Chilled water station system group control optimization method and system

Abstract

The present application relates to a kind of cold station system group control optimization method and system, belong to automation control technical field, method includes when cold station system runs, by mixed integer programming algorithm, with each equipment start-stop state as optimization variable, to the first stage group control optimization objective function is minimum optimization, control the start-stop state of each equipment of cold station system in the first stage group control;On the basis of the first stage group control, by model predictive control algorithm, with chilled water outlet temperature, refrigeration pump frequency, cooling pump frequency and cooling tower frequency as optimization variable, to the second stage group control optimization objective function is minimum optimization, control the running state of operating equipment of the cold station system in the second stage group control, to facilitate to reduce energy consumption as far as possible, improve energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and in particular to a group control optimization method and system for a chiller plant system. Background Technology

[0002] Central air conditioning chiller systems are a major source of energy consumption in large public buildings and data centers, typically accounting for 30% to 50% of the building's total energy consumption. Improving the energy efficiency of chiller systems has become a key breakthrough for building energy conservation.

[0003] Currently, chiller plant control systems mainly rely on traditional control logic, such as start-stop control based on fixed temperature differences or return water temperature thresholds, and PID setpoint regulation. While these methods are simple to implement, they are unable to cope with complex and ever-changing load demands, causing the system to operate under suboptimal conditions for extended periods, resulting in significant energy waste. Summary of the Invention

[0004] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a group control optimization method and system for a chiller plant system, which solves the technical problem of low energy efficiency in current chiller plant systems.

[0005] To achieve the above objectives, the main technical solutions adopted by the present invention include: The first aspect of this invention provides a method for optimizing the group control of a chiller plant system.

[0006] The cold storage system group control optimization method proposed in this embodiment of the invention includes: Based on the total energy consumption of the chilled water plant system, the cost of equipment start-up and shutdown, and the cost of equipment lifespan loss, a first-stage group control optimization objective function is constructed. Based on the total energy consumption of the chilled water plant system, the change in chilled water outlet temperature, the change in chilled water pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency, a second-stage group control optimization objective function is constructed. During the operation of the chiller system, a mixed integer programming algorithm is used to minimize the objective function of the first-stage group control optimization, with the start-stop state of each device as the optimization variable, so as to control the start-stop state of each device in the first-stage group control of the chiller system. Based on the first stage of group control, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables, so as to optimize the operating equipment status of the chiller plant system during the second stage of group control.

[0007] In some instances, when minimizing the objective function of the first-stage group control optimization, the control method for the start-stop state of each device in the chiller system during the first-stage group control is determined, and the start-stop state of each device in the chiller system is controlled based on the control method for the start-stop state of each device in the chiller system during the first-stage group control.

[0008] In some instances, when minimizing the objective function of the second-stage group control optimization, the control methods for the chilled water outlet temperature, chilled pump frequency, cooling pump frequency, and cooling tower frequency of each operating device in the chiller plant system during the second-stage group control are determined, and the operating equipment status of the chiller plant system is controlled based on the control methods for the chilled water outlet temperature, chilled pump frequency, cooling pump frequency, and cooling tower frequency of each operating device in the second-stage group control.

[0009] In some instances, the objective function for the first-stage group control optimization is... for: ;in, The total energy consumption of the chiller system, For equipment start-up and shutdown costs, Costs related to equipment lifespan depletion; All are weighting coefficients; ;in, The energy consumption of the chiller at time k. The energy consumption of the refrigeration pump and the cooling pump at time k. Let be the energy consumption of the cooling tower at time k; N Total number of devices; ;in, For the i-th device, the start / stop status for the t-th time. This represents the start-up and stop status of the i-th device during the (t-1)th iteration. ; The cumulative runtime of the i-th device, For the number of times the i-th device is started and stopped, , All are weighting coefficients.

[0010] In some instances, based on the first-stage group control, a model predictive control algorithm is used to minimize the objective function of the second-stage group control, with chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables. This optimizes the operating equipment status of the chiller plant system during the second-stage group control, including: Determine the second constraint condition for the optimization variables; Based on the first stage of group control, and combined with the second constraint of the optimization variables, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables; wherein, the second constraint of the optimization variables includes at least one of the following: chilled water outlet temperature constraint, chilled pump frequency constraint, cooling pump frequency constraint and cooling tower frequency constraint.

[0011] In some instances, before constructing the first-stage group control optimization objective function, the method includes: A COP model for chiller units is constructed based on chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling capacity, and the corresponding weighting coefficients for each quantity. Based on the chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling pump frequency, chilled water pump frequency, cooling capacity, and the correlation between these quantities, a chiller energy consumption model is constructed. A chilled pump energy consumption model is constructed based on chilled pump power, chilled pump frequency, actual flow rate, and chilled pump model coefficients. A cooling pump energy consumption model is constructed based on cooling pump power, cooling pump frequency, actual flow rate, and cooling pump model coefficients. A cooling tower energy consumption model is constructed based on cooling tower power, fan frequency, actual flow rate, and cooling tower model coefficients.

[0012] In some instances, including: Based on user configuration, the system periodically starts the online learning engine and uses the recursive least squares method to incrementally update the parameters of each model in the chiller COP model, chiller energy consumption model, chilled water pump energy consumption model, cooling pump energy consumption model, cooling tower energy consumption model, and cooling tower approximation model.

[0013] A second aspect of this invention provides a group control optimization system for a chiller plant system, comprising: The objective function construction unit is used to construct the first-stage group control optimization objective function based on the total energy consumption of the chiller plant system, the equipment start-up and shutdown costs, and the equipment life loss costs, and to construct the second-stage group control optimization objective function based on the total energy consumption of the chiller plant system, the change in chilled water outlet temperature, the change in chilled pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency. The first-stage group control unit is used to optimize the first-stage group control objective function by using a mixed integer programming algorithm with the start-stop state of each device as the optimization variable during the operation of the chiller system, thereby controlling the start-stop state of each device in the chiller system during the first-stage group control. The second-stage group control unit is used to optimize the operating status of the equipment in the chilled plant system during the second-stage group control by using a model predictive control algorithm with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables.

