A fresh air handling unit active energy-saving optimization control method based on machine learning and intelligent optimization

Abstract

The application discloses a fresh air handling unit active energy-saving optimization control method based on machine learning and intelligent optimization, and comprises the following steps: based on the historical operation data of the fresh air handling unit, a fresh air handling unit coil regulating valve opening-supply energy relationship data model is established by applying a machine learning algorithm, a fresh air handling unit system model is established in combination with a fresh air handling unit operation mechanism model, the supply air temperature and the supply air dew point are predicted, and the cold and hot water energy cost of the fresh air handling unit is predicted by the model; an intelligent optimization algorithm is applied to search for the best coil regulating valve opening to minimize the cold and hot water energy cost, so that active energy-saving optimization is realized; the fresh air handling unit is controlled according to the optimized coil regulating valve opening, the deviation between the actual value and the set value of the supply air temperature and the supply air dew point is calculated, and the deviation is corrected by a PID algorithm; and the outdoor air temperature and humidity conditions are monitored in real time, and energy-saving optimization is re-performed when the outdoor air temperature and humidity change. The application realizes active energy-saving optimization control of the fresh air handling unit according to the real-time change of the outdoor air conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy-saving optimization control, and in particular to an active energy-saving optimization control method for fresh air handling units based on machine learning and intelligent optimization. BACKGROUND

[0002] A fresh air handling unit is an important component of a plant air conditioning system, and is mainly used to process external air to meet certain temperature and humidity requirements before being sent indoors, so as to ensure that the indoor temperature and humidity environment meets the production process requirements. The regulation of air temperature and humidity by the fresh air handling unit is mainly achieved through four cold and hot water coils, i.e. preheating, precooling, reheating and recooling, and a humidifier. The cold and hot water provided by the four cold and hot water coils is controlled by electric regulating valves on the coils. At present, the PID control method is commonly used for controlling the regulating valves of the fresh air handling unit coils in an electronic plant. The regulating amount of the regulating valve of each coil is calculated and output by the PID algorithm based on the deviation between the measured value and the set value of the pre-processed air temperature (or enthalpy), the deviation between the measured value and the set value of the supply air temperature, and the deviation between the measured value and the set value of the supply air dew point, so as to control the opening degree of the regulating valve and achieve the set supply air temperature and dew point temperature.

[0003] At present, there are few energy-saving control methods for fresh air handling units. Referring to the energy-saving control methods of similar devices such as central air conditioners, the energy-saving control methods can be basically divided into PID-based energy-saving control methods and model prediction-based energy-saving control methods. The PID-based energy-saving control method is based on the commonly used PID control method described above, and the stability and self-adaptive ability of the PID algorithm are improved by optimizing the parameters in the PID control, so as to avoid energy loss caused by lag or fluctuation in the control process, such as the fuzzy PID-based energy-saving control method. The model prediction-based energy-saving control method refers to modeling the control system to predict the control results under different control parameters, and then selecting the control parameters that meet the control requirements and are relatively energy-saving. There are many system modeling and energy-saving optimization methods in the current model prediction-based control method. A common approach is to directly use historical operation data of the system, use the control parameters as the input variables of the model, and use the control results as the output variables, to establish a model reflecting the relationship between the control parameters and the control results, predict the control results under different control parameters, and directly select the most energy-saving control parameters based on this.

[0004] The commonly used PID control method and the PID-based energy-saving control method usually do not need to model the fresh air handling unit. The essence is to compare the actual values and set values of the target control variables such as the pre-processed air temperature or enthalpy, the supply air temperature, and the supply air dew point, calculate and output the corresponding control amount by the PID algorithm, so as to achieve stable control of the supply air temperature and the supply air dew point. The commonly used PID control method and the PID-based energy-saving control method have the following disadvantages:

[0005] 1) this kind of method belongs to passive control according to the feedback adjustment of controlled variable, and the reaction to the change of external air condition exists lag, and cannot actively adjust according to the real-time change of external air condition;

[0006] 2) for the system such as fresh air handling unit containing multiple control parameters, the PID control method cannot well solve the coupling between variables in the system, especially in the case of multiple PID controllers in series control, PID parameter setting is difficult, and energy waste problems such as over-humidification and dehumidification are easy to cause;

[0007] 3) the energy-saving control method based on PID is to realize energy-saving effect by optimizing PID control parameters, and the energy consumption of the system is not quantified and predicted, and there is lack of evaluation basis for energy-saving effect.

