Heat transfer performance on-line monitoring method and system for air cooling unit of thermal power generating unit

Abstract

The present application relates to the heat transfer performance on-line monitoring method of air cooling unit of thermal power generating unit, total flow is divided into two parts, the first part is data driven model training part, including obtaining historical data from data acquisition module, after heat transfer coefficient calculation, normalization, data driven model training is carried out;The second part is air cooling island heat transfer performance monitoring part, real-time data is obtained from data acquisition module, after time sliding window, normalizer, the heat transfer coefficient of each unit is calculated through data driven model, and then the visualization of heat transfer coefficient is carried out.Compared with the traditional calculation mode, the monitoring method based on data driving can comprehensively, accurately and timely monitor the heat transfer performance of each unit of air cooling island, and the heat transfer coefficient of each unit is calculated in real time, which provides important basis for formulating and implementing winter antifreeze measures and subsequent air cooling island fan economic optimization control of thermal power plant.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of online detection of air cooling systems, and particularly relates to a heat transfer performance online monitoring method and system for air cooling units of thermal power generating units. BACKGROUND

[0002] Real-time monitoring of the heat transfer performance of each unit in the air cooling island of the thermal power generating unit provides an important basis for the economic control of the axial flow fan of the air cooling island of the thermal power plant and the winter anti-freezing measures. The heat transfer performance is usually represented by the heat transfer coefficient, which can reflect the condensation effect of the air cooling condenser on water vapor. The air cooling island device is arranged in the open air, and each air cooling unit is easily affected by the flow of natural wind, and due to the change of the unit load, different condensation effect requirements are required for the air cooling island, and part of the axial flow fan is adjusted to stop or slow down, which may cause a significant difference in the temperature between the units of the air cooling island. In addition, the finned tube may freeze in the winter extremely cold temperature. Therefore, monitoring the heat transfer performance of the air cooling island is related to the safety and economy of the air cooling island.

[0003] The air cooling island is a complex thermodynamic system, and the condensation effect is usually calculated by a series of thermodynamic balance mechanism formulas to represent the total heat transfer coefficient of the air cooling island. The condensation effect of each unit of the air cooling island is affected by the unit operation process and environmental changes, and most of the calculation methods only calculate the total heat transfer coefficient of the air cooling island under a certain working condition, and cannot provide the heat transfer coefficient value of the current operation of the air cooling island in time.

[0004] There is a need in the market today for a method that can comprehensively, accurately and timely monitor the heat transfer performance of each unit of the air cooling island. SUMMARY

[0005] The purpose of the present application is to at least solve one of the deficiencies of the prior art, and provide a heat transfer performance online monitoring method and system for air cooling units of thermal power generating units.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme:

[0007] Specifically, a heat transfer performance online monitoring method for air cooling units of thermal power generating units is proposed, which comprises the following:

[0008] The first part is the training of the data-driven model, which comprises,

[0009] obtaining an infrared thermal image of each air cooling unit of the target air cooling island, and calculating the average temperature of each air cooling unit according to the infrared thermal image;

[0010] obtaining data as input features of data-driven and data for calculating the heat transfer coefficient through the DCS system;

[0011] The average temperature and the data obtained by the DCS system are taken as an input data set, and the data-driven model is trained by using the input data set to obtain a final data-driven model;

[0012] The second part, the heat transfer performance monitoring of the target air cooling island, comprises,

[0013] The infrared thermal image of the target air cooling island and the related data of the DCS system are obtained in real time and taken as an input data set, and the data of the input data set are normalized at the same time;

[0014] The normalized data are input into a time sliding window, and the time sliding window is taken as a real-time input sample and input into the data-driven model, and the model calculates the heat transfer coefficients of each unit in the thermal power generating unit.

[0015] Further, the method further comprises, after the model calculates the heat transfer coefficients of each unit in the thermal power generating unit, a heat map is formed according to the arrangement mode of the air cooling units in the thermal power generating unit to realize real-time visual display.

[0016] Further, specifically, the infrared thermal image of each air cooling unit of the target thermal power generating unit is obtained, and the average temperature of each air cooling unit is calculated according to the infrared thermal image, comprising,

[0017] The infrared thermal camera is installed on the leeward surface of each unit of the air cooling island, the infrared thermal image of each finned tube bundle is shot, and the obtained real-time finned tube bundle infrared image is transmitted to the back end for storage and calculation of the average temperature of the leeward surface.

