A method for frequency control of central air conditioning chilled and hot water pumps based on load forecasting

Abstract

This invention discloses a method for regulating the frequency of chilled and hot water pumps in a central air conditioning system based on load forecasting. The method includes the following steps: classifying winter and summer operating conditions; determining the optimal pump operating frequency for each classified winter and summer operating condition; identifying factors influencing cooling and heating loads; constructing and training a BP neural network load forecasting model based on these factors; inputting real-time data of these factors into the trained BP neural network load forecasting model to obtain predicted cooling and heating load values; and predictively regulating the pump frequency based on the optimal pump operating frequency for each winter and summer operating condition and the predicted cooling and heating load values. This invention achieves efficient operation of the air conditioning system by determining the optimal pump operating frequency corresponding to the maximum COP of the air conditioning system under different operating conditions. Simultaneously, by utilizing the forecasting model, it solves the problem of untimely regulation of traditional chilled and hot water pumps caused by the lag in the chilled and hot water system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of central air conditioning cold and hot water systems, and particularly relates to a central air conditioning cold and hot water pump frequency regulation method based on load prediction. BACKGROUND

[0002] The central air conditioning cold and hot water system mainly has two operation modes: constant flow operation and variable flow operation. The constant flow operation refers to keeping the flow in the supply and return water pipes constant during air conditioning operation, and the supply and return water temperature difference directly reflects the change of the building load. However, during system design, the designer selects the equipment according to the full load condition in order to meet the system flow demand. However, the air conditioning system is in part load operation most of the time during use, and the long-time full load operation of the water pump leads to "large flow, small temperature difference operation", and the phenomenon of mismatch between water flow and user load occurs, causing great energy waste.

[0003] The variable flow system changes the flow of the cold and hot water system to meet the user side load demand. In the past, the air conditioning operation and maintenance personnel manually reduced the number of water pumps started to realize variable flow operation. However, this method is highly dependent on the experience of the staff, and the long-time start and stop of the water pump will affect the service life of the equipment. In order to solve this problem, the designer gradually applies the frequency conversion device to the cold and hot water pump, and realizes the variable flow operation of the cold and hot water system by adjusting the speed of the water pump. The frequency conversion methods mainly include temperature difference frequency conversion and pressure difference frequency conversion. The two control methods take the temperature difference or pressure difference of the supply and return water as the input signal of the controller, compare with the set temperature difference or pressure difference value, and finally adjust the water pump frequency through the frequency converter to realize the water quantity change.

[0004] However, the above method mainly uses the PID controller to realize feedback control. Due to the strong hysteresis of the cold and hot water system, the water quantity cannot be timely adjusted to the load change, which affects the indoor thermal comfort environment and causes certain energy waste. SUMMARY

[0005] In view of the above problems in the prior art, the present application provides a central air conditioning cold and hot water pump frequency regulation method based on load prediction, which solves the problem of untimely traditional cold and hot water pump regulation caused by the hysteresis of the cold and hot water system.

[0006] In order to achieve the above-mentioned application purpose, the technical scheme adopted by the present application is as follows:

[0007] A central air conditioning cold and hot water pump frequency regulation method based on load prediction, comprising the following steps:

[0008] S1, determining the central air conditioning system operation hour average data and outdoor meteorological hour average data according to the experimental platform debugging, indoor temperature and humidity fluctuation and indoor environment feedback data;

[0009] S2. Classify the winter and summer operating conditions based on the average hourly operating data of the central air conditioning system in step S1.

[0010] S3. Based on the total energy consumption data of the air conditioning system equipment and the average hourly operating data of the central air conditioning system in step S1, determine the optimal water pump operating frequency for various winter and summer operating conditions after classification in step S2.

[0011] S4. Based on the outdoor meteorological hourly average data and the correlation between cooling and heating loads in step S1, determine the factors influencing cooling and heating loads and obtain historical data on these factors.

[0012] S5. Construct a BP neural network load prediction model and train the BP neural network load prediction model using the existing data of the factors affecting cold and hot loads in step S4.

[0013] S6. Obtain real-time data of factors affecting heating and cooling loads and input them into the BP neural network load prediction model trained in step S5 to obtain the predicted values ​​of heating and cooling loads.

[0014] S7. Based on the optimal pump operating frequency for various winter and summer operating conditions in step S3 and the predicted values ​​of cold and heat loads in step S6, predict and regulate the pump frequency.