[0014] The present invention provides a group control optimization method for a chiller plant system, comprising: constructing a first-stage group control optimization objective function based on the total energy consumption of the chiller plant system, equipment start-up and shutdown costs, and equipment lifespan loss costs; and constructing a second-stage group control optimization objective function based on the total energy consumption of the chiller plant system, changes in chilled water outlet temperature, changes in chilled pump frequency, changes in cooling pump frequency, and changes in cooling tower frequency; during the operation of the chiller plant system, using a mixed integer programming algorithm with the start-up and shutdown states of each device as optimization variables, minimizing the first-stage group control optimization objective function to control the start-up and shutdown states of each device in the chiller plant system during the first-stage group control; based on the first-stage group control, using a model predictive control algorithm with the chilled water outlet temperature, chilled pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables, minimizing the second-stage group control optimization objective function to control the operating states of the equipment in the chiller plant system during the second-stage group control. In this application, the most suitable equipment start-up and shutdown state in the chiller plant system is determined by minimizing the objective function of the first-stage group control optimization to reduce unnecessary start-ups. Based on the most suitable equipment start-up and shutdown state, the second-stage group control optimization objective function is minimized to determine the specific operating state of each piece of equipment, thereby helping to minimize energy consumption and improve energy efficiency. Attached Figure Description

[0015] Figure 1 A flowchart of a group control optimization method for a chiller plant system provided in an embodiment of the present invention; Figure 2 This is a block diagram of the overall architecture of the cooling plant system provided in an embodiment of the present invention; Figure 3 A flowchart of the two-stage optimization method provided in this embodiment of the invention; Figure 4 This is a schematic diagram of the optimization model of MPC provided in an embodiment of the present invention; Figure 5 A flowchart illustrating the workflow of the online learning module provided in this embodiment of the invention; Figure 6 A logic block diagram of a security verification module provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of the group control optimization device for a cooling plant system provided in an embodiment of the present invention. Detailed Implementation

[0016] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] The cold storage plant system group control optimization method proposed in this invention is used to solve the problem of low energy efficiency in current cold storage plant systems. Through the first stage of group control optimization objective function minimization optimization, the most suitable equipment start-up and shutdown state in the cold storage plant system is determined to reduce unnecessary start-up. Based on the most suitable equipment start-up and shutdown state, the second stage of group control optimization objective function minimization optimization is performed to determine the specific operating state of each equipment, thereby helping to minimize energy consumption and improve energy efficiency.

[0018] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0019] Figure 1 This is a flowchart illustrating a group control optimization method for a chiller plant system, provided as an embodiment of the present invention. Figure 1 As shown in the embodiment of the present invention, the group control optimization method for a chiller plant system includes: Step 100: Based on the total energy consumption of the chiller plant system, the cost of equipment start-up and shutdown, and the cost of equipment life loss, construct the first-stage group control optimization objective function, and based on the total energy consumption of the chiller plant system, the change in chilled water outlet temperature, the change in chilled pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency, construct the second-stage group control optimization objective function. Step 110: During the operation of the chiller system, a mixed integer programming algorithm is used to minimize the objective function of the first-stage group control optimization, with the start-stop state of each device as the optimization variable, so as to control the start-stop state of each device in the chiller system during the first-stage group control. Step 120: Based on the first stage of group control, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables, so as to optimize the operating equipment status of the chiller plant system during the second stage of group control.

[0020] In this exemplary embodiment, the MPC (Model Predictive Control) algorithm, as an advanced process control method, is introduced into the field of chiller plant group control. MPC establishes a dynamic model of the system, predicts the system behavior over a future period, and performs rolling optimization with the goal of minimizing energy consumption, thereby achieving globally optimal control.

[0021] In this application, the most suitable equipment start-up and shutdown state in the chiller plant system is determined by minimizing the objective function of the first-stage group control optimization to reduce unnecessary start-ups. Based on the most suitable equipment start-up and shutdown state, the second-stage group control optimization objective function is minimized to determine the specific operating state of each piece of equipment, thereby helping to minimize energy consumption and improve energy efficiency.

[0022] Figure 2 This is a block diagram of the overall architecture of a cooling plant system provided in an embodiment of the present invention. Figure 2 As shown, the overall architecture of the cooling plant system includes: The system consists of an application layer 01, a business logic layer 02, a data service layer 03, a device access layer 04, and a physical device layer 05. The application layer 01 is used for data visualization, alarm management, system configuration, and energy consumption analysis; the business logic layer 02 is used for a two-stage optimization engine, device combination optimization, MPC continuous optimization, a security verification online learning module, model training, parameter updates, and performance evaluation. Data service layer 03 is used for data preprocessing, feature engineering, storage management, and caching services; Device access layer 04 includes Modbus protocol, database interface, OPCUA (Open Platform Communications Unified Architecture) protocol and HTTP / REST protocol; Physical equipment layer 05 includes chiller units, chilled water pumps, cooling water pumps, and cooling towers.

[0023] This application solution can be applied to data center cold storage systems, wherein the data center cold storage system includes at least: At least 3 centrifugal chiller units, with N+1 redundancy configuration, each with a cooling capacity of ≥1000kW and a chilled water outlet temperature setting range of 7℃-12℃; At least four chilled water pumps and four cooling water pumps are equipped with variable frequency control, with a frequency adjustment range of 30Hz-50Hz. At least two cooling towers are used, each equipped with two variable frequency fans, with the fan frequency adjustment range being 20Hz-50Hz; The data center cooling station system operates 24 hours a day, 24 / 7, with cooling load demand fluctuating between 30% and 100%.

[0024] In some instances, during the operation of the chiller plant system, a mixed-integer programming algorithm is used to minimize the objective function of the first-stage group control, with the start-stop state of each device as the optimization variable, to control the start-stop state of each device in the chiller plant system during the first-stage group control, including: During the operation of the chiller plant system, a mixed integer programming algorithm is used to optimize the objective function of the first stage group control by minimizing the start and stop states of each device as optimization variables, and outputs the control mode of the start and stop states of each device in the first stage group control of the chiller plant system. The start-stop status of each device in the cooling plant system is controlled based on the control method of the start-stop status of each device during the first stage of group control.