[0018] Further, specifically, the data used as the input features of data driving and the data used for calculating the heat transfer coefficients are obtained by the DCS system, comprising,

[0019] The data obtained by the DCS system are of two kinds, one is used as the input features of data driving, comprising the air cooling island inlet steam temperature obtained at the monitoring point on the air cooling island, the outlet condensate water temperature of each unit of the air cooling island and the axial flow fan rotating speed; the other is the data used for calculating the heat transfer coefficients, comprising the exhaust steam pressure and the exhaust steam flow of the steam turbine outlet, the atmospheric temperature, i.e. the air cooling island inlet air temperature, the atmospheric pressure, and the heat transfer coefficients of each unit of the air cooling island are calculated.

[0020] Further, specifically, the calculation process of the heat transfer coefficients of each unit of the air cooling island comprises,

[0021] The water and water vapor physical properties and property standard formula IAPWS-IF97 are used to calculate the condensate water saturation temperature ts, the condensate water enthalpy hew and the steam enthalpy hew in the finned tube;

[0022] The steam condensation released heat Qn in the finned tube is calculated, and the formula used is:

[0023] Q n = G n (h ev -h ew )

[0024] In the formula, G n is the steam turbine exhaust flow rate;

[0025] The maximum temperature difference tem of the air cooling island air flow inlet and outlet is set, and the iteration interval of the air cooling island air flow outlet temperature ta2 is (ta1, ta1+tem ℃), and the interval is te, and the iteration calculation is carried out: for calculating the physical properties of air, first, the average temperature of the air cooling island air flow inlet and outlet is calculated The formula used is

[0026] The air is regarded as an ideal gas to calculate its density The formula used is

[0027] The specific heat capacity at constant pressure of air is calculated The formula used is:

[0028] The air outside the finned tube absorbs heat Qa, and the formula used is: In the formula, Ay is the air cooling island windward area, and vF is the wind speed of the windward surface;

[0029] When the heat absorbed by the air outside the finned tube is equal to the heat released by the condensation of steam inside the finned tube, that is, Qa=Qn, the output ta2 is determined, and the is the value at the current time;

[0030] The heat transfer unit number NTU of the air cooling island is calculated, and the formula used is:

[0031] The heat transfer coefficient K of the air cooling island is calculated, and the formula used is:

[0032] The heat transfer coefficient K of each unit of the air cooling island is calculated using the variable condition simplified formula ij , and the formula used is: Wherein, the subscripts i and j are the i-th row and j-th column of the air cooling unit;

[0033] The calculated heat transfer coefficient should correspond to the real time of the corresponding data.

[0034] Further, specifically, the data-driven model adopts an RNN structure neural network, normalizes the input features of the data set, sets parameters including the number of neural network layers L, the number of neurons ci (i = 1, 2, …, L) of different network layers, time step T, loss function, optimizer, and hyperparameter settings including learning rate LearnRate, batch size BatchSize, and iteration number epochs.

[0035] Further, specifically, the training process of the data-driven model includes,

[0036] According to the obtained monitoring points on the air cooling island, the inlet steam temperature of the air cooling island, the outlet condensate water temperature of each unit of the air cooling island, the rotational speed of the axial flow fan of each unit, and the average temperature of the backwind finned tube of each unit are obtained as input features, and strictly according to time as a vector X, and the input feature vector at time t is denoted as Xt; the sample output vector y is the one-dimensional arrangement of the heat transfer coefficients Kij of each unit of the air cooling island, strictly corresponding to the input feature time, denoted as yt. The input features are normalized by using the maximum and minimum normalization, and the formula is Xmax and Xmin are the maximum and minimum values of the input features, respectively. While normalizing, the maximum and minimum values of each feature in the sample set are saved as a normalizer. The input of the model at each time is the input sample of the last T time corresponding to a time t, that is, [Xt-(T-1), Xt-(T-2), …, Xt], and the output of the model is During model training, the input feature vector X and the sample output vector y are arranged in time sequence as the training sample set of the model;

[0037] The RNN adopts an LSTM neuron structure, and the calculation steps are: calculating the forgetting gate vector F t = σ(X t W xf + H t-1 W hf + b f ), calculating the input gate vector I t = σ(X t W xi + H t-1 W hi + b i ), calculating the output gate vector O t = σ(X t W xo + H t-1 W ho + b o ), calculating the temporary cell state vector calculating the cell state vector H t = O t ⊙ tanh(Ct ), wherein t is the time t, sigma is a sigmoid activation function, tanh is a tanh activation function, W represents a weight matrix for training, b represents a bias value for training, H is a hidden state,

[0038] The activation function used, wherein the formula of the sigmoid activation function is: The formula of the tanh activation function is:

[0039] The loss function adopts mean square error (MSE), and the formula is N is the number of samples, yi is the output value of the training sample, is the output value of the model;

[0040] The weight update of the neural network uses the gradient descent method:

[0041] A plurality of data-driven models are trained by setting a plurality of different neural network parameters and hyperparameters, and the training is stopped when the number of iterations is reached, and the model with the best prediction effect is selected through the validation set.