[0015] Further, step S1 includes the following sub-steps:

[0016] S11. Adjust the hot and cold water frequency according to the experimental platform, obtain the central air conditioning system operation data, and record the outdoor meteorological data;

[0017] S12. Calculate the hourly average values ​​of the central air conditioning system operation data and outdoor meteorological data in sub-step S11;

[0018] S13. Based on indoor temperature and humidity fluctuations and indoor environmental feedback data, filter the hourly average values ​​of the central air conditioning system operation data and the hourly average values ​​of the outdoor meteorological data from step S12 to determine the hourly average data of the central air conditioning system operation and the hourly average data of the outdoor meteorological data.

[0019] Furthermore, step S2 includes the following sub-steps:

[0020] S21. Calculate the cooling and heating load based on the average hourly operating data of the central air conditioning system in step S1, expressed as:

[0021] Q = cGΔT

[0022] Where: Q is the heating and cooling load, c is the specific heat capacity of hot and cold water, G is the flow rate of hot and cold water, and ΔT is the temperature difference between the supply and return water of hot and cold water;

[0023] S22. Calculate the load factor based on the heating and cooling loads in step S21, expressed as:

[0024]

[0025] Where: η is the heating and cooling load rate, Q max This represents the maximum building heating and cooling load during operation.

[0026] S23. Classify the winter and summer operating conditions according to the load rate in step S22.

[0027] Furthermore, step S3 includes the following sub-steps:

[0028] S31. Based on the average hourly operating data of the central air conditioning system in step S1, calculate the COP of the air conditioning system under various winter and summer operating conditions after classification in step S2.

[0029] S32. Sort the COP of the air conditioning system in sub-step S31 and determine the maximum value of the COP of the air conditioning system after sorting.

[0030] S33. Determine the optimal water pump operating frequency corresponding to the maximum value of the COP of the air conditioning system in sub-step S32.

[0031] 5. The method for frequency control of central air conditioning chilled and hot water pumps based on load forecasting according to claim 4, characterized in that step S31 includes the following sub-steps:

[0032] S311. Use a power quality analyzer to obtain the energy consumption of the main unit, water pump, and terminal equipment of the air conditioning system;

[0033] S312. Calculate the sum of the energy consumption of the air conditioning system main unit, water pump, and terminal equipment in step S311 to obtain the total energy consumption of the air conditioning system.

[0034] S313. Based on the average hourly operating data of the central air conditioning system in step S1 and the total energy consumption of the air conditioning system equipment in sub-step S312, calculate the COP of the air conditioning system, expressed as:

[0035]

[0036] Among them: COP sys Where COP is the COP of the air conditioning system, Q is the cooling and heating load, and P is the P value. sys This represents the total energy consumption of the air conditioning system equipment.

[0037] Furthermore, step S4 includes the following sub-steps:

[0038] S41. Based on the outdoor meteorological hourly average data and the correlation between cooling and heating loads in step S1, determine the factors influencing the cooling and heating loads of outdoor meteorology.

[0039] S42. Based on the correlation between cooling and heating loads, determine the factors affecting cooling and heating loads over time.

[0040] S43. Based on the factors affecting outdoor weather and heating load in step S41 and the factors affecting time and heating load in step S42, obtain historical data on the factors affecting heating load.

[0041] Furthermore, step S5 includes the following sub-steps:

[0042] S51. Construct a BP neural network load prediction model that includes forward propagation calculation and backward propagation calculation;

[0043] S52. Input the existing data of the factors affecting the heating and cooling loads in step S4 into the forward propagation calculation to obtain the training prediction values ​​of the heating and cooling loads.

[0044] S53. Determine whether the error between the training prediction value of the heating and cooling load in sub-step S52 and the actual value of the heating and cooling load is greater than the maximum error value; if so, proceed to sub-step S54, otherwise return to sub-step S52.

[0045] S54. Update the forward propagation calculation using backpropagation calculation.

[0046] Furthermore, step S52 includes the following sub-steps:

[0047] S521. Input the existing data of the influencing factors of heating and cooling loads in step S4 into the input layer, and use the input layer and the hidden layer to obtain the output value of the hidden layer, expressed as:

[0048]

[0049] Where: v j Let f(x) be the output value of the hidden layer, f(x) be the activation function of neuron j in the hidden layer, i be the neuron index in the input layer, m be the total number of neurons in the input layer, and w be the output value of the hidden layer. ij x represents the weights between the input layer and the hidden layer. i Let a be the input value of the i-th neuron. j is the threshold of the input layer and the hidden layer, j is the neuron index in the hidden layer, and p is the total number of neurons in the hidden layer;

[0050] S522. Based on the hidden layer output value in step S521, obtain the cold and heat load training prediction value using the hidden layer and the output layer, expressed as:

[0051]

[0052] Where: y k For the predicted values ​​of heating and cooling loads during training, wjk b represents the weights between the hidden layer and the output layer. k is the threshold between the hidden layer and the output layer, k is the neuron number in the output layer, and n is the total number of neurons in the output layer.

