy, parameters It is the reciprocal of the performance coefficient of the aggregated VRF system (e.g., 0.1). 0.25), and It is the time step The duration. Therefore, the item Indicates the time step Consumption during the period to meet the needs of outdoor units The total amount of energy consumed (e.g., kWh). Multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) is generated at the time step. The total cost of energy consumed during the period ($). The energy cost function can be included at each time step. The sum of energy costs during the period is used to determine the total cost of energy consumption during the optimization period.

[0164] The second term of the energy cost function considers demand charges. In some embodiments, This is the demand-based electricity rate (e.g., $ / kW). It is the peak outdoor unit demand during the demand electricity pricing period (e.g., (the maximum value), and It is the reciprocal of the performance coefficient of the aggregated VRF system. Therefore, the term This indicates that the peak outdoor unit demand is met. Peak power consumption. Multiplied by the demand rate. The total cost of generating demand electricity (e.g., $).

[0165] In some embodiments, the advanced MPC 608 generates an energy cost function that considers the cost per unit of energy consumed by the outdoor subsystem 638, the demand charge, and the energy consumption of the indoor subsystems 632 to 636. An example of such an energy cost function is shown in the following equation: The first part of the energy cost function considers the outdoor subsystem 638 at each time step during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). In some embodiments, In time step The cost of each unit of energy consumed. It is the time step The duration, and It is the reciprocal of the performance coefficient of the outdoor subsystem 638 (e.g., 0.2). Item Indicates the time step During this period, the outdoor subsystem 638 meets the requirements of the outdoor unit. The amount of power consumed (e.g., kW). Multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) and duration (For example, hours) generated by the outdoor subsystem 638 in time steps The total cost of energy consumed during the period ($). This can be used across all time steps of the optimization period. The first part of the energy cost function is summed to determine the total energy consumed by the outdoor subsystem 638 over the duration of the optimization period.

[0166] The first part of the energy cost function considers the time step of each indoor subsystem during the optimization period. The cost per unit of energy consumed during the process (e.g., $ / kWh). As mentioned above, It is the reciprocal of the performance coefficient of the indoor subsystem (e.g., 0.1), and It is the indoor subsystem targeting the indoor subsystem time step The heat load delivered at the point. (Item) Indicates indoor subsystem Power consumption of indoor VRF units It can be used across all indoor subsystems. And across all time steps The second part of the energy cost function is summed to determine the total power consumption of all indoor subsystems over the duration of the optimization period. This is then multiplied by the cost per unit of energy consumed. (e.g., $ / kWh) and duration (e.g., hours) generates the total cost ($) of energy consumed by the indoor subsystem over the duration of the time step.

[0167] The third part of the energy cost function considers demand charges. In some embodiments, It is the demand-based electricity rate (e.g., $ / kW), and This is the peak aggregate power consumption during the applicable demand-rate period, multiplied by the demand-rate rate. The total cost of generating demand electricity (e.g., $).

[0168] In some embodiments, performing advanced optimization in block 902 includes using an outdoor unit demand model to plot the outdoor unit demand distribution curve. Defined as the load distribution curve of the multiple indoor subsystems The function. Load distribution curves for each indoor subsystem. It can indicate the thermal energy distribution of one of the indoor subsystems at each of the plurality of time steps.

[0169] In some embodiments, the outdoor unit demand model represents the demand for outdoor subsystem 638 as the thermal load allocated to each indoor subsystem at each time step of the optimization period. The following equation illustrates an example of such an outdoor unit demand model: in, In time step Outdoor unit requirements at the location (e.g., at a time step) (Thermal or refrigerant status distribution of outdoor subsystem 638). In time step Assigned to indoor subsystem The heat load, and In time step The amount of heat energy stored in the TES tank during this period. The previous equation indicates the total demand for outdoor subsystem 638. It is allocated to each indoor subsystem thermal load The sum of these equations. This equation can be used by the advanced MPC 608 as an energy balance constraint to ensure that the outdoor subsystem 638 generates sufficient thermal energy at each time step. The load of the indoor subsystem is covered.

