A chiller plant control MPC optimization scheduling method based on COP maximization

Abstract

The application discloses a kind of based on COP maximization's cold water group control MPC optimization scheduling method, it is related to refrigeration system intelligent control technical field, to solve the problem that current cold water group control cannot overall plan the overall energy efficiency characteristics of multiple units collaborative operation, method includes: S1, complete unit hardware configuration, carry out system initialization, real-time acquisition operating parameter and carry out pretreatment;S2, based on historical data and environmental parameters, construct prediction model, predict the future time domain's cold load variation trend;S3, combined with mathematical model and constraint condition, construct MPC global optimization objective function, solve the scheduling scheme of COP maximization;S4, execute start-stop and load switching smooth transition logic, issue control instruction, real-time monitoring operating state and closed-loop correction parameter.The technical scheme of the application can significantly improve the overall refrigeration energy efficiency, guarantee cooling temperature stability, equipment smooth operation.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for refrigeration systems, specifically to a group control MPC optimization scheduling method for chillers based on maximizing COP. Background Technology

[0002] Centralized cooling systems typically employ parallel operation of multiple chiller units to meet dynamically changing cooling load demands. Currently, conventional group control schemes in the industry use a crude scheduling model with single-unit constant-temperature closed-loop control, fixed load distribution, and threshold-triggered start / stop. This model focuses solely on achieving the target cooling temperature at the terminal, neglecting to consider the overall energy efficiency characteristics of multiple units operating collaboratively. The current conventional group control schemes in the industry have the following shortcomings: Low overall energy efficiency: The overall system COP is not optimized, and the load distribution of units is out of sync with their own energy efficiency characteristics. Some units deviate from their high-efficiency operating range for a long time, resulting in a large amount of energy waste. Unreasonable load distribution: The use of fixed ratio or evenly distributed load mode does not take into account the aging of units and performance differences, which easily leads to problems such as unbalanced load, inefficient idling, and overload operation of some units. Large fluctuations in system operation: When the number of units is increased or decreased, or when the load changes drastically, the use of direct start-stop and step load adjustment methods leads to drastic fluctuations in cooling temperature and pipeline pressure. Increased equipment wear: Frequent load shocks and instantaneous start-stops exacerbate the wear of core components such as compressors and water pumps, shortening the service life of units. Poor adaptability: The control model is rigid and cannot adapt to the performance degradation of units, the drift of environmental conditions, and the sudden change of load, resulting in a continuous decline in control accuracy over a long period of time.

[0003] Chinese patent CN121968542A discloses a high-efficiency chiller room group control system based on multi-objective optimization. This system includes a group control center, multiple chiller units, multiple chilled water pumps, and multiple cooling towers and cooling water pumps, all signal-connected to the control center. It also includes a data acquisition module for real-time acquisition of operating parameters from the chiller units, chilled water pumps, and cooling towers and cooling water pumps; and a health prediction module signal-connected to the data acquisition module to receive the operating parameters. However, this system does not optimize the overall system COP, and its technical solution differs from that of this application. Summary of the Invention

[0004] This invention solves the problem that current chiller group control cannot coordinate the overall energy efficiency characteristics of multiple units operating in a coordinated manner. It proposes a chiller group control MPC optimization scheduling method based on maximizing COP, which improves the overall cooling energy efficiency and ensures stable cooling temperature and smooth equipment operation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] A chiller group control MPC optimization scheduling method based on COP maximization includes the following steps: S1: Complete the unit hardware configuration, perform system initialization, collect operating parameters in real time and perform preprocessing; S2, based on historical data and environmental parameters, constructs a predictive model to predict future trends in cooling load over time. S3. Combining mathematical models and constraints, construct the global optimization objective function of MPC and solve the scheduling scheme that maximizes COP. S4 executes the smooth transition logic for start-stop and load switching, issues control commands, monitors the operating status in real time, and corrects parameters in a closed loop.

