Parameter optimization design method and device for fully-open type flue gas total heat deep recovery system based on solution

Abstract

The invention discloses a solution-based full-open type flue gas total heat deep recovery system parameter optimization design method and equipment, and the method comprises the steps: building a system physical model through comprehensive experiment data, randomly generating a design parameter matrix through the Monte Carlo simulation idea, and carrying out the batch calculation of an accumulation database through the system physical model; constructing a neural network topological structure according to the input and output parameters of the sample and a database, and performing training effect evaluation; formulating a fitness function of the genetic algorithm from the angles of environment, energy and economy; coupling the neural network model, and determining the optimal design parameter combination configuration by using the strong optimization capacity of the genetic algorithm. According to the method, the consumption of computing resources is reduced by using the strong fitting capability of the neural network, and the engineering design period is shortened; according to the method, global optimization of system design parameters can be realized only by inputting external flue gas working condition parameters and freely selecting an optimization target, and the method provides a theoretical basis for optimization of a low-grade flue gas total heat deep recovery system.

Description

Technical Field

[0001] The present invention relates to the field of flue gas waste heat recovery, and in particular to a parameter optimization design method and equipment for a fully open flue gas full heat depth recovery system based on a solution. Background Art

[0002] With the development of social economy and industry, the demand for energy has increased significantly, and the greenhouse gas emissions from human activities have reached an unsustainable level. According to the International Energy Agency, the proportion of fossil fuels in the global energy structure has remained at around 80% in recent decades, and about 63% of the primary energy consumed is wasted in the combustion and heat transfer process, and is ultimately discharged into the environment mainly in the form of flue gas, water, etc. Among them, low-grade flue gas from power generation, heating, chemical industry and other processes accounts for a high proportion of waste heat resources. In addition, due to fuel moisture, fuel hydrogen base and production process, the discharged low-temperature flue gas not only has huge sensible heat, but also contains considerable latent heat and water resources. The direct emission of low-temperature and high-humidity flue gas will not only cause waste of energy and water, but also produce white wet smoke plumes, promote the secondary transformation of atmospheric pollutants, and cause double visual and environmental pollution. Therefore, the full heat depth recovery and quality improvement of low-temperature and high-humidity flue gas is crucial.

[0003] The existing low-temperature flue gas waste heat recovery methods can be divided into partition cooling condensation method, membrane indirect contact method, water direct contact method and solution direct contact method according to the contact method between flue gas and cooling medium. The recycled cooling media mainly include heat network return water, power plant condensate, compression / absorption heat pump evaporator cold water, and semi-open / fully open absorption heat pump absorber solution. Different cooling media are matched with different contact methods to achieve waste heat recovery of low-temperature flue gas. However, the above waste heat recovery methods all have certain limitations. For example, the partition cooling condensation method has the problem of heat exchanger corrosion and high temperature requirements for the cooling medium; the membrane indirect contact method has large heat and mass transfer resistance, easy to clog, and short service life; the water direct contact method is limited by the high water vapor partial pressure of pure water, and the flue gas is close to saturation after recovery; based on the solution contact method of the semi-open absorption heat pump, the regeneration device still needs to maintain a vacuum environment and needs high-grade heat drive. In summary, the fully open flue gas full heat recovery system based on solution is expected to become a low-temperature and high-humidity flue gas full heat depth recovery technology that can be promoted and utilized on a large scale. The system involves many parameters and has nonlinear and strong coupling characteristics, so the analysis and optimization of its design parameters is the basis for the promotion and application of technology. In view of the above-mentioned problem of waste heat recovery from low-grade flue gas, a method that can quickly determine the optimal combination configuration of design parameters under different external working conditions and optimization objectives is urgently needed. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide a parameter optimization design method and equipment for a solution-based fully open flue gas full heat depth recovery system, which can quickly determine the optimal combination configuration of design parameters.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0006] A method for optimizing the parameters of a fully open flue gas full heat depth recovery system based on a solution comprises the following steps:

[0007] (1) Based on the direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, a physical model of the flue gas full heat deep recovery system based on solution was established;

[0008] (2) Based on the Monte Carlo simulation concept, a number of design parameter matrices are randomly generated, and the physical model of the flue gas full heat depth recovery system is imported for simulation calculation to obtain a state performance parameter data set for characterizing the state and performance of the flue gas full heat depth recovery system; the state performance parameter data set includes state parameter data and performance index data, and the performance index data includes system comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB;

[0009] (3) removing outliers and normalizing the state performance parameter data set, and using the performance index data as labels, the corresponding design parameter matrix and the state parameter data as samples, and inputting them into the neural network model for training;

[0010] (4) With the goal of maximizing the performance indicator data, a genetic algorithm is used to perform global optimization in the system design parameter space to determine the optimal value of the design parameter matrix; during the global optimization, a trained neural network model is used to predict the performance indicator data corresponding to each design parameter matrix.