[0170] In some embodiments, the advanced optimization in block 902 includes generating an indoor subsystem temperature model for each of the plurality of indoor subsystems. Each indoor subsystem temperature model can define the heat energy distribution to the indoor subsystem. With the temperature of the indoor subsystem The relationship between them is illustrated in the following equation, which shows an example of such an indoor subsystem temperature model: in, and It is an index The specified heat capacity and temperature of the indoor subsystem, It is the indoor subsystem External ambient air temperature (e.g., outside air temperature). It is the indoor subsystem The heat transfer coefficient between the air and the ambient air. This is the amount of cooling applied to the indoor subsystem by the MPC system 600 (i.e., the amount of heat removed from the indoor subsystem), and It is the indoor subsystem The external loads, radiation, or disturbances experienced. If heating is supplied to the indoor subsystem instead of cooling, then... The sign on the symbol can be switched from negative to positive.

[0171] Advanced optimization can use the indoor subsystem disturbance estimates received from the low-level indoor controllers 612 to 616 to identify each indoor subsystem at each time step during the optimization period. External interference The appropriate value. In some embodiments, advanced optimization uses weather forecasts from weather service 604 and / or load and rate forecasts provided by load / rate predictor 602 to determine the appropriate value for each indoor subsystem at each time step during the optimization period. Ambient air temperature and / or external interference The appropriate value. and The value can be specified for the indoor subsystem. Parameters from the indoor subsystem The low-level indoor controller receives data from the user, retrieves it from memory, or otherwise provides it to the high-level MPC 608. Advanced optimization can be included for each indoor subsystem. Generate a temperature model, where, ,and This represents the total number of indoor subsystems.

[0172] In some embodiments, performing the advanced optimization in block 902 includes optimizing the energy cost and the load distribution curves of the plurality of indoor subsystems subject to constraints provided by the outdoor unit demand model and the temperature model of each indoor subsystem. Advanced optimization may use the energy cost model, the demand-cost model, the indoor subsystem temperature model, the outdoor unit demand model, and optimization constraints to formulate the optimization problem.

[0173] In some embodiments, the advanced optimization seeks to minimize the total energy cost (i.e., energy cost and demand charge) consumed by the outdoor subsystem 638, which is subject to indoor subsystem temperature constraints and other constraints provided by the advanced model described herein. For example, the advanced optimization can be formulated as follows: Subject to the following constraints: In some embodiments, advanced optimization seeks to minimize the total energy cost consumed by the aggregated VRF system, constrained by indoor subsystem temperature constraints and other constraints provided by the advanced model described herein. For example, advanced optimization can formulate the advanced optimization problem as: Subject to the following constraints: Still refer to Figure 9 Flowchart 900 is shown as including the optimal indoor subsystem load distribution curve. From a high-level MPC to multiple low-level indoor MPCs (Box 904). Each load distribution curve. It can send data from the high-level MPC 608 to one of the low-level indoor MPCs 612 to 616, such as... Figure 6 As shown in the diagram. In some embodiments, each low-level indoor MPC is configured to monitor and control a specific indoor subsystem. Each low-level indoor MPC can receive the optimal subsystem load distribution curve of the indoor subsystem monitored and controlled by the low-level indoor MPC.

[0174] Flowchart 900 is shown as including: performing low-level optimization at each point in the low-level indoor MPC to generate optimal refrigerant flow setpoints for each indoor VRF unit of the indoor subsystem (block 906). In some embodiments, the low-level optimization in block 906 includes each zone of each indoor subsystem. Generate a regional temperature model. A regional temperature model defines the temperature of a region. Thermal load distribution curve of the region The relationship between them. An example of such a regional temperature model is shown in the following equation: in, and It is a regional index The heat capacity and temperature of the specified building area, It is a region External ambient air temperature (e.g., outside air temperature). It is a region The heat transfer coefficient between the air and the ambient air. The MPC system 600 is applied to the building area. The amount of cooling (i.e., the amount of heat removed from the area), and It is a region The external loads, radiation, or disturbances experienced. If heating is provided to the area instead of cooling, then... The sign on the symbol can be switched from negative to positive.

[0175] parameter Characterization region With another region (For example, with the region) The degree of coupling between adjacent building areas. If the area and If the regions are not adjacent and / or cannot directly exchange heat with each other, the region temperature modeler 814 can be set. The value is zero. The regional temperature model can include building areas. With each other building area The sum of heat transfer between them is used as the regional temperature. and With coupling coefficient The function. In other embodiments, external disturbance estimates can be used. Let's consider heat transfer between regions.

[0176] Low-level optimization may include using area interference estimates received from area interference predictor 824 to identify each area at each time step during the optimization period. External interference The appropriate value. In some embodiments, low-level optimization uses weather forecasts from weather service 604 and / or load and rate forecasts provided by load / rate predictor 602 to determine the appropriate value for each region at each time step during the optimization period. Ambient air temperature and / or external interference The appropriate value. and The value can be specified as a parameter of the region. From the management of building areas The BMS is received from the user, retrieved from the memory 808, or otherwise provided to the area temperature modeler 814.

