Performance enhancing method of efficient molten salt heat storage and heat exchange system

Abstract

The invention relates to a performance enhancement method of an efficient molten salt heat storage and heat exchange system, which comprises the following steps: collecting system operation data and equipment parameters, preprocessing, constructing a performance analysis model based on energy conservation and heat transfer theory principles, predicting system performance in combination with a machine learning algorithm, and determining key optimization parameters and a range thereof. And an optimal parameter combination is searched by utilizing simulation software and an optimization algorithm, the system is transformed or adjusted according to the optimal parameter combination, and a real-time monitoring system is built. If the actual data and the simulation data have deviation, the reason is analyzed and dynamically adjusted; by predicting the operation condition of the system in advance, adjusting the cleaning period in time, accurately analyzing the energy conversion efficiency and the like, heat transfer temperature difference loss and energy loss can be reduced, the heat exchange efficiency is improved, the overall energy utilization rate is increased, and the adaptability of the system to complex working conditions is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of molten salt thermal energy storage and heat exchange systems, and specifically to a method for enhancing the performance of an efficient molten salt thermal energy storage and heat exchange system. Background Art

[0002] The molten salt thermal energy storage and heat exchange system is a core part for realizing stable power output. The prior art publication number CN109654756A discloses a molten salt thermal energy storage system for a solar thermal power plant and its heat exchange method, which consists of a high-temperature molten salt storage tank, a mixer, a preheater, a low-temperature molten salt storage tank, a mirror field, and multiple pipelines. Multiple loops are formed through different pipeline connection methods, and the mixer is used to mix molten salts from different sources. It is also equipped with multiple control valves, and different valve opening and closing states correspond to different operating loops. The system includes equipment such as an evaporator and a superheater, each having its heat exchange function, and there are submerged molten salt pumps at the tops of the high-temperature and low-temperature molten salt storage tanks. However, the prior art still has many limitations. Although the system is provided with heat exchange equipment such as a preheater, an evaporator, and a superheater, in actual operation, it is difficult to achieve an ideal heat exchange efficiency. On the one hand, there are temperature difference losses in the heat transfer process between the molten salt and the water working medium, and the thermal energy of the molten salt cannot be fully utilized; on the other hand, the heat loss of the equipment cannot be ignored, which affects the overall energy utilization efficiency of the system. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for enhancing the performance of an efficient molten salt thermal energy storage and heat exchange system to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides the following technical solutions: 1. A method for enhancing the performance of an efficient molten salt thermal energy storage and heat exchange system, including the following steps:

[0005] S1: Collect system operation data and equipment parameters, including molten salt temperature, flow rate, pressure, data related to heat exchangers, as well as external data such as environmental temperature and solar radiation intensity;

[0006] S2: Preprocess the collected data, calculate the mean value, standard deviation, remove outliers and normalize;

[0007] S3: Build a system performance analysis model, establish an energy balance model based on the principles of energy conservation and heat transfer, and build a performance prediction model in combination with machine learning algorithms;

[0008] S4: Determine the key optimization parameters and their ranges, including molten salt physical properties, heat storage tank and heat exchanger structure parameters, operation parameters, and set the target values and constraint conditions according to the design objectives and costs;

[0009] S5: Use simulation software to perform simulation calculations based on the analysis model and parameter range. Consider different operating condition factors, and find the optimal parameter combination by comparing indicators and using optimization algorithms.

[0010] S6: Modify or adjust the system according to the optimal parameter combination, and build a real-time monitoring system to compare the actual and simulated data.

[0011] S7: If there is a deviation, analyze the reasons, dynamically adjust the optimization parameters and operation strategies, continuously collect data for performance analysis and optimization, so as to continuously improve the system performance.

[0012] Further, in the above S1, the molten salt flow rate data Q(t) collected is measured by an electromagnetic flowmeter with a measurement accuracy of ±0.5%. Its measurement principle is based on Faraday's law of electromagnetic induction. The relationship between the induced electromotive force E, the molten salt flow velocity v, the pipe diameter D, and the magnetic induction intensity B is E = B·v·D. The molten salt flow rate is calculated by measuring the induced electromotive force.

[0013] Further, in the above S2, when calculating the data mean and standard deviation, the number of data samples n is determined according to the system operation stability. When the system operates stably, n takes the value of the past continuous 7 days, with 24 data points collected every day, that is, n = 7×24; if the system operates unstably, then n is the past continuous 3 days, with one data point collected every hour, that is, n = 3×24.

