A method and electronic device for predicting circulating cooling water flow in an indirect dry cooling system

Abstract

The present invention discloses a method and electronic device for predicting circulating cooling water flow in an indirect dry cooling system, the method comprising the following steps: step A: optimizing a BP neural network model using a genetic algorithm to obtain a GA-BP model; step B: predicting the outlet temperature of the circulating cooling water and the mass flow rate of the circulating cooling water using the GA-BP model; step C: inputting the outlet temperature of the circulating cooling water and the mass flow rate of the circulating cooling water into a condenser heat transfer model and an air-cooled heat exchanger heat transfer model to output the air outlet temperature and the mass flow rate of the air entering the tower; and step D: obtaining the optimal value of the circulating cooling water mass flow rate under variable output load. The entire prediction process of the present invention fully considers the influence of factors such as the terrain conditions around the unit, the plant and the cooling tower structure on the mass flow rate of the air entering the cooling tower, and is closer to the actual operation of the indirect dry cooling system.

Description

Technical Field

[0001] The present invention relates to the technical field of indirect dry cooling, and more particularly to a method and electronic equipment for predicting circulating cooling water flow in an indirect dry cooling system. Background Art

[0002] Indirect dry cooling systems effectively conserve water resources and are therefore widely used in thermal power plants in coal-rich, water-scarce regions. Natural draft dry cooling towers, as a crucial component of indirect dry cooling systems, have a cooling performance that significantly impacts the safe and economical operation of steam power units. The circulating cooling water mass flow rate is a key parameter in optimizing the operation of steam power units. Currently, most studies use analytical models of natural draft dry cooling towers and condensers for thermal analysis, optimizing the circulating cooling water mass flow rate of indirect dry cooling systems through iterative calculation methods. The air heat flow characteristics within the natural draft dry cooling towers are simulated and calculated using commercial computational fluid dynamics software. However, the use of commercial software often requires simplification of complex structures due to computational resource and functionality limitations. Most studies only consider the impact of cooling tower structure on air heat flow characteristics, with less consideration given to factors such as the surrounding terrain and plant structure.

[0003] Artificial neural networks are a data processing model based on biological neural networks. Using a collection of valid training data, they establish a nonlinear relationship between input and output. Their goal is to predict a system's expected output based on limited measured data. In theory, artificial neural networks can overcome the limitations of commercial computational fluid dynamics software, which requires a physical model of the system for simulation calculations. However, further research is needed to determine how to use artificial neural networks to predict the circulating cooling water flow rate of indirect dry cooling systems in a manner that more closely reflects their actual operation. Summary of the Invention

[0004] To this end, the technical problem to be solved by the present invention is to provide a method and electronic equipment for predicting the circulating cooling water flow of an indirect dry cooling system, so as to solve the problem in the prior art that, when predicting the air mass flow of the cooling tower, the influence of factors such as the terrain conditions around the thermal power unit, the plant and the cooling tower structure is not taken into consideration enough, resulting in poor accuracy of the prediction results.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A method for predicting circulating cooling water flow in an indirect dry cooling system comprises the following steps:

[0007] Step A: Optimize the BP neural network model using genetic algorithm to obtain the GA-BP model;

[0008] Step B: using the GA-BP model to predict the outlet temperature of the circulating cooling water and the mass flow rate of the circulating cooling water;

[0009] Step C: Input the circulating cooling water outlet temperature and mass flow rate into the condenser heat transfer model and the air-cooled heat exchanger heat transfer model, and output the air outlet temperature and inlet air mass flow rate. The condenser heat transfer model is used to determine the amount of heat absorbed by the cooling water in the condenser, which varies with load.

[0010] Step D: Obtain the optimal value of the circulating cooling water mass flow rate under variable output load.

[0011] This paper proposes a method for optimizing the mass flow rate of circulating cooling water under off-design operating conditions and variable output loads by coupling a GA-BP neural network model with the heat transfer models of the condenser and air-cooled heat exchanger for an indirect dry cooling system in a thermal power plant. The coupling algorithm is divided into three parts: (1) the nonlinear relationship of the heat transfer performance of the indirect dry cooling system is obtained by using the BP neural network optimized by the genetic algorithm; (2) the mass flow rate of air entering the cooling tower corresponding to the load and ambient temperature is predicted by coupling the BP neural network model optimized by the genetic algorithm with the heat transfer models of the condenser and air-cooled heat exchanger; and (3) the optimal value of the mass flow rate of circulating cooling water under variable output loads is calculated using the heat transfer models of the condenser and air-cooled heat exchanger under a certain turbine back pressure.