[0008] The system modeling method in the energy-saving control method based on model prediction currently mainly concentrates on single data-driven system modeling based on historical operation data, that is, the mapping relationship between control parameters and control results is directly established according to historical data, the energy consumption is predicted according to control parameters, and the most energy-saving control parameters are selected. The energy-saving control method based on model prediction currently has the following shortcomings:

[0009] 1) in the aspect of system modeling, the modeling description of the whole cycle system of the fresh air handling unit air treatment process is generally lacking, the energy conversion process in the fresh air handling unit air treatment process is not transparent, and the control parameters such as coil regulating valve opening degree are not associated with the coil energy supply condition, so that the relationship between the control parameters and the control results such as supply air temperature and supply air dew point cannot be described from the energy change angle;

[0010] 2) in the aspect of control parameter optimization, there is a lack of efficient optimization search method, and for the case that the control parameter combination is more, the optimization quality and optimization efficiency are low. SUMMARY

[0011] The technical problem to be solved by the present application is to provide a fresh air handling unit active energy-saving optimization control method based on machine learning and intelligent optimization, to realize the optimization search of the coil regulating valve opening degree, to reduce the cold and hot water energy consumption cost of the fresh air handling unit under the premise of meeting the supply air temperature and supply air dew point, and to realize the active energy-saving optimization control of the fresh air handling unit according to the real-time change of external air condition.

[0012] To solve the above technical problems, the present application provides a fresh air handling unit active energy-saving optimization control method based on machine learning and intelligent optimization, comprising the following steps:

[0013] Step 1, based on the historical operation data of the fresh air handling unit, a machine learning algorithm is applied to establish the mapping relationship between the coil regulating valve opening degree input variable and the coil cooling and heating capacity output variable, and a fresh air handling unit coil regulating valve opening degree-energy supply relationship data model is established.

[0014] Step 2, combine the coil regulating valve opening-energy relationship data model and the fresh air handling unit operation mechanism model to establish a fresh air handling unit system model, predict the supply air temperature and the supply air dew point according to the outdoor air temperature, the outdoor air humidity and the coil regulating valve opening; meanwhile, establish a cold and hot water energy cost model, predict the cold and hot water energy cost of the fresh air handling unit according to the unit energy cost of the cold and hot water and the coil regulating valve opening;

[0015] Step 3, according to the real-time outdoor air temperature and humidity conditions, take the cold and hot water energy cost as the optimization target, take the supply air temperature and the supply air dew point as the constraint conditions, take the coil regulating valve opening as the optimization variable, apply the intelligent optimization algorithm to search for the coil regulating valve opening that makes the cold and hot water energy cost minimum and meets the requirements of the supply air temperature and the supply air dew point, and realize the active energy-saving optimization;

[0016] Step 4, control the fresh air handling unit according to the optimized coil regulating valve opening, calculate the deviation amount of the measured value and the set value of the supply air temperature and the supply air dew point sensor, calculate the required regulating valve opening adjustment amount according to the deviation amount by the PID algorithm, and correct the supply air result;

[0017] Step 5, real-time monitor the outdoor air temperature and humidity conditions, and when the outdoor air temperature and humidity change, re-perform the energy-saving optimization.

[0018] Preferably, in step 1, the input variables of the machine learning algorithm are the historical time coil regulating valve opening, air mass flow rate, air temperature and humidity before the coil, and cold and hot water main pipe supply and return water temperature.

[0019] Preferably, in step 1, the output variable of the machine learning algorithm is the coil cooling and heating capacity, and the machine learning algorithm is used to establish the mapping relationship between the input variable and the output variable.

[0020] Preferably, in step 1, the coil regulating valve opening-energy relationship data model is specifically:

[0021] Q=PG(T,RH,m,TCV)

[0022] In the formula, m is the air mass flow rate in the fresh air handling unit, T is the air temperature before the coil, RH is the air humidity before the coil, TCV is the coil regulating valve opening, and Q is the coil cooling capacity or heating capacity.

[0023] Preferably, in step 2, the fresh air handling unit operation mechanism model is specifically:

[0024] Based on the air thermodynamic theorem, the energy conservation principle and the analysis of the air treatment process of the fresh air handling unit, the air state change process in the fresh air handling unit in the preheating mode is represented as

[0025] m*h0+Q1=m*h2

[0026] m*h2-Q3=m*h3+m*c 水 *(d w -d3)*T3

[0027] m*h3+Q4=m*h4

[0028]

[0029]

[0030]

[0031]

[0032] wherein m is the air mass flow rate in the fresh air handling unit, h0 is the enthalpy of the outdoor air, h1 is the enthalpy of the air after preheating, h3 is the enthalpy of the air after subcooling, h4 is the enthalpy of the supply air, Q1 is the heat supplied by the preheating coil, Q3 is the cooling supplied by the subcooling coil, Q4 is the heat supplied by the reheating coil, c is the specific heat capacity of water, d0 is the moisture content of the outdoor air, d1 is the moisture content of the air after preheating, d3 is the moisture content of the air after subcooling, d4 is the moisture content of the supply air, C8=-5800.2206, C9=1.3914993, C10=-0.04860239, C11=0.41764768*10-4, C12=-0.14452093*10-4. 水 w 10 11 -4 12 -7

[0033] The first three equations respectively represent the change of the air state after passing through the preheating coil, the subcooling coil and the reheating coil, the middle three equations represent the relationship between the parameters of the saturated air after cooling and dehumidification, and the last equation represents the relationship between the enthalpy of the supply air and the temperature and moisture content of the supply air.