[0177] In some embodiments, the low-level optimization in block 906 includes generating the total amount of heat energy delivered to the indoor subsystem. (For example, the total heating or cooling delivered to the indoor subsystem) as a separate zone load. The model of the function. In some embodiments, low-level optimization includes modeling the total indoor subsystem load using the following equations: in, It is the total amount of heat energy (e.g., heating or cooling) delivered to the indoor subsystem, and It is a specific area The rate at which heat energy is delivered (in units of power). Indoor subsystem load models can be used to calculate the heat load for each building area. Summation is performed to calculate the total indoor subsystem thermal load. The total indoor subsystem thermal load is the total amount of thermal energy delivered to the indoor subsystem. The derivative of .

[0178] In some embodiments, low-level optimization includes generating one or more models that measure the thermal load of each building area. Defined as the regional temperature shown in the following equation and area settings Functions: Among them, as mentioned above, the function It is based on the regional thermal load With area setting value (For example, zone temperature setpoint) or operating mode With refrigerant flow rate The model is defined by the relationship between the two parameters. The model generated by low-level optimization can be used as an optimization constraint to ensure the thermal load. It will not drop to the point where the regional temperature will be lower. Values ​​that deviate from the acceptable or comfortable temperature range.

[0179] In some embodiments, the low-level optimization in block 906 seeks to optimize the thermal energy used by the indoor subsystem, which is subject to regional temperature constraints and other constraints provided by the low-level indoor model described herein, during the optimization period. Minimize the total amount. For example, low-level optimization can be formulated as: Subject to the following constraints: Among them, as mentioned above, the function It is based on the regional thermal load With area setting value (For example, zone temperature setpoint) or operating mode With refrigerant flow rate It is defined by the relationship between them.

[0180] Still refer to Figure 9 Flowchart 900 is shown as including operating the indoor VRF unit in each indoor subsystem of the indoor subsystem using an optimal refrigerant flow setpoint (block 908). For example, each lower-level indoor MPC 612 to 616 may operate the indoor VRF unit located in the corresponding indoor subsystem 632 to 636. Operating the indoor VRF unit may include activating or deactivating the indoor VRF unit, adjusting the refrigerant flow rate, adjusting the fan speed, or otherwise controlling the indoor VRF unit to achieve the refrigerant flow setpoint.

[0181] Now refer to Figure 10 A flowchart 1000 is shown illustrating model predictive control techniques that can be performed by an MPC system 600 according to some embodiments. The flowchart 1000 is shown as including: performing advanced optimization at an advanced model predictive controller (MPC) to generate optimal temperature distribution profiles for each of multiple indoor subsystems in a variable refrigerant flow (VRF) system (box 1002). In some embodiments, the advanced optimization is performed by an advanced MPC 608, as shown in reference to... Figures 6 to 7 As described. Advanced optimizations can be compared with references. Figure 9 The advanced optimizations described are the same or similar.

[0182] The advanced optimization in box 1002 can generate optimal temperature distribution curves for each indoor subsystem. Optimal temperature distribution curve This can include the predicted temperature state vector for each lower-level indoor subsystem 632 to 636. Each predicted temperature state vector This can include each time step during the optimization period. Predicting indoor subsystem temperature status Temperature conditions can be predicted using any of a variety of forecasting techniques. The prediction technique includes, for example, a Kalman filter as described in U.S. Patent No. 9,235,657.

[0183] Flowchart 1000 is shown as including the optimal indoor subsystem temperature distribution curve. From a high-level MPC to multiple low-level indoor MPCs (Box 1004). Temperature distribution profiles for each subsystem. Data can be sent from the high-level MPC 608 to one of the low-level indoor MPCs 612 to 616. In some embodiments, each low-level indoor MPC is configured to monitor and control a specific indoor subsystem. Each low-level indoor MPC can receive the optimal subsystem temperature distribution curve of the indoor subsystem monitored and controlled by the low-level indoor MPC. .

[0184] Flowchart 1000 is shown as including: performing low-level optimization at each point in the low-level indoor MPC to generate an optimal temperature setpoint for each indoor VRF unit of the corresponding indoor subsystem (box 1006). The low-level optimization in box 1006 can be similar to the low-level optimization in box 906. However, the low-level optimization in box 1006 can formulate the low-level optimization problem to track the optimal temperature generated by the high-level optimization in box 1002, rather than minimizing the total amount of heat energy used.