[0012] In the above-mentioned method for predicting the circulating cooling water flow rate of the indirect dry cooling system, in step A, the construction of the GA-BP model includes the following steps:

[0013] Step A-1: ​​input data and process data;

[0014] Step A-2, determining the BP neural network topology;

[0015] Step A-3, determining the weight and threshold number;

[0016] Step A-4, using genetic algorithm to optimize the weights and thresholds of the BP neural network;

[0017] Step A-5, determining the optimal weights and thresholds of the BP neural network;

[0018] Step A-6: Perform BP neural network training, and obtain the GA-BP model after the training is completed.

[0019] In the above-mentioned method for predicting circulating cooling water flow of indirect dry cooling system, in step A-2, the topological structure of BP neural network includes three layers: input layer, hidden layer and output layer; the input layer includes unit load N d , turbine back pressure p b , circulating cooling water inlet tower temperature Twi and ambient temperature T am Four nodes; the output layer includes the circulating cooling water outlet temperature T wo and circulating cooling water mass flow Two nodes; the hidden layer consists of 8 nodes.

[0020] In the above-mentioned method for predicting the circulating cooling water flow of the indirect dry cooling system, the hidden layer activation function selects the tansig function, and the output layer activation function selects the purelin function, which are given by formula (1) and formula (2) respectively:

[0021]

[0022]

[0023] In formula (1) and formula (2), x i is the input value of the input layer node, x j is the input value of the hidden layer node, y k is the net output value, and b is the threshold.

[0024] In the above-mentioned method for predicting the circulating cooling water flow rate of the indirect dry cooling system, in step A-6, during the BP neural network training process, the update calculation formulas for the connection weights and thresholds of the hidden layer and the output layer are respectively:

[0025]

[0026] ω jk =ω jk +rx j e k (4);

[0027]

[0028] b k =b k +e k (6);

[0029] In the formula, r is the learning rate, which is 0.1;

[0030] During the BP neural network training process, two statistical indicators, the correlation coefficient R and the mean absolute percentage error MAPE, are used to evaluate the accuracy of the model prediction results; the calculation formula of the correlation coefficient R is:

[0031]

[0032] The calculation formula for the mean absolute percentage error MAPE is:

[0033]

[0034] In the formula, N is the number of test data, y k and Collect the real value and average value on site respectively. and are the model predicted value and the mean value, respectively.

[0035] In the above-mentioned method for predicting the circulating cooling water flow rate of an indirect dry cooling system, in step A-4, optimizing the weights and thresholds of the BP neural network using a genetic algorithm includes the following steps:

[0036] Step A-41, initializing weights and thresholds;

[0037] Step A-42: using the error calculated by the BP neural network model as the fitness value;

[0038] Step A-43, finding the individual corresponding to the optimal fitness value through selection, lengthening and mutation operations; the selection function adopts the roulette method based on fitness size, the crossover function adopts the real number crossover method, and the mutation operator adopts uniform mutation;

[0039] Step A-44: Calculate the fitness value and use the parameters of the optimal individual as the optimal weight and threshold of the BP neural network.

[0040] In the above-mentioned method for predicting the circulating cooling water flow rate of the indirect dry cooling system, in step A-42, the sum of the absolute values ​​of the errors between the prediction result and the expected result of the BP neural network model is used as the individual fitness value F. The calculation formula of the F value is:

[0041]

[0042] In formula (9), c is the coefficient;

[0043] The population size of the genetic algorithm is set to 100, the termination evolution generation is set to 1000, the crossover probability is 0.7, and the mutation probability is 0.05.

[0044] In the above-mentioned method for predicting the circulating cooling water flow rate of the indirect dry cooling system, in step C, the heat release Q of the circulating cooling water flowing through the air-cooled heat exchanger is calculated using the condenser heat transfer model and the air-cooled heat exchanger heat transfer model. w , the amount of heat absorbed by the cold air passing through the air-cooled heat exchanger Q a The circulating cooling water transfers heat Q to the cold air through the air-cooled heat exchanger ht ; When the following two conditions are met at the same time, the output air outlet temperature T ao and the mass flow rate of the air entering the tower

[0045] |Q w -Q a | / Qa <0.01%; and |Q ht -Q w | / Q w <0.01%;

[0046] When satisfied Under the condition, the output circulating cooling water mass flow rate Circulating cooling water inlet tower temperature T wi , circulating cooling water outlet temperature T wo and air outlet temperature T ao .