[0034] By using the above equations, the enthalpy of the outdoor air h0, the air mass flow rate m, the cooling and heating supplied by the coils Q1, Q3 and Q4 are known, and the supply air temperature T4 and the supply air moisture content d4 can be obtained by using the dichotomy method, and then the supply air dew point Td4 can be obtained.

[0035] Preferably, the supply air temperature and the supply air dew point of the fresh air handling unit are represented as functions of the outdoor air temperature, the outdoor air humidity, the air mass flow rate and the cooling and heating supplied by the coils, i.e. the operation mechanism model of the fresh air handling unit is represented as:

[0036] (T4, Td4)=MAU(T0, RH0, m, Q1, Q2, Q3, Q4).

[0037] ​​​​​​​Preferably, in step 2, the coil regulating valve opening-energy relationship data model and the fresh air handling unit operation mechanism model are combined, and the fresh air handling unit system model is represented as (T4, Td4) = MAU (T0, RH0, m, PG1 (TCV1), PG2 (TCV2), PG3 (TCV3), PG4 (TCV4))

[0038] In the formula, TCV1, TCV2, TCV3, and TCV4 represent the regulating valve openings of the preheating coil, the precooling coil, the reheating coil, and the reheating coil, respectively, and PG1, PG2, PG3, and PG4 represent the regulating valve opening-energy relationship models of the preheating coil, the precooling coil, the reheating coil, and the reheating coil, respectively. For convenience of representation, only the regulating valve opening TCV is written in the input variables of the regulating valve opening-energy relationship model of the coil. Through the model, the supply air temperature and the supply air dew point generated by different combinations of regulating valve openings under given outdoor air conditions are predicted.

[0039] Preferably, in step 2, the cold and hot water energy cost of the fresh air handling unit is specifically: due to the different types of cold and hot water flowing in different functional coils, the unit flow energy cost is different, and for the four-section combined fresh air handling unit involved, the cold and hot water energy cost E in the steady state is represented as

[0040] E = a * f 预冷 (TCV 预冷 ) + b * f 再冷 (TCV 再冷 ) + c * [f 再冷 (TCV 再冷 ) + f 再热 (TCV 再热 )]

[0041] In the formula, E is the cold and hot water energy cost of the fresh air handling unit in the steady state, with the unit of yuan / h; a, b, and c are respectively the unit flow costs of different cold and hot water, with the unit of m 3 / yuan; f(x) is the flow of the coil regulating valve at the opening x, with the unit of m 3 / h; and TCV is the opening of the coil regulating valve.

[0042] Through the cold and hot water energy cost model of the fresh air handling unit, the cold and hot water energy cost of the fresh air handling unit under different combinations of regulating valve openings is predicted.

[0043] Preferably, in step 3, the energy-saving optimization control of the fresh air handling unit is designed as an optimization problem, and then an intelligent optimization algorithm is used to solve it, so as to obtain the energy-saving control strategy of the fresh air handling unit; the optimization objective of the energy-saving optimization of the fresh air handling unit is to reduce the cold and hot water energy cost of the fresh air handling unit, the optimization variable is the regulating valve opening of the coil of the fresh air handling unit, and the constraint condition is that the supply air temperature and the supply air dew point meet the requirements.

[0044] Preferably, in step 3, the optimization problem is expressed as:

[0045] P: min E

[0046] s.t.

[0047] C1: MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = T set

[0048] C2: MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = Td set

[0049] Wherein, T set , Td set respectively represent the supply air temperature set value, the supply air dew point set value.

[0050] The beneficial effects of the present application are: (1) the modeling of the fresh air handling unit system adopts a modeling method of mixed driving of mechanism model and data model, first, the coil regulating valve opening-supply energy relationship modeling is performed based on the machine learning algorithm and the historical data of the fresh air handling unit, and then the fresh air handling unit system model is established in combination with the operation mechanism model of the fresh air handling unit, and the relationship between the coil regulating valve opening and the supply air temperature and other control results is represented based on the energy conversion process in the air treatment process of the fresh air handling unit; (2) the active energy saving optimization of the fresh air handling unit cold and hot water energy consumption is to establish a fresh air handling unit cold and hot water energy consumption cost model, based on the fresh air handling unit system model, according to the real-time outdoor air temperature and humidity conditions, the intelligent optimization algorithm is applied to actively search for the coil regulating valve control parameters that make the cold and hot water energy consumption optimal under the real-time outdoor air conditions, so as to realize the active energy saving optimization. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 It is the coil regulating valve opening-supply energy relationship model diagram of the present application.