[0185] In some embodiments, the low-level optimization in block 1006 can be formulated as follows: Subject to the following constraints: Among them, as mentioned above, the function It is based on the regional thermal load With area setting value (For example, zone temperature setpoint) or operating mode With refrigerant flow rate It is defined by the relationship between them.

[0186] Still refer to Figure 10 Flowchart 1000 is shown as including operating the indoor VRF unit in each indoor subsystem of the indoor subsystem using an optimal temperature setpoint (block 1008). For example, each lower-level indoor MPC 612 to 616 may operate the indoor VRF unit located in the corresponding indoor subsystem 632 to 636. Operating the indoor VRF unit may include activating or deactivating the indoor VRF unit, adjusting the refrigerant flow rate, adjusting the fan speed, or otherwise controlling the indoor VRF unit to achieve the refrigerant flow setpoint.

[0187] Configuration of exemplary embodiments The construction and arrangement of the systems and methods illustrated in the various exemplary embodiments are merely illustrative. Although only a few embodiments are described in detail in this disclosure, many modifications are possible (e.g., variations in the size, dimensions, structure, shape and proportions of various elements, parameter values, installation arrangements, use of materials, color, orientation, etc.). For example, the positions of elements may be reversed or otherwise varied, and the nature or number or position of discrete elements may be altered or changed. Therefore, all such modifications are intended to be included within the scope of this disclosure. The order or sequence of any process or method steps may be changed or reordered according to alternative embodiments. Other substitutions, modifications, alterations, and omissions may be made in terms of the design, operating conditions, and arrangements of the exemplary embodiments without departing from the scope of this disclosure.

[0188] This disclosure contemplates methods, systems, and program products on any machine-readable medium for performing various operations. Embodiments of this disclosure can be implemented using existing computer processors or by a dedicated computer processor combined with a suitable system for this or another purpose, or by a hard-wired system. Embodiments within the scope of this disclosure include program products comprising machine-readable media for carrying or having machine-executable instructions or data structures stored thereon. Such machine-readable media can be any available medium accessible by a general-purpose or special-purpose computer or other machine having a processor. For example, such machine-readable media may include RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disc storage devices, disk storage devices or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of machine-executable instructions or data structures and accessible by a general-purpose or special-purpose computer or other machine having a processor. Combinations of the foregoing are also included within the scope of machine-readable media. Machine-executable instructions include, for example, instructions and data that cause a general-purpose computer, special-purpose computer, or special-purpose processing machine to perform a function or group of functions.

[0189] Although the accompanying drawings show a specific order of method steps, the order of steps may differ from the depicted order. Two or more steps may also be performed simultaneously or partially simultaneously. This variation will depend on the selected software and hardware system and the designer's choices. All such variations are within the scope of this disclosure. Similarly, the software implementation can be implemented using standard programming techniques with rule-based logic and additional logic for implementing the various connection steps, processing steps, comparison steps, and decision steps.

Claims

1. A model predictive control system for optimizing the energy cost of a variable refrigerant flow (VRF) system including an outdoor subsystem and multiple indoor subsystems, the model predictive control system comprising: An advanced model predictive controller is configured to perform advanced optimization to generate an optimal indoor subsystem load distribution curve for each of the plurality of indoor subsystems, wherein (i) the optimal indoor subsystem load distribution curve includes the amount of energy allocated to each of the plurality of indoor subsystems during each of a plurality of time steps in an optimization period; and (ii) the advanced model predictive controller is configured to coordinate and stagger the amount of energy allocated to each of the plurality of indoor subsystems across the plurality of time steps when performing the advanced optimization to optimize the energy cost, including a demand charge based on peak demand; and Multiple low-level indoor model predictive controllers, each corresponding to one of the indoor subsystems and configured to perform low-level optimization to generate optimal indoor setpoints for one or more indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution curve of the corresponding indoor subsystem, wherein communication between the multiple low-level indoor model predictive controllers is not required to generate the optimal indoor setpoints. Each of the plurality of low-level indoor model predictive controllers is configured to operate the indoor VRF unit of the corresponding indoor subsystem using the optimal indoor setpoint.

2. The model predictive control system as described in claim 1, wherein: Each of the optimal indoor subsystem load distribution curves includes the heat distribution of one of the indoor subsystems at each of multiple time steps within the optimization period; and Each low-level indoor model predictive controller is configured to use the thermal energy allocation to the corresponding indoor subsystem as an optimization constraint when performing the low-level optimization.

