MPC-based large factory intelligent cooling system and cooling control method

Abstract

The invention relates to the field of factory intelligent cooling, and discloses an MPC-based large factory intelligent cooling system and a cooling control method, the system comprises a data acquisition module, an AI control module, a cooling tower control module, a cooling water pump control module and a cooling-water machine control module; the data acquisition module acquires outdoor environment temperature and humidity, cooling tower outlet water temperature, cooling water pump inlet and outlet water temperature difference and flow and water chiller operation state data in real time. The AI control module constructs a system model through real-time data, predicts and optimizes by using an MPC algorithm, and calculates optimal values of the running frequency of a cooling tower fan, the running frequency of a cooling water pump and the power of a cooling-water machine compressor by adopting sequential quadratic programming, so as to realize cooperative dynamic control of equipment in the system; the overall operation efficiency of the cooling system under the outdoor temperature and humidity change condition is effectively improved, and the overall energy consumption of the system is minimized in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of factory intelligent cooling, and in particular to a large-scale factory intelligent cooling system and a cooling control method based on MPC (Model Predictive Control). Background Art

[0002] With rapid socioeconomic development, the demand for cooling in industrial buildings and large-scale projects is rapidly increasing. Industrial cooling systems, as a core component for ensuring the proper operation of industrial production equipment, are becoming increasingly critical infrastructure. Industrial cooling systems typically consist of cooling towers, cooling water pumps, chillers, chilled water pumps, and other equipment. Their overall operating performance directly determines the system's overall energy efficiency and operating costs. In recent years, stricter requirements have been placed on the management of industrial energy consumption, particularly with regard to the energy efficiency of industrial refrigeration systems. Therefore, achieving energy conservation and emission reduction, lowering operating costs, and improving operational efficiency in industrial refrigeration systems has become a critical issue for industrial enterprises.

[0003] At present, some technical solutions have been proposed for energy efficiency management and energy-saving optimization of industrial cooling systems:

[0004] Patent application CN202311412390.9 proposes a non-invasive smart factory energy consumption monitoring method and system. This method establishes an energy consumption model based on real-time data such as power consumption, temperature, and power consumption of the smart factory. Through model fitting and coefficient processing, the energy efficiency level of the smart factory is obtained. This method facilitates load monitoring in industrial scenarios and saves installation and maintenance costs. However, its monitoring and control processes are relatively independent and do not form a complete closed-loop control. It is impossible to optimize and control industrial equipment in real time and dynamically.

[0005] Patent application CN202411018384.X proposes an intelligent refrigeration control system for slaughterhouses. This system uses an information acquisition module to obtain real-time temperature information, refrigeration pipeline information, and cold storage door opening hours. It also monitors and optimizes the compressor motor speed, analyzes freezing efficiency and abnormal conditions, and achieves coordinated control of the compressor speed and cooling water circulation speed, thereby improving refrigeration efficiency. However, this system is mainly targeted at the specific scenarios of slaughterhouses and does not fully consider the coordinated optimization control requirements between external temperature and humidity changes in industrial scenarios and equipment such as cooling towers and cooling water pumps.

[0006] Patent application CN202323001941.0 proposes an energy-saving factory cooling water circulation system that utilizes underground cooling water tanks for centralized cooling water circulation management. This system uses a primary chilled water circulation pump, an air-cooled wind-water unit, and a secondary chilled water supply pump to achieve centralized cooling of factory equipment. This reduces the procurement and maintenance costs of individual chillers and overall cooling energy consumption. However, this technical solution employs a simple physical cooling method and fails to fully consider the refined dynamic optimization control between devices. It also fails to demonstrate the advantages of intelligent response to real-time changes in ambient temperature and humidity and real-time energy efficiency optimization.

[0007] Although the above existing technical solutions have improved the energy saving level of the system to a certain extent, they still have the following technical defects and deficiencies:

[0008] First, most existing technologies use single-device or simple device combination control schemes, without considering the real-time coordinated optimization of the entire system equipment, especially the dynamic real-time optimization of the system under changes in the external environment;

[0009] Second, existing solutions typically lack accurate prediction models for system operating status, making it impossible to effectively optimize energy consumption in a forward-looking manner. This makes it difficult to further reduce energy consumption during actual operation.

[0010] Third, existing technical solutions usually do not have a complete real-time closed-loop control mechanism and fail to dynamically correct model deviations in real time, resulting in large errors between actual operation and optimized predictions, making it difficult to achieve the best overall energy-saving effect.

[0011] Therefore, how to establish an overall collaborative optimization method and system for industrial cooling systems, fully consider the changes in external environmental temperature and humidity, dynamically optimize the equipment operating parameters in real time and accurately, and be able to correct deviations in a closed loop in real time during actual operation, thereby achieving overall energy-saving operation and stable refrigeration of the cooling system, has become a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the Invention

[0012] In response to the shortcomings of the existing technology, the present invention provides a large-scale factory intelligent cooling system and cooling control method based on MPC. By collecting outdoor environmental temperature and humidity data and operating status data of equipment such as cooling towers, cooling water pumps, and chillers in real time, the model predictive control algorithm is used to predict the operating status of the system in real time, and the sequential quadratic programming algorithm is used to accurately solve the optimal setting values ​​of the system operating parameters, dynamically optimize the equipment operating power combination, achieve the lowest overall energy consumption of the system, and ensure the stable operation of the industrial refrigeration system.