[0052] Figure 2 It is the fresh air handling unit operation mechanism model diagram of the present application.

[0053] Figure 3 It is the fresh air handling unit energy saving control system schematic diagram of the present application.

[0054] Figure 4 It is the fresh air handling unit energy saving control method flowchart schematic diagram of the present application. DETAILED DESCRIPTION

[0055] As Figure 4 shown, a fresh air handling unit active energy saving optimization control method based on machine learning and intelligent optimization comprises the following steps:

[0056] Step 1, based on the historical operation data of the fresh air handling unit, a machine learning algorithm is applied to establish a fresh air handling unit coil regulating valve opening-supply energy relationship data model;

[0057] Step 2, a fresh air handling unit system model is established in combination with the fresh air handling unit operation mechanism model and the coil regulating valve opening-supply energy relationship data model, the supply air temperature and the supply air dew point are predicted according to the outdoor air temperature, the outdoor air humidity and the coil regulating valve opening; at the same time, a cold and hot water energy cost model is established, and the cold and hot water energy cost of the fresh air handling unit is predicted according to the unit energy cost of the cold and hot water and the coil regulating valve opening;

[0058] Step 3, according to the real-time outdoor air temperature and humidity conditions, taking the cold and hot water energy cost as the optimization target, taking the supply air temperature and the supply air dew point as the constraint conditions, and taking the coil regulating valve opening as the optimization variable, an intelligent optimization algorithm is applied to search for the coil regulating valve opening that makes the cold and hot water energy cost minimum and meets the requirements of the supply air temperature and the supply air dew point, so as to realize active energy-saving optimization;

[0059] Step 4, the fresh air handling unit is controlled according to the optimized coil regulating valve opening, the deviation amount of the measured value and the set value of the supply air temperature and the supply air dew point sensor is calculated, the regulating valve opening adjustment amount required by the PID algorithm is calculated according to the deviation amount, and the supply air result is corrected;

[0060] Step 5, the outdoor air temperature and humidity conditions are monitored in real time, and when the outdoor air temperature and humidity change, energy-saving optimization is re-performed.

[0061] In step 1, the input variables of the machine learning algorithm are the historical time coil regulating valve opening, air mass flow rate, coil front air temperature and humidity, and cold and hot water main pipe supply and return water temperature, and the output variable is the coil cooling and heating capacity, the mapping relationship between the input variables and the output variables is established by the machine learning algorithm, and the coil regulating valve opening-supply energy relationship data model is established.

[0062] As shown in Figure 1 , in step 1, the coil regulating valve opening-supply energy relationship data model is specifically:

[0063] Q=PG(T,RH,m,TCV)

[0064] In the formula, m is the air mass flow rate in the fresh air handling unit, T is the air temperature before the coil, RH is the air humidity before the coil, TCV is the coil regulating valve opening, and Q is the coil cooling capacity or heating capacity;

[0065] The coil energy supply (cooling capacity or heating capacity) is related to multiple factors, including the coil regulating valve opening degree, the coil cold and hot water inlet and outlet water temperature, the coil heat exchange efficiency, the coil material, the cold and hot water system water supply temperature, and the air state passing through the coil, wherein the coil regulating valve opening degree, the air temperature and humidity and mass flow rate passing through the coil, and the cold and hot water system water supply temperature are the main factors affecting the coil energy supply. In order to represent the relationship between the above main factors and the coil energy supply, the present application applies a machine learning algorithm to analyze data based on the historical operation data of the fresh air handling unit, and finally obtains a coil regulating valve opening degree-energy supply relationship data model. The required historical operation data of the fresh air handling unit at least includes: outdoor air temperature, outdoor air humidity, air speed, pre-processed air temperature and humidity, supply air temperature, supply air dew point, and each coil regulating valve opening degree. Based on the above historical data, the cooling capacity or heating capacity of each coil at the historical time is calculated and taken as the output variable of the machine learning algorithm, and the main influencing factors such as the coil regulating valve opening degree, the air temperature and humidity and mass flow rate passing through the coil, and the cold and hot water system water supply temperature are taken as the input variables of the machine learning algorithm, and the mapping relationship between the input variables and the output variables is established through the algorithm, and the establishment of the coil regulating valve opening degree-energy supply relationship data model is completed.