[0025] In this exemplary embodiment, when performing minimum optimization on the objective function of the first-stage group control, the control method for the start-stop state of each device in the cooling plant system during the first-stage group control is determined, and the start-stop state of each device in the cooling plant system is controlled based on the control method for the start-stop state of each device in the cooling plant system during the first-stage group control.

[0026] In this exemplary embodiment, the control method for the start / stop status of each device in the first-stage group control of the chiller plant system includes whether each device in the first-stage group control is in an on state or a stopped state. For example, the first device is in an on state, the second device is in a stopped state, and so on.

[0027] In this exemplary embodiment, based on the first-stage group control, the step of using a model predictive control algorithm to minimize the objective function of the second-stage group control, with chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables, to control the operating state of the equipment in the chiller plant system during the second-stage group control includes: Based on the first stage of group control, the model predictive control algorithm is used to minimize the objective function of the second stage of group control by using chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables, and outputs the control mode of chilled water outlet temperature, chilled pump frequency and cooling tower frequency of the equipment operating in the chiller plant system during the second stage of group control. The operating status of the chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency of the chilled water plant system during the second stage of group control is controlled based on the control method of the chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency of the operating equipment.

[0028] In this exemplary embodiment, when performing minimum optimization on the objective function of the second-stage group control, the control methods of chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency of each operating device of the chiller plant system during the second-stage group control are determined, and the operating equipment status of the chiller plant system is controlled based on the control methods of chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency of each operating device of the chiller plant system during the second-stage group control.

[0029] In this exemplary embodiment, the control method for the chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency of the operating equipment in the second-stage group control of the chiller system includes determining the set value of the chilled water outlet temperature output from the chiller, determining the chilled water pump frequency, cooling pump frequency, and cooling tower frequency, etc. Thus, by minimizing the objective function of the first-stage group control optimization, the most suitable equipment start-up and shutdown states in the chiller system are determined to reduce unnecessary start-ups. Based on the most suitable equipment start-up and shutdown states, the second-stage group control optimization objective function is minimized to determine the specific operating states of each piece of equipment, including the most suitable chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency, etc., thereby helping to minimize energy consumption and improve energy efficiency.

[0030] In this exemplary embodiment, the step of using a mixed integer programming algorithm to minimize the objective function of the first-stage group control during the operation of the chiller system, with the start-stop state of each device as the optimization variable, and outputting the control method of the start-stop state of each device in the first-stage group control of the chiller system, includes: Step A: Obtain the input data required for optimization: Collect real-time operating data and equipment topology connections of the chiller plant system at the current moment; real-time operating data includes the operating status, cumulative runtime, and number of start-ups and shutdowns of each device; Obtain the predicted cooling load demand for the future control cycle output by the load forecasting module; determine the balance intensity coefficient based on the predicted cooling load demand, setting a low balance intensity coefficient for high load periods and a high balance intensity coefficient for low load periods; Based on the topological connection relationship, the device grouping boundary is determined, and devices with physical connection relationships are divided into the same balanced group; Within each equalization group, a device priority sequence is generated using a weighted round-robin mechanism based on the cumulative runtime and the equalization intensity coefficient. The device priority sequence is used as the input for generating candidate device start-up and shutdown schemes; Read the updated device energy consumption model parameters online for accurate estimation of the total system energy consumption in the objective function; Step B: Construct a mixed-integer programming optimization problem: Using the start / stop status of each device x_i∈{0,1} as the optimization variable, i=1,2,...,N, where N is the total number of devices, construct the first-stage group control optimization objective function; Set constraints for the optimization problem, including at least cooling capacity constraints, minimum number of devices to be activated constraints, device priority sequence, and device state logic constraints. Cooling capacity constraint means that in a cooling plant system, the total cooling capacity of all operating chillers must be greater than or equal to the current or predicted cooling load demand in order to ensure that the system can meet the cooling needs of end users. Equipment state logic constraints refer to the state association constraints between various devices in a chiller plant system due to physical connection relationships, control logic, or operating rules. In other words, the start-up and shutdown states of certain devices must satisfy specific logical relationships with the states of other devices. Step C: Perform mixed-integer programming solution: Invoke the mixed integer programming solver to search for the optimal solution that minimizes the objective function in the feasible region consisting of 2^N combinations of devices; When the number of devices N is large, resulting in excessive computational cost for the full-space search, the branch and bound method or heuristic algorithm can be used to accelerate the solution process. Set the solution time limit T_max, with a value range of 5-30 seconds, to ensure that the optimization process meets the control cycle requirements; Step D: During the operation of the chiller plant system, a mixed integer programming algorithm is used, with the start / stop status of each device as the optimization variable, and the constraints of cooling capacity, minimum number of devices to be turned on, device priority sequence, and device state logic as the constraints, to minimize the objective function of the first-stage group control optimization, and output the control method of the start / stop status of each device in the chiller plant system during the first-stage group control: The optimal solution obtained from the solution is then parsed into specific control commands for each device: For devices with x_i=1, the control mode is "on"; for devices with x_i=0, the control mode is "off" or "remain off". For devices that need to change their state, record their target state and estimated execution time; Step E: Output control mode and prepare for execution: The generated device start / stop status control method is encapsulated into a standard format control instruction set; The control instruction set includes device identifier, target status, execution priority, and execution timestamp; The control instruction set is transmitted to the instruction issuing module for subsequent security verification and actual execution.