[0013] To achieve the above objectives, the present invention is implemented through the following technical solutions: A large-scale factory intelligent cooling system based on MPC, comprising:

[0014] Data acquisition module, used to collect real-time outdoor environment temperature and humidity data and cooling system equipment operating status data;

[0015] An AI control module is in communication with the data acquisition module and is used to build an MPC model based on the collected data and calculate the optimal operating parameters of the cooling system equipment;

[0016] An equipment control module, which is in communication with the AI ​​control module and includes a cooling tower control unit, a cooling water pump control unit, and a chiller control unit, and is configured to receive and execute the optimal operating parameters output by the AI ​​control module;

[0017] Equipment execution units, including cooling tower fans, cooling water pumps, and chiller compressors, are driven by the equipment control module to adjust operating states;

[0018] Among them, the data acquisition module feeds back the actual operating status of the equipment execution unit to the AI ​​control module to form a closed-loop control and realize the dynamic optimization operation of the cooling system.

[0019] Preferably, the data acquisition module includes:

[0020] Outdoor temperature and humidity sensor, used to collect ambient dry-bulb temperature and wet-bulb temperature;

[0021] Temperature sensors are installed at the cooling tower outlet, cooling water pump inlet and outlet, and chiller condenser inlet and outlet to collect water temperature data;

[0022] A flow sensor is provided at the outlet of the cooling water pump to collect cooling water flow data;

[0023] Power sensor, installed on the chiller compressor, used to collect real-time power data;

[0024] The data acquisition unit is used to receive data collected by the outdoor temperature and humidity sensor, temperature sensor, flow sensor, and power sensor, and transmit the data to the AI ​​control module in real time.

[0025] Preferably, when the AI ​​control module performs MPC model predictive control operations on the data, a discrete state space prediction model is constructed, and the state prediction formula is:

[0026] X(k+1)=AX(k)+BU(k)+Ed(k)

[0027] Among them, X(k) is the system state variable vector at the current moment; x(k+1) is the system state variable vector at the next moment; U(k) is the control input vector at the current moment, including the adjustment instructions of the cooling tower fan frequency, cooling water pump frequency and chiller compressor power; d(k) is the interference input vector, which is composed of environmental change data obtained by the outdoor temperature and humidity sensor and represents the changes in the outdoor temperature and humidity environment; A, B, and E are the state matrix, control input matrix, and interference input matrix, respectively, which are obtained through offline or online system identification using historical data including cooling tower outlet water temperature, cooling water pump inlet and outlet temperature difference, chiller inlet and outlet temperature, and outdoor temperature and humidity.

[0028] Preferably, the AI ​​control module establishes an objective function J, and the calculation formula of the objective function is:

[0029]

[0030] Among them, N p is the predicted time domain length; P total is the total system power, which is the sum of the cooling tower fan power, cooling water pump power and chiller compressor power; T set is the target temperature setting value of the chilled water; T chilled (k+i|k) is the predicted chilled water supply temperature; u(k+i|k) is the control input variable at the k+i prediction time; w1, w2, w3 are adjustment coefficients.

[0031] Preferably, the optimization constraints of the objective function include:

[0032] Equipment operating range constraints: The operating frequency of the cooling tower fan, the operating frequency of the cooling water pump, and the operating power of the chiller compressor must all be between their respective predefined minimum and maximum values;

[0033] Temperature safety limit constraint: The cooling tower outlet water temperature must be maintained within the preset safety temperature lower and upper limits;

[0034] Control adjustment rate limit constraint: The real-time adjustment range of the cooling tower fan frequency, cooling water pump frequency and chiller compressor power, and the single-step change amount must not exceed the maximum allowable threshold.

[0035] Preferably, the equipment operating range is limited to:

[0036] f tower,min ≤f tower (k+i|k)≤f tower,max

[0037] f pump,min ≤f pump (k+i|k)≤f pump,max

[0038] P chiller,min ≤P chiller (k+i|k)≤P chiller,max

[0039] Where k represents the current time index; i represents the i-th prediction step from the current time forward; f tower (k+i|k) represents the predicted value of the cooling tower fan operating frequency at time k+i under the conditions observed at time k; f tower,min and f tower,max Respectively represent the minimum and maximum operating frequencies of the cooling tower fan; f pump (k+i|k) represents the predicted value of the cooling water pump operating frequency at time k+i under the conditions observed at time k; f pump,min and f pump,max Respectively represent the lowest and highest frequencies at which the cooling water pump can operate; P chiller (k+i|k) represents the predicted value of the chiller compressor power at the k+ith moment under the conditions observed at the time k; P chiller,min and P chiller,max Respectively represent the minimum and maximum power at which the chiller compressor can operate.

[0040] Preferably, the temperature safety limit constraint is:

[0041] T ct,out,min ≤T ct,out (k+i|k)≤T ct,out,max

[0042] Among them, T ct,out (k+i|k) is the predicted value of the cooling tower outlet water temperature at time k+i under the observation at time k, T ct,out,min and T ct,out,max They respectively represent the lower and upper limits of the safe operating range of the cooling tower outlet water temperature.

[0043] Preferably, the control adjustment rate limit constraint is:

[0044] |Δu(k+i|-k)|≤Δu max

[0045] Where Δu(k+ik)| represents the change value of the control input quantity in the k+i step under the observation condition at time k; Δu max Indicates the maximum absolute value allowed for the rate of change of the controlled variable.

[0046] Preferably, the AI ​​control module uses a sequential quadratic programming algorithm to solve the optimization problem, specifically including:

[0047] Expand the objective function and constraints into a quadratic programming problem using the second-order Taylor expansion;

[0048] The optimal setting values ​​of cooling tower fan frequency, cooling water pump frequency and chiller compressor power are output through iterative solution using the effective set method or interior point method.