[0042] The application also proposes an online monitoring system for heat transfer performance of a thermal power unit air cooling unit, comprising the following:

[0043] The data-driven model training device comprises,

[0044] The training data acquisition module is used to acquire the infrared thermal image of each air cooling unit of the target air cooling island, calculate the average temperature of each air cooling unit according to the infrared thermal image, acquire the data serving as the input features of data driving and the data for calculating the heat transfer coefficient through the DCS system, and acquire the data serving as the input features of data driving and the data for calculating the heat transfer coefficient through the DCS system.

[0045] The model training module is used to take the average temperature and the data acquired by the DCS system as the input data set, train the data-driven model by using the input data set, and obtain the final data-driven model.

[0046] The heat transfer performance monitoring device of the target air cooling island comprises,

[0047] The real-time data acquisition module is used to acquire the infrared thermal image of the target air cooling island and the related data of the DCS system in real time as the input data set, and normalize the data of the input data set at the same time.

[0048] The heat transfer coefficient calculation module is used to input the normalized data into the time sliding window, input the time sliding window as the real-time input sample into the data-driven model, and calculate the heat transfer coefficient of each unit in the thermal power unit by the model.

[0049] Further, the system further comprises,

[0050] A result display module is configured to display the heat transfer coefficients of the units in the thermal power generating unit in real time in the form of a heat map according to the arrangement of the air cooling units in the thermal power generating unit.

[0051] The application further provides a computer readable storage medium storing a computer program, and the computer program is executed by a processor to implement the steps of the heat transfer performance online monitoring method for the air cooling units of the thermal power generating unit.

[0052] The application has the following advantages:

[0053] The heat transfer performance online monitoring method for the air cooling units of the thermal power generating unit divides the whole process into two parts, the first part is a data-driven model training part, including obtaining historical data from the data acquisition module, and training the data-driven model after heat transfer coefficient calculation and normalization; the second part is an air cooling island heat transfer performance monitoring part, obtaining real-time data from the data acquisition module, and calculating the heat transfer coefficients of the units through the data-driven model after time sliding window and normalization, and then visualizing the heat transfer coefficients. Compared with the traditional calculation method, the data-driven monitoring method can comprehensively, accurately and timely monitor the heat transfer performance of the units of the air cooling island, and calculate the heat transfer coefficients of the units in real time, thereby providing an important basis for formulating and implementing winter anti-freezing measures and subsequent air cooling island fan economic optimization control for the thermal power plant. BRIEF DESCRIPTION OF DRAWINGS

[0054] The above and other features of the present disclosure will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which like reference numerals denote like or similar elements, and wherein:

[0055] Figure 1 Fig. 1 shows a flowchart of the heat transfer performance online monitoring method for the air cooling units of the thermal power generating unit;

[0056] Figure 2 Fig. 4 shows an iterative calculation principle diagram of the air outlet air temperature of the air cooling island in the heat transfer performance online monitoring method for the air cooling units of the thermal power generating unit;

[0057] Figure 3 Fig. 5 shows an RNN network structure diagram used in the heat transfer performance online monitoring method for the air cooling units of the thermal power generating unit;

[0058] Figure 4An LSTM diagram used by the heat transfer performance online monitoring method of the air cooling unit of the thermal power generating unit is shown.

[0059] Figure 5 A time sliding window schematic diagram used by the heat transfer performance online monitoring method of the air cooling unit of the thermal power generating unit is shown.

[0060] Figure 6 An air cooling island heat transfer coefficient visual display schematic diagram of the heat transfer performance online monitoring method of the air cooling unit of the thermal power generating unit is shown. DETAILED DESCRIPTION

[0061] The concept, specific structure and generated technical effects of the present application will be described clearly and completely in combination with embodiments and drawings to fully understand the purpose, scheme and effects of the present application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The same reference signs used in the drawings indicate the same or similar parts.