3. The model predictive control system as described in claim 1, wherein, The optimal indoor setting value includes at least one of the following: The refrigerant flow rate setpoint for each of the indoor VRF units; and Temperature setpoints for one or more building zones controlled by the indoor VRF unit.

4. The model predictive control system as described in claim 1, wherein: Each indoor subsystem comprises multiple building areas; and Each of the low-level indoor model predictive controllers is configured to generate an optimal indoor temperature setpoint for each of the plurality of building areas in the respective indoor subsystem.

5. The model predictive control system as described in claim 1, wherein: Each indoor subsystem includes multiple indoor VRF units; and Each of the low-level indoor model predictive controllers is configured to: Generate an optimal refrigerant flow decision for each of the plurality of indoor VRF units in the respective indoor subsystem; and The optimal refrigerant flow decision is used to calculate at least one of the flow setpoint, temperature setpoint, and valve setpoint for each of the plurality of indoor VRF units.

6. The model predictive control system as described in claim 5, wherein, Each of the low-level indoor model predictive controllers is configured to: Identify the relationship between the thermal load and refrigerant flow rate of each of the indoor VRF units in the corresponding indoor subsystem; The optimal refrigerant flow decision is generated for each of the indoor VRF units in the corresponding indoor subsystem using the identified relationship between thermal load and refrigerant flow. as well as The optimal indoor setpoint is calculated based on the optimal refrigerant flow decision.

7. The model predictive control system as described in claim 1, wherein: The advanced model predictive controller is configured to generate a predicted temperature distribution curve for each indoor subsystem, derived from the optimal indoor subsystem load distribution curve; and The low-level indoor model predictive controller is configured to generate the optimal indoor setpoint by minimizing the error between the indoor subsystem temperature and the predicted temperature distribution curve.

8. The model predictive control system as described in claim 1, wherein, Each of the indoor subsystems represents a separate building, which is thermally decoupled from each other so that no direct heat exchange occurs between the indoor subsystems.

9. The model predictive control system as described in claim 1, wherein, The advanced model predictive controller is configured to generate optimal outdoor unit demand distribution curves for the outdoor subsystem. The system further includes a low-level outdoor model predictive controller, which is configured to perform low-level optimization to generate optimal outdoor setpoints for one or more outdoor VRF units of the demand-constrained outdoor subsystem based on the optimal outdoor unit demand distribution curve. The low-level outdoor model prediction controller is configured to operate the outdoor VRF unit using the optimal outdoor setpoint.

10. The model predictive control system as described in claim 1, wherein, The advanced model prediction controller is configured to determine: The optimal thermal energy distribution for each of the plurality of indoor subsystems at each of the multiple time steps during the optimization period; as well as The optimal thermal or refrigerant state allocation of the outdoor subsystem at each of the plurality of time steps during the optimization period.

11. A method for optimizing the energy cost of a variable refrigerant flow (VRF) system comprising an outdoor subsystem and multiple indoor subsystems, the method comprising: Advanced optimization is performed at the advanced model predictive controller to generate an optimal indoor subsystem load distribution curve for each of the plurality of indoor subsystems, wherein (i) the optimal indoor subsystem load distribution curve includes the amount of energy allocated to each of the plurality of indoor subsystems during each of a plurality of time steps in the optimization period; and (ii) the advanced model predictive controller is configured to coordinate and stagger the amount of energy allocated to each of the plurality of indoor subsystems across the plurality of time steps when performing the advanced optimization to optimize the energy cost, including demand charges based on peak demand. The optimal indoor subsystem load distribution curve is provided from the high-level model predictive controller to a plurality of low-level indoor model predictive controllers, each of which corresponds to one of the plurality of indoor subsystems; Low-level optimization is performed at each of the low-level indoor model predictive controllers to generate optimal indoor setpoints for one or more indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution curve of the corresponding indoor subsystem, wherein communication between the plurality of low-level indoor model predictive controllers is not required to generate the optimal indoor setpoints; and The indoor VRF unit in each of the plurality of indoor subsystems is operated using the optimal indoor settings.

12. The method of claim 11, wherein: Each of the optimal indoor subsystem load distribution curves includes the heat distribution of one of the indoor subsystems at each of multiple time steps within the optimization period; and Performing the low-level optimization includes using the thermal energy allocation of the corresponding indoor subsystem as an optimization constraint.