[0049] To achieve the above objectives, the present invention further provides a cooling control method using the above system, which comprises the following steps:

[0050] Through multiple types of sensors, real-time data on ambient temperature and humidity, cooling tower outlet water temperature, cooling water pump inlet and outlet temperature difference and flow rate, chiller condenser inlet and outlet water temperature difference, and compressor power are collected and transmitted to the AI ​​control module;

[0051] Filter the collected data and compare it with history to calculate the current status parameters of the system;

[0052] Input real-time data and historical data into the MPC algorithm, predict the future system state based on the prediction model, and generate a combination of equipment control parameters with the minimum total energy consumption and stable chilled water temperature as the optimization goals;

[0053] The sequential quadratic programming algorithm is used to solve the optimization problem and output the optimal setting values ​​of the cooling tower fan frequency, cooling water pump frequency and chiller compressor power;

[0054] The set value is sent to the device control module for execution. After execution is completed, the data acquisition module once again collects the operating data of each execution device in real time, and feeds back the status after execution to the AI ​​control module for real-time calibration and closed-loop optimization of the model.

[0055] The present invention provides an MPC-based intelligent cooling system and cooling control method for large-scale factories, which have the following beneficial effects:

[0056] 1. The present invention adopts an intelligent optimization algorithm based on model predictive control (MPC), which can dynamically predict and optimize the cooling tower fan operating frequency, cooling water pump operating frequency and chiller compressor operating power in real time and accurately according to the real-time data of external environmental temperature and humidity changes and the internal operating status of the system, forming a complete real-time closed-loop optimization control, effectively overcoming the problems of control response lag and insufficient control accuracy in the existing technology.

[0057] 2. The present invention constructs a complete real-time data acquisition and feedback closed-loop system, which can automatically calibrate the model prediction error in real time, effectively improving the accuracy and robustness of the system model prediction, and further improving the stability and operational reliability of the overall cooling system.

[0058] 3. The present invention incorporates multiple refrigeration equipment such as cooling towers, cooling water pumps, and chillers into overall collaborative optimization control. Through a global real-time optimization algorithm, it achieves optimal collaborative control among various devices, significantly reducing the overall energy consumption level of the system. Compared with the existing single device or simple combination control method, the overall energy saving efficiency is significantly improved.

[0059] 4. The present invention adopts the sequential quadratic programming (SQP) algorithm to solve the system optimization model, which can quickly and efficiently calculate the optimal control parameters, ensure that the system can respond to external temperature changes in a timely manner, improve the response speed and control accuracy of the overall system operation, and further enhance the economic benefits and energy-saving effects of the system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a system structure diagram of a large-scale factory intelligent cooling system based on MPC in the present invention;

[0061] Figure 2 This is an operational flow chart of an MPC-based large-scale factory intelligent cooling system of the present invention. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] Example 1:

[0064] See also Figure 1-Figure 2 This invention provides a large-scale intelligent cooling system for industrial plants based on model predictive control (MPC). Through real-time data acquisition, multi-device collaborative optimization, and a closed-loop feedback mechanism, it achieves efficient energy conservation and stable operation of industrial cooling systems. The following describes the system's structure and operating principles in detail, along with specific implementations.

[0065] The system includes data acquisition module, AI control module, equipment control module and equipment execution unit. Each module is connected through a communication interface (such as RS485, Modbus TCP or OPC UA protocol) to form a closed-loop control circuit. The specific architecture is as follows Figure 1 shown.

[0066] In this embodiment, the data acquisition module is the core component of the perception layer of the intelligent cooling system, which is used to collect the operating status parameters and external environmental parameters of each key node of the cooling system in real time, and provide multi-dimensional input data for model predictive control.

[0067] The data acquisition module realizes the data acquisition function through a distributed sensor network, receives the data collected by each sensor through the data acquisition unit, and transmits it to the AI ​​control module in real time.

[0068] The sensor types, installation locations, and state variables collected by the data acquisition module include:

[0069] Outdoor temperature and humidity sensor: installed at the air inlet of the cooling tower to monitor the ambient dry bulb temperature T in real time ambient,dry (k) and wet bulb temperature T ambient,wet (k), where k represents the discretized time index. The dry-bulb and wet-bulb temperature data are used to quantify the dynamic impact of the environment on the cooling tower's heat dissipation efficiency. Its correlation with the cooling tower's outlet water temperature is embedded in the subsequent MPC prediction algorithm via a heat transfer model.

[0070] Cooling tower outlet temperature sensor: installed in the cooling tower outlet pipe, using a contact temperature sensor (preferably PT100 platinum resistance temperature sensor) to collect the cooling tower outlet water temperature T in real time ct,out (k), which directly reflects the heat dissipation performance of the cooling tower and is input into the state space equation of the MPC model as a state variable.

[0071] Cooling water pump inlet and outlet temperature difference sensor: Install high-precision temperature sensors on the inlet and outlet pipes of the cooling water pump respectively, and obtain the inlet and outlet temperature difference ΔT of the cooling water pump in real time through difference calculation pump (k) = T pump,in (k)-T pump,out (k), used to evaluate the heat exchange efficiency of the water pump during operation and the effect of cooling water flow on the temperature difference.

[0072] Cooling water flow sensor: Installed in the cooling water pump outlet pipe, using electromagnetic flowmeter or ultrasonic flowmeter to collect cooling water flow Q in real time pump (k), the flow data is used to calculate the pump power consumption and perform closed-loop verification with the frequency regulation command.

[0073] Chiller condenser temperature difference sensor: Install temperature sensors on the inlet and outlet pipes of the chiller condenser to collect the real-time inlet and outlet water temperature difference ΔT on the condenser side. con (k) = T con,in (k)-T con,out (k), this parameter is used to evaluate the cooling efficiency of the chiller and the heat load status of the condenser.