[0011] Furthermore, step (1) comprises:

[0012] (1.1) obtaining direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, wherein the direct contact heat and mass transfer experimental data include the inlet and outlet temperature, humidity, concentration and flow rate of flue gas and solution, and the inlet and outlet temperature, humidity and flow rate of flue gas and condensed water under different temperatures, humidity, concentrations and flow rates;

[0013] (1.2) According to the direct contact heat and mass transfer experimental data, the heat and mass transfer coefficients between flue gas and solution, and between flue gas and condensed water are fitted to obtain the fitting expressions of the heat and mass transfer coefficients;

[0014] (1.3) According to the fitting expression of heat and mass transfer coefficients, the equivalent energy flow model of absorber, regenerator, condenser and intermediate wall heat exchanger is constructed, and the equivalent energy flow model of each device is coupled through the corresponding isothermal points and energy and mass conservation to obtain the physical model of solution-based flue gas full heat deep recovery system.

[0015] Furthermore, the fitting expression of the heat and mass transfer coefficient is specifically:

[0016] h c =1.114G s 1.019 ·G g 1.002

[0017] h d =0.9953G s 1.038 ·G g 1.014

[0018] h c =1.534G w 0.9609 ·G g 1.007

[0019] h d =1.311G w 1.026 ·G g 1.003

[0020] h c =1.75G s 0.8333 ·G g 0.814

[0021] h d =2.481G s 0.7756 ·G g 0.8101

[0022] In the formula, G s , G w , G g Represent the mass flow rate of solution, condensed water and flue gas respectively, h c represents the heat transfer coefficient, h d represents the mass transfer coefficient.

[0023] Furthermore, the physical model of the flue gas full heat depth recovery system is specifically as follows:

[0024] Equivalent energy flow model of absorber and regenerator:

[0025] -Cp s ·m s ·dt s -Cp s ·dm s ·t s =m g ·dh g +h dil ·m g ·dω g

[0026] dm s =-m g ·dω g

[0027] h c ·πR 2 ·dH·α(t s -t g )=m g ·(Cp g +ω g ·Cp v )·dt g

[0028] h d ·πR 2 ·dH·α(ω s -ω g )=m g ·dω g

[0029]

[0030] X s ·m s =(X s +dX s )(m s +dm s )

[0031] Condenser equivalent energy flow model:

[0032] -Cp w ·m w ·dt w -Cp w ·dm w ·t w =m g ·dh g

[0033] dm w =-m g ·dω g

[0034] hc ·πR 2 ·dH·α(t w -t g )=m g ·(Cp g +ω g ·Cp v )·dt g

[0035] h d ·πR 2 ·dH·α(ω w -ω g )=m g ·dω g

[0036]

[0037] Equivalent energy flow model of inter-wall heat exchanger:

[0038] Q hx =Cp w ·m w ·(t w_i -t w_o )

[0039] Q hx =K hx ·A hx LMTD hx

[0040]

[0041] In the formula, m s ,t s , Cp s , P s ,ω s , X s Respectively represent the mass flow rate, temperature, specific heat, water vapor pressure, equivalent moisture content, and concentration of the solution, m g ,t g , Pg g ,ω g 、h g Respectively represent the mass flow rate, temperature, water vapor pressure, equivalent moisture content, enthalpy of flue gas, m w ,t w , Cp w , P w ,ω w , X w They represent the mass flow rate, temperature, specific heat, water vapor pressure, equivalent moisture content, and concentration of the solution, respectively. dilrepresents the dilution enthalpy, P0 represents the local atmospheric pressure, α and R represent the specific surface area and radius of the ceramic structured packing material, H represents the height of the open spray packing tower (absorber, regenerator or condenser), Cp v represents the specific heat of water vapor, Cp g represents the specific heat of dry air, d represents the derivative, Q hx , K hx , A hx LMTD hx They represent the heat, heat transfer coefficient, area and logarithmic mean temperature difference of the inter-wall heat exchanger, t w_i ,t w_o Respectively represent the inlet and outlet water temperatures, t s_i ,t s_o Represent the inlet and outlet solution temperatures respectively.

[0042] Furthermore, the design parameter matrix in step (2) is specifically:

[0043] A desigb =[H abs ,H reg ,H con ,m g_reg_i ,m g_rec ,m w_con_i ,m s_reg_i ]

[0044] In the formula, A design represents the design parameter matrix, H abs , H reg , H con Represent the height vectors of the absorber, regenerator, and condenser, respectively, m g_reg_i 、m g_rec 、m w_con_i 、m s_reg_i They respectively represent the driving flue gas mass flow vector, the recovered flue gas mass flow vector, the condensate water mass flow vector at the condenser inlet, and the solution mass flow vector at the regenerator inlet.

[0045] The state parameter data specifically include the state parameter data of each position point of the system flue gas, solution working fluid and condensed water under different driving and recovered flue gas input conditions, specifically temperature, humidity, flow rate and concentration, as well as the heat exchange capacity of each component.

[0046] Furthermore, after obtaining the state performance parameter data set in step (2), a sensitivity analysis of the design parameters and the state performance parameters is performed, and the sensitivity analysis is achieved by calculating the Pearson correlation coefficient between the design parameter X and the state performance parameter Y.