[0062] REFERENCE Figure 1 , embodiment 1, the present application proposes a heat transfer performance online monitoring method of an air cooling unit of a thermal power generating unit, including the following:

[0063] The first part is data-driven model training, including,

[0064] Obtaining an infrared thermal image of each air cooling unit of the target air cooling island, calculating the average temperature of each air cooling unit according to the infrared thermal image;

[0065] Obtaining data as input features of data-driven and data for calculating heat transfer coefficient through the DCS system;

[0066] Taking the average temperature and the data obtained by the DCS system as an input data set, training a data-driven model with the input data set to obtain a final data-driven model;

[0067] The second part is the heat transfer performance monitoring of the target air cooling island, including,

[0068] Real-time acquisition of the infrared thermal image of the target air cooling island and the related data of the DCS system as the input data set, normalization of the data of the input data set at the same time;

[0069] Input the normalized data into the time sliding window, input the time sliding window as a real-time input sample into the data-driven model, and the model calculates and outputs the heat transfer coefficient of each unit in the thermal power generating unit.

[0070] In embodiment 1, the total flow is divided into two parts, the first part is the data-driven model training part, including obtaining historical data from the data acquisition module, after heat transfer coefficient calculation and normalization, the data-driven model is trained; the second part is the air cooling island heat transfer performance monitoring part, real-time data is obtained from the data acquisition module, after time sliding window and normalizer, the heat transfer coefficient of each unit is calculated through the data-driven model, and then the heat transfer coefficient is visualized. Compared with the traditional calculation method, the monitoring method based on data driving can comprehensively, accurately and timely monitor the heat transfer performance of each unit of the air cooling island, and real-time calculation of the heat transfer coefficient of each unit, which provides an important basis for formulating and implementing winter anti-freezing measures and subsequent air cooling island fan economic optimization control of the thermal power plant.

[0071] As a preferred embodiment of the present application, the method further comprises, after the model calculates the heat transfer coefficient of each unit in the thermal power unit, a heat map is formed according to the arrangement mode of the air cooling unit in the thermal power unit to perform real-time visual display.

[0072] As a preferred embodiment of the present application, specifically, the infrared thermal image of each air cooling unit of the target thermal power unit is obtained, and the average temperature of each air cooling unit is calculated according to the infrared thermal image, including,

[0073] The infrared thermal camera is installed on the leeward side of each unit of the air cooling island, the infrared thermal image of each finned tube bundle is shot, and the obtained real-time finned tube bundle infrared image is transmitted to the back end for storage and calculation of the average temperature of the leeward side.

[0074] The infrared thermal imager is a custom-made version for the actual situation of the air cooling island, has all-weather passive thermal imaging function, and can be used in a wide range of environmental temperature. The field of view of up to 56° can observe the effective measured target of 9.04*6.78 (m2) at a distance of 10 meters. The infrared thermal imager corresponding to the current air cooling unit is installed on the outer side opposite to the fin of the current air cooling unit, and on the lower side of the steam pipe of the air cooling island. Two outer sides of one air cooling unit each correspond to one infrared thermal imager. The infrared temperature image of the adjacent air cooling unit is obtained in real time, the obtained infrared temperature image is transmitted to the back end for storage through the bus, and the average temperature of the two sides of the leeward side of each air cooling unit is extracted.

[0075] As a preferred embodiment of the present application, specifically, the data as the input features of data driving and the data for calculating the heat transfer coefficient are obtained through the DCS system, including,

[0076] The data obtained by the DCS system has two kinds, one is as a data-driven input feature, including obtaining the steam temperature at the inlet of the air cooling island, the outlet condensate water temperature of each unit of the air cooling island and the rotational speed of the axial flow fan; the other is the data for calculating the heat transfer coefficient, including the exhaust steam pressure and flow rate at the outlet of the steam turbine, the atmospheric temperature, i.e., the air temperature at the inlet of the air cooling island, the atmospheric pressure, and the heat transfer coefficient of each unit of the air cooling island is calculated.