13. The method of claim 11, wherein, The optimal indoor setting value includes at least one of the following: The refrigerant flow rate setpoint for each of the indoor VRF units; and Temperature setpoints for one or more building zones controlled by the indoor VRF unit.

14. The method of claim 11, wherein: Each indoor subsystem comprises multiple building areas; and Performing the low-level optimization includes generating optimal indoor temperature setpoints for each of the plurality of building areas in the respective indoor subsystems.

15. The method of claim 11, wherein: Each indoor subsystem includes multiple indoor VRF units; and Performing the aforementioned low-level optimizations includes: Generate an optimal refrigerant flow decision for each of the plurality of indoor VRF units in the respective indoor subsystem; as well as The optimal refrigerant flow decision is used to calculate at least one of the flow setpoint, temperature setpoint, and valve setpoint for each of the plurality of indoor VRF units.

16. The method of claim 15, wherein, Performing the aforementioned low-level optimizations includes: Identify the relationship between the thermal load and refrigerant flow rate of each of the indoor VRF units in the corresponding indoor subsystem; The optimal refrigerant flow decision is generated for each of the indoor VRF units in the corresponding indoor subsystem using the identified relationship between thermal load and refrigerant flow; and The optimal indoor setpoint is calculated based on the optimal refrigerant flow decision.

17. The method of claim 11, wherein: Performing the advanced optimization includes generating a predicted temperature distribution curve for each indoor subsystem, derived from the optimal indoor subsystem load distribution curve; and Performing the low-level optimization involves generating the optimal indoor setpoint by minimizing the error between the indoor subsystem temperature and the predicted temperature distribution curve.

18. The method of claim 11, wherein, Performing the advanced optimization includes generating an optimal outdoor unit demand distribution curve for the outdoor subsystem, the method further comprising: Low-level optimization is performed at the low-level outdoor model predictive controller to generate optimal outdoor setpoints for one or more outdoor VRF units of the demand-constrained outdoor subsystem based on the optimal outdoor unit demand distribution curve; and The outdoor VRF unit is operated using the aforementioned optimal outdoor settings.

19. The method of claim 11, wherein, Performing the advanced optimization includes determining the following: The optimal thermal energy distribution for each of the plurality of indoor subsystems at each of the multiple time steps during the optimization period; as well as The optimal thermal or refrigerant state allocation of the outdoor subsystem at each of the plurality of time steps during the optimization period.

20. A variable refrigerant flow (VRF) system, comprising: Multiple indoor subsystems, each indoor subsystem including one or more indoor VRF units; An outdoor subsystem, comprising one or more outdoor VRF units; An advanced model predictive controller is configured to perform advanced optimization to generate an optimal indoor subsystem load distribution curve for each of the plurality of indoor subsystems, wherein (i) the optimal indoor subsystem load distribution curve includes the amount of energy allocated to each of the plurality of indoor subsystems during each of a plurality of time steps in an optimization period; and (ii) the advanced model predictive controller is configured to coordinate and stagger the amount of energy allocated to each of the plurality of indoor subsystems across the plurality of time steps when performing the advanced optimization to optimize the energy cost, including a demand charge based on peak demand. as well as Multiple low-level indoor model predictive controllers, each corresponding to one of the indoor subsystems and configured to perform low-level optimization to generate optimal indoor setpoints for the indoor VRF units of the corresponding indoor subsystem using the optimal indoor subsystem load distribution curve of the corresponding indoor subsystem, wherein communication between the multiple low-level indoor model predictive controllers to generate the optimal indoor setpoints is not required.

21. The system of claim 20, wherein each of the plurality of low-level indoor model predictive controllers is configured to generate optimal indoor setpoints for the indoor VRF units of the respective indoor subsystem based on the optimal indoor subsystem load distribution curve of the respective indoor subsystem and a model of the thermal load and refrigerant flow rate of each of the indoor VRF units of the respective indoor subsystem; and Each of the plurality of low-level indoor model predictive controllers is configured such that the respective indoor subsystem operates according to the optimal indoor setpoint for the indoor VRF unit of the respective indoor subsystem.

22. The model predictive control system of claim 1, wherein the advanced model predictive controller is configured to perform the advanced optimization based at least on a demand charge model and an outdoor subsystem demand model to generate the optimal indoor subsystem load distribution curve for each of the plurality of indoor subsystems.

23. A method comprising any one of the technical features of claims 11-19 or any combination of the technical features.

Images