[0047] Furthermore, the system comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB are calculated as follows:

[0048]

[0049] IC sys =1.4[(IC abs +IC con +IC reg +IC hx +IC pum +IC fan +IC s +IC w ) / 0.85]

[0050] In the formula, Q reg , Q abs , Q con Represent the heat of the regenerator, absorber, and condenser, respectively, h g_reg_i 、h w_con_i , g g_abs_o 、h g_rec 、h env They represent the flue gas enthalpy at the regenerator inlet, the condensed water enthalpy at the condenser inlet, the flue gas enthalpy at the absorber outlet, the recovered flue gas enthalpy and the ambient air enthalpy, respectively. g_reg_i 、m w_con_i 、m w_con They represent the flue gas flow rate at the regenerator inlet, the condensed water flow rate at the condenser inlet, and the water flow rate condensed from the flue gas in the condenser. abs 、IC con 、IC reg 、IC hx 、IC pun 、IC fan 、IC s 、IC w Respectively represent the initial investment of absorber, condenser, regenerator, heat exchanger, pump, fan, solution and water, P heat , P ele Represent the price of heat and electricity respectively, Q total Indicates the total heat supply, IC sys is the total investment, W sys is the total power.

[0051] Furthermore, step (4) specifically includes:

[0052] (4.1) The design parameters in a design parameter matrix are organized into a chromosome, each design parameter is used as a gene of the chromosome, and the population is randomly initialized;

[0053] (4.2) According to the population information, the trained neural network model is input to predict the performance index, and the performance index is used as the fitness value;

[0054] (4.3) After evaluating and comparing the fitness values, the chromosomes are crossed and mutated until the iterative setting conditions are met and the chromosome with the best fitness is obtained. The selection operator uses the roulette wheel selection method;

[0055] (4.4) Decode the chromosome with the best fitness and obtain the optimal design parameter matrix.

[0056] Furthermore, after step (4), the method further includes:

[0057] (5) The optimal design parameter matrix is ​​introduced into the physical model of the flue gas full heat deep recovery system for reverse verification. The physical model of the flue gas full heat deep recovery system calculates the state parameters of each position point and explains and analyzes it from a thermodynamic perspective.

[0058] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above method.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] (1) The strong fitting ability of neural networks is used to predict system performance parameters, which reduces the calculation time and resource consumption of the system physical model, effectively improves the efficiency of engineering design optimization, and shortens the design cycle;

[0061] (2) Multidimensional optimization objectives, setting comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB evaluation indicators, taking into account both investment cost and benefits as well as energy recovery rate;

[0062] (3) Genetic algorithms are used to collaboratively optimize the design parameters of the absorber, regenerator and condenser heights, condensate, solution, driving flue gas and recovered flue gas flow rates, avoiding over-optimization of a certain design parameter while ignoring other design parameters, and avoiding large-scale exhaustive search methods to determine the optimal design parameter combination.

[0063] (4) The parameter sensitivity analysis results before the optimization search provide a rough assessment of the impact of the parameters on the model results and a theoretical basis for subsequent design optimization. The energy flow diagram drawn at the end of the optimization search explains the data-driven model results from a thermodynamic perspective, helping engineers understand and trust the underlying model. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a flow chart of a parameter optimization design method of a fully open flue gas full heat depth recovery system based on a solution provided in an embodiment of the present invention;

[0065] Figure 2 It is a schematic diagram of the topological structure of the neural network model established by the present invention;

[0066] Figure 3 This is a graph showing the training effect of a neural network model;

[0067] Figure 4 is a diagram of the optimization process of the genetic algorithm model; wherein, (a) is a diagram of the optimization iterative process of the COP objective function, (b) is a diagram of the optimization iterative process of the LCB objective function, and (c) is a diagram of the optimization iterative process of the η objective function;

[0068] Figure 5 is the energy flow diagram; where (a) is the energy flow diagram of the control group, (b) is the energy flow diagram of the COP optimization group, (c) is the energy flow diagram of the η optimization group, and (d) is the energy flow diagram of the LCB group. DETAILED DESCRIPTION

[0069] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0070] Embodiment 1

[0071] The embodiment of the present invention provides a method for optimizing the parameters of a fully open flue gas full heat depth recovery system based on a solution, such as Figure 1 As shown, the following steps are included:

[0072] (1) Based on the direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, a physical model of the solution-based flue gas full heat deep recovery system was established.

[0073] This step specifically includes:

[0074] (1.1) obtaining direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, wherein the direct contact heat and mass transfer experimental data include the inlet and outlet temperature, humidity, concentration and flow rate of flue gas and solution, and the inlet and outlet temperature, humidity and flow rate of flue gas and condensed water under different temperatures, humidity, concentrations and flow rates;

[0075] (1.2) According to the direct contact heat and mass transfer experimental data, the heat and mass transfer coefficients between flue gas and solution, and between flue gas and condensed water are fitted, and the fitting formulas of heat and mass transfer coefficients under absorption (flue gas and solution), condensation (flue gas and condensed water) and regeneration (flue gas and solution) conditions are obtained as follows:

[0076] h c =1.114G s 1.019 ·H g 1.002

[0077] h d =0.9953G s 1.038 ·Gg 1.014

[0078] h c =1.534G w 0.9609 ·G g 1.007

[0079] h d =1.311G w 1.026 ·G g 1.003

[0080] h c =1.75G s 0.8333 ·G g 0.814

[0081] h d =2.481G s 0.7756 ·G g 0.8101

[0082] In the formula, G s , G w , G g Represent the mass flow rate of solution, condensed water and flue gas respectively, h c represents the heat transfer coefficient, h d represents the mass transfer coefficient.