[0066] As shown in Figure 2 , the fresh air handling unit operation mechanism model is a description of the air treatment process of the fresh air handling unit based on the principle of energy conservation. The present application takes the cold and heat provided by the cold and hot coils as the main factor affecting the air state. According to the principle of energy conservation, the air enthalpy after the preheating coil or the reheating coil is equal to the sum of the air enthalpy before the coil and the coil heat supply; the air enthalpy after the pre-cooling coil or the re-cooling coil is equal to the difference between the air enthalpy before the coil and the coil cooling capacity; the air enthalpy after the pre-cooling coil or the re-cooling coil is equal to the difference between the air enthalpy before the coil and the coil cooling capacity. Based on the above principle, when the humidifying section is isenthalpic humidification and the humidification efficiency is 100%, the supply air temperature and the supply air dew point can be expressed as a function of the outdoor air temperature and humidity and the coil cooling capacity or heating capacity, which is the fresh air handling unit operation mechanism model.

[0067] In step 2, the fresh air handling unit operation mechanism model is specifically:

[0068] Based on the air thermodynamic theorem, the principle of energy conservation and the analysis of the air treatment process of the fresh air handling unit, the air state change process in the preheating mode of the fresh air handling unit is represented as

[0069] m*h0+Q1=m*h2

[0070] m*h2-Q3=m*h3+m*c 水 *(d w -d3)*T3

[0071] m*h3+Q4=m*h4

[0072]

[0073]

[0074]

[0075]

[0076] wherein m is the air mass flow rate in the fresh air handling unit, h0 is the enthalpy of the outdoor air, h1 is the enthalpy of the air after preheating, h3 is the enthalpy of the air after recooling, h4 is the enthalpy of the supply air, Q1 is the heat supply of the preheating coil, Q3 is the cooling supply of the recooling coil, Q4 is the heat supply of the reheating coil, c 水 is the specific heat capacity of water, d w is the moisture content of the air after humidification, d3 is the moisture content of the air after recooling, d4 is the moisture content of the air after reheating, C8=-5800.2206, C9=1.3914993, C 10 =-0.04860239, C 11 =0.41764768*10 -4 , C 12 =-0.14452093*10 -7 .

[0077] wherein the first three equations respectively represent the state changes of the air passing through the preheating coil, the recooling coil and the reheating coil, the middle three equations represent the relationships between the parameters of the saturated air after cooling and dehumidification, and the last equation represents the relationship between the enthalpy of the supply air and the temperature and moisture content of the supply air;

[0078] By combining the above equations, the supply air temperature T4 and the supply air dew point d4 can be obtained by using the bisection method, given the enthalpy of the outdoor air h0, the air mass flow rate m and the cooling and heating supplies Q1, Q3 and Q4 of the coils, and the supply air dew point Td4 can be further obtained;

[0079] Therefore, the supply air temperature and the supply air dew point of the fresh air handling unit are represented as functions of the outdoor air temperature, the outdoor air humidity, the air mass flow rate and the cooling and heating supplies of the coils, i.e. the operation mechanism model of the fresh air handling unit can be represented as

[0080] (T4,Td4) = MAU(T0,RH0,m,Q1,Q2,Q3,Q4).

[0081] The system model of the fresh air handling unit established by combining the above two models can obtain the supply air temperature and the supply air humidity under the current outdoor air condition and control parameters by calculation, given the outdoor air temperature, the outdoor air humidity and the opening degrees of the regulating valves of the coils.

[0082] Combining with the coil model, the fresh air handling unit system model is expressed as (T4, Td4) = MAU (T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4))

[0083] In the formula, TCV1, TCV2, TCV3 and TCV4 respectively represent the opening degrees of the regulating valves of the preheating coil, the precooling coil, the reheating coil and the reheating coil, and PG1, PG2, PG3 and PG4 respectively represent the regulating valve opening-power relationship data models of the preheating coil, the precooling coil, the reheating coil and the reheating coil. For convenience of representation, only the regulating valve opening TCV is written in the input variables of the coil regulating valve opening-power relationship data model. Through the model, the supply air temperature and the supply air dew point generated by different combinations of regulating valve openings under a given outdoor air condition are predicted.

[0084] In step 2, the chilled and hot water energy cost of the fresh air handling unit is specifically: since the types of the chilled and hot water flowing in the different functional coils are different, the unit flow energy cost is different, and for the four-section combined fresh air handling unit involved in the application, the chilled and hot water energy cost E in the steady state is expressed as

[0085] E = a * f 预冷 (TCV 预冷 ) + b * f 再冷 (TCV 再冷 ) + c * [f 再冷 (TCV 再冷 ) + f 再热 (TCV 再热 )]

[0086] In the formula, E is the chilled and hot water energy cost of the fresh air handling unit in the steady state, with the unit of yuan / h; a, b and c are respectively different chilled and hot water unit flow costs, with the unit of m 3 / yuan; f(x) is the flow of the coil regulating valve at the opening degree x, with the unit of m 3 / h; and TCV is the opening degree of the coil regulating valve.