[0047] In the above-mentioned indirect dry cooling system circulating cooling water flow prediction method, the circulating cooling water transfers heat Q to the cold air through the air-cooled heat exchanger. ht for:

[0048] Q ht =K e A e△ T av (10);

[0049] In formula (10), K e is the total heat transfer coefficient, W / (m 2 K); A e is the heat transfer surface area on the air side of the air-cooled heat exchanger, m 2 ; ΔT av is the average heat transfer temperature difference, K;

[0050] Air-side heat transfer surface area A of air-cooled heat exchanger e It is equal to the sum of the fin area and the exposed tube surface area. The calculation formula is derived as follows:

[0051] A e =L t (A t +A f ) (11);

[0052] A t =πd o (1-δ f N f )N t (12);

[0053]

[0054] In formula (11), formula (12) and formula (13), A t is the surface area of ​​the base tube without fins per unit length of the air-cooled heat exchanger, m 2 / m;A f is the surface area per unit length of the air-cooled heat exchanger fin, m 2 / m;Lt is the base tube length, m; d o is the outer diameter of the base tube, m; δ f is the fin thickness, m; N f is the number of fins per unit length of the air-cooled heat exchanger, fins / m; N t is the number of base tubes of air-cooled heat exchanger, root;

[0055] In formula (10), the average heat transfer temperature difference ΔT av The calculation formula is:

[0056]

[0057] In formula (14), is the conversion coefficient of the cross flow of the two processes; T wi 、T wo are the inlet and outlet temperatures of the circulating cooling water, K; T ao is the air outlet temperature, K; T am is the ambient temperature, K;

[0058] In formula (10), the total heat transfer coefficient K e The calculation formula is:

[0059]

[0060] In formula (17), k w is the convection heat transfer coefficient between the circulating cooling water and the inner wall of the base tube, W / (m 2 ·K); k a is the convection heat transfer coefficient between air and heat exchanger wall, W / (m 2 ·K); β is the wing ratio; N t is the number of base tubes, root; λ t is the thermal conductivity of the base tube material, W / (m·K); η0 is the total efficiency of the fin wall, %; C ash is the dust accumulation coefficient assuming that dust accumulates only on the upper surface of the fin; δ ash is the thickness of dust deposited evenly on the outer surface of the base tube and the upper surface of the fin, m; λ ash is the thermal conductivity of dust accumulation, W / (m·K);

[0061] The calculation formula of wing ratio β is:

[0062]

[0063] In formula (18), A in is the inner surface area of ​​the base tube, m 2 ;

[0064] Convection heat transfer coefficient k between circulating cooling water and the inner wall of the base tube wDetermined by the Dittus-Boelter formula:

[0065]

[0066] In formula (19), λ w is the thermal conductivity of circulating cooling water, W / (m·K); Re di is the Reynolds number; Pr is the Prandtl number; Re di ≥10 4 , Pr=0.6-160,L t / d i ≥10;

[0067] Convective heat transfer coefficient k between air and heat exchanger wall a The calculation formula is:

[0068]

[0069] In formula (20), λ a is the thermal conductivity of air, W / (m·K); d eq is the equivalent diameter of the air passing across the fin side, m; is the air mass flow rate at the minimum cross section when air flows through the finned tube, kg / (m 2 ·s); μ a is the dynamic viscosity of air, Pa·s;

[0070] Equivalent diameter d of the air passing across the fin side eq The calculation formula is:

[0071]

[0072] The calculation formula for the total fin wall efficiency is:

[0073]

[0074] In formula (22), η f is the fin efficiency, which is the ratio of the actual heat transfer of the fin surface to the heat transfer when the fin surface is assumed to be on the base tube surface, and is taken as 0.933;

[0075] The heat released by the circulating cooling water flowing through the air-cooled heat exchanger Q w for:

[0076]

[0077] The amount of heat absorbed by the cold air passing through the air-cooled heat exchanger Q a for:

[0078]

[0079] In formula (15) and formula (16), and are the mass flow rates of circulating cooling water and air entering the tower, kg / s; c w and c a are the specific heat capacities of circulating cooling water and air, J / (kg·K), respectively.

[0080] An electronic device for predicting the circulating cooling water flow of an indirect dry cooling system. The electronic device stores a computer program, which implements the above-mentioned method for predicting the circulating cooling water flow of an indirect dry cooling system when executed by a processor.

[0081] The technical solution of the present invention achieves the following beneficial technical effects:

[0082] This method uses a BP neural network optimized by a genetic algorithm, replacing commercial computational fluid dynamics software. By collecting extensive data from actual chiller operation, it predicts the air mass flow rate entering the cooling tower. The prediction process fully considers the impact of factors such as the chiller's surrounding terrain, the plant building, and the cooling tower structure on the air mass flow rate entering the cooling tower, resulting in a prediction that more closely approximates the actual operation of an indirect dry cooling system. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 A schematic structural diagram of an indirect dry cooling system for a unit according to an embodiment of the present invention;

[0084] Figure 2 A schematic structural diagram of a heat exchanger in an embodiment of the present invention;

[0085] Figure 3 A schematic diagram of the topological structure of a BP neural network in an embodiment of the present invention;

[0086] Figure 4 The computational block diagram of the genetic algorithm optimizing the BP neural network in the embodiment of the present invention;