[0083] (1.3) According to the fitting expression of heat and mass transfer coefficient, the equivalent energy flow model of absorber, regenerator, condenser and partition heat exchanger is constructed, and the equivalent energy flow model of each device is coupled through the corresponding isothermal point and energy and mass conservation, and the physical model of flue gas full heat depth recovery system based on solution is obtained as follows:

[0084] Equivalent energy flow model of absorber and regenerator:

[0085] -Cp s ·m s ·dt s -Cp s ·dm s ·t s =m g ·dh g +h dil ·m g ·dω g

[0086] dm s =-m g ·dωg

[0087] h c ·πR 2 ·dH·α(t s -t g )=m g ·(Cp g +oh g ·Cp v )·dt g

[0088] h d ·πR 2 ·dH·α(ω s -oh g )=m g ·dω g

[0089]

[0090] X s ·m s =(X s +dX s (m) s +dm s )

[0091] Refrigerant equivalent energy flow model:

[0092] -Cp w ·m w ·dt w -Cp w ·dm w ·t w =m g ·dh g

[0093] dm w =-m g ·dω g

[0094] h c ·πR 2 ·dH·α(t w -t g 0=m g ·(Cp g +oh g ·Cp v 0·dt g

[0095] h d ·πR 2 ·dH·α(ω w -oh g )=mg ·dω g

[0096]

[0097] Equivalent energy flow model of inter-wall heat exchanger:

[0098] Q hx =Cp w ·m w ·(t w_i -t w_o )

[0099] Q hx =K hx ·A hx LMTD hx

[0100]

[0101] In the formula, m s ,t s , Cp s , P s ,ω s , X s Respectively represent the mass flow rate, temperature, specific heat, water vapor partial pressure, equivalent moisture content, and concentration of the solution, m g ,t g , Pg t ,ω g 、h t Respectively represent the mass flow rate, temperature, water vapor partial pressure, equivalent moisture content, enthalpy of flue gas, m w ,t w , Cp w , P w ,ω w Respectively represent the mass flow rate, temperature, specific heat, water vapor pressure, and equivalent moisture content of water, h dil represents the dilution enthalpy, P0 represents the local atmospheric pressure, α and R represent the specific surface area and radius of the ceramic structured packing material, H represents the height of the open spray packing tower (absorber, regenerator and condenser), Cp v represents the specific heat of water vapor, Cp g represents the specific heat of dry air, d represents the derivative, Q hx , K hx , A hx LMTD hx They represent the heat, heat transfer coefficient, area and logarithmic mean temperature difference of the inter-wall heat exchanger, t w_i ,t w_o Respectively represent the inlet and outlet water temperatures, t s_i ,t s_oRepresent the inlet and outlet solution temperatures respectively.

[0102] (2) Based on the Monte Carlo simulation concept, several design parameter matrices are randomly generated, and the physical model of the flue gas full heat deep recovery system is imported for simulation calculation to obtain a state performance parameter data set used to characterize the state and performance of the flue gas full heat deep recovery system.

[0103] Among them, Monte Carlo simulation is to generate a large number of random numbers in the specified design parameter space to achieve the effect of approximation traversal. The design parameter matrix generated by Monte Carlo simulation is specifically:

[0104] A design =[H abs ,H reg ,H con ,m g_reg_i ,m g_rec ,m w_con_i ,m s_reg_i ]

[0105] In the formula, A design represents the design parameter matrix, H abs , H reg , H con Represent the height vectors of the absorber, regenerator, and condenser, respectively, m g_reg_i 、m g_rec 、m w_con_i 、m s_reg_i They respectively represent the driving flue gas mass flow vector, the recovered flue gas mass flow vector, the condensate water mass flow vector at the condenser inlet, and the solution mass flow vector at the regenerator inlet.

[0106] The state performance parameter data set obtained by simulation calculation includes state parameter data and performance index data, specifically including the state parameter data of each position point of the system flue gas, solution working fluid and condensed water under different driving and recovered flue gas input conditions, specifically temperature, humidity, flow rate and concentration, heat exchange capacity of each component, and performance index data, the performance index data including the system comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB, state parameter data.

[0107] In addition, after obtaining the state performance parameter data set, a sensitivity analysis of the design parameters and the state performance parameters can also be performed. The sensitivity analysis is performed by calculating the Pearson correlation coefficient between the design parameter X and the state performance parameter Y. The calculation formula is:

[0108]

[0109] Among them, X i is the true value of the design parameter, Y i is the true value of the state performance parameter, and are the average values ​​of X and Y respectively, and N is the number of groups in the data set; the value of the correlation coefficient will be between -1 and 1. When |r|>0.8, the state performance parameters and design parameters are considered to be strongly correlated, while |r|<0.2 indicates weak correlation, and |r|=0 is irrelevant.