[0087] Through the chilled and hot water energy cost model of the fresh air handling unit, the chilled and hot water energy cost of the fresh air handling unit under different combinations of regulating valve openings is predicted.

[0088] In step 3, the energy-saving optimization control of the fresh air handling unit is designed as an optimization problem, and then an intelligent optimization algorithm is used to solve the problem, so as to obtain the energy-saving control strategy of the fresh air handling unit; the optimization objective of the energy-saving optimization of the fresh air handling unit is to reduce the chilled and hot water energy cost of the fresh air handling unit, the optimization variable is the opening degree of the coil regulating valve of the fresh air handling unit, and the constraint condition is that the supply air temperature and the supply air dew point meet the requirements, so that the optimization problem is expressed as:

[0089] P: min E

[0090] s.t.

[0091] C1: MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = T set

[0092] C2: MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = Td set

[0093] wherein, T set , Td set respectively represent the supply air temperature set value, the supply air dew point set value.

[0094] The energy-saving control of the fresh air handling unit is to give a set of coil regulating valve opening combinations, so as to meet the supply air temperature and the supply air dew point while achieving the energy-saving effect. The present application selects the cold and hot water heat as the energy-saving object, first calculates the fresh air handling unit cold and hot water energy cost according to the unit energy cost of different kinds of cold and hot water, and then minimizes the cold and hot water energy cost as the optimization target, takes the supply air temperature and the supply air dew point as the constraint condition, and takes the coil regulating valve opening as the optimization variable, and proposes the optimization problem of the fresh air handling unit cold and hot water energy cost. The intelligent optimization algorithm is applied to solve the optimization problem, so as to obtain the coil regulating valve opening combination which meets the supply air temperature and the supply air dew point and has the minimum cold and hot water energy cost.

[0095] In step 4, the coil regulating valve opening obtained by solving according to the intelligent optimization algorithm controls the fresh air handling unit, calculates the deviation amount of the supply air temperature and the supply air dew point sensor measured value from the set value, and applies the PID algorithm to calculate the required regulating valve opening adjustment amount by linear combination according to the given proportional value, integral value and differential value, and corrects the regulating valve opening to meet the supply air temperature and the supply air dew point set value.

[0096] In step 5, the outdoor air conditions are monitored in real time, and when the outdoor air temperature or the outdoor air humidity deviates by more than 5% compared with the outdoor air temperature or the outdoor air humidity when the last optimization action occurs, the optimization action described in steps 1 to 4 is re-executed.

[0097] As Figure 3As shown, the new fan unit energy-saving control system designed by the application comprises two basic function modules: a new fan unit system modeling module and a new fan unit energy-saving optimization control module. The new fan unit system modeling module first preprocesses the historical operation data of the new fan unit in the database, and uses a machine learning method to establish a coil regulating valve opening-supply energy relationship data model, and then combines a new fan unit operation mechanism model and a cold and hot water energy consumption cost model to establish a new fan unit supply air temperature, supply air dew point and cold and hot water energy consumption cost prediction model. The new fan unit energy-saving optimization control module is based on the above prediction model, combines the real-time outdoor temperature and humidity conditions collected by the field sensor, calls an intelligent optimization algorithm, takes the cold and hot water energy consumption cost as the optimization target, the supply air temperature and the supply air dew point as the constraint condition, searches for the coil regulating valve opening that optimizes the cold and hot water energy consumption cost under the current outdoor conditions, and outputs to the new fan unit controller to complete the energy-saving control.

[0098] Compared with the single data-driven system modeling method in the existing model prediction-based energy-saving control method, the new fan unit system modeling method proposed in the application considers the common sensor configuration scheme and actual application situation in the new fan unit control of the electronic factory, combines the mechanism modeling method and the data modeling method, makes the two methods complementary to each other, solves the problems of difficulty in obtaining the coil-related parameters such as coil heat exchange efficiency and coil inlet and outlet water temperature in the mechanism modeling method and the problems of insufficient model reliability and interpretability in the data modeling method, and associates the control parameters such as the coil regulating valve opening in the new fan unit with the control results such as the supply air temperature and the supply air dew point, realizes the systematic description of the relationship between the control parameters and the control results from the energy angle, and provides a basis for the next step of coil regulating valve opening optimization.

[0099] In view of the problems that there is a lack of quantitative energy-saving effect evaluation index in the existing energy-saving control method and the energy-saving optimization cannot be actively carried out according to the real-time change of the outdoor conditions, the application selects the cold and hot water (cold and heat) energy consumption which is relatively important and controllable in the energy consumption type related to the new fan unit as the object, proposes an evaluation method of the cold and hot water energy consumption cost of the new fan unit, quantifies the cold and hot water energy consumption index of the new fan unit, and then takes the cold and hot water energy consumption cost as the optimization target, searches for the energy-saving control parameters based on the new fan unit system model and according to the real-time outdoor temperature and humidity conditions, and realizes the active energy-saving optimization of the cold and hot water energy consumption by using the intelligent optimization algorithm.