[0074] Chiller compressor power sensor: integrated in the compressor power distribution cabinet, real-time measurement of compressor input power P through current transformer and voltage sensor chiller , and based on the electric power formula P chiller= V(k) × I(k) × cosφ to calculate the actual power consumption, where V(k) and I(k) are the real-time voltage and current, respectively, and cosφ is the power factor.

[0075] The data acquisition unit of the data acquisition module transmits sensor data to the AI ​​control module in real time through the industrial communication protocol. During the transmission process, the following preprocessing strategies are used to ensure data validity:

[0076] Data synchronization mechanism: For sensor data with different sampling frequencies (for example, ambient temperature and humidity are collected every 5 minutes, and equipment operating parameters are collected every 1 minute), timestamp alignment technology is used to ensure that the data input into the MPC model at the same time has temporal consistency.

[0077] Abnormal data filtering: The original data is smoothed based on the sliding window mean filtering algorithm to eliminate instantaneous noise interference; at the same time, a threshold range is set (for example, the reasonable range of cooling tower outlet water temperature is 25℃–40℃), and data exceeding the threshold is triggered by abnormal alarm and automatically eliminated.

[0078] Data normalization: The original data of different dimensions (such as temperature in °C, flow in m 3 / h, the power unit is kW) is normalized to the range of [0,1] to improve the computational stability and convergence speed of the MPC model.

[0079] The state variables output by the data acquisition module constitute the core input of the state space equation of the MPC model. The specific relationship is as follows:

[0080] Cooling tower outlet water temperature T ct,out (k) and cooling tower fan frequency f tower (k) together constitute the control variables of the cooling tower's heat dissipation capacity, and their dynamic relationship is expressed through the differential equation Description, where α and β are the heat transfer coefficient and fan efficiency coefficient.

[0081] Cooling water pump inlet and outlet temperature difference ΔT pump (k) and flow Q pump The product of (k) is used to calculate the heat transfer power of the water pump, that is, P pump,heat (k) = c p ρQ pump (k)ΔT pump (k), where c p is the specific heat capacity of water, ρ is the density of water, and this parameter and the electric power consumption of the water pump together constitute the energy efficiency evaluation index of the water pump.

[0082] Chiller condenser temperature difference ΔT con (k) and compressor power P chillerThe ratio of (k) is used to characterize the instantaneous energy efficiency ratio (COP) of the chiller, that is, where Q evap (k) is the cooling capacity on the evaporator side, which is indirectly calculated through the condenser heat balance equation.

[0083] After the device execution unit completes the control command, the data acquisition module collects updated operating status data in real time (such as the cooling tower outlet water temperature after adjusting the fan frequency or the flow rate change after adjusting the pump frequency). This data is fed back to the AI ​​control module through the communication interface, forming a closed-loop control loop. This feedback data is used to dynamically calibrate the parameters of the MPC prediction model to eliminate the cumulative error between the model's predicted values ​​and the actual operating status.

[0084] In this embodiment, the AI ​​control module is the core decision-making unit of the intelligent cooling system. It realizes multi-device collaborative optimization based on the model predictive control (MPC) framework and generates device control instructions through dynamic modeling, rolling optimization and closed-loop feedback mechanism.

[0085] The AI ​​control module describes the system dynamic characteristics through a discrete state space model, and its state prediction equation is defined as:

[0086] X(k+1)=AX(k)+BU(k)+Ed(k)

[0087] Where:

[0088] X(k) represents the system state vector at the current moment, specifically including: cooling tower outlet water temperature T ct,out (k) Temperature difference between the inlet and outlet of the cooling water pump T pump (k) The temperature difference between the inlet and outlet water of the chiller condenser ΔT con (k) Chilled water supply temperature T chilled (k);

[0089] U(k) represents the control input vector, which is composed of the cooling tower fan frequency f tower (k), cooling water pump frequency f pumf (k), chiller compressor power P chilled (k) composition;

[0090] d(k) represents the external interference input vector, and the ambient dry-bulb temperature t provided by the data acquisition module ambient,dry (k) and wet bulb temperature T ambient,wet (k) composition, representing the changes in outdoor temperature and humidity environment;

[0091] A, B, and E are the state transfer matrix, control input matrix, and disturbance input matrix, respectively. They are obtained through offline identification of historical operating data and are dynamically updated online using the recursive least squares method (preferred) to eliminate model mismatch caused by equipment aging or environmental mutations.

[0092] The AI ​​control module takes the lowest total system energy consumption and the stable chilled water temperature as the optimization goals and constructs the following multi-objective optimization function:

[0093]

[0094] Where:

[0095] N p Indicates the length of the prediction horizon, which is used to define the number of future optimization steps;

[0096] P total (k+i|k)=P tower (k+i|k)+P pump (k+i|k)+P chiller (k+i|k) is the predicted value of the total system power, where the power of each device is associated with the control variable through an empirical formula;

[0097] T set Indicates the target temperature setting value of the chilled water, which is determined by user requirements or process conditions;

[0098] T chilled (k+i|k) is the predicted value of the chilled water supply temperature, which is obtained by recursive calculation of the state space model;

[0099] ΔU(k+i|k)=U(k+i|k)-U(k+i-1|k) represents the change in the control input;

[0100] w1, w2, and w3 are weight coefficients used to adjust the priority of energy consumption, temperature deviation, and control quantity change (preferably, the weight values ​​are determined by trial and error or Pareto front analysis).