[0110] (3) The state performance parameter data set is subjected to outlier elimination and normalization processing, and the performance indicator data is used as labels, and the corresponding design parameter matrix and state parameter data are used as samples, which are input into the neural network model for training.

[0111] Among them, data preprocessing mainly involves drawing scatter plots to remove outliers and normalizing the maximum and minimum values. The calculation formula is:

[0112]

[0113] Among them, y normalize is the normalized data; y is the original data of various types; y min and max They are the minimum and maximum values ​​of each type of raw data. 80% of the preprocessed data set is divided into a training set and 20% is divided into a test set to train and verify the neural network model.

[0114] The performance index calculation formula is as follows:

[0115]

[0116] IC sys =1.4[(IC abs +IC con +IC reg +IC hs +IC pum +IC fan +IC s +IC w ) / 0.85]

[0117] In the formula, Q reg , Q abs , Q con Represent the heat of the regenerator, absorber, and condenser, respectively, h g_reg_i 、h w_con_i 、h g_abs_o 、h g_rec 、h env They represent the flue gas enthalpy at the regenerator inlet, the condensed water enthalpy at the condenser inlet, the flue gas enthalpy at the absorber outlet, the recovered flue gas enthalpy and the ambient air enthalpy, respectively. g_reg_i 、m w_con_i 、m w_conThey represent the flue gas flow rate at the regenerator inlet, the condensed water flow rate at the condenser inlet, and the water flow rate condensed from the flue gas in the condenser. abs 、IC con 、IC reg 、IC hx 、IC pum 、IC fan 、IC s 、IC w Respectively represent the initial investment of absorber, condenser, regenerator, heat exchanger, pump, fan, solution and water, P heat , P ele Represent the price of heat and electricity respectively, Q total Indicates the total heat supply, IC sys is the total investment, W sys The default system life cycle is 20 years, the bank discount rate is 4.9%, the pipe and processing costs account for 8% to 10% of the total system cost, the electric control cost accounts for about 5%, the profit is estimated at 30% to 40% of the total cost price, and the system maintenance cost ratio is 0.2.

[0118] The constructed neural network model includes input layer, output layer and hidden layer, such as Figure 2 As shown in the figure, the number of neurons in the input layer and the output layer are 11 and 3 respectively, and the input is the design parameter of the design parameter matrix (H abs ,H reg ,H con ,m g_reg_i ,m g_rec ,m w_con_i ,m s_reg_i ) and state parameters (condenser temperature t t_rec , condenser humidity w g_rec , regenerator inlet temperature t g_reg_i , Regenerator inlet humidity w g_reg_i ), double hidden layers and the number of hidden layer neurons is 10, the training function adopts Levenberg-Marquardt algorithm, the learning rate, number of iterations and target error are set to 0.1, 1000 and 0.001% respectively; the constructed neural network model is trained with the preprocessed data set until the convergence condition is met; the root mean square error RMSE and the goodness of fit R are selected 2 As a model evaluation indicator, the deviation between the predicted value and the actual value reflects the model prediction accuracy. The calculation formula is:

[0119]

[0120] Among them, y j is the true value of the output data, is the predicted value of the output data, is the average value of the true value of the output data, and n is the number of output predicted output data.

[0121] For neural network models that meet the accuracy requirements, model weights and thresholds need to be exported in a timely manner for subsequent coupled genetic algorithm model optimization.

[0122] (4) With the goal of maximizing the performance indicator data, a genetic algorithm is used to perform global optimization in the system design parameter space to determine the optimal value of the design parameter matrix; when the genetic algorithm is used for optimization, a trained neural network model is used to predict the performance indicator data corresponding to each design parameter matrix.

[0123] The calculation formulas for the system comprehensive performance coefficient COP and the recovery efficiency η are shown in step (3), and the calculation formula for the life cycle benefit LCB is as follows:

[0124] This step specifically includes:

[0125] (4.1) The design parameters in a design parameter matrix are organized into a chromosome, each design parameter is used as a gene of the chromosome, and the population is randomly initialized; the number of constructed populations is 20, and the number of evolution generations is 400;

[0126] (4.2) According to the population information, the trained neural network model is input to predict the performance index, and the performance index is used as the fitness value;

[0127] (4.3) After evaluating and comparing the fitness values, the chromosomes are crossed and mutated until the iterative setting conditions are met and the chromosome with the best fitness is obtained. The selection operator uses the roulette wheel selection method; the mutation and crossover probabilities are 0.3 and 0.5 respectively;

[0128] (4.4) Decode the chromosome with the best fitness and obtain the optimal design parameter matrix.

[0129] (5) The optimal design parameter matrix is ​​introduced into the physical model of the flue gas full heat deep recovery system for reverse verification. The physical model of the flue gas full heat deep recovery system calculates the state parameters of each position point and explains and analyzes it from a thermodynamic perspective.

