Optimal control method for energy consumption of water-cooled air conditioner

Abstract

The invention discloses an energy consumption optimization control method for a water-cooled air conditioner. According to the method, key factors such as air-conditioning box air inlet temperature, chilled water inlet and outlet water temperature, cooling water inlet and outlet water temperature, chilled water outlet flow and refrigerating machine power are comprehensively considered, and an XGBoost machine learning model is introduced to establish a refrigerating machine power prediction model and a chilled water outlet water temperature prediction model. By adjusting the freezing water outlet temperature, the power of the refrigerator is dynamically optimized, the energy efficiency ratio of the water-cooled air conditioner is maximized, and relative minimization of the energy consumption of the water-cooled air conditioner is achieved. Dynamic control over the outlet water temperature of the refrigerating machine in the water-cooled air conditioning system is achieved, the power of the refrigerating machine can be optimized according to real-time and predicted data, and therefore low-energy-consumption control over the temperature of a factory workshop is achieved, system energy consumption is remarkably reduced, and the service life of equipment is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation and energy-saving control, and in particular to a method for optimizing the energy consumption of water-cooled air conditioners in factory production workshops by dynamically controlling the operation of water-cooled air conditioners using artificial intelligence and machine learning technologies. Background Art

[0002] With the continuous advancement of industrialization, whether the temperature control in factory workshops meets the standards has become one of the key factors affecting production efficiency and product quality. For factory workshops that require low-temperature operation, temperature control is usually achieved through water-cooled air conditioners, which achieve cooling effects through heat exchange of water.

[0003] like Figure 1 As shown in the figure, the water circulation system of the water-cooled air conditioner includes a cooling pump, a cooling tower, a refrigerator, a refrigeration pump and an air conditioning box. The refrigerator lowers the temperature of the chilled water in the pipeline by lowering the cooling water inlet temperature, and transports the cooled chilled water to the air conditioning box coil. The air in the air conditioning box (external air intake) exchanges heat with the chilled water through the coil, and the chilled water that takes away the heat is transported to the refrigerator again through the refrigeration pump.

[0004] The heat in the chilled water is transferred to the cooling water inside the freezer, and then the cooling water discharges the heat to the outside atmosphere through the cooling tower. The cooled water returns to the freezer again to complete the cycle.

[0005] Traditional water-cooled air conditioner control methods mainly rely on simple switch operation, proportional-integral-differential (PID) algorithm, valve adjustment, timed operation and other control methods. These methods generally lack intelligence and flexibility, and cannot achieve dynamic setting and precise adjustment of water temperature and water flow according to the actual working conditions inside the workshop, resulting in high energy consumption, inaccurate adjustment and insufficient adaptability. Therefore, a new water-cooled air conditioner control method is needed that can dynamically and efficiently adjust the water temperature while taking into account energy saving. Summary of the invention

[0006] The present invention proposes a water-cooled air conditioner energy consumption optimization control method based on machine learning, which comprehensively considers key factors such as outdoor air temperature, chilled water inlet and outlet water temperature, cooling water inlet and outlet water temperature, chilled water outlet flow rate and chiller power, and introduces the XGBoost machine learning model to establish a chiller power prediction model and a chilled water outlet temperature prediction model. By adjusting the chilled water outlet temperature, the chiller power is dynamically optimized, the energy efficiency ratio of the water-cooled air conditioner is maximized, and the energy consumption is relatively minimized. It is particularly suitable for controlling the optimal energy consumption of water-cooled air conditioners in industrial production workshops that require low-temperature operation.

[0007] The present invention specifically comprises the following steps:

[0008] Step 1: Sensor installation and data collection:

[0009] Temperature sensors are installed at the chilled water inlet and outlet of the freezer and the cooling water inlet and outlet, and a flow sensor is installed at the chilled water outlet to measure the water temperature and flow at time t in real time; temperature sensors are installed at the air inlet of the air conditioner and inside the factory workshop to record the air inlet temperature at time t in real time. and the temperature inside the workshop All the above sensors are connected to the industrial gateway.

[0010] The refrigerator power P at time t t , Cooling pump power P_CW t , refrigeration pump power P_CH t and cooling tower power P_CT t Send to the industrial gateway via MODBUS RTU protocol;

[0011] The industrial gateway uses the MQTT protocol to transmit data to the cloud server in real time.