[0087] Figure 5 A comparison chart of the results of the circulating cooling water outlet tower temperature prediction using the GA-BP model in the embodiment of the present invention and the actual value;

[0088] Figure 6 A comparison chart of the results of the circulating cooling water mass flow rate prediction using the GA-BP model in the embodiment of the present invention and the actual value;

[0089] Figure 7 Schematic diagram of the coupling mode of the GA-BP model with the condenser and air-cooled heat exchanger heat transfer models in an embodiment of the present invention;

[0090] Figure 8 Calculation flow chart of the heat transfer model in an embodiment of the present invention. DETAILED DESCRIPTION

[0091] This embodiment takes the indirect dry cooling system of a 350MW coal-fired unit in Northwest China as an example to illustrate the indirect dry cooling system circulating cooling water flow prediction method - coupling algorithm proposed by the present invention. Figure 1 As shown, it consists of a surface condenser and a natural draft dry cooling tower equipped with a vertically arranged air-cooled heat exchanger. The air-cooled heat exchanger uses the sixth generation Forgo heat exchanger, which is an all-aluminum round tube finned tube heat exchanger. The heat exchanger structure is as follows: Figure 2 The unit is equipped with three circulating water pumps with a total rated flow of 37,020 t / h. The flow can be adjusted during operation using variable frequency speed regulation technology.

[0092] 1 Method

[0093] 1.1 BP neural network based on genetic algorithm optimization

[0094] 1.1.1 Model

[0095] This embodiment adopts a three-layer BP neural network, the topology is as follows Figure 3 As shown. The input layer has four nodes, including unit load (N d ), turbine back pressure (p b ), circulating cooling water inlet tower temperature (T wi ), ambient temperature (T am The output layer has two nodes, including the circulating cooling water outlet tower temperature (T wo ) and circulating cooling water mass flow

[0096] The hidden layer activation function is the tansig function, and the output layer activation function is the purelin function, which are given by equations (1) and (2).

[0097]

[0098]

[0099] Where x i 、x j are the input values ​​of the input layer and hidden layer nodes, y k is the net output value, and b is the threshold.

[0100] Since the Levenberg-Marquardt (LM) back propagation algorithm has the fastest convergence speed and a smaller mean square error, the tranilm algorithm is selected as the training function in this embodiment.

[0101] During network training, the update formula for the connection weights and thresholds of the hidden layer and the output layer is:

[0102]

[0103] ω jk =ω jk +rx j e k (4);

[0104]

[0105] b k =b k +e k (6);

[0106] Where r is the learning rate, which is set to 0.1 in this embodiment.

[0107] The correlation coefficient (R) and mean absolute percentage error (MAPE) are two statistical indicators used to evaluate the accuracy of the model prediction results. R is a measure of the deviation between the predicted value and the test value. The R value between the test value and the predicted value is defined as:

[0108]

[0109] The mean absolute percentage error (MAPE) is calculated as:

[0110]

[0111] Where N is the number of test data, y k and Collect the real value and average value on site respectively. and are the model predicted value and the mean value, respectively.

[0112] In this embodiment, the neural network model selects one hidden layer. Comparing the prediction results in Table 1, when the number of hidden layer nodes is 8, the training result is closer to the actual value.

[0113] Table 1

[0114]

[0115] This embodiment introduces a genetic algorithm (GA) to optimize the initial weights and thresholds of the BP neural network, improving the stability and efficiency of searching for the optimal global solution. The BP neural network structure in this embodiment is 4-8-2, meaning the input layer has 4 nodes, the hidden layer has 8 nodes, and the output layer has 2 nodes. With a total of 48 nodes and 10 thresholds, the genetic algorithm optimization parameter is 58.

[0116] The connection weights and thresholds in the initial population are assigned to the BP neural network. After training the BP network, the system output is predicted. The sum of the absolute values ​​of the errors between the predicted results and the expected results is taken as the individual fitness value F, which is calculated as follows:

[0117]

[0118] In formula (9), c is a coefficient.

[0119] In the genetic algorithm, the selection function uses a roulette wheel method based on fitness scale, the crossover function uses real-valued crossover, and the mutation operator uses uniform mutation. The genetic algorithm uses selection, crossover, and mutation to find the individual with the optimal fitness value. The parameters of the optimal individual are used as the initial weights and thresholds for the BP neural network, which is then trained to obtain prediction results.

[0120] The calculation block diagram of genetic algorithm optimization BP neural network is as follows Figure 4 The calculation block diagram is divided into three parts: BP neural network structure determination, genetic algorithm optimization and BP neural network prediction.