[0130] The following simulation verification is carried out on the embodiment of the present invention. The simulation takes the typical working conditions of driving flue gas temperature and humidity parameters of 300℃ and 60g / kg and recovery flue gas temperature and humidity parameters of 70℃ and 110g / kg as an example. After randomly setting a group of control groups, the design optimization analysis is carried out with COP, η and LCB as the optimization targets. The optimized design parameters are specifically regeneration, absorption and condenser height (H reg , H abs , H con ), condenser inlet condensate and regenerator inlet solution flow (m w_con_i 、ms_reg_i ) and driving flue gas and recovered flue gas flow (m g_reg_i 、m g_rec ).

[0131] The training effect of the neural network model is shown in 3, and the fitting degree R 2 All are greater than 0.99, and the prediction parameter Q total The root mean square errors of , COP and η are 1.179kW, 0.080 and 0.004 respectively, indicating that the model can predict the system performance results well.

[0132] Coupled with the neural network model, the objective functions are set as COP, η and LCB respectively. The genetic algorithm optimization process is as follows: Figure 4 All optimization processes achieved convergence within 400 generations. Convergence to different values ​​indicates that the search has stabilized on the global minimum of the objective function. The changes in design parameters and performance parameters after optimization are shown in Table 1.

[0133] Table 1 Design parameters and performance parameters under different objective function optimization

[0134]

[0135] The control group and the three groups of optimized design parameters in Table 1 were respectively brought into the physical model for reverse verification and the energy flow diagrams were drawn as shown in Figure 5 As shown. Figure 5 From the analysis, it can be seen that the system performance coefficient is improved from 1.3427 to 1.8583 by taking the performance coefficient COP as the objective function for global optimization. The driving flue gas flow rate is reduced from 0.4kg / s to 0.3kg / s, which leads to a simultaneous decrease in the concentration and temperature of the dilute calcium chloride solution sprayed in the regenerator. However, the increase in the height of the regenerator (1.00m increased to 1.36m) and the condenser (1.1m increased to 1.57m) prolongs the contact time between the flue gas and the solution, and between the flue gas and pure water, and increases the heat and mass exchange contact area and the temperature and humidity gradient difference. Under the condition of a 25% reduction in the driving flue gas flow rate, the heat output of the regenerator and the condenser is reduced by 19%. The concentration of the concentrated calcium chloride solution circulating in the absorber is reduced, the surface water vapor partial pressure is significantly increased, and the heat and mass transfer rate between the flue gas and the solution is reduced. At the same time, the increase in the flow rate of flue gas to be recovered (from 0.2kg / s to 0.4974kg / s) ensured that the output heat supply of the absorber dropped from 56.3273KW to 55.1414KW, a slight decrease of only 2.1054%, while the proportion of driving flue gas heat consumption was reduced by 44.7933%. The final exhaust enthalpy of the system increased, the total driving heat consumption was greatly reduced, and the performance coefficient was significantly improved.

[0136] Taking the recovery efficiency η as the objective function for global optimization, the flue gas recovery efficiency was improved from 0.5429 to 0.7221. The driving flue gas flow rate increased from 0.4kg / s to 0.547kg / s, and the concentration and temperature of the dilute calcium chloride solution in the generator increased significantly after regeneration, and the temperature and humidity of the driving flue gas discharged from the generator increased synchronously. The increase in the condenser height and the condensation water flow rate caused the waste heat and moisture in the driving flue gas to be recovered in large quantities in the condenser, among which the sensible heat recovery and latent heat recovery increased by 8.4328KW and 19.6192KW respectively, and the latent heat percentage increased slightly by 0.4557%. The state parameters of the driving flue gas discharged after condensation changed little compared with the control group. At the same time, the concentrated calcium chloride solution circulating to the absorber inlet remained basically unchanged due to the heat exchange through two wall-type heat exchangers. The increase in concentration caused the equivalent moisture content of the solution to decrease by 43.4498%, that is, the absorption capacity of the solution increased significantly. After optimization, the flow rate of flue gas to be recovered was reduced from 0.2kg / s to 0.1kg / s. The reduction in the flow rate of recovered flue gas, combined with the increase in the absorber height (from 1.2m to 1.7524) and the increase in the absorption capacity of the solution, enhanced the heat and mass transfer efficiency between the flue gas and the solution. The total recovered heat in the absorber increased from 56.3273KW to 94.3390KW. The humidity of the flue gas finally discharged from the system was reduced by 22.6404g / kg (44.1392%), and the enthalpy value was reduced by 58.1046kj / kg (31.5076%), thereby greatly increasing the flue gas recovery efficiency.