[0012] The cloud server receives key data sent by the industrial gateway, including the chilled water outlet temperature of the freezer Chilled water inlet temperature of the refrigerator Cooling water outlet temperature of refrigerator Cooling water inlet temperature of refrigerator Workshop temperature Air conditioning box inlet temperature Refrigerator power P t , Chilled water outlet flow F t , refrigeration pump power P_CH t , Cooling pump power P_CW t , Cooling tower power P_CT t ;

[0013] Calculate the system energy efficiency ratio Scop based on the above data t , and its calculation formula is as follows:

[0014]

[0015] Among them, Q t is the instantaneous cooling capacity at time t, C p is the specific heat capacity of chilled water (1.163kw / m3 / h·℃).

[0016] Step 2: The cloud server obtains the temperature at time t+1 through the standard API interface of the National Meteorological Administration

[0017] Step 3: Set the time interval for the cloud server to receive key data sent by the industrial gateway and the target temperature T of the factory workshoptarget , power reduction of the refrigerator ΔP%, temperature difference threshold ΔT in the startup state fac-th , the minimum temperature difference threshold ΔT of the equilibrium heating state air-th and temperature fluctuation threshold ΔT target .

[0018] Step 4: Use the data collected by the sensors in step 1 and use the XGBoost machine learning model to build a freezer power prediction model and a chilled water outlet temperature prediction model:

[0019] Chiller power prediction model: The chilled water outlet temperature As input parameters, chilled water inlet temperature Cooling water outlet temperature Cooling water inlet temperature Workshop temperature Air conditioning box inlet temperature Target temperature T of the factory floor target , Chilled water outlet flow F t , system energy efficiency ratio Scop t As characteristic data, the refrigerator power P t as a predicted value.

[0020] Chilled water outlet temperature prediction model: The chiller power P t As an input parameter, the chilled water inlet temperature Cooling water outlet temperature Cooling water inlet temperature Workshop temperature Air conditioning box inlet temperature Target temperature T of the factory floor target , Chilled water outlet flow F t , system energy efficiency ratio Scop t As characteristic data, the chilled water outlet temperature as a predicted value.

[0021] The cloud server uses the XGBoost machine learning model and combines it with hyperparameter tuning to train the model. After the model training is completed:

[0022] Input any chilled water outlet temperature into the chiller power prediction model, and the corresponding chiller power will be output;

[0023] Input any chiller power into the chilled water outlet temperature prediction model, and the corresponding chilled water outlet temperature will be output.

[0024] Step 5: System operation status judgment and control:

[0025] Start-up state: When the workshop temperature With the target temperature Ttarget There are large differences between The system enters the startup state. At this time, the cloud server sets the chilled water outlet temperature to the minimum value CH_TP min The control command is sent to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol CH_TP min , to quickly lower the indoor temperature.

[0026] As a preferred option, the minimum chilled water outlet temperature CH_TP min The value range is [T target ×30%,T target ×60%], default CH_TP min =T target ×40%.

[0027] Stable fluctuation state: When the workshop temperature is close to the target temperature, When the system enters a stable fluctuation state, compare the current chilled water outlet temperature The temperature at the next moment Using the set temperature difference range and step size, call the refrigerator power prediction model, traverse the power consumption at different temperatures, and select the temperature value that minimizes the power consumption As the chilled water outlet temperature The cloud server sends control instructions to the industrial gateway via the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature via the Modbus protocol. The temperature corresponding to the minimum power

[0028] Further, perform the following steps:

[0029] a. Temperature trend judgment: Compare the current chilled water outlet temperature The temperature at the next moment

[0030] like The temperature is on the rise, and the temperature difference step length T is set. step , and define the temperature difference range from the current temperature The ambient temperature at the next moment

[0031] like The temperature is decreasing, and the temperature difference step length T is set. step , and define the temperature difference range from the current temperature The ambient temperature at the next moment

[0032] like Then the temperature remains unchanged, and the temperature difference step length T is set step , and define the temperature difference range from the current temperature arrive

[0033] b. Call the freezer power prediction model, use the set temperature difference range and step size to traverse the power consumption at different temperatures, and select the temperature value that minimizes the power consumption As the chilled water outlet temperature

[0034] c. The cloud server sends control instructions to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol The temperature corresponding to the minimum power

[0035] As a preference, when the temperature is rising or remains constant, set the temperature difference step length T step =0.1; if the temperature is decreasing, set the temperature difference step length T step =-0.1.