[0121] The GA-BP model was constructed using MATLAB 2021. The BP neural network was established using the newff function, and the training function was trainlm. The maximum number of training cycles was 1000, and the convergence error was 0.001. To ensure computational accuracy and accelerate convergence, the genetic algorithm population size was set to 100, the number of termination generations was set to 1000, the crossover probability was set to 0.7, and the mutation probability was set to 0.05.

[0122] 1.1.2 Data Collection Methods and Model Validation

[0123] Considering the delayed nature of heat transfer in air-cooled heat exchangers, it's impossible to optimize the circulating cooling water mass flow rate based on real-time wind speed fluctuations during actual operation. Therefore, the unit operating data collected on-site is for a specific time period (usually greater than 0.5 hours) when the unit load, ambient temperature, circulating cooling water inlet and outlet temperatures, and circulating cooling water mass flow rate remain constant and the ambient wind speed is near zero.

[0124] The GA-BP model is used to predict the temperature and mass flow of circulating cooling water out of the tower. The comparison between the predicted results and the actual values ​​is shown in the following figure: Figure 5 and Figure 6 As shown in the figure, the prediction results of the GA-BP model are very close to the field data. wo The R and MAPE were 0.997 and 0.017 respectively; m w The R and MAPE are 0.981 and 0.036 respectively, indicating that the prediction results obtained by the GA-BP model constructed in this embodiment are credible.

[0125] 1.2 Method for determining air mass flow

[0126] 1.2.1 Heat transfer model of air-cooled heat exchanger

[0127] The circulating cooling water transfers heat Q to the cold air through the air-cooled heat exchanger ht for:

[0128] Q ht =K e A e△ T av (10);

[0129] In formula (10), K e is the total heat transfer coefficient, W / (m 2 K); A e is the heat transfer surface area on the air side of the air-cooled heat exchanger, m 2 ; ΔT av is the average heat transfer temperature difference, K.

[0130] Air-side heat transfer surface area A of air-cooled heat exchanger e It is equal to the sum of the fin area and the exposed tube surface area. The calculation formula is derived as follows:

[0131] A e =L t (A t +A f ) (11);

[0132] A t =πd o (1-δ f N f )N t (12);

[0133]

[0134] Where A t is the surface area of ​​the base tube without fins per unit length of the air-cooled heat exchanger, m 2 / m;A f is the surface area per unit length of the air-cooled heat exchanger fin, m 2 / m;L t is the base tube length, m; d o is the outer diameter of the base tube, m; δ f is the fin thickness, m; N f is the number of fins per unit length of the air-cooled heat exchanger, fins / m; N t is the number of base tubes of the air-cooled heat exchanger, roots.

[0135] The flow between the circulating cooling water and air in the tube bundle of the air-cooled heat exchanger belongs to a two-process cross flow, so the calculation formula for the average heat transfer temperature difference is:

[0136]

[0137] Where, is the conversion coefficient of the cross flow of the two processes; T wi 、T wo are the inlet and outlet temperatures of the circulating cooling water, K; T ao is the air outlet temperature, K; T am is the ambient temperature, K.

[0138] The heat released by the circulating cooling water flowing through the air-cooled heat exchanger Q w for:

[0139]

[0140] The amount of heat absorbed by the cold air passing through the air-cooled heat exchanger Q a for:

[0141]

[0142] Where, and are the mass flow rates of circulating cooling water and air entering the tower, kg / s; c w and c a are the specific heat capacities of circulating cooling water and air, J / (kg·K), respectively.

[0143] The calculation formula for the total heat transfer coefficient based on the heat transfer surface area on the air side of the air-cooled heat exchanger is:

[0144]

[0145] Where k w is the convection heat transfer coefficient between the circulating cooling water and the inner wall of the base tube, W / (m 2 ·K); k a is the convection heat transfer coefficient between air and heat exchanger wall, W / (m 2 ·K); β is the wing ratio; N t is the number of base tubes, root; λ t is the thermal conductivity of the base tube material, W / (m·K); η0 is the total efficiency of the fin wall, %; C ash is the dust accumulation coefficient assuming that dust accumulates only on the upper surface of the fin; δ ash is the thickness of dust deposited evenly on the outer surface of the base tube and the upper surface of the fin, m; λ ash is the thermal conductivity of dust accumulation, W / (m·K).

[0146] The calculation formula of wing ratio β is:

[0147]

[0148] Where A in is the inner surface area of ​​the base tube, m 2 .

[0149] The heat transfer between the circulating cooling water and the wall surface in the base tube of the air-cooled heat exchanger belongs to forced convection heat transfer in the tube. The heat transfer coefficient can be determined by the Dittus-Boelter formula:

[0150]

[0151] Where λ w is the thermal conductivity of circulating cooling water, W / (m·K); Re di is the Reynolds number; Pr is the Prandtl number. The applicable range of this formula is Re di ≥10 4 , Pr=0.6-160,L t / d i ≥10.