[0137] Taking the system life cycle benefit LCB as the objective function for global optimization, the total benefit increased from 583568CNY to 1140091CNY, an increase of 95.3656%. Since the system adopts direct contact heat and mass transfer, there is no need to maintain a vacuum environment. Compared with the initial investment and operation and maintenance costs, the heating benefit occupies a dominant position. Therefore, the main core of maximizing the system's life cycle benefits is to achieve the maximum output heat supply. The system heat supply comes from the sensible heat and latent heat recovered from the driving and recovered flue gases. The optimized driving flue gas and recovered flue gas flow rates increased to 0.5476kg / s and 0.4887kg / s respectively, the concentration level in the generator increased, the equivalent moisture content of the solution decreased from 45.8g / kg to 37.6g / kg, and the absorption capacity increased. At the same time, the height of the generator, condenser and absorber also increased synchronously. The increase in the amount of heat source and the enhancement of the heat and mass transfer effect caused the system heat supply to increase from 163.23KW to 288.7768KW, and the system life cycle benefits were significantly improved.

[0138] Embodiment 2

[0139] An embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method of the first embodiment.

[0140] The memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as "hard drives"). A program / utility having a set (at least one) of program modules may be stored in, for example, memory 301, such program modules including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The computer executable program of the program module typically performs the functions and / or methods in the embodiments described in the present invention.

[0141] The processor executes various functional applications and data processing by running the program stored in the memory, such as implementing the method provided in the first embodiment of the present invention.

[0142] The code of the computer executable program can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages.

[0143] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. A parameter optimization design method for a fully open flue gas full heat depth recovery system based on solution, characterized in that: The steps include: (1) Based on the direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, a physical model of the flue gas full heat deep recovery system based on solution was established; (2) Based on the Monte Carlo simulation concept, a number of design parameter matrices are randomly generated, and the physical model of the flue gas full heat depth recovery system is imported for simulation calculation to obtain a state performance parameter data set for characterizing the state and performance of the flue gas full heat depth recovery system; the state performance parameter data set includes state parameter data and performance index data, and the performance index data includes system comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB; (3) removing outliers and normalizing the state performance parameter data set, and using the performance index data as labels, the corresponding design parameter matrix and the state parameter data as samples, and inputting them into the neural network model for training; (4) With the goal of maximizing the performance indicator data, a genetic algorithm is used to perform global optimization in the system design parameter space to determine the optimal value of the design parameter matrix; during the global optimization, a trained neural network model is used to predict the performance indicator data corresponding to each design parameter matrix.

2. The parameter optimization design method of the solution-based full-open flue gas full heat depth recovery system according to claim 1 is characterized in that: Step (1) comprises: (1.1) obtaining direct contact heat and mass transfer experimental data between flue gas and solution, and between flue gas and condensed water, wherein the direct contact heat and mass transfer experimental data include the inlet and outlet temperature, humidity, concentration and flow rate of flue gas and solution, and the inlet and outlet temperature, humidity and flow rate of flue gas and condensed water under different temperatures, humidity, concentrations and flow rates; (1.2) According to the direct contact heat and mass transfer experimental data, the heat and mass transfer coefficients between flue gas and solution, and between flue gas and condensed water are fitted to obtain the fitting expressions of the heat and mass transfer coefficients; (1.3) According to the fitting expression of heat and mass transfer coefficients, the equivalent energy flow model of absorber, regenerator, condenser and intermediate wall heat exchanger is constructed, and the equivalent energy flow model of each device is coupled through the corresponding isothermal points and energy and mass conservation to obtain the physical model of solution-based flue gas full heat deep recovery system.

3. The parameter optimization design method of the solution-based fully open flue gas full heat depth recovery system according to claim 2 is characterized in that: The fitting expression of the heat and mass transfer coefficient is specifically: h c =1.114G s 1.019 ·G g 1.002 h d =0.9953G s 1.038 ·G g 1.014 h c =1.534G w 0.9609 ·G g 1.007 h d =1.311G w 1.026 ·G g 1.003 h c =1.75G s 0.8333 ·G g 0.814 h d =2.481G s 0.7756 ·G g 0.8101 In the formula, G s , G w , G g Represent the mass flow rate of solution, condensed water and flue gas respectively, h c represents the heat transfer coefficient, h d represents the mass transfer coefficient.