[0036] Balanced heating state: When the air inlet temperature of the air conditioner is close to the target temperature, When the system enters the balanced heating state, compare the current air inlet temperature of the air conditioner The air inlet temperature of the air conditioner at the previous moment Calculate the temperature change.

[0037] If the temperature change satisfies the condition That is, if the temperature change is within the allowable target fluctuation range, the power of the refrigerator is reduced by ΔP%, and the chilled water outlet temperature prediction model is called; the optimized power is input into the chilled water outlet temperature prediction model, and the corresponding chilled water outlet temperature is output.

[0038] like That is, the temperature change exceeds the target fluctuation range, and the chilled water outlet temperature prediction model is directly called to calculate the chiller power P at this time. t Input the chilled water outlet temperature prediction model and output the corresponding chilled water outlet temperature

[0039] The cloud server sends control instructions to the industrial gateway via the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature via the Modbus protocol. The corresponding temperature value

[0040] Preferably, the industrial gateway adopts the HMI1002-Lite-E gateway of Shanghai Xunrao Automation Technology Co., Ltd.

[0041] Preferably, the cloud server saves all data to a database, and uses an XGBoost machine learning algorithm combined with random grid search to perform hyperparameter tuning to train a freezer power prediction model and a chilled water outlet temperature prediction model every hour.

[0042] The present invention realizes dynamic control of refrigerator power by introducing the XGBoost machine learning model, which can not only accurately adjust the temperature of the factory workshop, but also significantly reduce the energy consumption of water-cooled air conditioners through the optimization algorithm. In addition, the optimized control method helps to extend the service life of the equipment and improve the overall operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a structural diagram of the water circulation system of a water-cooled air conditioner;

[0044] Figure 2 This is a schematic diagram of the overall structure of a water-cooled air conditioner in an embodiment;

[0045] Figure 3 A comparison chart of actual energy efficiency and predicted energy efficiency;

[0046] Figure 4 Energy consumption optimization diagram optimized for energy saving;

[0047] Figure 5 This is a comparison chart for system energy consumption optimization. DETAILED DESCRIPTION

[0048] The specific implementation steps of the air conditioning energy consumption optimization method of the present invention are described in detail below with reference to the accompanying drawings.

[0049] A method for optimizing energy consumption of a water-cooled air conditioner based on machine learning. Specifically, the method comprises the following steps:

[0050] Step 1: Sensor installation and data collection:

[0051] like Figure 2 As shown in the figure, AOSUN ATH20 temperature sensors are installed at the chilled water inlet and outlet and cooling water inlet and outlet of the freezer, and temperature sensors of the same type are installed at the air inlet of the air-conditioning box and inside the factory workshop; Polyrun HRTX-S100E flow sensor of Shanghai Pujian Measurement and Control Instrument Co., Ltd. is installed at the outlet of the freezer; the output data of all sensors are connected to the HMI1002-Lite-E industrial gateway of Shanghai Xunrao Automation Technology Co., Ltd.

[0052] The refrigerator power P t , Cooling pump power P_CW t , refrigeration pump power P_CH t and cooling tower power P_CT tSend to the industrial gateway via MODBUS RTU protocol;

[0053] In this embodiment, the industrial gateway sends real-time temperature, flow and power data to the cloud server through the MQTT protocol every 1 minute; the cloud server receives key data from the HMI1002-Lite-E gateway every 1 minute, including the chilled water outlet temperature. Chilled water inlet temperature Cooling water outlet temperature Cooling water inlet temperature Factory floor temperature Air inlet temperature Chilled water outlet flow F t , Refrigerator power P t , refrigeration pump power P_CH t , cooling pump power P_CW t , cooling tower power P_CT t ;

[0054] Calculate the system energy efficiency ratio Scop based on the above data t , and its calculation formula is as follows:

[0055]

[0056] Among them, Q t is the instantaneous cooling capacity at time t, C p is the specific heat capacity of chilled water (1.163kw / m3 / h·℃).