[0031] In this exemplary embodiment, the model predictive control algorithm includes a load forecasting model: Based on the first-stage group control, a model predictive control algorithm is used to minimize the objective function of the second-stage group control using chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables. The resulting output shows the control methods for the chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency of the operating equipment in the chiller plant system during the second-stage group control, including: Based on historical cooling load data, outdoor meteorological information, and calendar characteristics, a load forecasting model is used to predict the cooling load demand sequence for the next N control cycles; N is 6-8. The predicted cooling load is used as the core constraint for model predictive control, i.e., the cooling capacity ≥ the predicted load demand. When the load prediction error exceeds a preset threshold, the model will be automatically retrained. The optimization of the objective function for the second-stage group control includes minimizing the objective function, which involves: Model predictive control algorithms can employ multi-objective optimization strategies to optimize objective functions, including: A multi-objective optimization strategy is adopted to optimize the first objective, the second objective, and the third objective. The first objective is to minimize the total system energy consumption in the prediction time domain based on the online-updated device energy consumption model. The second objective includes minimizing the changes in control variables, including changes in chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency. The third objective includes using soft constraints to penalize control actions that exceed the preset rate of change, minimizing the impact of equipment start-up and shutdown switching on system stability. The three objective items are summed by a dynamic weighting coefficient, which is adaptively adjusted according to the system's operating conditions.

[0032] In this exemplary embodiment, the adaptive adjustment strategy for the dynamic weight coefficients includes: When the system load change rate exceeds 10% / 15 minutes, increase the weight of the second objective term to suppress drastic fluctuations in the control variable; When the system is running in the high-efficiency zone (COP≥5.0), reduce the weight of the first objective item to prioritize system stability; When the number of equipment start-ups and shutdowns approaches the lifespan limit, increase the weight of the third objective item to extend the equipment's lifespan. The weighting coefficient is adjusted in the range of 0.1-0.9, and a smooth transition is achieved by using fuzzy control rules.

[0033] In this exemplary embodiment, the model predictive control algorithm employs a rolling time-domain optimization strategy, including: The prediction time domain P is set to 6-12 control cycles, and the control time domain M is set to 1-3 control cycles; In each control cycle, based on the current system state and the prediction model, the optimal control sequence for the next P cycles is solved, and the control command for the first cycle of the control sequence is issued and executed. In the next control cycle, the system state is updated, and the optimization solution is re-performed to achieve rolling optimization.

[0034] The prediction time domain P and the control time domain M are adaptively adjusted according to the dynamic characteristics of the system: For periods of slow load change (nighttime or transitional seasons), increase the prediction time domain P to 12 cycles to improve the optimization foresight; For periods of drastic load changes (daytime peak hours), reduce the prediction time domain P to 6 cycles to improve the optimization response speed; The control time domain M is dynamically adjusted according to the system inertia. For systems with slow water temperature response, M is increased to 3 cycles; for systems with fast water temperature response, M is decreased to 1 cycle.

[0035] Figure 3 A flowchart illustrating the two-stage optimization method provided in this embodiment of the invention. Figure 3 As shown, the two-stage optimization method includes: Step 30: Trigger MPC optimization periodically; Step 31: Obtain data; Step 32: Obtain the device operating status; Step 33: Obtain the topology connection relationship; Step 34: Load the load prediction model; Step 35: Balance equipment operation; Step 36: Generate equipment start-up and shutdown plans; Step 37: Construct the objective function and constraints; Step 38: Solve the optimization problem; Step 39: Obtain the variable setting value; Step 40: Issue control commands; Step 41: Record the optimization results.

[0036] Figure 4 This is a schematic diagram illustrating the optimization of the MPC optimization model provided in an embodiment of the present invention. Figure 4 As shown, it includes: Step 50: Obtain real-time measurement values, such as temperature, pressure, flow rate, power, and cooling capacity; Step 51: Obtain historical data, such as operational data, environmental data, and load data; Step 52: Update the parameters of the prediction model, chiller energy consumption model, water pump energy consumption model, and cooling tower energy consumption model using real-time measurements and historical data; Step 53: Optimize the objective function by using an optimized solver, based on the constraints and the solution algorithm. Step 54: Output control commands, including setpoints, frequency, and start / stop commands; Step 55: Control the chiller plant system, including controlling the chiller units, water pumps, cooling towers, and valves.

[0037] In some instances, the objective function for the first-stage group control optimization is... for: ; wherein, the The total energy consumption of the chiller system, For equipment start-up and shutdown costs, Costs related to equipment lifespan depletion; All are weighting coefficients; equipment start-up and shutdown costs are penalties for changes in equipment state, that is, the costs incurred when the equipment changes from on to off or from off to on; ;in, The energy consumption of the chiller at time k. The energy consumption of the refrigeration pump and the cooling pump at time k. Let be the energy consumption of the cooling tower at time k; ;in, For the i-th device, the start / stop status for the t-th time. This represents the start-up and stop status of the i-th device during the (t-1)th iteration. ; The cumulative runtime of the i-th device, For the number of times the i-th device is started and stopped, , All are weighting coefficients.

[0038] In this exemplary embodiment, during the operation of the chiller system, the first-stage group control optimization objective function is minimized using a mixed-integer programming algorithm, with the start-stop state of each device as the optimization variable. This includes: Determine the first constraint condition for the optimization variables; During the operation of the chiller plant system, in conjunction with the first constraint condition of the optimization variables, a mixed integer programming algorithm is used to minimize the objective function of the first-stage group control optimization, with the start-stop state of each device as the optimization variable; wherein, the first constraint condition of the optimization variables includes: ; The actual cooling capacity of the chiller Forecast the demand for refrigeration systems; This represents the total number of chillers; This is the margin coefficient; ; This represents the start / stop status of the i-th device; ; This represents the minimum number of chillers that can be turned on.

[0039] In this exemplary embodiment, by using a mixed integer programming algorithm based on the first constraint condition, and taking the start-up and shutdown status of each device as the optimization variable, the objective function of the first-stage group control optimization is minimized. This helps to better determine the most suitable start-up and shutdown status of the devices in the chiller plant system, thereby reducing unnecessary startups.