[0101] The AI ​​control module imposes the following constraints during the optimization process to ensure system operation safety and device physical limitations:

[0102] 1. Equipment operating range constraints:

[0103] f tower,min ≤f tower (k+i|k)≤f tower,max

[0104] f pump,min ≤f pump (k+i|k)≤f pump,max

[0105] P chiller,min ≤P chiller (k+i|k)≤P chiller,max

[0106] in:

[0107] k represents the current moment index;

[0108] i represents the i-th prediction step from the current moment forward;

[0109] f tower (k+i|k) represents the predicted value of the cooling tower fan operating frequency at time k+i under the conditions observed at time k;

[0110] f tower,min and f tower,max Respectively represent the minimum and maximum frequencies at which the cooling tower fan can operate;

[0111] f pump (k+i|k) represents the predicted value of the cooling water pump operating frequency at time k+i under the conditions observed at time k;

[0112] f pump,min and f pump,max Respectively represent the minimum and maximum frequencies at which the cooling water pump can operate;

[0113] P chiller (k+i|k) represents the predicted value of the chiller compressor power at time k+i under the conditions observed at time k;

[0114] P chiller,min and P chiller,max Respectively represent the minimum and maximum power at which the chiller compressor can operate.

[0115] 2. Temperature safety constraints:

[0116] T ct,out,min ≤T ct,out (k+i|k)≤T ct,out,max

[0117] in:

[0118] T ct,out (k+i|k) is the predicted value of the cooling tower outlet water temperature at time k+i under the observation at time k;

[0119] T ct,out,min and T ct,out,max They respectively represent the lower and upper limits of the safe operating range of the cooling tower outlet water temperature.

[0120] 3. Control variable change rate constraint:

[0121] |Δu(k+i|-k)|≤Δu max

[0122] in:

[0123] Δu(k+ik)| represents the change in the control input at step k+i under the observation condition at time k;

[0124] Δu max Indicates the maximum absolute value allowed for the rate of change of the controlled variable.

[0125] The AI ​​control module uses a sequential quadratic programming algorithm to solve constrained nonlinear optimization problems. The specific steps are as follows:

[0126] 1. Based on the real-time acquired data and state variables, construct the objective function and constraints, and determine the variables to be optimized.

[0127] 2. Perform a second-order Taylor expansion of the objective function and constraints around the current control variable value to form a standard quadratic programming problem, which is calculated as follows:

[0128]

[0129]

[0130] Among them, ΔU is the correction amount of the control variable in the current iteration step; is the gradient vector of the objective function, which indicates the sensitivity of the objective function to changes in the control variables; is the Hessian matrix of the objective function, reflecting the curvature of the objective function, g(U)>0 represents the inequality constraint of the system, and h(U) represents the equality constraint of the system;

[0131] 3. Use the active set method or interior point method to solve the above quadratic programming problem and obtain the correction value ΔU of the control variable;

[0132] 4. Use the correction value ΔU to update the control variables and determine the convergence conditions. When the convergence conditions are met, output the final optimized set values ​​of the cooling tower fan operating frequency, cooling water pump operating frequency, and chiller compressor power.

[0133] After each control instruction is executed, the AI ​​control module receives the actual operating status data fed back by the data acquisition module (such as the adjusted cooling tower outlet water temperature T ct,out (k+1)), dynamic calibration of model parameters is achieved through the following mechanism:

[0134] Prediction error calculation: Compare the model predicted value X(k+1|k) with the actual measured value X(k+1), and calculate the residual vector e(k+1)=X(k+1)-X(k+1|k).

[0135] Parameter online update: The state matrix A, control matrix B, and interference matrix E are updated based on the recursive least squares method. The specific formula is:

[0136] θ(k+1)=θ(k)+K(k+1)[e(k+1)-φ T (k+1)θ(k)]

[0137] Where θ = [A, B, E] T is the parameter matrix to be identified, φ(k+1) is the regression matrix, and K(k+1) is the Kalman gain matrix.

[0138] The AI ​​control module integrates fault detection logic. When it identifies a device anomaly (such as a pump flow sensor failure), it performs the following actions:

[0139] Switch to backup sensor data or continue optimization calculations based on model estimates;

[0140] If a device fails completely (e.g., a compressor shuts down), the load is redistributed to redundant devices and the MPC model constraints are updated to ensure system degradation and operational stability.

[0141] In this embodiment, the equipment control module is the core component of the execution layer of the intelligent cooling system. It is responsible for converting the optimization control instructions output by the AI ​​control module into equipment executable signals, and driving the cooling tower fan, cooling water pump and chiller compressor to operate according to the set parameters.

[0142] The equipment control module consists of a cooling tower control unit, a cooling water pump control unit, and a chiller control unit. Each unit is connected to the AI ​​control module through an industrial communication interface, receiving control instructions and providing feedback on execution status in real time:

[0143] Cooling tower control unit: receives the fan frequency setting value issued by the AI ​​control module The fan motor is driven by the frequency converter to adjust the fan speed to control the air flow of the cooling tower. (where p is the number of motor pole pairs) generates a PWM modulation signal to achieve precise speed control.

[0144] Cooling water pump control unit: Water pump frequency setting value based on the output of the AI ​​control module The frequency converter is used to adjust the speed of the water pump motor and dynamically adjust the cooling water flow. The relationship between flow and frequency is shown in the figure below. pump (k) = k Q ·f pump (k) Determine (k Q is the pump characteristic coefficient), ensuring that the flow response matches the frequency instruction linearly.

[0145] Chiller control unit: according to the compressor power setting value The compressor input current is adjusted by the PID controller and the power regulation module (preferably, IGBT inverter) to achieve precise power control. The power regulation model is Pchiller (k)=V dc (k) I comp (k)·η, where V dc is the DC bus voltage, I comp is the compressor current, and η is the inverter efficiency coefficient.