[0057] Step 2: The cloud server obtains the temperature at time t+1 through the standard API interface of the National Meteorological Administration

[0058] Step 3: Set the target temperature T of the factory floor target , power reduction of the refrigerator ΔP%, temperature difference threshold ΔT in the startup state fac-th , the minimum temperature difference threshold ΔT of the equilibrium heating state air-th and temperature fluctuation threshold ΔT target .

[0059] Step 4: Use the data collected by the sensors in step 1 and use the XGBoost machine learning model to build a freezer power prediction model and a chilled water outlet temperature prediction model:

[0060] Refrigerator power prediction model: For input, T target , F t , Scop t is the characteristic data, the refrigerator power P tTo predict the value, the XGBoost machine learning model is used in combination with random grid search (RandomizedSearchCV) for hyperparameter tuning.

[0061] Chilled water outlet temperature prediction model: based on P t For input, T target , F t , Scop t is the characteristic data, refrigerator power To predict the value, the XGBoost machine learning model is used in combination with random grid search (RandomizedSearchCV) for hyperparameter tuning.

[0062] The cloud server saves all data to the database and trains the refrigerator power prediction model and the chilled water outlet temperature prediction model every hour.

[0063] Figure 3 and Figure 4 The comparison results between the actual energy efficiency and predicted energy efficiency of the refrigerator power prediction model and the chilled water outlet temperature prediction model are respectively explained.

[0064] Step 5: Adjust the chilled water outlet temperature according to different operating conditions, optimize the energy consumption ratio of the water-cooled air conditioner, and minimize energy consumption.

[0065] Start-up state: When the workshop temperature With the target temperature T target There are large differences between The system enters the startup state. At this time, set the chilled water outlet temperature to the minimum value CH_TP min (Its value range is [T target ×30%,T target ×60%], in this embodiment, the default situation CH_TP is selected min =T target ×40%), and sends the control command to the industrial gateway through the MQTT protocol. The industrial gateway sets the chilled water outlet temperature through the Modbus protocol to quickly reduce the indoor temperature.

[0066] Stable fluctuation state: when the workshop temperature is close to the target temperature , the system enters a stable fluctuation state. In this state, the cloud server calls the chiller power prediction model, uses the set temperature difference range and step size, traverses the power consumption at different temperatures, and selects the temperature value that minimizes power consumption as the chilled water outlet temperature. The cloud server sends a control instruction to the industrial gateway through the MQTT protocol to set the chilled water outlet temperature to the temperature corresponding to the minimum power.

[0067] Specifically:

[0068] a. Temperature trend judgment: Compare the current chilled water outlet temperature The temperature at the next moment

[0069] like The temperature is on the rise, and the temperature difference step length T is set. step = 0.1, and define the temperature difference range from the current temperature The temperature at the next moment

[0070] like The temperature is decreasing, and the temperature difference step length T is set. step = -0.1, and define the temperature difference range from the current temperature The temperature at the next moment

[0071] like Then the temperature remains unchanged, and the temperature difference step length T is set step = 0.1, and define the temperature difference range from the current temperature arrive

[0072] b. Call the freezer power prediction model, use the set temperature difference range and step size to traverse the power consumption at different temperatures, and select the temperature value that minimizes the power consumption As the chilled water outlet temperature

[0073] c. The cloud server sends control instructions to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol The temperature corresponding to the minimum power

[0074] Balanced heating state: When the air inlet temperature of the air conditioner is close to the target temperature When the system enters the balanced heating state, the following steps are performed:

[0075] i. Compare the current air inlet temperature of the air conditioner The air inlet temperature of the air conditioner at the previous moment Calculate the temperature change.