[0077] As a preferred embodiment of the present application, specifically, the calculation process of the heat transfer coefficient of each unit of the air cooling island includes the exhaust steam pressure Pc and flow rate Gn at the outlet of the steam turbine, the atmospheric temperature ta1, the atmospheric pressure pa, and the calculation of the heat transfer coefficient of each unit of the air cooling island, the calculation process is as follows:

[0078] The condensate water saturation temperature ts, the condensate water enthalpy hew and the steam enthalpy hew in the finned tube are calculated using the water and water vapor property and property standard formula IAPWS-IF97. The calculation function of the IAPWS-IF97 standard formula in the third-party library iapws of python and CoolProp can be calculated by calling the related function combined with the known parameters of the exhaust steam pressure Pc. The formula used for calculating the heat release Qn when the steam condenses in the finned tube is:

[0079] Q n =G n (h ev -h ew )

[0080] In the formula, Gn is the exhaust steam flow rate of the steam turbine.

[0081] Referring to Figure 2 , the outlet temperature ta2 of the air cooling island and its corresponding air physical properties such as Figure 2 are calculated by iterative calculation. The maximum temperature difference tem of the inlet and outlet of the air cooling island is set to 40℃, the iterative interval of the outlet temperature ta2 of the air cooling island is (ta1, ta1+tem℃), the iterative interval is te=0.01℃, and the iterative calculation is performed: for calculating the air physical properties, first, the average temperature of the inlet and outlet of the air cooling island is calculated Then the air is regarded as an ideal gas to calculate its density The formula used is Secondly, the specific heat capacity at constant pressure of the air is calculated The formula used is: Then the heat absorbed by the air outside the finned tube Qa is calculated, and the formula used is: In the formula, Ay is the windward area of the air-cooled island, and vF is the wind speed of the windward surface (the average wind speed of the air-cooled unit corresponding to the rotational speed of the axial flow fan of the air-cooled island). Finally, the calculation result is judged. When the heat absorbed by the air outside the finned tube is equal to the heat released by the condensation of the steam inside the finned tube, that is, Qa=Qn, the iteration of this time is determined. If Qa=Qn is not established, iteration is continued. In the actual iteration process, due to calculation errors and measurement errors, the condition of Qa=Qn is usually difficult to establish. Therefore, in the execution process, it is judged by Qe=|Qa-Qn|<Qt, wherein Qe is the heat difference, Qt is the heat error tolerance, and Qt is set to 5% of Qn. In order to increase the accuracy, the entire iteration process is carried out, and among all the results meeting the condition of Qe<Qt, the air outlet temperature ta2 corresponding to the smallest Qe and the corresponding air physical properties are selected.

[0082] After the air outlet temperature ta2 of the air-cooled island and the corresponding air physical properties are calculated through iteration, the heat transfer performance of the entire air-cooled island is determined by calculating NTU, and the heat transfer coefficient of each unit of the air-cooled island is calculated using the variable operating condition laboratory formula. First, the heat transfer unit number NTU of the air-cooled island is calculated, and the formula used is:

[0083] Then, the heat transfer coefficient K of the air-cooled island is calculated, and the formula used is: Finally, the heat transfer coefficient Kij of each unit of the air-cooled island is calculated using the variable operating condition laboratory simplified formula, and the formula used is: Wherein, the subscripts i and j are the i-th row and j-th column air-cooled units. The calculated heat transfer coefficient corresponds to the real time of the corresponding data.

[0084] As a preferred embodiment of the present application, specifically, the data-driven model adopts an RNN structure neural network, such as Figure 3 The structure shown in the figure is a "multiple input single output structure", that is, multiple time step inputs and output at the last time step. The input features of the training data set are normalized, and the neural network model is set. The number of neural network layers is L=4 layers, the number of neurons in different network layers is c=[256, 512, 512, 256], the time step length T=360, the time interval is 5s (that is, the historical half-hour data is used as the input data), the loss function is MSE, and the optimizer is the gradient descent method. The hyperparameters include learning rate LearnRate, batch size BatchSize, and iteration number epochs.

[0085] First, from the training data set obtained in the previous step, the monitoring points on the air cooling island obtain the air cooling island inlet steam temperature, the outlet condensate water temperature of each unit of the air cooling island, the speed of the axial flow fan of each unit, and the average temperature of the back wind surface finned tube of each unit as input features, and strictly according to the time as a vector X, and the input feature vector at time t is denoted as Xt; the sample output vector y is the one-dimensional arrangement of the heat transfer coefficients Kij of each unit of the air cooling island, that is, y = [K11, …, K1m, K21, …, K2m, …, Kn1, …, Knm], where n and m are the number of rows and columns of the air cooling island unit array respectively, and y strictly corresponds to the input feature time, denoted as yt. Normalize the input features, use the maximum and minimum normalization, and the formula is Xmax and Xmin are the maximum and minimum values of the input features, respectively, and the calculation is by element calculation rather than matrix operation. While normalizing, save the maximum and minimum values of each feature in the sample set as a normalizer. The input of the model is the input sample of the corresponding T recent time points at time t, that is, [Xt-T+1, Xt-T+2, …, Xt], and the output of the model is During model training, the input feature vector X and the sample output vector y are arranged in time sequence as the sample set of the model, 80% of which is used as the training set, 10% as the test set, and 10% as the validation set.