4. The method for optimizing the parameters of the solution-based fully open flue gas full heat depth recovery system according to claim 2 is characterized in that: The physical model of the flue gas full heat depth recovery system is specifically as follows: Equivalent energy flow model of absorber and regenerator: -Cp s ·m s ·dt s -Cp s ·dm s ·t s =m g ·dh g +h dil ·m g ·dω g dm s =-m g ·dω g h c ·πR 2 ·dH·α(t s -t g )=m g ·(Cp g +ω g ·Cp v )·dt g h d ·πR 2 ·dH·α(ω s -oh g )=m g ·dω g X s ·m s =(X s +dX s )(m s +dm s ) Condenser equivalent energy flow model: -Cp w ·m w ·dt w -Cp w ·dm w ·t w =m g ·dh g dm w =-m g ·dω g h c ·πR 2 ·dH·α(t w -t g )=m g ·(Cp g +ω g ·Cp v )·dt g h d ·πR 2 ·dH·α(ω w -oh g )=m g ·dω g Equivalent energy flow model of inter-wall heat exchanger: Q hx =Cp w m w ·(t w_i -t w_o ) Q hx =K hx ·A hx ·LMTD hx In the formula, m s ,t s , Cp s , P s ,ω s , X s Respectively represent the mass flow rate, temperature, specific heat, water vapor partial pressure, equivalent moisture content, and concentration of the solution, m g ,t g , P g ,ω g 、h g Respectively represent the mass flow rate, temperature, water vapor partial pressure, equivalent moisture content, and enthalpy of flue gas, m w ,t w , Cp w , P w ,ω w They represent the mass flow rate, temperature, specific heat, water vapor partial pressure, and equivalent moisture content of water, respectively. dil represents the dilution enthalpy, P0 represents the local atmospheric pressure, α and R represent the specific surface area and radius of the ceramic structured packing material, H represents the height of the open spray packing tower, that is, the height of the absorber, regenerator or condenser, Cp v represents the specific heat of water vapor, Cp g represents the specific heat of dry air, d represents the derivative, Q hx , K hx , A hx LMTD hx They represent the heat, heat transfer coefficient, area and logarithmic mean temperature difference of the inter-wall heat exchanger, t w_i ,t w_o Respectively represent the inlet and outlet water temperatures, t s_i ,t s_o Represent the inlet and outlet solution temperatures respectively.

5. The method for optimizing the parameters of the solution-based fully open flue gas full heat depth recovery system according to claim 1 is characterized in that: The design parameter matrix in step (2) is specifically: A design =[H abs ,H reg ,H con ,m g_reg_i ,m g_rec ,m w_con_i ,m s_reg_i ] In the formula, A design represents the design parameter matrix, H abs , H reg , H con Represent the height vectors of the absorber, regenerator, and condenser, respectively, m g_reg_i 、m g_rec 、m w_con_i 、m s_reg_i They represent the driving flue gas mass flow vector, the recovered flue gas mass flow vector, the condensate water mass flow vector at the condenser inlet, and the solution mass flow vector at the regenerator inlet respectively; The state parameter data specifically include the state parameter data of each position point of the system flue gas, solution working fluid and condensed water under different driving and recovered flue gas input conditions, specifically temperature, humidity, flow rate and concentration, as well as the heat exchange capacity of each component.

6. The method for optimizing the parameters of the fully open flue gas full heat depth recovery system based on solution according to claim 1 is characterized in that: After the state performance parameter data set is obtained in step (2), a sensitivity analysis of the design parameters and the state performance parameters is performed. The sensitivity analysis is achieved by calculating the Pearson correlation coefficient between the design parameter X and the state performance parameter Y.

7. The method for optimizing the parameters of the fully open flue gas full heat depth recovery system based on solution according to claim 1 is characterized in that: The calculation formulas for the system comprehensive performance coefficient COP, recovery efficiency η and life cycle benefit LCB are as follows: IC sys =1.4[(IC abs +IC con +IC reg +IC hx +IC pum +IC fan +IC s +IC w ) / 0.85] In the formula, Q reg , Q abs , Q con Represent the heat of the regenerator, absorber, and condenser, respectively, h g_reg_i 、h w_con_i 、h g_abs_o 、h g_rec 、h env They represent the flue gas enthalpy at the regenerator inlet, the condensed water enthalpy at the condenser inlet, the flue gas enthalpy at the absorber outlet, the recovered flue gas enthalpy and the ambient air enthalpy, respectively. g_reg_i 、m w_con_i 、m w_con They represent the flue gas flow rate at the regenerator inlet, the condensed water flow rate at the condenser inlet, and the water flow rate condensed from the flue gas in the condenser. abs 、IC con 、IC reg 、IC hx 、IC pum 、IC fan 、IC s 、IC w Respectively represent the initial investment of absorber, condenser, regenerator, heat exchanger, pump, fan, solution and water, P heat , P ele represent the price of heat and electricity respectively, Q total Indicates the total heat supply, IC sys is the total investment, W sys is the total power.

8. The method for optimizing the parameters of a fully open flue gas heat recovery system based on solution according to claim 1 is characterized in that: Step (4) specifically includes: (4.1) The design parameters in a design parameter matrix are organized into a chromosome, each design parameter is used as a gene of the chromosome, and the population is randomly initialized; (4.2) According to the population information, the trained neural network model is input to predict the performance index data, and the performance index data is used as the fitness value; (4.3) After evaluating and comparing the fitness values, the chromosomes are crossed and mutated until the iterative setting conditions are met and the chromosome with the best fitness is obtained. The selection operator uses the roulette wheel selection method; (4.4) Decode the chromosome with the best fitness and obtain the optimal design parameter matrix.

9. The method for optimizing the parameters of a fully open flue gas heat recovery system based on solution according to claim 1 is characterized in that: After step (4), the method further includes: (5) The optimal design parameter matrix is ​​introduced into the physical model of the flue gas full heat deep recovery system for reverse verification. The physical model of the flue gas full heat deep recovery system calculates the state parameters of each position point and explains and analyzes it from a thermodynamic perspective.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the computer program to implement the method according to any one of claims 1 to 9.

Images