[0152] The air-cooled heat exchanger is a circular finned tube heat exchanger. The tube bundles in the air flow direction are staggered. Therefore, the convection heat transfer coefficient between the air and the heat exchanger wall can be calculated using the heat transfer coefficient formula for air flowing horizontally through the transverse rectangular finned tube bundle outside the staggered circular tubes:

[0153]

[0154] Where λ a is the thermal conductivity of air, W / (m·K); d eq is the equivalent diameter of the air passing across the fin side, m; is the air mass flow rate at the minimum cross section when air flows through the finned tube, kg / (m 2 ·s); μ a is the dynamic viscosity of air, Pa·s.

[0155] Equivalent diameter d of the air passing across the fin side eq The calculation formula is:

[0156]

[0157] The calculation formula for the total fin wall efficiency is:

[0158]

[0159] Where η f is the fin efficiency, which is the ratio of the actual heat transfer of the fin surface to the heat transfer when the fin surface is assumed to be on the base tube surface, and is taken as 0.933.

[0160] 1.2.2 Condenser heat transfer model

[0161] Under stable operating conditions, the heat released by the circulating cooling water in the cooling tower is equal to the heat absorbed in the condenser, which can be obtained through the turbine exhaust parameters and the circulating cooling water parameters. The heat released by the turbine exhaust in the condenser is:

[0162] Q c =D c (h c -h cw ) (twenty three);

[0163] Where D c is the mass flow rate of steam exhaust from the turbine, kg / s; h c is the exhaust enthalpy of the steam turbine, J / kg; h cw is the enthalpy of condensed water, which can be considered equal to the saturated water enthalpy under the unit back pressure condition, J / kg.

[0164] The amount of heat absorbed by the circulating cooling water in the condenser is:

[0165]

[0166] In the condenser, the heat transfer between the circulating cooling water and the turbine exhaust steam through the wall is:

[0167]

[0168] Where K c is the heat transfer coefficient of the condenser, which can be calculated according to the HEI standard, W / (m 2 K); A c is the heat transfer area of ​​the condenser, m 2 ;T s is the saturation temperature of the turbine exhaust, ℃; T ci and T co is the temperature of the circulating cooling water entering and leaving the condenser, °C. Ignoring the heat dissipation in the pipes, T ci It can be considered to be equal to the circulating cooling water outlet temperature T wo , T co Equal to the circulating cooling water inlet tower temperature T wi .

[0169] 1.2.3 Method for determining air mass flow

[0170] This embodiment establishes a coupling algorithm between the GA-BP model and the condenser and air-cooled heat exchanger heat transfer model to obtain the air mass flow rate entering the cooling tower. The specific steps are as follows: (1) Collecting the unit operation data; (2) Using the unit output load, turbine back pressure, circulating cooling water inlet tower temperature, and ambient temperature as input parameters, the GA-BP model is used to predict the circulating cooling water outlet tower temperature and mass flow rate; (3) Using the predicted circulating cooling water outlet tower temperature, mass flow rate, circulating cooling water inlet temperature, and ambient temperature as known data, the condenser and air-cooled heat exchanger heat transfer model is used to calculate the air mass flow rate entering the cooling tower and the air outlet tower temperature. The coupling method of the GA-BP model with the condenser and air-cooled heat exchanger heat transfer model is as follows: Figure 7 shown.

[0171] 1.2.4 Model Validity Verification

[0172] The calculation results of the GA-BP model developed in this example, coupled with the condenser and air-cooled heat exchanger heat transfer models, were compared with field data. The comparison results are shown in Table 2. The unit output load, unit backpressure, circulating cooling water mass flow rate, circulating cooling water inlet and outlet temperatures, ambient temperature, and wind speed are the unit's existing monitoring data, while the air outlet temperature is obtained from temporarily installed temperature measurement points. It can be seen that the circulating cooling water outlet temperature and air outlet temperature calculated by the coupled algorithm developed in this example are highly accurate, with an average relative error of 2.88%.

[0173] Table 2 Comparison of coupling algorithm calculation results and field acquisition results

[0174]

[0175] 1.3 Optimization method of circulating cooling water mass flow

[0176] In the optimization calculation of circulating cooling water mass flow, in order to ensure that the back pressure of the unit is close to the design back pressure, the minimum value of the circulating cooling water inlet tower temperature T is set. wi,min (i.e. the minimum value of the circulating cooling water outlet condenser temperature T co,min ); Considering the antifreeze requirements of the air-cooled heat exchanger in winter, the minimum temperature of the circulating cooling water out of the tower is set to T wo,min (i.e. the minimum value of the circulating cooling water inlet condenser temperature T ci,min ).