[0100] The application selects the intelligent optimization algorithm to solve the above optimization problem, searches for the coil regulating valve opening combination that optimizes the cold and hot water energy consumption cost, reduces the cold and hot water energy consumption cost of the new fan unit to the maximum extent on the basis of meeting the supply air temperature and supply air dew point demand, and realizes the energy-saving optimization.

Claims

1. A fresh air handling unit active energy-saving optimization control method based on machine learning and intelligent optimization, characterized in that, The method comprises the following steps: Step 1: Based on the historical operation data of the fresh air handling unit, a machine learning algorithm is applied to establish a mapping relationship between the input variable of the coil regulating valve opening degree and the output variable of the coil cooling and heating capacity, and a coil regulating valve opening degree-energy supply relationship data model of the fresh air handling unit is established; Step 2: A fresh air handling unit system model is established in combination with the coil regulating valve opening degree-energy supply relationship data model and the fresh air handling unit operation mechanism model, the supply air temperature and the supply air dew point are predicted according to the outdoor air temperature, the outdoor air relative humidity and the regulating valve opening degrees of the coils, and a cooling and heating water energy cost model is established, the cooling and heating water energy cost of the fresh air handling unit is predicted according to the unit energy cost of the cooling and heating water and the regulating valve opening degrees of the coils; Step 3: According to the real-time outdoor air temperature and humidity conditions, the cooling and heating water energy cost is taken as the optimization target, the supply air temperature and the supply air dew point are taken as the constraint conditions, the coil regulating valve opening degree is taken as the optimization variable, an intelligent optimization algorithm is applied, the coil regulating valve opening degree that makes the cooling and heating water energy cost minimum and meets the requirements of the supply air temperature and the supply air dew point is searched, and active energy-saving optimization is realized; Step 4: The fresh air handling unit is controlled according to the optimized coil regulating valve opening degree, the deviation amount of the measured value and the set value of the supply air temperature and the supply air dew point is calculated, the regulating valve opening degree adjustment amount required is calculated by a PID algorithm according to the deviation amount, and the supply air result is corrected; Step 5: The outdoor air temperature and humidity conditions are monitored in real time, when the outdoor air temperature and humidity change, energy-saving optimization is re-performed.

2. The machine learning and intelligent optimization-based fresh air handling unit active energy-saving optimization control method of claim 1, wherein, In step 1, the input variables of the machine learning algorithm are the coil regulating valve opening degree, the air mass flow rate, the air temperature and humidity before the coil, and the cooling and heating water main pipe supply and return water temperature at historical time, and the output variable of the machine learning algorithm is the coil cooling and heating capacity.

3. The machine learning and intelligent optimization-based fresh air handling unit active energy-saving optimization control method of claim 1, wherein, In step 1, the machine learning algorithm is used to establish the mapping relationship between the input variable and the output variable.

4. The machine learning and intelligent optimization based fresh air handling unit active energy saving optimization control method of claim 1, wherein, In step 1, the coil regulating valve opening degree-energy supply relationship data model is specifically: Q = PG (T, RH, m, TCV) In the formula, m is the air mass flow rate in the fresh air handling unit, T is the air temperature before the coil, RH is the air humidity before the coil, TCV is the coil regulating valve opening degree, and Q is the coil cooling capacity or heating capacity.

5. The machine learning and intelligent optimization-based fresh air handling unit active energy-saving optimization control method of claim 1, wherein, In step 2, the fresh air handling unit operation mechanism model is specifically: Based on the air thermodynamic theorem, the energy conservation principle and the analysis of the air treatment process of the fresh air handling unit, the air state change process in the fresh air handling unit in the preheating mode is represented as m * h0 + Q1 = m * h2 m*h2 - Q3 = m*h3 + m*c 水 *(d w -d3)*T3 m * h3 + Q4 = m * h4 wherein m is the air mass flow rate in the fresh air handling unit, h0 is the enthalpy of outside air, h2 is the enthalpy of preheated air, h3 is the enthalpy of re-cooled air, h4 is the enthalpy of supply air, Q1 is the heat supply of preheating coil, Q3 is the cooling supply of re-cooling coil, Q4 is the heat supply of re-heating coil, c 水 is the specific heat capacity of water, d w is the moisture content of preheated air, d3 is the moisture content of re-cooled air, d4 is the moisture content of re-heated air, C8 = -5800.2206, C9 = 1.3914993, C 10 = -0.04860239, C 11 = 0.41764768*10 -4 , C 12 = -0.14452093*10 -7 ; T3 is the temperature of re-cooled air, Pq is the saturated water vapor pressure, C 13 = 6.5459673; The first three formulas respectively represent the air state changes after passing through the preheating coil, the reheating coil and the reheating coil, the middle three formulas represent the relationships between the parameters of the saturated air after cooling and dehumidification, and the last formula represents the relationship between the supply air enthalpy and the supply air temperature and the supply air humidity content; By simultaneously solving the above formulas, the supply air temperature T4 and the supply air humidity content d4 are obtained by applying the bisection method under the conditions of the known outdoor air enthalpy h0, the air mass flow rate m, the cooling and heating capacities Q1, Q3 and Q4 provided by the coils, and the supply air temperature and the supply air humidity content are obtained.