[0053] Furthermore, step S54 includes the following sub-steps:

[0054] S541. Update the weights and thresholds between the hidden layer and the output layer, expressed as follows:

[0055] w′ jk =λv j e k ,

[0056] b′ k =b k +e k

[0057] Where: w′ jk The updated weights are the hidden and output layers, λ is the learning rate of the BP neural network load prediction model, and v is the weights between the hidden and output layers. j e represents the hidden layer output value. k The predicted value y for cold and heat load training k and the actual value of heating and cooling load Y k The error between them, b′ k b is the threshold between the updated hidden layer and the output layer. k The threshold between the hidden layer and the output layer;

[0058] S542. Update the weights and thresholds between the input layer and the hidden layer, as follows:

[0059]

[0060]

[0061] Where: w′ ij w represents the updated weights between the input layer and the hidden layer. ij x represents the weights between the input layer and the hidden layer. i Let w be the input value of the i-th neuron, k be the neuron index in the output layer, n be the total number of neurons in the output layer, and w be the input value of the i-th neuron. jk a′ represents the weights between the hidden layer and the output layer. j For the updated thresholds of the input layer and hidden layer, a j The threshold values ​​are for the input layer and the hidden layer.

[0062] Furthermore, step S7 includes the following sub-steps:

[0063] S71. Determine the pump frequency prediction and control value based on the optimal pump operating frequency for various winter and summer operating conditions in step S3 and the predicted cold and heat load values ​​in step S6.

[0064] S72. Predict and regulate the pump frequency according to the pump frequency prediction and control value in step S71.

[0065] The beneficial effects of this invention are as follows:

[0066] (1) This invention classifies operating conditions by cooling and heating load rates and determines the optimal water pump operating frequency corresponding to the maximum COP of the air conditioning system under different operating conditions, thereby achieving efficient operation of the air conditioning system.

[0067] (2) This invention uses a BP neural network load prediction model to predict the heating and cooling loads and determines the pump frequency prediction and control value based on the predicted heating and cooling load values, thus solving the problem of untimely control of traditional hot and cold water pumps caused by the lag of the hot and cold water system. Attached Figure Description

[0068] Figure 1 This is a flowchart of a central air conditioning chilled and hot water pump frequency control method based on load forecasting.

[0069] Figure 2 This is a COP distribution diagram of an air conditioning system under different load rates during summer cooling.

[0070] Figure 3 This is a diagram showing the COP distribution of an air conditioning system under different load rates during winter heating.

[0071] Figure 4 The topology diagram of the BP neural network load prediction model;

[0072] Figure 5 The graph shows the cooling load prediction results of the BP neural network load prediction model under summer operating conditions.

[0073] Figure 6 The graph shows the heat load prediction results of the BP neural network load prediction model under winter operating conditions.

[0074] Figure 7 A graph showing the changes in air conditioning system energy consumption before and after optimization for summer operating conditions;

[0075] Figure 8 A graph showing the changes in air conditioning system energy consumption before and after optimization for winter operation. Detailed Implementation

[0076] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0077] like Figure 1 As shown, a method for frequency control of central air conditioning chilled and hot water pumps based on load forecasting includes steps S1-S7, as detailed below:

[0078] S1. Based on the experimental platform debugging, indoor temperature and humidity fluctuations, and indoor environmental feedback data, determine the average hourly data of the central air conditioning system and the average hourly data of outdoor weather.

[0079] In an optional embodiment of the present invention, the present invention fully adjusts the frequency of the hot and cold water pumps based on an existing experimental platform, obtains the operating data of the central air conditioning system, records outdoor data, and calculates the hourly average value of the above data. Then, the hourly average value of the central air conditioning system operating data and the hourly average value of the outdoor meteorological data are filtered to determine the hourly average data of the central air conditioning system and the hourly average data of the outdoor meteorological data.

[0080] Step S1 includes the following sub-steps:

[0081] S11. Adjust the hot and cold water frequencies according to the experimental platform, obtain the central air conditioning system operation data, and record outdoor meteorological data.