[0040] In some instances, the objective function for the second-stage group control optimization is... for: ;in, This is the current sampling time; For prediction of the time domain (number of steps), the value ranges from 6 to 12. To control the time domain (number of steps), the value ranges from 1 to 3; The predicted cooling capacity (kW) for step i is estimated by combining the predicted power from the energy consumption model with the COP. The predicted cooling load (kW) for step i is provided by the load forecasting model (such as a time-series large model); For example, when the cooling penalty coefficient is applied, the penalty coefficient is calculated as follows: (Sufficient cooling supply) It can be 1.0, when (Insufficient cooling) This could be 2.0 (with heavier penalties and priority given to cooling supply), but this is just an example and has no limiting effect. The total power of the system in step i; To control the weight diagonal matrix; ; These are the diagonal elements of the control weights diagonal matrix R, and the vector of changes in the control variables. The four components correspond one-to-one; Weighting for cooling capacity tracking; Energy consumption weighting; It is a vector consisting of the changes in chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency. ,in, This represents the change in the outlet temperature of the chilled water. This represents the frequency change of the refrigeration pump. This represents the frequency variation of the cooling pump. This refers to the frequency variation of the cooling tower. ;in, The energy consumption of the chiller at time k. The energy consumption of the refrigeration pump and the cooling pump at time k. Let be the energy consumption of the cooling tower at time k.

[0041] In some instances, based on the first-stage group control, a model predictive control algorithm is used to minimize the objective function of the second-stage group control, with chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables. This optimizes the operating equipment status of the chiller plant system during the second-stage group control, including: Determine the second constraint condition for the optimization variables; Based on the first stage of group control, and combined with the second constraint of the optimization variables, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables; wherein, the second constraint of the optimization variables includes at least one of the following: chilled water outlet temperature constraint, chilled pump frequency constraint, cooling pump frequency constraint and cooling tower frequency constraint.

[0042] The temperature limits for chilled water outlet include: ; Let the chilled water outlet temperature be at time k. The lower limit of the chilled water outlet temperature at time k. This represents the upper limit of the chilled water outlet temperature at time k. ;in, Let k be the change in chilled water outlet temperature at time k. This represents the upper limit of the change in chilled water outlet temperature at time k.

[0043] The frequency limits for chilled pumps include: ;in, Let the frequency of the chilled pump be at time k. The lower limit of the refrigerant pump frequency at time k. This represents the upper limit of the refrigerant pump frequency at time k. ;in, The change in the frequency of the refrigeration pump at time k is... This represents the upper limit of the change in the refrigeration pump frequency at time k.

[0044] Cooling pump frequency limitations include: ;in, The frequency of the cooling pump at time k is... The lower limit of the cooling pump frequency at time k. This represents the upper limit of the cooling pump frequency at time k. ;in, The change in cooling pump frequency at time k. This represents the upper limit of the frequency change of the cooling pump at time k.

[0045] Cooling tower frequency limits include: ;in, Let the frequency of the cooling tower be at time k. The lower limit of the cooling tower frequency at time k. This represents the upper limit of the cooling tower frequency at time k. ;in, The change in cooling tower frequency at time k. This represents the upper limit of the frequency change of the cooling tower at time k.

[0046] In some instances, the second constraint further includes: an approximation temperature limit for the cooling tower outlet water; wherein the approximation temperature limit for the cooling tower outlet water is determined to be that the approximation temperature of the cooling tower outlet water at time K is less than or equal to a pre-set maximum approximation temperature of the cooling tower outlet water.

[0047] For example, ;in, The approximation temperature of the cooling tower outlet water at time k. This represents the maximum approximation temperature of the cooling tower outlet water.

[0048] In this exemplary embodiment, based on the second constraint of the optimization variables, the model predictive control algorithm is used to minimize the objective function of the second-stage group control optimization by using the chilled water outlet temperature, chilled water pump frequency, cooling pump frequency and cooling tower frequency as optimization variables. This helps to better determine a more suitable chilled water outlet temperature setpoint of the chiller, determine the chilled water pump frequency, cooling pump frequency and cooling tower frequency, etc., thereby helping to minimize energy consumption and improve energy efficiency.

[0049] In this exemplary embodiment, the CasADi framework and IPOPT solver are used to solve the optimal control sequence within 6 to 12 steps in the prediction time domain and 1 to 3 steps in the control time domain.

[0050] After outputting the control methods for the start and stop status of each device in the first stage of group control of the chiller plant system, safety verification and instruction generation can be performed.

[0051] For example, the safety verification module performs a triple check on the optimization results output by MPC to confirm that the set value is within a safe range (such as the chilled water outlet temperature). The system determines whether the change in the control variable exceeds the threshold and verifies that the instruction is consistent with the current operating status of the equipment. If the rate of change exceeds the limit, a step-by-step adjustment strategy is initiated, decomposing the target value into multiple small steps to gradually approach it, thus avoiding shocks.

[0052] The verified commands are generated by the control command module into standard BACnet write commands, which are then sent to the edge controller via the industrial network to achieve closed-loop control. The system also supports a RESTful API, allowing integration with other energy management systems.

[0053] In some instances, before constructing the first-stage group control optimization objective function, the method includes: A COP model for chiller units is constructed based on chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling capacity, and the corresponding weighting coefficients for each quantity. Based on the chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling pump frequency, chilled water pump frequency, cooling capacity, and the correlation between these quantities, a chiller energy consumption model is constructed. A chilled pump energy consumption model is constructed based on chilled pump power, chilled pump frequency, actual flow rate, and chilled pump model coefficients. A cooling pump energy consumption model is constructed based on cooling pump power, cooling pump frequency, actual flow rate, and cooling pump model coefficients. A cooling tower energy consumption model is constructed based on cooling tower power, fan frequency, actual flow rate, and cooling tower model coefficients. A cooling tower approximation model is constructed based on the cooling water outlet temperature, wet-bulb temperature, fan frequency, cooling capacity, and cooling tower approximation model coefficients.

[0054] In some instances, including: Construct a COP model for the chiller unit, an energy consumption model for the chiller, an energy consumption model for the water pump, an energy consumption model for the cooling tower, and an approximation model for the cooling tower; The chiller unit COP model includes: ;in, This refers to the inlet temperature of the chilled water. This refers to the outlet temperature of the chilled water. This refers to the inlet temperature of the cooling water. This refers to the outlet temperature of the cooling water. This refers to the cooling capacity; a 0、 a 1、 a 2、 a 3、 a 4 and a 5 are all weighting coefficients.