6. The machine learning and intelligent optimization based fresh air handling unit active energy saving optimization control method of claim 5, wherein, The supply air temperature and the supply air dew point of the fresh air handling unit are represented as functions of the outdoor air temperature, the outdoor air humidity, the air mass flow rate and the coil cooling and heating capacity, that is, the fresh air handling unit operation mechanism model is represented as (T4, Td4) = MAU (T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) where T0 is the outdoor air temperature, RH0 is the outdoor air humidity, T4 is the supply air temperature, Td4 is the supply air dew point, TCV1, TCV2, TCV3, TCV4 represent the opening degree of the preheating coil, precooling coil, reheating coil, reheating coil respectively, PG1, PG2, PG3, PG4 represent the opening degree-supply energy relationship model of the preheating coil, precooling coil, reheating coil, reheating coil respectively, and the input variable of the coil opening degree-supply energy relationship data model is only written as the opening degree TCV. Through the model, the supply air temperature and the supply air dew point generated under different opening degree combinations of the adjusting valve under the given outdoor air condition are predicted.

7. The machine learning and intelligent optimization-based fresh air handling unit active energy-saving optimization control method of claim 1, wherein, In step 2, the cold and hot water energy cost model of the fresh air handling unit is specifically: since the types of cold and hot water flowing in different functional coils are different, the unit flow energy cost is different. For the four-section combined fresh air handling unit involved, the cold and hot water energy cost E in the steady state is expressed as Through the cold and hot water energy cost model of the fresh air handling unit, the cold and hot water energy cost of the fresh air handling unit under different opening degree combinations of the adjusting valve is predicted. In step 3, the energy-saving optimization control of the fresh air handling unit is designed as an optimization problem, and then an intelligent optimization algorithm is used to solve it, so as to obtain the coil adjusting valve opening degree that makes the cold and hot water energy cost of the fresh air handling unit optimal; the optimization objective of the energy-saving optimization of the fresh air handling unit is to reduce the cold and hot water energy cost of the fresh air handling unit, the optimization variable is the coil adjusting valve opening degree of the fresh air handling unit, and the constraint condition is that the supply air temperature and the supply air dew point meet the demand.

8. The machine learning and intelligent optimization based fresh air handling unit active energy saving optimization control method of claim 1, wherein, In step 3, the cold and hot water energy cost optimization problem of the fresh air handling unit is expressed as: E = a * f 预冷 (TCV 预冷 ) + b * f 再冷 (TCV 再冷 ) + c * [f 再冷 (TCV 再冷 ) + f 再热 (TCV 再热 ) ] In the formula, E is the energy cost of the fresh air handling unit under stable state, in yuan / h; a, b, and c are respectively the unit flow cost of different cold and hot water, in yuan / m 3 ; f(x) is the flow of the coil regulating valve under the opening degree of x, in m 3 / h; TCV is the opening degree of the coil regulating valve. P: min E 9. The machine learning and intelligent optimization based fresh air handling unit active energy saving optimization control method of claim 1, wherein, s.t.

10. The machine learning and intelligent optimization based fresh air handling unit active energy saving optimization control method of claim 1, wherein, ​ ​ ​ C1 : MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = T set C2: MAU(T0, RH0, m, PG1(TCV1), PG2(TCV2), PG3(TCV3), PG4(TCV4)) = Td set wherein T set , Td set represent the supply air temperature set value and the supply air dew point set value, respectively, m is the air mass flow rate in the fresh air handling unit, TCV1, TCV2, TCV3, TCV4 represent the opening degree of the regulating valve of the preheating coil, the precooling coil, the subcooling coil, and the reheating coil, respectively, PG1, PG2, PG3, PG4 represent the regulating valve opening-supply energy relationship model of the preheating coil, the precooling coil, the subcooling coil, and the reheating coil, respectively, E is the energy cost of the fresh air handling unit under stable state, T0 is the outdoor air temperature, and RH0 is the outdoor air humidity.

Images