[0007] In this technical solution, the overall process is roughly as follows: system initialization, data acquisition and preprocessing, cooling load prediction, MPC optimization solution, switching judgment, cooling execution and feedback, periodic iteration, and finally continuous rolling optimization scheduling. In this application, the global COP of the chiller group is maximized as the core optimization objective, replacing the traditional single-unit constant temperature control objective. A mathematical model for optimal load allocation of multiple units is constructed based on MPC model predictive control to achieve dynamic load matching. Smooth transition logic for unit start-up / load switching is designed to eliminate system fluctuations and equipment impacts.

[0008] The present invention is further configured such that step S1 includes: Configure unit parameters, set constraint thresholds, and initialize MPC prediction time domain and control parameters; The unit's internal and external operating parameters are collected, and moving average filtering is used to remove noise. Linear interpolation is used to complete missing data.

[0009] In this technical solution, the preprocessing mainly involves filtering and denoising, as well as filling in missing data.

[0010] The present invention is further configured such that: the construction of the prediction model includes: Using future cooling load demand and environmental condition changes within a preset time domain as prediction inputs, and combining the performance model of a single unit, a state-space prediction model for a chiller group is established to predict the system operating status under different load distribution schemes.

[0011] In this technical solution, while constructing the prediction model, multi-dimensional safety constraints are also set, including: upper and lower limits of unit load rate constraints, minimum start-up and shutdown duration constraints, evaporation / condensation temperature safety constraints, and single-step load fluctuation amplitude constraints.

[0012] The present invention is further configured such that: the MPC global optimization objective function takes into account both the global COP maximization and operational stability objectives, and the objective function J is equal to the sum of the first term and the second term, wherein the first term is the cumulative sum of the predicted global COP of the system, and the second term is the cumulative sum of -λ multiplied by the predicted change in unit load rate.

[0013] In this technical solution, the dual objectives of maximizing global COP and ensuring operational stability are considered, and an MPC rolling optimization objective function is constructed.

[0014] The present invention is further configured such that step S4 includes: Determine whether start-up, shutdown, or load switching is required. If so, execute the smooth transition logic; otherwise, execute the steady-state control mode. The unit executes control commands, and the operating status is fed back in a closed loop to determine whether the control cycle has ended. If yes, the cycle ends; otherwise, the process returns to step S1.

[0015] The present invention is further configured such that the scheduling scheme for solving the COP maximization problem includes: Iterate through all unit start-up and shutdown combinations to select effective solutions that meet the balance between cooling load supply and demand. For each start-stop combination, the optimal load factor of each unit is calculated to obtain the maximum local COP under that combination; By comparing the local optimal solutions of all effective combinations, the system's globally optimal start / stop state and load distribution scheme are selected.

[0016] The present invention is further configured such that: the COP is the ratio of the total cooling capacity to the total input power of the parallel chiller group, the total cooling capacity is the sum of the cooling capacities of each unit, and the total input power is the sum of the input power of each unit.

[0017] In this technical solution, the ratio of the total cooling capacity to the total input power of the parallel chiller group is the core optimization objective of this invention, and the total cooling capacity and total input power satisfy the principle of unit superposition.

[0018] The present invention is further configured such that: the single unit performance model includes a nonlinear general mapping model of single unit cooling capacity and operating parameters and a nonlinear general mapping model of single unit input power and operating parameters.

[0019] The present invention is further configured such that: the smooth transition logic performs a smooth transition according to the load gradation formula, wherein the load rate of the i-th unit at time t is equal to the load rate of the i-th unit in the previous control cycle plus or minus the product of the load gradation slope and the control step size.

[0020] The present invention is further configured such that steps S1 to S4 are repeated for each control step, and the scheduling is continuously optimized in a rolling manner.

[0021] The present invention provides a chiller group control MPC optimization scheduling method based on COP maximization, which can bring the following beneficial effects: 1. Significantly improved energy efficiency: With the core objective of maximizing global COP, all units are driven to operate in the high-efficiency range, completely solving the energy waste problem of traditional temperature control mode; 2. Precise load distribution: Dynamically matches the differences in unit energy efficiency, eliminates unbalanced load and inefficient idling, and achieves the global optimal match between cooling load and unit performance; 3. Smooth and shock-free operation: The smooth transition logic eliminates temperature and pressure fluctuations caused by start-up, shutdown, and load adjustment, reducing equipment wear and extending the service life of the unit; 4. High versatility: No specific equipment model or parameter restrictions, adaptable to various parallel chiller unit groups, convenient on-site modification and low deployment cost; 5. Excellent adaptive capability: The model can be adaptively corrected online to resist the effects of unit aging and operating condition drift, and maintain high-precision optimized control effect over a long period of time. Attached Figure Description

[0022] Figure 1 This is a flowchart of an MPC optimization scheduling method for chiller group control based on maximizing COP, according to the present invention.