[0055] The chiller energy consumption model includes: ;in, f () represents the relationship between the chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, and cooling pump frequency, chilled water pump frequency, and cooling capacity. f () can be a polynomial regression function; for example... ; This refers to the inlet temperature of the chilled water. This refers to the outlet temperature of the chilled water. This refers to the inlet temperature of the cooling water. This refers to the outlet temperature of the cooling water. This refers to the frequency of the cooling pump. This refers to the frequency of the chilled pump. This refers to the cooling capacity; , , , , , , and All of these are model parameters.

[0056] The cooling / refrigeration pump energy consumption model includes: ,in These are the model coefficients (obtained through online learning); where, This refers to the power of the refrigeration / cooling pump; This refers to the frequency of the cooling / refrigeration pump.

[0057] The cooling tower energy consumption model includes: Among them, model coefficientsk fan Dynamic adjustments are made through online learning; For cooling tower power; This refers to the frequency of the wind turbine.

[0058] The cooling tower approximation model includes: ;in, This refers to the cooling water outlet temperature; Wet-bulb temperature; For the fan frequency; The approximation temperature of the cooling tower outlet water; All are model coefficients.

[0059] In this exemplary embodiment, during the operation of the chiller plant system, a mixed integer programming algorithm is used to minimize the objective function of the first-stage group control optimization, with the start-stop state of each device as the optimization variable. This is to ensure that before controlling the start-stop state of each device in the first-stage group control of the chiller plant system, the method includes: Based on user configuration, the system periodically starts the online learning engine and uses the recursive least squares method to incrementally update the parameters of each model in the chiller COP model, chiller energy consumption model, chilled water pump energy consumption model, cooling pump energy consumption model, cooling tower energy consumption model, and cooling tower approximation model corresponding to the operating equipment in the chiller plant system.

[0060] Specifically, according to user configuration, the system periodically activates the online learning engine and uses the recursive least squares method to incrementally update the parameters of each model in the chiller COP model, chiller energy consumption model, chilled water pump energy consumption model, cooling pump energy consumption model, cooling tower energy consumption model, and cooling tower approximation model corresponding to the operating equipment in the chiller plant system. This includes: During the initial cold start phase of system deployment, based on the acquired physical architecture information of the cold station system, the initial parameters, forgetting factor range, and parameter update frequency of the recursive least squares method are dynamically configured; wherein, the physical architecture information of the cold station system includes equipment type, number of equipment, topology connection relationship, communication protocol type, and controller computing resources; Model parameters are updated based on initial parameters, the range of forgetting factor values, and parameter update frequency. The initial parameters of the recursive least squares method are configured differently based on the device type and the number of devices: For different types of equipment, differentiated initial model parameter vectors are set based on their rated power, service life, and historical performance degradation characteristics; For a large number of devices of the same type, clustering is used to divide the devices into several categories, and representative initial parameters are set for each category of devices to reduce the degree of freedom in parameter identification. The range of the forgetting factor is dynamically configured based on the communication protocol type and the controller's computing resources. For systems that use high-speed communication protocols and have sufficient computing power, the forgetting factor is set to a range of 0.95 to 0.99 to balance the memory of historical data with the response speed to new data. For systems using low-speed communication protocols or with limited computing power, the forgetting factor is set to a range of 0.90 to 0.95 to reduce the dependence on high-frequency data updates and improve algorithm stability. For scenarios where device performance degrades rapidly, the forgetting factor is dynamically adjusted to a lower value to accelerate model convergence to the current system characteristics. The parameter update frequency is dynamically configured based on the controller's computing resources and communication protocol type. For example, when the controller's CPU utilization is detected to be below 30% and the communication latency is less than 50ms, the parameter update frequency is configured to be once every 10-30 seconds. When the controller CPU utilization rate is detected to be higher than 70% or the communication delay is greater than 200ms, the parameter update frequency is reduced to once every 60-120 seconds to avoid system overload or communication blockage caused by high frequency updates. For devices using slow communication protocols, a lower update frequency is set separately and processed differently from devices using fast protocols; The topological connectivity is used to guide the parameter update structure of the recursive least squares method: Construct a device association matrix based on the physical connection relationships between devices; In the parameter update process of recursive least squares, a topological constraint matrix is ​​introduced to update the coupling parameters only for devices with physical connections, thereby reducing redundant calculations. The step size and range of coupling parameter updates are dynamically adjusted based on the complexity of the topological connections.

[0061] In this exemplary embodiment, data acquisition and preprocessing can be performed before constructing the first-stage group control optimization objective function.

[0062] The system collects real-time operating data from chillers, pumps, cooling towers, and environmental sensors in the chiller plant system via industrial communication protocols. This data includes: chiller inlet and outlet water temperatures, chiller power, chiller status, pump frequency, pump flow rate, pump current, cooling tower fan frequency, inlet and outlet water temperatures, and outdoor dry and wet bulb temperatures.

[0063] The collected data is first sent to the data preprocessing module for cleaning, including: removing outliers caused by communication interruptions, and using... The criteria identify and correct sensor drift, and use a moving average method to smooth high-frequency noise. Then, standardization is performed to unify data of different dimensions into the [0,1] interval (normalization), facilitating subsequent modeling and optimization.

[0064] The collected data may include equipment operation data and environmental and external data. Equipment operation data includes chiller inlet and outlet water temperatures, chiller power, chiller status, water pump frequency, water pump flow rate, water pump current, cooling tower fan frequency, and inlet and outlet water temperatures.

[0065] Environmental and external data, including outdoor dry and wet bulb temperatures, and calendar information (weekdays / holidays).

[0066] The preprocessing steps for the collected data include outlier cleaning, drift correction, noise smoothing, and standardization.

[0067] Outlier cleaning includes removing NaN and Inf values ​​caused by communication interruptions.

[0068] Drift correction includes using the 3σ criterion to identify sensor drift, combined with mid-range or linear interpolation correction within a sliding window.

[0069] Noise smoothing includes using moving averages or low-pass filters to eliminate high-frequency noise.

[0070] Standardization includes normalizing data of different dimensions to the [0,1] interval.