[0082] Specifically, the central air conditioning system operation data acquired in this invention includes the set frequency of the hot and cold water pumps, the supply and return water temperatures of the hot and cold water system, and the hot and cold water flow rates. The recorded outdoor meteorological data includes the outdoor dry-bulb temperature and the outdoor relative humidity. This invention records outdoor meteorological data at 5-minute intervals.

[0083] S12. Calculate the hourly average values ​​of the central air conditioning system operation data and outdoor meteorological data in sub-step S11.

[0084] Specifically, this invention statistically analyzes the operation data of the central air conditioning system and outdoor meteorological data, wherein the outdoor meteorological data is recorded at 5-minute intervals, and then calculates the hourly average value of the operation data of the central air conditioning system and the outdoor meteorological data.

[0085] S13. Based on indoor temperature and humidity fluctuations and indoor environmental feedback data, filter the hourly average values ​​of the central air conditioning system operation data and the hourly average values ​​of the outdoor meteorological data from step S12 to determine the hourly average data of the central air conditioning system operation and the hourly average data of the outdoor meteorological data.

[0086] Specifically, this invention uses indoor temperature and humidity fluctuations and indoor environmental feedback data—that is, it statistically analyzes indoor temperature and humidity fluctuations and staff feedback on the indoor environment—to eliminate data that does not meet human comfort requirements, and finally determines the average hourly operating data of the central air conditioning system and the average hourly outdoor weather data.

[0087] S2. Classify the winter and summer operating conditions based on the average hourly operating data of the central air conditioning system in step S1.

[0088] In an optional embodiment of the present invention, the present invention calculates the cooling and heating load based on the average hourly operating data of the central air conditioning system in step S1, and then classifies the winter operating conditions and summer operating conditions according to the cooling and heating load.

[0089] Step S2 includes the following sub-steps:

[0090] S21. Calculate the cooling and heating load based on the average hourly operating data of the central air conditioning system in step S1, expressed as:

[0091] Q = cGΔT

[0092] Where: Q is the heating and cooling load, c is the specific heat capacity of the hot and cold water, G is the flow rate of the hot and cold water, and ΔT is the temperature difference between the supply and return water of the hot and cold water.

[0093] S22. Calculate the load factor based on the heating and cooling loads in step S21, expressed as:

[0094]

[0095] Where: η is the heating and cooling load rate, Q max This represents the maximum building heating and cooling load during operation.

[0096] Specifically, the summer cooling load is mainly concentrated between 375-675kW, and its calculated load rate is mainly concentrated between 50%-90%; the winter heating load is mainly concentrated between 180-600kW, and its calculated load rate is mainly concentrated between 30%-70%.

[0097] S23. Classify the winter and summer operating conditions according to the load rate in step S22.

[0098] Specifically, this invention classifies summer operating conditions into four categories based on the load rate in step S22: 50%-60%, 60%-70%, 70%-80%, and 80%-90%; and classifies winter operating conditions into four categories based on the load rate in step S22: 30%-40%, 40%-50%, 50%-60%, and 60%-70%.

[0099] S3. Based on the total energy consumption data of the air conditioning system equipment and the average hourly operating data of the central air conditioning system in step S1, determine the optimal water pump operating frequency for various winter and summer operating conditions after classification in step S2.

[0100] In an optional embodiment of the present invention, the data on the winter and summer operating conditions of the central air conditioning system are filtered according to the classification of winter and summer operating conditions in step S2. Since the operating frequency of the hot and cold water pumps affects the energy consumption of the main unit, it is unreasonable to judge the efficiency of the air conditioning system solely based on the energy consumption of the pumps. Therefore, the present invention uses the COP of the air conditioning system as an indicator, which is the ratio of the air conditioning cooling (heating) capacity to the total energy consumption of the air conditioning system. The maximum system COP and its corresponding hot and cold water pump frequency are determined, and the corresponding hot and cold water pump frequency is taken as the optimal pump operating frequency under that cooling and heating load rate.

[0101] Step S3 includes the following sub-steps:

[0102] S31. Based on the average operating hours of the central air conditioning system and the total energy consumption data of the air conditioning system equipment in step S1, calculate the COP of the air conditioning system under various winter and summer operating conditions after classification in step S2.

[0103] Step S31 includes the following sub-steps:

[0104] S311. Use a power quality analyzer to obtain the energy consumption of the air conditioning system's main unit, water pump, and terminal equipment.

[0105] S312. Calculate the sum of the energy consumption of the air conditioning system main unit, water pump, and terminal equipment in step S311 to obtain the total energy consumption of the air conditioning system.