[0086] The RNN of the data-driven model uses an LSTM neuron structure as Figure 4 The calculation steps are: calculate the forgetting gate vector F t = σ(X t W xf +H t-1 W hf +b f ), calculate the input gate vector I t = σ(X t W xi +H t-1 W hi +b i ), calculate the output gate vector O t = σ(X t W xo +H t-1 W ho +b o ), calculate the temporary cell state vector Calculate the cell state vector H t = O t ⊙tanh(C t). Wherein, t is t time, sigma is sigmoid activation function, tanh is tanh activation function, W represents the weight matrix of training, b represents the bias value of training, H is the hidden state. The activation function used, wherein the formula of sigmoid activation function is: The formula of tanh activation function is: The loss function adopts mean square error MSE, and the formula is N is the number of samples, yi is the output value of the training sample, The output value of the model is the output value of the model. The weight update of the neural network uses gradient descent method: Through the method, the weight of each neuron of the model is finally confirmed. A plurality of data-driven models are trained by setting a plurality of different neural network parameters and hyperparameters, and the training is stopped when the iteration number is reached. The model with the best prediction effect is selected through the validation set.

[0087] Referring to Figure 5 And Figure 6 , the heat transfer performance monitoring operation process of the second part of the air cooling island is as follows:

[0088] Real-time data acquisition module, that is, from the DCS system and the infrared camera, the air cooling island inlet steam temperature, the air cooling island unit outlet condensate water temperature, the air cooling island unit axial flow fan speed and the average temperature of the leeward side of the air cooling unit are obtained. The obtained normalizer is used for maximum and minimum normalization of the data. Input into the time sliding window, such as Figure 5 Xt+1 in the figure does not exist at t time, and is only used for display. When Xt is input into the time sliding window, it is inserted before Xt-1, and Xt-T is deleted. At this time, the data in the time sliding window is used as the real-time input sample and input into the data-driven model. The model is as follows: Figure 4 And Figure 3 The heat transfer coefficient of each unit of the air cooling island is calculated. The data is saved, and the real-time data interface is retained to provide for the staff. The heat transfer coefficient is visualized in real time using a heat map according to the arrangement of the air cooling unit. Taking an 8x8 air cooling island as an example, Figure 6 As shown in the figure, the heat map and the numerical value are combined to display the color of the heat map from white to red, and the larger the heat transfer coefficient, the redder the color of the corresponding unit.

[0089] The present application also proposes an online monitoring system for heat transfer performance of a thermal power unit air cooling unit, comprising the following:

[0090] The data-driven model training device comprises,

[0091] The training data acquisition module is configured to acquire an infrared thermal image of each air cooling unit of the target air cooling island, calculate an average temperature of each air cooling unit according to the infrared thermal image, acquire data serving as input features of data driving through a DCS system, and acquire data for calculating a heat transfer coefficient;

[0092] The model training module is configured to train a data driving model by taking the average temperature and the data acquired by the DCS system as an input data set, and obtain a final data driving model.

[0093] The heat transfer performance monitoring device of the target air cooling island comprises,

[0094] The real-time data acquisition module is configured to acquire, in real time, the infrared thermal image of the target air cooling island and related data of the DCS system as an input data set, and normalize data of the input data set at the same time.

[0095] The heat transfer coefficient calculation module is configured to input the normalized data into a time sliding window, input the time sliding window as a real-time input sample into the data driving model, and calculate and output heat transfer coefficients of each unit in the thermal power generating unit by the model.

[0096] As a preferred embodiment of the present application, the system further comprises,

[0097] The result display module is configured to, after calculating and outputting the heat transfer coefficients of each unit in the thermal power generating unit by the model, further form a heat map according to an arrangement mode of the air cooling units in the thermal power generating unit, and perform real-time visual display.

[0098] The present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize steps of the heat transfer performance online monitoring method of the air cooling unit of the thermal power generating unit.