[0023] Figure 2 This is a system schematic diagram of a chiller group control MPC optimization scheduling method based on COP maximization according to the present invention. Detailed Implementation

[0024] Example 1:

[0025] To address the technical problems of low overall energy efficiency, unreasonable load distribution, large system operation fluctuations, increased equipment wear, and poor adaptability in existing chiller group control technologies, this embodiment proposes a chiller group control MPC optimization scheduling method based on maximizing COP. With maximizing the global COP of the chiller group as the core optimization objective, it breaks through the limitations of traditional single-unit constant temperature control, improving overall cooling energy efficiency at the system level. A multi-unit coupled load distribution mathematical model is constructed, and dynamic optimal allocation of cooling load is achieved through MPC rolling optimization, accurately matching the real-time energy efficiency characteristics of each unit. A smooth transition mechanism for unit start-up and shutdown and load switching is established to eliminate control shocks, ensuring stable cooling temperature and smooth equipment operation. It achieves universal and adaptive control, adapting to chiller groups of different brands, numbers, and models, and possesses long-term stable optimization and control capabilities.

[0026] refer to Figure 1 and Figure 2,for Figure 1 This is a flowchart illustrating a chiller group control MPC optimization scheduling method based on maximizing COP; for Figure 2 This is a system schematic diagram of a chiller group control MPC optimization scheduling method based on COP maximization; the chiller group control MPC optimization scheduling method based on COP maximization in this embodiment includes the following steps.

[0027] Step S1: Complete the unit hardware configuration, perform system initialization, and collect and preprocess operating parameters in real time. That is, in this embodiment, the system hardware is configured first, followed by system initialization, and then data collection and preprocessing are performed.

[0028] More specifically, step S1 above mainly includes the following processes: configuring unit parameters, setting constraint thresholds, and initializing MPC prediction time domain and control parameters; collecting internal and external operating parameters of the unit, using moving average filtering to denoise, and using linear interpolation to complete missing data; for MPC, it is an abbreviation for Model Predictive Control.

[0029] Before step S2, the global COP mathematical model and the single-unit performance model are defined.

[0030] For the global COP mathematical model: COP is an abbreviation for the global energy efficiency ratio of the system. The global energy efficiency ratio of the system is defined as the ratio of the total cooling capacity of the parallel chiller group to the total input power, which is the core optimization objective of this invention.

[0031] Specifically, the system global energy efficiency ratio (COP) is equal to the total cooling capacity Q of the unit group divided by the total input power P of the unit group, where the units of the total cooling capacity Q and the total input power P of the unit group are both kW.

[0032] Furthermore, the total cooling capacity Q of the unit group and the total input power P of the unit group satisfy the superposition principle of the units, that is, the total cooling capacity Q of the unit group is the cooling capacity Q of each unit. i The summation, the total input power P is the input power P of each unit. i The sum of; where i equals 1, 2, ..., n, n is the total number of parallel chiller units, Q i The real-time cooling capacity of the i-th unit is expressed in kW and P. i This represents the real-time input power of the i-th generating unit, expressed in kW.

[0033] For the single-unit performance model: Based on the thermodynamic operating characteristics of the chiller unit, a nonlinear general mapping model of the cooling capacity, input power and operating parameters of the single unit is established. Specific equipment parameters are eliminated. The single-unit performance model includes a nonlinear general mapping model of the cooling capacity and operating parameters of the single unit and a nonlinear general mapping model of the input power and operating parameters of the single unit.

[0034] Q i equals f1(α) i T z T l ), P i equals f2(α) i T z T l ), where α i Let T be the load factor of the i-th unit (the ratio of real-time cooling capacity to rated cooling capacity). z The temperature on the evaporator side of the unit is expressed in °C (T). l The condenser side temperature of the unit is expressed in °C. f1 and f2 are nonlinear fitting functions of cooling capacity and input power, respectively, which are obtained based on the unit performance curve.