[0076] If the temperature change satisfies the condition That is, the temperature change is within the allowable target fluctuation range, then the power of the refrigerator P t Optimize and reduce the refrigerator power by ΔP%. After optimization, the refrigerator power Call the chilled water outlet temperature prediction model; optimize the power Input the chilled water outlet temperature prediction model and output the corresponding chilled water outlet temperature

[0077] like That is, the temperature change exceeds the target fluctuation range, and the chilled water outlet temperature prediction model is directly called to calculate the chiller power P at this time. t Input the chilled water outlet temperature prediction model and output the corresponding chilled water outlet temperature

[0078] ii. The server sends control instructions to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol For the corresponding

[0079] The formal description of the above method is as follows:

[0080]

[0081]

[0082]

[0083] Among them: ST_START indicates the system startup state; ST_STABLE indicates the stable fluctuation state; ST_BALANCE indicates the balanced heating state; ΔT fac-th Indicates the temperature difference threshold for the startup state, the default is 3°C; ΔT air-th Indicates the minimum temperature difference threshold for balanced heating state, the default is 1°C; ΔT target Indicates the temperature fluctuation threshold allowed in the factory workshop, the default value is 0.2℃; CH_TP min Indicates that the chilled water outlet temperature is the minimum value, the default value is ΔT target ×40%;T target Indicates the target temperature of the factory workshop, the default is 24°; ρ indicates the density of water 1000kg / m 3 ; Cp represents the specific heat capacity of water, 4.18 kJ / (kg·℃); ForecastAPI represents the API function for obtaining weather forecast.

[0084] like Figure 5 As shown, the present invention performs energy consumption calculation and analysis based on data collected by actual factory equipment. The above energy consumption optimization method is used to control the optimal working state of the chiller by accumulating 185,234 minutes of operation data from April to October. Before optimization, the total energy consumption of the water-cooled air conditioner was 115.208 million kilowatts, and after optimization, the energy consumption was reduced to 108.2977 million kilowatts, achieving an energy consumption optimization rate of 5.998%.

[0085] The present invention realizes the dynamic control of the chilled water outlet temperature in the water-cooled air-conditioning refrigerator by introducing the XGBoost machine learning model, and can optimize the refrigerator power according to the real-time and predicted data, thereby realizing low-energy control of the factory workshop temperature, significantly reducing the system energy consumption, and extending the service life of the equipment. The method has broad application prospects and significant economic benefits in industrial production environments.