[0146] When the device control module executes AI control instructions, it uses the following technical implementation methods:

[0147] Frequency command smoothing: To avoid mechanical stress on the equipment due to frequency mutation, the received frequency setting value Perform ramp function filtering to limit the single-step frequency change rate Δf max , the specific formula is:

[0148] f actual (k) = f actual (k-1)+sat(f target (k)-f actual (k-1),Δf max )

[0149] Among them, sat(X,Δ) is a saturation function that limits the input change to the range of [-Δ,Δ].

[0150] Power closed-loop control: The chiller control unit adjusts the compressor input voltage in real time through the current closed-loop feedback mechanism to ensure the actual power P chiller,actual (k) Tracking set value Power deviation After calculation by the PID controller, the PWM duty cycle adjustment signal is output. The formula is:

[0151]

[0152] Among them, K p , K i , K d are the proportional, integral and differential coefficients, and Δt is the control period.

[0153] After executing the control instructions, the device control module implements execution effect verification and closed-loop calibration through the following mechanisms:

[0154] Execution status acquisition: real-time reading of the actual operating parameters of the equipment, such as fan speed n tower,actual (k), water pump flow Q pump,actual (k), compressor current I comp,actual (k) and fed back to the data acquisition module through the communication interface.

[0155] Instruction tracking error calculation: Compare the deviation between the set value and the actual value, such as If the deviation continues to exceed the allowable range, the control parameter self-tuning process is triggered to update the PID controller coefficients or the inverter response curve.

[0156] In this embodiment, the device execution unit is the physical actuator of the intelligent cooling system, which directly responds to the optimization instructions issued by the device control module and realizes dynamic optimization operation of the cooling system by adjusting the cooling tower fan speed, cooling water pump flow and chiller compressor power.

[0157] The equipment execution unit consists of a cooling tower variable frequency fan, a cooling water pump variable frequency motor, and a chiller compressor. Each device implements the physical execution of control instructions through electrical connections and mechanical transmission mechanisms:

[0158] Cooling tower variable frequency fan: A three-phase asynchronous motor is used to drive the axial flow fan blades. Its speed is linearly related to the input frequency. The PWM signal output by the inverter adjusts the motor input frequency, thereby controlling the fan speed to meet the heat dissipation requirements.

[0159] Cooling water pump variable frequency motor: The motor input frequency is adjusted through the inverter to change the speed of the water pump impeller, thereby controlling the cooling water flow.

[0160] Chiller compressor: Using variable frequency drive or magnetic bearing technology, the refrigerant circulation volume is controlled by adjusting the compressor input power, thereby adjusting the heat load of the evaporator and condenser.

[0161] like Figure 2 As shown, the present invention also provides a cooling control method using the above system, which achieves efficient and stable operation of the cooling system through data-driven modeling, dynamic optimization and real-time feedback. The specific implementation process is described in detail below:

[0162] Step S1: Multi-dimensional data collection and transmission

[0163] In this step, the system collects cooling system operating parameters and environmental data in real time through a distributed sensor network, including:

[0164] Ambient temperature and humidity data: Temperature and humidity sensors installed at the cooling tower air inlet provide real-time access to outdoor dry-bulb and wet-bulb temperatures, which are used to quantify the dynamic impact of the environment on the cooling tower's heat dissipation efficiency.

[0165] Equipment operating parameters:

[0166] Cooling tower outlet water temperature: Use high-precision temperature sensor to monitor the cooling tower outlet water temperature to reflect the heat dissipation performance;

[0167] Cooling water pump inlet and outlet temperature difference and flow rate: Use a temperature difference sensor and an electromagnetic flowmeter to measure the pump inlet and outlet temperature difference and cooling water flow rate to evaluate the pump's heat exchange efficiency;

[0168] Chiller condenser inlet and outlet water temperature difference: Use temperature difference sensor to monitor the heat load status on the condenser side;

[0169] Compressor power data: The power sensor collects the compressor input power in real time for energy consumption calculation.

[0170] Data is transmitted to the AI ​​control module via industrial communication protocols (such as Modbus RTU or OPC UA). During the transmission process, sliding window mean filtering and outlier removal are performed to ensure data validity.

[0171] Step S2: Data preprocessing and state assessment

[0172] In this step, the AI ​​control module performs the following processing on the received real-time data:

[0173] Data alignment and synchronization: For sensor data with different sampling frequencies (such as environmental data updated every 5 minutes and device data updated every 1 minute), timestamp alignment technology is used to ensure that the data input into the optimization model at the same time has temporal consistency.

[0174] Historical data comparison: Compare current data with historical operating data from the previous period (e.g., the past 24 hours) to identify abnormal operating conditions (such as a sudden drop in pump flow or abnormal fluctuation in compressor power) and trigger an alarm.

[0175] State parameter calculation: Based on indicators such as the cooling tower outlet water temperature change rate, the water pump flow-temperature difference product, and the compressor power-cooling capacity relationship, the system's current energy efficiency status and heat load level are comprehensively calculated.

[0176] Step S3: Model prediction and multi-objective optimization

[0177] In this step, the AI ​​control module generates device control instructions based on the model predictive control (MPC) framework. The specific implementation methods include:

[0178] Prediction model construction: Utilize current system status data and historical operation data to construct a discrete-time state-space model to predict the system status (such as cooling tower outlet water temperature and chilled water temperature) for multiple time steps in the future, and introduce ambient temperature and humidity as interference variables to improve prediction accuracy.

[0179] Multi-objective optimization decision-making: With the dual optimization objectives of minimizing total system energy consumption and maintaining stable chilled water temperature, this system continuously optimizes the combination of cooling tower fan frequency, cooling water pump frequency, and chiller compressor power over the forecast horizon. During the optimization process, the priority of energy consumption and temperature control is dynamically adjusted using weight coefficients, taking into account physical limitations of the equipment (such as frequency range and power limit) and constraints on the rate of change of the controlled variable.