[0106] S313. Based on the average hourly operating data of the central air conditioning system in step S1 and the total energy consumption of the air conditioning system equipment in sub-step S312, calculate the COP of the air conditioning system, expressed as:

[0107]

[0108] Among them: COP sys Where COP is the COP of the air conditioning system, Q is the cooling and heating load, and P is the P value. sys This represents the total energy consumption of the air conditioning system equipment.

[0109] Specifically, this invention calculates the COP of the air conditioning system under various winter and summer operating conditions after classification in step S2 using the above formula, thereby obtaining the COP of the air conditioning system under different cooling and heating load rates during summer cooling, i.e., under different operating conditions, such as... Figure 2 As shown; similarly, the present invention obtains the COP of the air conditioning system under different load rates, i.e., different operating conditions, during winter heating, such as... Figure 3 As shown.

[0110] S32. Sort the COP of the air conditioning system in sub-step S31 and determine the maximum value of the COP of the air conditioning system after sorting.

[0111] S33. Determine the optimal water pump operating frequency corresponding to the maximum value of the COP of the air conditioning system in sub-step S32.

[0112] Specifically, this invention determines the optimal water pump operating frequency for winter and summer operating conditions based on the maximum value of the COP of the air conditioning system in step S31. The optimal water pump operating frequencies for different load rates, i.e., different operating conditions, during summer cooling are shown in Table 1.

[0113] Table 1 Optimal pump frequency under different load rates during summer cooling.

[0114] Category Load value distribution Load rate distribution Maximum COP Optimal water pump frequency 1 375-450 50％-60％ 3.92 35 Hz 2 450-525 60％-70％ 3.24 35 Hz 3 525-600 70％-80％ 3.28 35 Hz 4 600-675 80％-90％ 2.95 30 Hz

[0115] Table 2 shows the optimal pump operating frequency under different load rates, i.e., different operating conditions, during winter heating.

[0116] Table 2 Optimal Pump Frequency under Different Load Rates for Winter Heating

[0117] Category Load value distribution Load rate distribution Maximum COP Optimal water pump frequency 1 180-240 30％-40％ 3.22 30 Hz 2 240-300 40％-50％ 3.05 30 Hz 3 300-360 50％-60％ 3.01 30 Hz 4 360-420 60％-70％ 2.82 40 Hz

[0118] The present invention determines the optimal pump operating frequency for winter and summer operating conditions using Tables 1 and 2 above.

[0119] S4. Based on the outdoor meteorological hourly average data and the correlation between cooling and heating loads in step S1, determine the factors influencing cooling and heating loads and obtain historical data on these factors.

[0120] In an optional embodiment of the present invention, the present invention analyzes the correlation between the outdoor hourly average data in step S1 and the heating / cooling load based on the outdoor hourly average data in step S1, and analyzes the correlation between the heating / cooling load and the working time and working day date to determine the influencing factors of the heating / cooling load. The influencing factors of the heating / cooling load determined by the present invention include both outdoor weather and time. After determining the influencing factors of the heating / cooling load, the present invention obtains historical data of the influencing factors of the heating / cooling load based on the influencing factors of the heating / cooling load.

[0121] Step S4 includes the following sub-steps:

[0122] S41. Based on the outdoor meteorological hourly average data and the correlation between cooling and heating loads in step S1, determine the factors influencing the cooling and heating loads of outdoor meteorology.

[0123] Specifically, based on the hourly average outdoor meteorological data and the correlation between cooling and heating loads, this invention ultimately determines that the factors influencing the cooling and heating loads of outdoor meteorology include outdoor dry-bulb temperature and outdoor relative humidity.

[0124] S42. Based on the correlation between cooling and heating loads, determine the factors affecting cooling and heating loads over time.

[0125] Specifically, based on the correlation between heating and cooling loads, this invention ultimately determines the time-related factors influencing heating and cooling loads, including working hours and working day dates. Working hours are from 9:00 to 17:00, which this invention represents using values ​​from 9 to 17. Working day dates are Monday to Friday, which this invention represents using values ​​from 1 to 5.

[0126] S43. Based on the factors affecting outdoor weather and heating load in step S41 and the factors affecting time and heating load in step S42, obtain historical data on the factors affecting heating load.

[0127] S5. Construct a BP neural network load prediction model and train the BP neural network load prediction model using the existing data of the factors affecting cold and hot loads in step S4.