[0071] In this exemplary embodiment, load and parameter prediction can be based on historical load data, outdoor meteorological information (temperature, humidity), and calendar characteristics (weekdays / holidays) to use a time-series large model to predict the cooling load demand and wet-bulb temperature trend for the next 6 to 12 time periods (sampling period 10 to 30 minutes). Cooling load (cooling capacity ≥ demand) is the core constraint in MPC optimization. Wet-bulb temperature is input into the cooling tower model for approximation calculation and affects COP estimation.

[0072] Input data may include historical cooling load sequences, outdoor weather information (temperature, humidity), and calendar features (weekdays / holidays).

[0073] Prediction models: large time series models (Transformer), LSTM, ARIMA, etc.

[0074] Output: Cooling load demand sequence for the next 6-12 time periods (each time period is 10-30 minutes). Future wet-bulb temperature trend ;in, This represents the cooling load demand for the t+N period. Let t be the future wet-bulb temperature during the time period t+N.

[0075] In this exemplary embodiment, during online model learning and updating, the model management module incorporates a chiller COP model, a chiller energy consumption model, a water pump energy consumption model, a cooling tower energy consumption model, and a cooling tower approximation model. Based on user configuration, the system periodically (e.g., hourly) starts the online learning engine, using recursive least squares to incrementally update the model parameters, including a forgetting factor. The value is set to 0.98 to balance historical data with recent performance. The updated model is automatically deployed after evaluation on the validation set, ensuring that the model always reflects the current system characteristics.

[0076] In this exemplary embodiment, during model updates and deployments, the new model can be evaluated on the validation set (data from the most recent hour) for root mean square error (RMSE) and goodness of fit (R²). If the performance improvement is ≥5%, the old model is automatically replaced, the model version is archived, and manual review and rollback are supported.

[0077] Figure 5 This is a flowchart illustrating the workflow of the online learning module provided in an embodiment of the present invention. Figure 5 As shown, the workflow of the online learning module includes: Step 60: Trigger training tasks periodically; Step 61: Obtain historical data; Step 62: Data Processing and Feature Engineering; Step 63: Select the model architecture; Step 64: Train the model; Step 65: Model performance evaluation; Step 66: Determine if the model performance meets the standards; Step 67: If the model performance meets the target, deploy the new model; otherwise, proceed to step 71 to memorize the training results. Step 68: Monitor model performance; Step 69: Determine if the model performance is stable; Step 70: If the model performance is stable, the model update is complete. Step 71: If the model performance is unstable, roll back to the old model.

[0078] Figure 6 This is a logic block diagram of a security verification module provided in an embodiment of the present invention. Figure 6 As shown, the verification logic of the security verification module includes: Step 72, Input: Optimize control command; Step 73: Multiple security checks, including boundary checks, rate of change limits, and state matching; Step 74: Determine whether the comprehensive safety assessment meets the standards; Step 75: If the comprehensive safety assessment meets the standards, then issue the instruction; Step 76: If the comprehensive safety assessment fails to meet the standards, adjust the strategy step by step.

[0079] This application first determines the optimal equipment start-up and shutdown combination through discrete optimization, and then performs continuous variable MPC optimization based on this combination to maximize global energy efficiency. Recursive least squares (RLS) or neural networks are used to identify and update the energy consumption model online, solving the MPC model mismatch problem. Multiple safety checks and step-by-step adjustment strategies are employed to verify the optimization results for boundary conditions, rate of change, and state matching. Step-by-step approximation is used when changes are drastic to ensure control safety. Equipment lifespan loss costs are incorporated into the optimization objective, and a polling mechanism is used to balance the operating time of multiple devices, extending the overall lifespan. Furthermore, the modular system architecture of this application can include independent modules such as data acquisition, preprocessing, model management, two-stage optimization, safety checks, and instruction generation, supporting flexible deployment and system integration.

[0080] This application also includes the use of a state machine model internally by the control command executor to execute commands during actual equipment control. The execution of each command consists of multiple steps. The detailed process is as follows; 1. Instruction Reception and Arbitration: Upon receiving the instruction, generate a unique execution ID and record the start time.

[0081] Compare the priority of the new instruction with that of the currently executing instruction: If the new instruction has a higher priority, the current instruction is interrupted (the remaining steps are discarded), and the new instruction begins execution.

[0082] Interrupted instructions are marked as "overwritten".

[0083] If the priority of the new instruction is the same as or lower than that of the currently executed instruction, it is added to the waiting queue (if queuing is allowed), or it is rejected directly.

[0084] 2. Generate the operation sequence: If the instruction is for a combination of target devices (e.g., operating chillers A and B), then: Obtain the current device status and determine the list of devices that need to be started and stopped.

[0085] Power-on sequence (fixed): Cooling tower butterfly valve → chiller butterfly valve → cooling water pump → chilled water pump → cooling tower → chiller unit; Shutdown sequence (fixed): Chiller unit → Cooling tower → Chilled water pump → Cooling water pump → Cooling tower butterfly valve → Chiller butterfly valve (addition / reduction sequence); Addition process: refrigeration pump → cooling pump → chiller butterfly valve → chiller; Reduced load: Refrigeration unit → Refrigeration unit butterfly valve → Cooling pump → Refrigeration pump.

[0086] This invention provides a group control optimization system for a cooling plant system. Figure 7 This is a schematic diagram of the structure of the group control optimization device for a chiller plant system provided in an embodiment of the present invention. Figure 7 As shown, the chiller plant system group control optimization system includes: The objective function construction unit 80 is used to construct the first-stage group control optimization objective function based on the total energy consumption of the chiller system, the equipment start-up and shutdown costs, and the equipment life loss costs, and to construct the second-stage group control optimization objective function based on the total energy consumption of the chiller system, the change in chilled water outlet temperature, the change in chilled pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency. The first-stage group control unit 81 is used to optimize the first-stage group control objective function by using a mixed integer programming algorithm with the start-stop state of each device as the optimization variable during the operation of the chiller system, thereby controlling the start-stop state of each device in the chiller system during the first-stage group control. The second-stage group control unit 82 is used to optimize the operating status of the equipment in the chilled plant system during the second-stage group control by using a model predictive control algorithm with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables.

[0087] In this exemplary embodiment, the MPC (Model Predictive Control) algorithm, as an advanced process control method, is introduced into the field of chiller plant group control. MPC establishes a dynamic model of the system, predicts the system behavior over a future period, and performs rolling optimization with the goal of minimizing energy consumption, thereby achieving globally optimal control.