[0035] The aforementioned model has an adaptive correction mechanism. Specifically, it adopts a dual-mode approach of offline pre-identification and online adaptive correction to ensure the long-term accuracy of the model: in the offline stage, the initial model parameters are fitted using the unit's factory performance curve and standard operating condition data; in the online operation stage, the unit's operating data is collected in real time, and the model parameters are iteratively corrected using the least squares method to eliminate control errors caused by unit aging, heat exchange attenuation, and operating condition drift.

[0036] Step S2: Based on historical data and environmental parameters, construct a prediction model to predict the future trend of cooling load changes in the time domain.

[0037] The specific process for building a predictive model includes the following steps.

[0038] Using future cooling load demand and environmental condition changes within a preset time domain as predictive inputs, and combining a single unit performance model, a state-space prediction model for a chiller cluster is established to predict the system operating state under different load distribution schemes. This model must satisfy the cooling load supply-demand balance constraint, which is: the sum of the unit cooling capacities ΣQ. i Not less than the system's real-time cooling load requirement Q load The system's real-time cooling load demand Q load The unit is kW. Meanwhile, multi-dimensional safety constraints are set, including but not limited to upper and lower limits of unit load rate constraints, minimum start-up and shutdown duration constraints, evaporation / condensation temperature safety constraints, and single-step load fluctuation amplitude constraints.

[0039] Step S3: Combining the mathematical model and constraints, construct the global optimization objective function of MPC and solve the scheduling scheme that maximizes COP.

[0040] The global optimization objective function of MPC takes into account both the maximization of global COP and the stability of operation. The objective function J is equal to the sum of the first and second terms. The first term is the cumulative sum of the predicted global COP of the system, and the second term is the cumulative sum of -λ multiplied by the predicted change in unit load rate.

[0041] In this technical solution, the dual objectives of maximizing global COP and ensuring operational stability are considered. An MPC rolling optimization objective function is constructed, which is expressed as: maxJ equals ΣCOP minus λ·ΣΔα. k Where k is 1, 2, ..., N, J is the global optimization objective function; N is the time domain length of the MPC prediction; COP is the global COP of the system at the k-th prediction step; Δα k λ represents the change in load rate of the i-th unit at the k-th prediction step; λ is the load fluctuation penalty coefficient, which can balance energy efficiency and stability.

[0042] The specific process of solving the scheduling scheme that maximizes COP, i.e. the logic of optimal load allocation for multiple units, is as follows: Based on the MPC rolling optimization algorithm, the global optimal load allocation scheme is obtained by solving layer by layer. The execution steps are as follows.

[0043] The system collects real-time data on cooling load, environmental conditions, unit operating status, and power parameters, and updates the model input after filtering and noise reduction preprocessing.

[0044] We traversed all feasible unit start-up and shutdown combinations and selected the effective solutions that met the supply and demand balance of cooling load.

[0045] For each start-stop combination, a nonlinear optimization algorithm is used to solve for the optimal load rate of each unit, and the maximum local COP under that combination is obtained.

[0046] By comparing the local optimal solutions of all effective combinations, the system's globally optimal start / stop state and load distribution scheme are selected.

[0047] The optimal load command is sent to the local controller of each unit to execute precise load control.

[0048] Step S4: Execute the smooth transition logic for start-up, shutdown, and load switching, issue control commands, monitor the operating status in real time, and correct parameters in a closed loop.

[0049] Step S4 above includes the following process: determining whether start-up, shutdown or load switching is required; if yes, then executing smooth transition logic; if no, then executing steady-state control mode; the unit executes control commands, and the operating status is fed back in a closed loop to determine whether the control cycle has ended; if yes, then ending; if no, returning to step S1.

[0050] The smooth transition logic performs a smooth transition based on the load gradation formula, in which the load rate α of the i-th unit at time t is...t Equal to the load rate α of the i-th unit in the previous control cycle t-1 Add or subtract the product of the load gradient slope and the control step size.