[0099] The modules described as separate components may or may not be physically separate, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. According to actual needs, part or all of the modules can be selected to achieve the purpose of the scheme in the embodiment.

[0100] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software functional module.

[0101] The integrated module, if implemented in the form of a software function module and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0102] Although the description of the present application has been quite detailed and particularly described with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, but should be considered to be a broad interpretation of the claims in view of the prior art, so as to effectively encompass the intended scope of the present application. In addition, the present application is described above in embodiments that the inventors can foresee, and the purpose is to provide a useful description, and those non-essential modifications to the present application that have not yet been foreseen can still represent equivalent modifications of the present application.

[0103] The above is only the preferred embodiment of the present application, and the present application is not limited to the above-mentioned embodiments, as long as the same means achieve the technical effect of the present application, it should belong to the protection scope of the present application. The technical solutions and / or embodiments within the protection scope of the present application can have various modifications and changes.

Claims

1. A method for on-line monitoring of heat transfer performance of an air cooling unit of a thermal power generating unit, characterized in that, Comprise the following: The first part, data-driven model training, comprising, Obtain the infrared thermal image of each air cooling unit of the target air cooling island, and calculate the average temperature of each air cooling unit according to the infrared thermal image; Obtain the data as the input features of data driving and the data for calculating the heat transfer coefficient through the DCS system; The average temperature and the data obtained by the DCS system are used as the input data set, and the data set is used to train the data-driven model, and the final data-driven model is obtained; The second part, the heat transfer performance monitoring of the target air cooling island, comprising, Real-time acquisition of the infrared thermal image of the target air cooling island and the related data of the DCS system as the input data set, and normalization of the data of the input data set; The normalized data is input into the time sliding window, and the data in the time sliding window is used as the real-time input sample to input into the data-driven model, and the model calculates the heat transfer coefficient of each unit in the thermal power generating unit; The data-driven model adopts an RNN structure neural network, and the RNN adopts an LSTM neuron structure, Specifically, the data as the input features of data driving and the data for calculating the heat transfer coefficient are obtained through the DCS system, comprising, There are two kinds of data obtained by the DCS system, one is the input features of data driving, including the air cooling island inlet steam temperature, the air cooling island unit outlet condensate temperature and the axial flow fan speed obtained at the monitoring points on the air cooling island; the other is the data for calculating the heat transfer coefficient, including the exhaust steam pressure and flow of the steam turbine outlet, the atmospheric temperature, i.e. the air cooling island inlet air temperature, the atmospheric pressure, and the heat transfer coefficient of each unit of the air cooling island is calculated; The training process of the data-driven model comprises, According to the obtained monitoring points on the air cooling island, the inlet steam temperature of the air cooling island, the outlet condensate water temperature of each unit of the air cooling island, the rotational speed of the axial flow fan of each unit, and the average temperature of the finned tube on the leeward surface of each unit are obtained as input features, and strictly according to time as a vector X, and the input feature vector at time t is denoted as ; the sample output vector y is the one-dimensional arrangement of the heat transfer coefficients of each unit of the air cooling island , which strictly corresponds to the input feature time, and is denoted as , Each input of the model is the input sample of the latest T time points corresponding to time t, that is, , , …, ], and the output of the model is . During the training of the model, the input feature vector X and the sample output vector y are arranged in time sequence as the training sample set of the model.

2. The method for on-line monitoring of heat transfer performance of air cooling unit of thermal power generating unit according to claim 1, characterized in that, The method further comprises, after the model calculates the heat transfer coefficient of each unit in the thermal power generating unit, a heat map is formed according to the arrangement mode of the air cooling unit in the thermal power generating unit to realize real-time visual display.

3. The method for on-line monitoring of heat transfer performance of air cooling unit of thermal power generating unit as claimed in claim 1 wherein, Specifically, the infrared thermal image of each air cooling unit of the target thermal power generating unit is obtained, and the average temperature of each air cooling unit is calculated according to the infrared thermal image, comprising, The infrared thermal camera is installed on the leeward side of each unit of the air cooling island, the infrared thermal image of each finned tube bundle is shot, and the real-time finned tube bundle infrared image obtained is transmitted to the back end for storage and calculation of the average temperature of the leeward side.