[0177] 2 Optimization results analysis

[0178] According to the above coupling algorithm, Table 3 gives the T wo,min =20℃、T wi,min =40℃, the recommended value of the circulating cooling water mass flow rate corresponding to the ambient temperature and the corresponding circulating cooling water inlet and outlet temperatures.

[0179] Table 3 Recommended values ​​of circulating cooling water mass flow corresponding to ambient temperature

[0180]

[0181] Table 4 Ambient temperature range and corresponding mass flow rate of circulating cooling water mass flow rate

[0182]

[0183] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the claims of this patent application.

Claims

1. A method for predicting circulating cooling water flow in an indirect dry cooling system, characterized in that: The steps include: Step A: Optimize the BP neural network model using genetic algorithm to obtain the GA-BP model; Step B: using the GA-BP model to predict the outlet temperature of the circulating cooling water and the mass flow rate of the circulating cooling water; Step C: Input the outlet temperature of the circulating cooling water and the mass flow rate of the circulating cooling water into the condenser heat transfer model and the air-cooled heat exchanger heat transfer model, and output the air outlet temperature and the mass flow rate of the air entering the tower; In step C, the heat release Q of the circulating cooling water flowing through the air-cooled heat exchanger is calculated using the condenser heat transfer model and the air-cooled heat exchanger heat transfer model. w , the amount of heat absorbed by the cold air passing through the air-cooled heat exchanger Q a The circulating cooling water transfers heat Q to the cold air through the air-cooled heat exchanger ht ; When the following two conditions are met at the same time, the output air outlet temperature T ao and the mass flow rate of the air entering the tower |Q w -Q a | / Q a <0.01%; and |Q ht -Q w | / Q w <0.01%; When satisfied Under the condition, the output circulating cooling water mass flow rate Circulating cooling water inlet tower temperature T wi , circulating cooling water outlet temperature T wo and air outlet temperature T ao ;m wr Refers to the rated mass flow rate of the circulation pump; Step D: Obtain the optimal value of the circulating cooling water mass flow rate under variable output load; To ensure that the unit back pressure is close to the design back pressure, set the minimum value of the circulating cooling water inlet tower temperature T wi,min Considering the antifreeze requirements of the air-cooled heat exchanger in winter, the minimum temperature of the circulating cooling water out of the tower is set to T wo,min .

2. The method for predicting circulating cooling water flow of an indirect dry cooling system according to claim 1, characterized in that: In step A, the construction of the GA-BP model includes the following steps: Step A-1: ​​input data and process data; Step A-2, determining the BP neural network topology; Step A-3, determining the weight and threshold number; Step A-4, using genetic algorithm to optimize the weights and thresholds of the BP neural network; Step A-5, determining the optimal weights and thresholds of the BP neural network; Step A-6: Perform BP neural network training, and obtain the GA-BP model after the training is completed.

3. The method for predicting circulating cooling water flow of an indirect dry cooling system according to claim 1, characterized in that: In step A-2, the topological structure of the BP neural network includes three layers: input layer, hidden layer and output layer; the input layer includes the unit load N d , turbine back pressure p b , circulating cooling water inlet tower temperature T wi and ambient temperature T am Four nodes; the output layer includes the circulating cooling water outlet temperature T wo and circulating cooling water mass flow Two nodes; the hidden layer consists of 8 nodes.

4. The method for predicting circulating cooling water flow of an indirect dry cooling system according to claim 2, characterized in that: The hidden layer activation function selects the tansig function, and the output layer activation function selects the purelin function, which are given by formula (1) and formula (2) respectively: In formula (1) and formula (2), x i is the input value of the input layer node, x j is the input value of the hidden layer node, y k is the net output value, and b is the threshold.

5. The method for predicting circulating cooling water flow of an indirect dry cooling system according to claim 3, characterized in that: In step A-6, during the BP neural network training process, the update calculation formulas for the connection weights and thresholds of the hidden layer and the output layer are: oh jk =ω jk +rx j e k (4); b k =b k +e k (6); In the formula, r is the learning rate, which is 0.1; During the BP neural network training process, two statistical indicators, the correlation coefficient R and the mean absolute percentage error MAPE, are used to evaluate the accuracy of the model prediction results; the calculation formula of the correlation coefficient R is: The calculation formula for the mean absolute percentage error MAPE is: In the formula, N is the number of test data, y k and Collect the real value and average value on site respectively. and are the model predicted value and the mean value, respectively.

6. The method for predicting circulating cooling water flow in an indirect dry cooling system according to claim 2, characterized in that: In step A-4, optimizing the weights and thresholds of the BP neural network using a genetic algorithm includes the following steps: Step A-41, initializing weights and thresholds; Step A-42: using the error calculated by the BP neural network model as the fitness value; Step A-43, finding the individual corresponding to the optimal fitness value through selection, lengthening and mutation operations; the selection function adopts the roulette method based on fitness size, the crossover function adopts the real number crossover method, and the mutation operator adopts uniform mutation; Step A-44: Calculate the fitness value and use the parameters of the optimal individual as the optimal weight and threshold of the BP neural network.