[0180] Step S4: Optimization solution and closed-loop feedback

[0181] In this step, the system realizes the generation of control instructions and closed-loop calibration through the following mechanisms:

[0182] Sequential Quadratic Programming (SQP) solution: This nonlinear optimization problem is transformed into quadratic programming subproblems. The optimal settings for the cooling tower fan frequency, cooling water pump frequency, and chiller compressor power are determined through iterative calculations. During the solution process, the active set method is used to address inequality constraints (such as upper and lower limits on equipment frequency) to ensure the optimality and feasibility of the solution.

[0183] Instruction issuance and execution: The optimized set value is sent to the actuator through the equipment control module, driving the cooling tower fan to adjust the speed, the cooling water pump to adjust the flow, and the chiller compressor to dynamically match the power demand.

[0184] State feedback and model calibration: After the instruction is executed, the data acquisition module obtains the actual operating data of the equipment in real time (such as the adjusted cooling tower outlet water temperature and water pump flow), and feeds it back to the AI ​​control module. The prediction model parameters are updated online through the recursive least squares method to eliminate the cumulative error between the model prediction value and the actual state, forming a closed-loop optimization control loop.

[0185] In general, through the real-time dynamic optimization closed-loop operation of the above system, accurate real-time optimization control of the industrial cooling system under the constantly changing external environment can be achieved, fully achieving the technical goals of reducing the overall energy consumption of the system and ensuring the stable cooling effect of the system. It can be widely used in industrial production, building cooling, data centers and other fields.

[0186] Example 2:

[0187] This implementation uses an industrial cooling system as an example. This cooling system in this embodiment is equipped with a cooling tower, cooling water pump, chiller, and other equipment. Before implementation, the system used traditional fixed-frequency control (without intelligent optimization control). After implementation, the system uses the MPC-based intelligent optimization control system proposed in this invention for dynamic optimization control. The following table compares the actual operating data of the system under typical summer operating conditions, as the outdoor ambient temperature changes:

[0188] Table 1 Operating parameters and power consumption of each device before implementation (traditional control method)

[0189]

[0190] The data in Table 1 show that as the outdoor temperature gradually rises, the power of the system's cooling tower fan, cooling water pump, and chiller compressor continues to increase, resulting in a significant increase in the system's overall operating energy consumption. The comprehensive energy efficiency deteriorates from 0.610kW / ton to 0.683kW / ton, and the energy efficiency decreases significantly.

[0191] Table 2 Operating parameters and power consumption of MPC intelligent optimization control mode

[0192]

[0193] As shown in Table 2, after the implementation of the present invention, under the same conditions of increased outdoor temperature and humidity, the power of the cooling tower fan, cooling water pump and chiller are all finely dynamically controlled through the MPC intelligent optimization control system provided by the present invention: the power of the cooling tower fan is reduced from 44kW to 36kW when the wet-bulb temperature is 28°C; the power of the cooling water pump is reduced from 48kW to 45kW when the wet-bulb temperature is 28°C; the power of the chiller compressor is reduced from 260kW to 237kW when the wet-bulb temperature is 28°C;

[0194] More importantly, the precise control increases the system's chilled water temperature difference to approximately 6°C, correspondingly improving the chiller's COP and significantly increasing the system's overall operating efficiency. As shown in Table 3, the system's overall efficiency improved significantly after implementation, rising from 0.683 kW / ton to 0.580 kW / ton, achieving an overall energy saving rate of approximately 15%.

[0195] Table 3 Energy saving analysis and result comparison after implementing the intelligent optimization control system of the present invention

[0196]

[0197] The intelligent optimization control system provided by the present invention collects environmental temperature and humidity data and the operating parameters of each device in real time, applies the MPC prediction model and SQP optimization solution, and finely adjusts the equipment operating power combination, thereby significantly improving the energy efficiency of the overall refrigeration system. The present invention not only ensures the stability and reliability of the system's cooling effect, but also significantly reduces the system's overall energy consumption and operating costs. The actual operating data in the above implementation case shows that the technical solution of the present invention has achieved significant economic and social benefits in the actual application of industrial cooling systems, and has the value of widespread promotion and application.

[0198] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A large-scale factory intelligent cooling system based on MPC, characterized by: include: Data acquisition module, used to collect real-time outdoor environment temperature and humidity data and cooling system equipment operating status data; An AI control module is in communication with the data acquisition module and is used to build an MPC model based on the collected data and calculate the optimal operating parameters of the cooling system equipment; An equipment control module, which is in communication with the AI ​​control module and includes a cooling tower control unit, a cooling water pump control unit, and a chiller control unit, and is configured to receive and execute the optimal operating parameters output by the AI ​​control module; Equipment execution units, including cooling tower fans, cooling water pumps, and chiller compressors, are driven by the equipment control module to adjust operating states; Among them, the data acquisition module feeds back the actual operating status of the equipment execution unit to the AI ​​control module to form a closed-loop control and realize the dynamic optimization operation of the cooling system.

2. The MPC-based large-scale factory intelligent cooling system according to claim 1 is characterized in that: The data acquisition module includes: Outdoor temperature and humidity sensor, used to collect ambient dry-bulb temperature and wet-bulb temperature; Temperature sensors are installed at the cooling tower outlet, cooling water pump inlet and outlet, and chiller condenser inlet and outlet to collect water temperature data; A flow sensor is provided at the outlet of the cooling water pump to collect cooling water flow data; Power sensor, installed on the chiller compressor, used to collect real-time power data; The data acquisition unit is used to receive data collected by the outdoor temperature and humidity sensor, temperature sensor, flow sensor, and power sensor, and transmit the data to the AI ​​control module in real time.