[0051] More specifically, α i equal to α t-1 ±kΔt, for the plus or minus sign, is selected according to the unit's loading / unloading conditions; where α t Let α be the load factor of the i-th unit at time t. t-1 Δt represents the load rate of the i-th unit in the previous control cycle, k represents the load gradient slope, and Δt represents the control step size.

[0052] The specific execution process of the start-stop and load switching smooth transition logic includes the following.

[0053] Based on the predicted load, the start-up and shutdown needs of the units are anticipated in advance, and sufficient transition time is reserved to prohibit instantaneous start-up and shutdown.

[0054] Units awaiting shutdown should be gradually deloaded, while units awaiting startup should be gradually loaded, with strict control over the magnitude of single-step load changes.

[0055] Real-time monitoring of cooling temperature deviation and dynamic correction of transition slope ensure that temperature fluctuations remain within acceptable limits.

[0056] Once the system parameters stabilize, the unit's operating status and load distribution are locked, and the system enters the steady-state optimization phase. Repeat steps S1 to S4 for each control step, continuously optimizing the scheduling.

[0057] For the chiller group control MPC optimization scheduling method based on COP maximization in this embodiment, the overall process is roughly as follows: system initialization, data acquisition and preprocessing, cooling load prediction, MPC optimization solution, switching judgment, cooling execution and feedback, periodic iteration, and finally continuous rolling optimization scheduling.

[0058] System initialization mainly includes: configuring unit parameters, setting constraint thresholds, and initializing MPC prediction time domain and control parameters.

[0059] For data acquisition and preprocessing, it mainly includes: real-time acquisition of various operating parameters, removal of outliers, and completion of missing data; For cooling load forecasting, it mainly includes: predicting the future time-domain cooling load change trend based on historical data and environmental parameters.

[0060] The optimization solution for MPC mainly includes: combining mathematical models and constraints to find a scheduling scheme that maximizes COP.

[0061] The switching decision mainly includes: determining whether start / stop / load switching is required, and selecting a smooth transition or steady-state control mode.

[0062] Command execution and feedback mainly include: issuing control commands, monitoring the operating status in real time, and correcting parameters in a closed loop.

[0063] For periodic iteration, it mainly includes repeating the above process for each control step to achieve continuous rolling optimization scheduling.

[0064] In this embodiment, the core optimization objective is to maximize the global COP of the chiller group, replacing the traditional single-unit constant temperature control objective; a mathematical model for optimal load allocation of multiple units is constructed based on MPC model predictive control to achieve dynamic load matching; and a smooth transition logic for unit start-up / load switching is designed to eliminate system fluctuations and equipment impacts.

[0065] This embodiment, through the above-described technical solution, can achieve the following technical effects.

[0066] First, significant improvement in energy efficiency: With the core objective of maximizing global COP, all units are driven to operate in the high-efficiency range, completely solving the energy waste problem of traditional temperature control mode.

[0067] Secondly, precise load allocation: dynamically matches the differences in unit energy efficiency, eliminates unbalanced loads and inefficient idling, and achieves the global optimal match between cooling load and unit performance.

[0068] Thirdly, the operation is stable and shock-free: the smooth transition logic eliminates temperature and pressure fluctuations caused by start-up, shutdown and load adjustment, reduces equipment wear and extends the service life of the unit.

[0069] Fourthly, it is highly versatile: there are no specific equipment models or parameter restrictions, it is compatible with various parallel chiller groups, and on-site modification is convenient and deployment costs are low.

[0070] Fifthly, it has excellent adaptive capabilities: the model can be adaptively corrected online to resist the effects of unit aging and operating condition drift, and maintain high-precision optimized control performance over a long period of time.

[0071] Example 2:

[0072] The overall process of a chiller group control MPC optimization scheduling method based on COP maximization includes the following: system initialization, data acquisition and preprocessing, cooling load prediction, MPC optimization solution, switching judgment, cooling execution and feedback, periodic iteration, and finally continuous rolling optimization scheduling.

[0073] System initialization includes configuring unit parameters, setting constraint thresholds, and initializing MPC prediction time domain and control parameters.

[0074] For data acquisition and preprocessing, it includes: real-time acquisition of various operating parameters, removal of outliers, and completion of missing data.

[0075] For cooling load forecasting, it includes: predicting future time-domain cooling load trends based on historical data and environmental parameters.