7. The method for predicting circulating cooling water flow in an indirect dry cooling system according to claim 6, characterized in that: In step A-42, the sum of the absolute values ​​of the errors between the prediction result and the expected result of the BP neural network model is used as the individual fitness value F. The calculation formula of the F value is: In formula (9), c is the coefficient; The population size of the genetic algorithm is set to 100, the termination evolution generation is set to 1000, the crossover probability is 0.7, and the mutation probability is 0.

05.

8. The method for predicting circulating cooling water flow in an indirect dry cooling system according to claim 1, characterized in that: The circulating cooling water transfers heat Q to the cold air through the air-cooled heat exchanger ht for: Q ht =K e A e△ T av (10); In formula (10), K e is the total heat transfer coefficient, W / (m 2 K); A e is the heat transfer surface area on the air side of the air-cooled heat exchanger, m 2 ; ΔT av is the average heat transfer temperature difference, K; Air-side heat transfer surface area A of air-cooled heat exchanger e It is equal to the sum of the fin area and the exposed tube surface area. The calculation formula is derived as follows: A e =L t (A t +A f ) (11); A t =πd o (1-d f N f )N t (12); In formula (11), formula (12) and formula (13), A t is the surface area of ​​the base tube without fins per unit length of the air-cooled heat exchanger, m 2 / m;A f is the surface area per unit length of the air-cooled heat exchanger fin, m 2 / m;L t is the base tube length, m; d o is the outer diameter of the base tube, m; δ f is the fin thickness, m; N f is the number of fins per unit length of the air-cooled heat exchanger, fins / m; N t is the number of base tubes of air-cooled heat exchanger, root; In formula (10), the average heat transfer temperature difference ΔT av The calculation formula is: In formula (14), is the conversion coefficient of the cross flow of the two processes; T wi 、T wo are the inlet and outlet temperatures of the circulating cooling water, K; T ao is the air outlet temperature, K; T am is the ambient temperature, K; In formula (10), the total heat transfer coefficient K e The calculation formula is: In formula (17), k w is the convection heat transfer coefficient between the circulating cooling water and the inner wall of the base tube, W / (m 2 ·K); k a is the convection heat transfer coefficient between air and heat exchanger wall, W / (m 2 ·K); β is the wing ratio; N t is the number of base tubes, roots; λ t is the thermal conductivity of the base tube material, W / (m·K); η0 is the total efficiency of the fin wall, %; C ash is the dust accumulation coefficient assuming that dust accumulates only on the upper surface of the fin; δ ash is the thickness of dust deposited evenly on the outer surface of the base tube and the upper surface of the fin, m; λ ash is the thermal conductivity of dust accumulation, W / (m·K); The calculation formula of wing ratio β is: In formula (18), A in is the inner surface area of ​​the base tube, m 2 ; Convection heat transfer coefficient k between circulating cooling water and the inner wall of the base tube w Determined by the Dittus-Boelter formula: In formula (19), λ w is the thermal conductivity of circulating cooling water, W / (m·K); is the Reynolds number; Pr is the Prandtl number; Pr=0.6-160,L t / d i ≥10; Convective heat transfer coefficient k between air and heat exchanger wall a The calculation formula is: In formula (20), λ a is the thermal conductivity of air, W / (m·K); d eq is the equivalent diameter of the air passing across the fin side, m; is the air mass flow rate at the minimum cross section when air flows through the finned tube, kg / (m 2 ·s); μ a is the dynamic viscosity of air, Pa·s; Equivalent diameter d of the air passing across the fin side eq The calculation formula is: The calculation formula for the total fin wall efficiency is: In formula (22), η f is the fin efficiency, which is the ratio of the actual heat transfer of the fin surface to the heat transfer when the fin surface is assumed to be on the base tube surface, and is taken as 0.933; The heat released by the circulating cooling water flowing through the air-cooled heat exchanger Q w for: The amount of heat absorbed by the cold air passing through the air-cooled heat exchanger Q a for: In formula (15) and formula (16), and are the mass flow rates of circulating cooling water and air entering the tower, kg / s; c w and c a are the specific heat capacities of circulating cooling water and air, J / (kg·K), respectively.

9. An electronic device for predicting the circulating cooling water flow rate of an indirect dry cooling system, characterized in that: The electronic device stores a computer program, and when the computer program is executed by the processor, the method for predicting the circulating cooling water flow rate of the indirect dry cooling system according to any one of claims 1 to 8 is implemented.

Images