Claims

1. A method for optimizing energy consumption of a water-cooled air conditioner, characterized in that: The following steps are involved: Step 1: Install temperature sensors at the chilled water inlet and outlet of the freezer and the cooling water inlet and outlet, and install a flow sensor at the chilled water outlet to measure the water temperature and flow at time t in real time; install temperature sensors at the air inlet of the air conditioner and inside the factory workshop to measure the air inlet temperature at time t in real time. and the temperature inside the workshop All sensors are connected to the industrial gateway; the refrigerator power, cooling pump power, freezing pump power and cooling tower power at time t are sent to the industrial gateway in real time through the MODBUS RTU protocol; The cloud server receives key data sent by the industrial gateway through the MQTT protocol, including the chilled water outlet temperature. Chilled water inlet temperature Cooling water outlet temperature Cooling water inlet temperature Temperature inside the workshop Air conditioning box inlet temperature Refrigerator power P t , chilled water outlet flow F t , refrigeration pump power P_CH t , cooling pump power P_CW t , cooling tower power P_CT t ; and calculate the system energy efficiency ratio Scop based on the above data t ; Step 2: The cloud server obtains the temperature at time t+1 through the standard API interface of the National Meteorological Administration Step 3: When the system is initialized, set the time interval for the cloud server to receive key data sent by the industrial gateway and the target temperature T of the factory workshop. target , power reduction of the refrigerator ΔP%, temperature difference threshold ΔT in the startup state fac-th , the minimum temperature difference threshold ΔT of the equilibrium heating state air-th and temperature fluctuation threshold ΔT target The chiller lowers the water temperature in the pipes and delivers chilled water to the air conditioning box coils, and the air in the air conditioning box exchanges heat through the coils; Step 4: Using the data collected by the sensor in step 1, use the XGBoost machine learning model to establish a freezer power prediction model and a chilled water outlet temperature prediction model; The refrigerator power prediction model uses the chilled water outlet temperature as input parameters, the chilled water inlet temperature, the cooling water outlet temperature, the cooling water inlet temperature, the workshop internal temperature, the air conditioning box inlet temperature, the target temperature, the chilled water outlet flow rate, and the system energy consumption ratio as characteristic data, and the refrigerator power as the predicted value; The chilled water outlet temperature prediction model uses the chiller power as input parameter, the chilled water inlet temperature, the cooling water outlet temperature, the cooling water inlet temperature, the workshop internal temperature, the air conditioning box inlet temperature, the target temperature, the chilled water outlet flow rate, and the system energy consumption ratio as characteristic data, and the chilled water outlet temperature as the predicted value; The cloud server uses the XGBoost machine learning model and combines it with hyperparameter tuning to train the model. After the model training is completed: Input any chilled water outlet temperature into the chiller power prediction model, and the corresponding chiller power will be output; Input any chiller power into the chilled water outlet temperature prediction model, and the corresponding chilled water outlet temperature will be output; Step 5: Adjust the power of the refrigerator according to the different operating conditions of the system, optimize the system energy consumption ratio, and minimize energy consumption; Start-up state: When the workshop temperature With the target temperature T target There are large differences between When the system enters the startup state, the cloud server sets the chilled water outlet temperature to the minimum value CH_TP min The control command is sent to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol CH_TP min , to quickly reduce the indoor temperature; Stable fluctuation state: When the workshop temperature is close to the target temperature, When the system enters a stable fluctuation state; compare the current chilled water outlet temperature The temperature at the next moment Using the set temperature difference range and step size, call the refrigerator power prediction model, traverse the power consumption at different temperatures, and select the temperature value that minimizes the power consumption As the chilled water outlet temperature The cloud server sends control instructions to the industrial gateway via the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature via the Modbus protocol. The temperature corresponding to the minimum power Balanced heating state: When the air inlet temperature of the air conditioner is close to the target temperature, When the system enters the balanced heating state, compare the current air inlet temperature of the air conditioner The air inlet temperature of the air conditioner at the previous moment Calculate the temperature change; If the temperature change satisfies the condition That is, if the temperature change is within the allowable target fluctuation range, the power of the chiller is reduced by ΔP%, and the chilled water outlet temperature prediction model is called; the optimized power is input into the chilled water outlet temperature prediction model, and the corresponding chilled water outlet temperature is output; like That is, the temperature change exceeds the target fluctuation range, and the chilled water outlet temperature prediction model is directly called to calculate the chiller power P at this time. t Input the chilled water outlet temperature prediction model and output the corresponding chilled water outlet temperature The cloud server sends control instructions to the industrial gateway via the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature via the Modbus protocol. The corresponding temperature value 2. The method according to claim 1, characterized in that: The industrial gateway adopts the HMI1002-Lite-E gateway of Shanghai Xunrao Automation Technology Co., Ltd.

3. The method according to claim 1, characterized in that: The cloud server saves all data to the database, and uses the XGBoost machine learning algorithm combined with random grid search to perform hyperparameter tuning and training of the refrigerator power prediction model and the chilled water outlet temperature prediction model every hour.

4. The method according to claim 1, characterized in that: In the step 5 startup state, the minimum chilled water outlet temperature CH_TP min The value range is [T target ×30%,T target ×60%].

5. The method according to claim 1, characterized in that: The step 5 described above stabilizes the fluctuation state by performing the following steps: a. Temperature trend judgment: Compare the current chilled water outlet temperature The temperature at the next moment like The temperature is on the rise, and the temperature difference step length T is set. ste, , and define the temperature difference range from the current temperature The ambient temperature at the next moment like The temperature is decreasing, and the temperature difference step length T is set. ste, , and define the temperature difference range from the current temperature The ambient temperature at the next moment like The temperature remains unchanged, and the temperature difference step length T is set ste, , and define the temperature difference range from the current temperature arrive b. Call the freezer power prediction model, use the set temperature difference range and step size to traverse the power consumption at different temperatures, and select the temperature value that minimizes the power consumption As the chilled water outlet temperature c. The cloud server sends control instructions to the industrial gateway through the MQTT protocol, and the industrial gateway sets the chilled water outlet temperature through the Modbus protocol The temperature corresponding to the minimum power 6. The method according to claim 5, characterized in that: When the temperature is rising or remains constant, set the temperature difference step length T ste, =0.1; if the temperature is decreasing, set the temperature difference step length T ste, =-0.1.

Images