[0128] In an optional embodiment of the present invention, the present invention utilizes the factors influencing cooling and heating loads in step S4 to obtain existing data on these factors, uses the existing data as input layer parameters, and uses the predicted values ​​of cooling and heating loads as output values ​​to construct a BP neural network load prediction model. The BP neural network load prediction model constructed by the present invention includes an input layer, a hidden layer, and an output layer, as follows: Figure 4 As shown in the figure. The number of neurons in the hidden layer is 13.

[0129] Specifically, this invention uses MATLAB software to write a BP neural network load prediction model. Its algorithm learning process mainly includes two parts: forward propagation and backward propagation. In the forward propagation calculation part, the activation function of hidden layer neuron j is determined to be f(x), and the weights between the input layer and the hidden layer are w. ij The weights between the hidden layer and the output layer are w jk The threshold values ​​for the input layer and hidden layer are a. j And the threshold between the hidden layer and the output layer is b k .

[0130] Step S5 includes the following sub-steps:

[0131] S51. Construct a BP neural network load prediction model that includes forward propagation calculation and backward propagation calculation.

[0132] S52. Input the existing data of the factors affecting the heating and cooling loads in step S4 into the forward propagation calculation to obtain the training prediction values ​​of the heating and cooling loads.

[0133] Step S52 includes the following sub-steps:

[0134] S521. Input the existing data of the influencing factors of heating and cooling loads in step S4 into the input layer, and use the input layer and the hidden layer to obtain the output value of the hidden layer, expressed as:

[0135]

[0136] Where: v j Let f(x) be the output value of the hidden layer, f(x) be the activation function of neuron j in the hidden layer, i be the neuron index in the input layer, m be the total number of neurons in the input layer, and w be the output value of the hidden layer. ij x represents the weights between the input layer and the hidden layer. i Let a be the input value of the i-th neuron. j is the threshold of the input layer and the hidden layer, j is the neuron number in the hidden layer, and p is the total number of neurons in the hidden layer.

[0137] S522. Based on the hidden layer output value in step S521, obtain the cold and heat load training prediction value using the hidden layer and the output layer, expressed as:

[0138]

[0139] Where: y k For the predicted values ​​of heating and cooling loads during training, w jk b represents the weights between the hidden layer and the output layer. k is the threshold between the hidden layer and the output layer, k is the neuron number in the output layer, and n is the total number of neurons in the output layer.

[0140] S53. Determine whether the error between the training predicted value of the heating and cooling load in sub-step S52 and the actual value of the heating and cooling load is greater than the maximum error value; if so, proceed to sub-step S54; otherwise, return to sub-step S52 and continue to obtain the training predicted value of the heating and cooling load using the existing data of the factors affecting the heating and cooling load.

[0141] Specifically, the predicted value y for cold and heat load training. k and the actual value of heating and cooling load Y k The error e between k , is represented as:

[0142] e k =Y k -y k ;

[0143] If the error of the final prediction result is higher than the given maximum error, proceed to step S54; otherwise, return to step S52 to continue using the algorithm.

[0144] S54. Update the forward propagation calculation using backpropagation calculation.

[0145] Specifically, this invention adjusts the weights w of the input layer and the hidden layer according to the learning rate of the set BP neural network load prediction model. ij The weights w between the hidden layer and the output layer jk Thresholds a of the input layer and hidden layer j And the threshold b between the hidden layer and the output layer k Update until the error meets the set requirements.

[0146] Step S54 includes the following sub-steps:

[0147] S541. Update the weights and thresholds between the hidden layer and the output layer, expressed as follows:

[0148] w′ jk =λv j e k ,

[0149] b′ k =b k +e k

[0150] Where: w′ jk The updated weights are the hidden and output layers, λ is the learning rate of the BP neural network load prediction model, and v is the weights between the hidden and output layers. j e represents the hidden layer output value. k The predicted value y for cold and heat load training k and the actual value of heating and cooling load Y k The error between them, b′ k b is the threshold between the updated hidden layer and the output layer. k This is the threshold between the hidden layer and the output layer.

[0151] S542. Update the weights and thresholds between the input layer and the hidden layer, as follows:

[0152]

[0153]

[0154] Where: w′ ij w represents the updated weights between the input layer and the hidden layer. ij x represents the weights between the input layer and the hidden layer. iLet w be the input value of the i-th neuron, k be the neuron index in the output layer, n be the total number of neurons in the output layer, and w be the input value of the i-th neuron. jk a′ represents the weights between the hidden layer and the output layer. j For the updated thresholds of the input layer and hidden layer, a j The threshold values ​​are for the input layer and the hidden layer.