4. The method for on-line monitoring of heat transfer performance of air cooling unit of thermal power generating unit according to claim 3, characterized in that, Specifically, the heat transfer coefficient calculation process of each unit of the air cooling island comprises, The condensing water saturation temperature in the finned tube is calculated using the standard equation IAPWS-IF97 for the thermodynamic properties of water and water vapor , the condensing water enthalpy and the vapor enthalpy ; Computing heat released by condensation of steam in finned tube The formula used is: ； In the formula, is the steam turbine exhaust flow rate; Setting the maximum temperature difference of the air flow inlet and outlet of the air cooling island , the air flow outlet temperature of the air cooling island , the iteration interval is , + , the interval is , and the iteration calculation is performed: firstly, the average temperature of the air flow inlet and outlet of the air cooling island is calculated , and the formula used is ; Calculating the density of air as an ideal gas , the formula used is ; The specific heat at constant pressure of air is calculated using the formula: ; Computing the heat absorbed by the air outside the finned tube The formula used is: where is the windward area of the air-cooled island, is the wind speed on the windward side; When the heat absorbed by the air outside the finned tube is equal to the heat released by the condensation of the steam inside the finned tube, i.e. = , the output , the current time value of the , , of this iteration is determined. The heat transfer unit number NTU of the air cooling island is calculated by using the following formula: ; The heat transfer coefficient K of the air cooling island is calculated by using the following formula: ; The heat transfer coefficient of each unit air cooling island is calculated by using the variable working condition simplified formula , and the formula is: Wherein, the subscripts i and j are the i-th row and j-th column air cooling units respectively The calculated heat transfer coefficient corresponds to the real time of the corresponding data.

5. The method for on-line monitoring of heat transfer performance of air cooling unit of thermal power generating unit as claimed in claim 1 wherein, Specifically, the data set input features are normalized, and the parameters include the number of neural network layers L, the number of neurons c in different network layers i , i = 1,2, …, L, time step T, loss function, optimizer, and hyperparameter settings include learning rate LearnRate, batch size BatchSize, and number of iterations epochs.

6. The method for on-line monitoring of heat transfer performance of air cooling unit of thermal power generating unit as claimed in claim 5 wherein, Specifically, the training process of the data-driven model further comprises, The input features are normalized by using maximum and minimum normalization, and the formula is , , are the maximum and minimum values of the input features, respectively. At the same time of normalization, the maximum and minimum values of each feature in the sample set are saved as a normalizer. The calculation steps are: calculating a forget gate vector , calculating an input gate vector , calculating an output gate vector , calculating a temporary cell state vector , calculating a cell state vector , wherein t is the t moment, is a sigmoid activation function, tanh is a tanh activation function, W represents a weight matrix being trained, b represents a bias value being trained, and H is a hidden state The activation function used, where the formula for the sigmoid activation function is: The formula for the tanh activation function is: ​ The loss function adopts mean square error (MSE), and the formula is N is the sample number, is the output value of the training sample, is the model output value; The weight update of the neural network uses the gradient descent method: ; A plurality of data-driven models are trained by setting a plurality of different neural network parameters and hyperparameters, and the training is stopped when the iteration number is reached, and the model with the best prediction effect is selected through the validation set.

7. The heat transfer performance on-line monitoring system of the air cooling unit of the thermal power generating unit, characterized in that, The system comprises the steps of the method of any one of claims 1-6, Comprise the following: The data-driven model training device comprises, The training data acquisition module is configured to obtain the infrared thermal image of each air cooling unit of the target air cooling island, calculate the average temperature of each air cooling unit according to the infrared thermal image, and obtain the data as the input features of data driving and the data for calculating the heat transfer coefficient through the DCS system; The model training module is configured to train a data-driven model by using the average temperature and data obtained by the DCS system as an input data set, and obtain a final data-driven model. The heat transfer performance monitoring device of the target air cooling island comprises, The real-time data acquisition module is configured to acquire an infrared thermal image of the target air cooling island and related data of the DCS system as an input data set, and normalize data of the input data set at the same time. The heat transfer coefficient calculation module is configured to input the normalized data into a time sliding window, input data in the time sliding window as real-time input samples into the data-driven model, and calculate and output heat transfer coefficients of each unit in the thermal power generating unit by the model.

8. The heat transfer performance on-line monitoring system of the air cooling unit of a thermal power generating unit according to claim 7, characterized in that, The system further comprises, The result display module is configured to display a heat map in real time according to an arrangement of the air cooling units in the thermal power generating unit after the model calculates and outputs the heat transfer coefficients of each unit in the thermal power generating unit.

9. A computer-readable storage medium storing a computer program, the computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1-6.

Images