[0076] The optimization solution for MPC includes: combining mathematical models and constraints to find a scheduling scheme that maximizes COP.

[0077] The switching decision includes: determining whether a start / stop / load switching is required, and selecting a smooth transition or steady-state control mode.

[0078] Command execution and feedback include: issuing control commands, monitoring operating status in real time, and correcting parameters in a closed loop.

[0079] For periodic iterations, the process involves repeating the above steps for each control step to achieve continuous rolling optimization scheduling.

[0080] In this embodiment, the core optimization objective is to maximize the global COP of the chiller group, replacing the traditional single-unit constant temperature control objective; a mathematical model for optimal load allocation of multiple units is constructed based on MPC model predictive control to achieve dynamic load matching; and a smooth transition logic for unit start-up / load switching is designed to eliminate system fluctuations and equipment impacts.

[0081] Based on this, this embodiment also proposes a system for implementing a chiller group control MPC optimization scheduling method based on COP maximization, referring to... Figure 2 It includes a data acquisition module, a data preprocessing module, a cooling load forecasting module, an MPC optimization solution module, a smooth switching judgment module, a smooth transition control module, a unit actuator, a status control module, and a status feedback module. The data acquisition module and the data preprocessing module can perform data acquisition and preprocessing functions. The cooling load forecasting module can perform cooling load forecasting. The MPC optimization solution module can perform MPC optimization solutions. The smooth switching judgment module and the smooth transition control module can perform switching judgment and mode selection. The unit actuator, the status control module, and the status feedback module can perform command execution and feedback.

[0082] This technical solution, through the synergistic action of the data acquisition module, data preprocessing module, cooling load prediction module, MPC optimization solution module, smooth switching judgment module, smooth transition control module, unit actuator, status control module, and status feedback module, can significantly improve energy efficiency. With maximizing global COP as the core objective, it drives all units to operate in the high-efficiency range, completely solving the energy waste problem of traditional temperature control modes; it makes load distribution more precise, dynamically matching the differences in unit energy efficiency, eliminating unbalanced loads and inefficient idling, and achieving the global optimal match between cooling load and unit performance; it makes operation more stable and shock-free, and the smooth transition logic eliminates temperature and pressure fluctuations caused by start-up, shutdown, and load adjustment, reducing equipment wear and extending the service life of the units.

[0083] Example 3:

[0084] Based on Example 1 or Example 2, this example proposes a specific implementation process of a chiller group control MPC optimization scheduling method based on COP maximization.

[0085] This embodiment of a chiller group control MPC optimization scheduling method based on COP maximization is mainly applied to centralized cooling stations in commercial complexes. It sequentially performs system hardware configuration, system initialization, data acquisition and preprocessing, cooling load prediction, model adaptive online correction, MPC constraint and optimization solution, smooth transition and instruction execution, and periodic iteration.

[0086] The system hardware configuration includes: Chiller units: #1 centrifugal (1000RT), #2 centrifugal (1000RT), #3 screw (500RT); Data acquisition module hardware: temperature / pressure / flow / energy meter, sampling period 1s; Control unit: Industrial PLC and edge computing gateway, control cycle 10s; Communication: Modbus TCP / IP, supporting remote control and data acquisition of the unit.

[0087] System initialization includes: Unit parameters: load rate range [20%, 100%], minimum start-up and shutdown time 300s, evaporation temperature 4 / 8℃, condensation temperature 28 / 36℃; MPC parameters: prediction time N is 60 steps (10 min), load gradient slope k is 2% / step, and penalty coefficient λ is 0.8; Model initialization: Fit initial nonlinear mapping functions f1 and f2.

[0088] For data acquisition and preprocessing, the following are included: collecting outdoor temperature and humidity, supply and return water temperature, flow rate, unit operating status, power, and evaporation / condensation temperature; using moving average filtering to remove noise and linear interpolation to complete missing data.

[0089] For cooling load forecasting, this embodiment uses an LSTM time series model to achieve the forecast. It uses 72 hours of historical data, weather, and time period as input to predict the cooling load for the next 60 steps, and smooths outliers.