[0155] S6. Obtain real-time data of factors affecting heating and cooling loads and input them into the BP neural network load prediction model trained in step S5 to obtain the predicted values ​​of heating and cooling loads.

[0156] In an optional embodiment of the present invention, the present invention acquires real-time data of factors influencing heating and cooling loads, and obtains predicted values ​​of heating and cooling loads by inputting the real-time data of these factors into a trained BP neural network load prediction model. The BP neural network load prediction model predicts a maximum relative error of approximately 10% and a minimum relative error of approximately 0.03% under summer operating conditions, with an overall average relative error of 5.7%. Figure 5 As shown, the BP neural network load forecasting model predicts a maximum relative error of 11% and a minimum relative error of only 0.02% under winter operating conditions, with an average relative error of 5.1%. Figure 6 As shown.

[0157] S7. Based on the optimal pump operating frequency for various winter and summer operating conditions in step S3 and the predicted values ​​of cold and heat loads in step S6, predict and regulate the pump frequency.

[0158] In an optional embodiment of the present invention, based on the optimal pump operating frequencies for various winter and summer operating conditions in step S3, i.e., the optimal pump operating frequencies corresponding to different summer and winter operating conditions in Tables 1 and 2, the specific values ​​of the optimal pump operating frequencies under different heating and cooling loads can be determined. Therefore, the present invention can obtain the pump frequency prediction and control value, i.e., the optimal pump operating frequency corresponding to the heating and cooling load prediction value, based on the heating and cooling load prediction value in step S6, and then predictively control the pump frequency based on the pump frequency prediction and control value.

[0159] Step S7 includes the following sub-steps:

[0160] S71. Based on the optimal pump operating frequency for various winter and summer operating conditions in step S3 and the predicted cold and hot load values ​​in step S6, determine the predicted pump frequency control value.

[0161] S72. Predict and regulate the pump frequency according to the pump frequency prediction and control value in step S71.

[0162] Specifically, this invention predictively regulates the water pump frequency based on a predicted control value, thereby optimizing the water pump frequency control and ultimately reducing the energy consumption of the air conditioning system. The changes in air conditioning system energy consumption before and after optimization under summer operating conditions are shown below. Figure 7 As shown in the figure, the changes in air conditioning system energy consumption before and after optimization under winter operating conditions are as follows: Figure 8 As shown.

[0163] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A central air conditioning cold and hot water pump frequency regulation method based on load prediction, characterized in that, Comprise the following steps: S1, according to the experimental platform debugging, indoor temperature and humidity fluctuation and indoor environment feedback data, determine the central air conditioning system operating hours average data and outdoor weather hour average data; S2, according to the central air conditioning system operating hours average data in step S1, the winter and summer operation condition is classified, specifically including the following steps: S21, according to the central air conditioning system operating hours average data in step S1, the cooling and heating load is calculated, expressed as: wherein: is the cold thermal load, is the specific heat capacity of the cold and hot water, is the flow rate of the cold and hot water, is the supply and return water temperature difference of the cold and hot water; S22, according to the cooling and heating load in step S21, the load rate is calculated, expressed as: wherein: is the cold thermal load ratio, is the maximum building cold thermal load during operation; S23, according to the load rate in step S22, the winter and summer operation condition is classified; S3, according to the total energy consumption data of air conditioning system equipment and the central air conditioning system operating hours average data in step S1, the optimal water pump operating frequency of various winter and summer operation condition classified in step S2 is determined, specifically including the following steps: S31, according to the central air conditioning system operating hours average data in step S1, the air conditioning system COP of various winter and summer operation condition classified in step S2 is calculated; S32, the air conditioning system COP in step S31 is sorted, the maximum value of the sorted air conditioning system COP is determined; S33, the optimal water pump operating frequency corresponding to the maximum value of the air conditioning system COP in step S32 is determined; S4, according to the outdoor weather hour average data in step S1 and the cooling and heating load correlation, the cooling and heating load influencing factors are determined and the historical data of the cooling and heating load influencing factors are obtained; S5, a BP neural network load prediction model is constructed, and the existing data of the cooling and heating load influencing factors in step S4 is used to train the BP neural network load prediction model; S6, the real-time data of the cooling and heating load influencing factors are obtained and input into the trained BP neural network load prediction model in step S5, and the cooling and heating load prediction value is obtained; S7, according to the optimal water pump operating frequency of various winter and summer operation condition in step S3 and the cooling and heating load prediction value in step S6, the water pump frequency is predicted and controlled.

2. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 1, characterized in that, Step S1 includes the following steps: S11, according to the experimental platform, the cooling and heating water frequency is debugged, the central air conditioning system operating data is obtained, and the outdoor weather data is recorded; S12, the hourly average value of the central air conditioning system operating data and the outdoor weather data in step S11 is calculated; S13, according to the indoor temperature and humidity fluctuation and the indoor environment feedback data, the central air conditioning system operating data hourly average value and the outdoor weather data hourly average value in step S12 are filtered, and the central air conditioning system operating hours average data and the outdoor weather hour average data are determined.

3. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 1, characterized in that, Step S31 includes the following steps: S311, the energy quality analyzer is used to obtain the air conditioning system host energy consumption, water pump energy consumption and terminal device energy consumption; S312, the sum of the air conditioning system host energy consumption, water pump energy consumption and terminal device energy consumption in step S311 is calculated, and the total energy consumption of air conditioning system equipment is obtained; S313, according to the central air conditioning system operating hours average data in step S1 and the total energy consumption of air conditioning system equipment in step S312, the air conditioning system COP is calculated, expressed as: wherein: COP for the air conditioning system, cooling and heating loads, total energy consumption of the air conditioning system equipment.

4. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 1, characterized in that, Step S4 includes the following steps: S41, determining the cold and heat load influencing factors related to outdoor weather according to the outdoor weather hourly average data in step S1 and the cold and heat load correlation; S42, determining the cold and heat load influencing factors related to time according to the cold and heat load correlation; S43, obtaining the historical data of the cold and heat load influencing factors according to the cold and heat load influencing factors related to outdoor weather in sub-step S41 and the cold and heat load influencing factors related to time in sub-step S42.

5. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 1, characterized in that, Step S5 comprises the following sub-steps: S51, constructing a BP neural network load prediction model comprising forward propagation calculation and back propagation calculation; S52, inputting the existing data of the cold and heat load influencing factors in step S4 into the forward propagation calculation to obtain the cold and heat load training prediction value; S53, judging whether the error between the cold and heat load training prediction value in sub-step S52 and the real value of the cold and heat load is greater than the maximum error value; if yes, entering sub-step S54, otherwise returning to sub-step S52; S54, updating the forward propagation calculation by using the back propagation calculation.

6. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 5, characterized in that, Step S52 comprises the following sub-steps: S521, inputting the existing data of the cold and heat load influencing factors in step S4 into the input layer, and obtaining the hidden layer output value by using the input layer and the hidden layer, which is expressed as: in: The hidden layer output value. Hidden layer neurons activation function, i The sequence number of the neuron in the input layer. m This represents the total number of neurons in the input layer. The weights between the input layer and the hidden layer. For the first i The input values ​​of each neuron. The threshold values ​​for the input layer and the hidden layer are... j The sequence number of the neurons in the hidden layer. p This represents the total number of neurons in the hidden layer; S522, obtaining the cold and heat load training prediction value by using the hidden layer and the output layer according to the hidden layer output value in sub-step S521, which is expressed as: wherein: is the cold-heat load training prediction value, is the weight between the hidden layer and the output layer, is the threshold between the hidden layer and the output layer, k is the neuron number in the output layer, n is the total number of neurons in the output layer.

7. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 3, characterized in that, Step S54 comprises the following sub-steps: S541, updating the weights and thresholds between the hidden layer and the output layer, which is expressed as: ， wherein: is the updated weight between the updated hidden layer and the output layer, is the learning rate of the BP neural network load prediction model, is the hidden layer output value, is the error between the cold and heat load training prediction value and the cold and heat load true value is the updated threshold between the updated hidden layer and the output layer, is the threshold between the hidden layer and the output layer;​ S542, updating the weights and thresholds between the input layer and the hidden layer, which is expressed as: ， wherein: is the updated weight between the input layer and the hidden layer, is the weight between the input layer and the hidden layer, is the input value of the th neuron, k is the neuron number in the output layer, n is the total number of neurons in the output layer, is the weight between the hidden layer and the output layer, is the updated threshold value between the input layer and the hidden layer, is the threshold value between the input layer and the hidden layer.

8. The central air conditioning cold and hot water pump frequency regulation method based on load prediction according to claim 1, characterized in that, Step S7 comprises the following sub-steps: S71, determining the water pump frequency prediction control value according to the optimal water pump running frequency of various winter and summer operation conditions in step S3 and the cold and heat load prediction value in step S6; S72, predicting and controlling the water pump frequency according to the water pump frequency prediction control value in sub-step S71.

Images