3. The large-scale factory intelligent cooling system based on MPC according to claim 1 is characterized in that: When the AI ​​control module performs MPC model predictive control operations on the data, a discrete state space prediction model is constructed. The state prediction formula is: X(k+1)=AX(k)+BU(k)+Ed(k) Among them, X(k) is the system state variable vector at the current moment; X(k+1) is the system state variable vector at the next moment; U(k) is the control input vector at the current moment, including the adjustment instructions for the cooling tower fan frequency, cooling water pump frequency and chiller compressor power; d(k) is the interference input vector, which is composed of environmental change data obtained by the outdoor temperature and humidity sensor and represents the changes in the outdoor temperature and humidity environment; A, B, and E are the state matrix, control input matrix, and interference input matrix, respectively, which are obtained through offline or online system identification using historical data including cooling tower outlet water temperature, cooling water pump inlet and outlet temperature difference, chiller inlet and outlet temperature, and outdoor temperature and humidity.

4. The MPC-based large-scale factory intelligent cooling system according to claim 3 is characterized in that: The AI ​​control module establishes the objective function J, and the calculation formula of the objective function is: Among them, N p is the predicted time domain length; P total is the total system power, which is the sum of the cooling tower fan power, cooling water pump power and chiller compressor power; T set is the target temperature setting value of the chilled water; T chilled (k+i|k) is the predicted chilled water supply temperature; u(k+i|k) is the control input variable at the k+i prediction time; w1, w2, w3 are adjustment coefficients.

5. The MPC-based large-scale factory intelligent cooling system according to claim 4 is characterized in that: The optimization constraints of the objective function include: Equipment operating range constraints: The operating frequency of the cooling tower fan, the operating frequency of the cooling water pump, and the operating power of the chiller compressor must all be between their respective predefined minimum and maximum values; Temperature safety limit constraint: The cooling tower outlet water temperature must be maintained within the preset safety temperature lower and upper limits; Control adjustment rate limit constraint: The real-time adjustment range of the cooling tower fan frequency, cooling water pump frequency and chiller compressor power, and the single-step change amount must not exceed the maximum allowable threshold.

6. The MPC-based large-scale factory intelligent cooling system according to claim 5 is characterized in that: The equipment operating range is limited to: f tower,min ≤f tower (k+i|k)≤f tower,max f pump,min ≤f pump (k+i|k)≤f pump,max P chiller,min ≤P chiller (k+i|k)≤P chiller,max Where k represents the current time index; i represents the i-th prediction step from the current time forward; f tower (k+i|k) represents the predicted value of the cooling tower fan operating frequency at time k+i under the conditions observed at time k; f tower,min and f tower,max Respectively represent the minimum and maximum operating frequencies of the cooling tower fan; f pump (k+i|k) represents the predicted value of the cooling water pump operating frequency at time k+i under the conditions observed at time k; f pump,min and f pump,max Respectively represent the lowest and highest frequencies at which the cooling water pump can operate; P chiller (k+i|k) represents the predicted value of the chiller compressor power at the k+ith moment under the conditions observed at the time k; P chiller,min and P chiller,max Respectively represent the minimum and maximum power at which the chiller compressor can operate.

7. The MPC-based large-scale factory intelligent cooling system according to claim 5, characterized in that: The temperature safety limit constraint is: T ct,out,min ≤T ct,out (k+i|k)≤T ct,out,max Among them, T ct,out (k+i|k) is the predicted value of the cooling tower outlet water temperature at time k+i under the observation at time k; T ct,out,min and T ct,out,max They respectively represent the lower and upper limits of the safe operating range of the cooling tower outlet water temperature.

8. The MPC-based large-scale factory intelligent cooling system according to claim 5, characterized in that: The control adjustment rate limit constraint is: |Δu(k+i|-k)|≤Δu max Where Δu(k+ik)| represents the change value of the control input quantity in the k+i step under the observation condition at time k; Δu max Indicates the maximum absolute value allowed for the rate of change of the controlled variable.

9. The MPC-based large-scale factory intelligent cooling system according to claim 4, characterized in that: The AI ​​control module uses a sequential quadratic programming algorithm to solve the optimization problem, specifically including: Expand the objective function and constraints into a quadratic programming problem using the second-order Taylor expansion; The optimal setting values ​​of cooling tower fan frequency, cooling water pump frequency and chiller compressor power are output through iterative solution using the effective set method or interior point method.

10. A cooling control method for a large-scale factory intelligent cooling system based on MPC according to claim 1, characterized in that: The following steps are involved: The data acquisition module uses multiple sensors to collect data on ambient temperature and humidity, cooling tower outlet water temperature, cooling water pump inlet and outlet temperature difference and flow rate, chiller condenser inlet and outlet water temperature difference, and compressor power in real time, and transmits this data to the AI ​​control module. Filter the collected data and compare it with history to calculate the current status parameters of the system; Input real-time data and historical data into the MPC algorithm, predict the future system state based on the prediction model, and generate a combination of equipment control parameters with the minimum total energy consumption and stable chilled water temperature as the optimization goals; The sequential quadratic programming algorithm is used to solve the optimization problem and output the optimal setting values ​​of the cooling tower fan frequency, cooling water pump frequency and chiller compressor power; The set value is sent to the device control module for execution. After execution is completed, the data acquisition module once again collects the operating data of each execution device in real time, and feeds back the status after execution to the AI ​​control module for real-time calibration and closed-loop optimization of the model.

Images