[0090] For adaptive online correction of the model, it includes: collecting stable running data and updating the model coefficients using the recursive least squares method; and forcibly correcting the error when it exceeds 5% for three consecutive periods.

[0091] For MPC constraints and optimization solutions, it includes: Establish constraints on supply and demand balance, load factor, start / stop duration, temperature, and fluctuation. The optimal load factor is solved by iterating through 7 unit combinations and using SQP sequential quadratic programming. The globally optimal solution is selected based on maximizing the objective function J.

[0092] For smooth transitions and instruction execution, it includes: Start-up / Stop / Load Adjustment: Anticipate 60 seconds in advance and gradually increase / decrease load by ±2% per step; Dynamic correction slope when temperature deviation exceeds ±0.3℃; Once the system stabilizes, it locks into a stable state, issues commands, and provides closed-loop feedback.

[0093] For periodic iterations, it includes: A rolling optimization of MPC is completed every 10 seconds, with continuous adaptive scheduling.

Claims

1. A method for optimizing scheduling of a chiller plant control (MPC) based on COP maximization, characterized in that, Includes the following steps: S1: Complete the unit hardware configuration, perform system initialization, collect operating parameters in real time and perform preprocessing; S2, based on historical data and environmental parameters, constructs a predictive model to predict future trends in cooling load over time. S3. Combining mathematical models and constraints, construct the global optimization objective function of MPC and solve the scheduling scheme that maximizes COP. S4 executes the smooth transition logic for start-stop and load switching, issues control commands, monitors the operating status in real time, and corrects parameters in a closed loop.

2. The method of claim 1, wherein the method is based on COP maximization. Step S1 includes: Configure unit parameters, set constraint thresholds, and initialize MPC prediction time domain and control parameters; The unit's internal and external operating parameters are collected, and moving average filtering is used to remove noise. Linear interpolation is used to complete missing data.

3. The method of claim 1 or 2, wherein, The construction of the prediction model includes: Using future cooling load demand and environmental condition changes within a preset time domain as prediction inputs, and combining the performance model of a single unit, a state-space prediction model for a chiller group is established to predict the system operating status under different load distribution schemes.

4. The method of claim 3, wherein, The MPC global optimization objective function takes into account both the global COP maximization and operational stability. The objective function J is equal to the sum of the first and second terms. The first term is the cumulative sum of the predicted global COP of the system, and the second term is the cumulative sum of -λ multiplied by the predicted change in unit load rate.

5. The method of claim 1 or 4, wherein, Step S4 includes: Determine whether start-up, shutdown, or load switching is required. If so, execute the smooth transition logic; otherwise, execute the steady-state control mode. The unit executes control commands, and the operating status is fed back in a closed loop to determine whether the control cycle has ended. If yes, the cycle ends; otherwise, the process returns to step S1.

6. The method of claim 1 or 4, wherein, The scheduling scheme for maximizing COP includes: Iterate through all unit start-up and shutdown combinations to select effective solutions that meet the balance between cooling load supply and demand. For each start-stop combination, the optimal load factor of each unit is calculated to obtain the maximum local COP under that combination; By comparing the local optimal solutions of all effective combinations, the system's globally optimal start / stop state and load distribution scheme are selected.

7. The method of claim 1, wherein the method is a method of optimal dispatching of a chiller plant based on COP maximization. The COP is the ratio of the total cooling capacity to the total input power of the parallel chiller group, where the total cooling capacity is the sum of the cooling capacities of each chiller unit and the total input power is the sum of the input power of each chiller unit.

8. The method of claim 3, wherein the method is a method of optimal dispatching of a chiller plant based on COP maximization. The single-unit performance model includes a nonlinear general mapping model of single-unit cooling capacity and operating parameters, and a nonlinear general mapping model of single-unit input power and operating parameters.

9. The method of claim 5, wherein the method is a method of optimal dispatching of a chiller plant based on COP maximization. The smooth transition logic performs a smooth transition based on the load gradation formula, in which the load rate of the i-th unit at time t is equal to the load rate of the i-th unit in the previous control cycle plus or minus the product of the load gradation slope and the control step size.

10. The method of claim 1, wherein, Repeat steps S1 to S4 for each control step, continuously optimizing the scheduling.

Images