[0088] In this application, the most suitable equipment start-up and shutdown state in the chiller plant system is determined by minimizing the objective function of the first-stage group control optimization to reduce unnecessary start-ups. Based on the most suitable equipment start-up and shutdown state, the second-stage group control optimization objective function is minimized to determine the specific operating state of each piece of equipment, thereby helping to minimize energy consumption and improve energy efficiency.

[0089] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0090] This invention provides a computer-readable storage medium storing a cooling plant system group control optimization program thereon. When the cooling plant system group control optimization program is executed by a processor, it implements the cooling plant system group control optimization method described in the above embodiments.

[0091] In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0092] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0093] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first and second features are in direct contact, or that they are in indirect contact through an intermediate medium. Furthermore, "above," "over," or "on top" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," or "beneath" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0094] In the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0095] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make modifications, alterations, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A group control optimization method for a chiller plant system, characterized in that, include: Based on the total energy consumption of the chilled water plant system, the cost of equipment start-up and shutdown, and the cost of equipment lifespan loss, a first-stage group control optimization objective function is constructed. Based on the total energy consumption of the chilled water plant system, the change in chilled water outlet temperature, the change in chilled water pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency, a second-stage group control optimization objective function is constructed. During the operation of the chiller system, a mixed integer programming algorithm is used to minimize the objective function of the first-stage group control optimization, with the start-stop state of each device as the optimization variable, so as to control the start-stop state of each device in the first-stage group control of the chiller system. Based on the first stage of group control, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables, so as to optimize the operating equipment status of the chiller plant system during the second stage of group control.

2. The group control optimization method for a chiller plant system as described in claim 1, characterized in that, When performing minimum optimization on the objective function of the first stage group control, the control method for the start-stop state of each device in the first stage group control of the cooling plant system is determined, and the start-stop state of each device in the cooling plant system is controlled based on the control method for the start-stop state of each device in the first stage group control of the cooling plant system.

3. The group control optimization method for a chiller plant system as described in claim 2, characterized in that, When performing minimum optimization on the objective function of the second-stage group control, the control methods for the chilled water outlet temperature, chilled pump frequency, cooling pump frequency, and cooling tower frequency of each operating device in the chiller plant system during the second-stage group control are determined, and the operating equipment status of the chiller plant system is controlled based on the control methods for the chilled water outlet temperature, chilled pump frequency, cooling pump frequency, and cooling tower frequency of each operating device in the second-stage group control.

4. The group control optimization method for a chiller plant system as described in claim 2, characterized in that, The first stage of group control optimization objective function for: ;in, The total energy consumption of the chiller system, For equipment start-up and shutdown costs, Costs related to equipment lifespan depletion; All are weighting coefficients; ;in, The energy consumption of the chiller at time k. The energy consumption of the refrigeration pump and the cooling pump at time k. Let be the energy consumption of the cooling tower at time k; N Total number of devices; ;in, For the i-th device, the start / stop status for the t-th time. This represents the start-up and stop status of the i-th device during the (t-1)th iteration. ; The cumulative runtime of the i-th device, For the number of times the i-th device is started and stopped, , All are weighting coefficients.

5. The group control optimization method for a chiller plant system as described in claim 3, characterized in that, Based on the first-stage group control, the second-stage group control objective function is minimized using a model predictive control algorithm, with chilled water outlet temperature, chilled water pump frequency, cooling pump frequency, and cooling tower frequency as optimization variables. This optimizes the operating equipment status of the chiller plant system during the second-stage group control, including: Determine the second constraint condition for the optimization variables; Based on the first stage of group control, and combined with the second constraint of the optimization variables, the model predictive control algorithm is used to minimize the objective function of the second stage of group control, with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables; wherein, the second constraint of the optimization variables includes at least one of the following: chilled water outlet temperature constraint, chilled pump frequency constraint, cooling pump frequency constraint and cooling tower frequency constraint.

6. The group control optimization method for a chiller plant system as described in claim 1, characterized in that, Before constructing the first-stage group control optimization objective function, the method includes: A COP model for chiller units is constructed based on chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling capacity, and the corresponding weighting coefficients for each quantity. Based on the chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, cooling water outlet temperature, cooling pump frequency, chilled water pump frequency, cooling capacity, and the correlation between these quantities, a chiller energy consumption model is constructed. A chilled pump energy consumption model is constructed based on chilled pump power, chilled pump frequency, actual flow rate, and chilled pump model coefficients. A cooling pump energy consumption model is constructed based on cooling pump power, cooling pump frequency, actual flow rate, and cooling pump model coefficients. A cooling tower energy consumption model is constructed based on cooling tower power, fan frequency, actual flow rate, and cooling tower model coefficients.

7. The group control optimization method for a chiller plant system as described in claim 6, characterized in that, include: Based on user configuration, the system periodically starts the online learning engine and uses the recursive least squares method to incrementally update the parameters of each model in the chiller COP model, chiller energy consumption model, chilled water pump energy consumption model, cooling pump energy consumption model, cooling tower energy consumption model, and cooling tower approximation model.

8. A group control optimization system for a chiller plant system, characterized in that, include: The objective function construction unit is used to construct the first-stage group control optimization objective function based on the total energy consumption of the chiller plant system, the equipment start-up and shutdown costs, and the equipment life loss costs, and to construct the second-stage group control optimization objective function based on the total energy consumption of the chiller plant system, the change in chilled water outlet temperature, the change in chilled pump frequency, the change in cooling pump frequency, and the change in cooling tower frequency. The first-stage group control unit is used to optimize the first-stage group control objective function by using a mixed integer programming algorithm with the start-stop state of each device as the optimization variable during the operation of the chiller system, thereby controlling the start-stop state of each device in the chiller system during the first-stage group control. The second-stage group control unit is used to optimize the operating status of the equipment in the chilled plant system during the second-stage group control by using a model predictive control algorithm with chilled water outlet temperature, chilled pump frequency, cooling pump frequency and cooling tower frequency as optimization variables.

Images