[0014] Further, in the above S3, in the energy balance model, the logarithmic mean temperature difference ΔT hx (t) in the heat transfer formula Q m (t) of the heat exchanger is calculated by the following formula:

[0015] where ΔT1(t) is the temperature difference between the hot fluid inlet and the cold fluid outlet of the heat exchanger, and ΔT2(t) is the temperature difference between the hot fluid outlet and the cold fluid inlet.

[0016] Further, in the above S3, the machine learning algorithm uses a backpropagation neural network to construct a performance prediction model; this network includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is determined according to the input data dimension. The number of nodes N h in the hidden layer is calculated through the empirical formula

[0017] where i is the number of nodes in the input layer, o is the number of nodes in the output layer, and a is a constant between 1 and 10; when training the network, the mean square error MSE is used as the loss function, is the number of training samples, i is the true value, is the predicted value.

[0018] Furthermore, in said S4, the specific heat capacity C of the molten salt is p The optimization target value is determined according to different temperature ranges; in the 300–400°C range, the target value is 1.5-1.8 kJ / (kg·K); in the 400-500°C range, the target value is 1.8-2.0 kJ / (kg·K), and its value must meet the requirements of the overall energy balance equation of the system for heat storage capacity.

[0019] Furthermore, in the above S5, when ANSYS Fluent is used for simulation, the finite volume method is used to discretize the control equations; taking the two-dimensional steady-state heat transfer problem as an example, for the heat conduction equation The discretized equation is a p T p =∑ n a n T n , a p and a n is the coefficient of dispersion, T p is the central node temperature, T n is the temperature of the adjacent nodes, and the temperature distribution is obtained by iteratively solving the discrete equation.

[0020] Furthermore, in said S5, when the genetic algorithm is used to find the optimal parameter combination, the parameter coding adopts binary coding, and the parameter value range is [x min , x max ], the encoding length is l, then the formula for converting the binary encoding value b corresponding to the parameter x into the actual value is

[0021] The selection operation of the genetic algorithm adopts the roulette wheel selection method, and the probability of an individual being selected is f i is the fitness value of individual i, and N is the population size.

[0022] Furthermore, in the above S6, the temperature sensor of the real-time monitoring system adopts a K-type thermocouple with a measurement accuracy of ±1°C. The thermoelectric potential E of the thermocouple is related to the temperature T of the measuring end and the reference end. O The relationship is S(T) is the Seebeck coefficient; the pressure sensor uses a strain gauge pressure sensor, and its output voltage U O The relationship with pressure P is U O =kP, k is the sensor sensitivity.

[0023] Furthermore, in S7, when the actual heat exchange efficiency η a Lower than the simulation optimization value η s The method to judge whether the heat exchanger is fouled is to calculate the fouling thermal resistance

[0024] K a is the actual overall heat transfer coefficient, K d is the designed overall heat transfer coefficient; if R f > 0.0001 m 2 ·K / W, it is determined that the heat exchanger is fouled and the cleaning cycle needs to be adjusted. The new cleaning cycle t n is calculated according to the fouling rate r R f，max is the maximum allowable fouling thermal resistance.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0026] By constructing a system performance analysis model and combining simulation optimization calculations, the present invention can predict in advance the operating conditions of the system under different working conditions and take measures in advance to avoid failures. When the heat exchanger may be fouled, the cleaning cycle is adjusted in a timely manner according to the calculated fouling thermal resistance. A performance prediction model is constructed using a suitable machine learning algorithm to accurately analyze the energy conversion efficiency under different working conditions, find the optimal parameter combination, reduce the heat transfer temperature difference loss and energy loss, improve the heat exchange efficiency, enable the system to make more full use of the thermal energy of molten salt, and improve the overall energy utilization rate. Continuously collect external data such as ambient temperature and solar radiation intensity, and conduct comprehensive analysis in combination with the system operation data, so that the system can automatically adjust the operation strategy according to the changes in external conditions. In extremely cold weather, automatically increase the heating power or adjust the flow rate of molten salt to prevent the molten salt from freezing; in high-temperature weather, optimize the heat dissipation measures to ensure the normal operation of the system and enhance the adaptability of the system to complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0029] Please refer to Figure 1 , a method for enhancing the performance of an efficient molten salt heat storage and heat exchange system provided by the present invention includes the following steps:

[0030] S1: Collect system operation data and equipment parameters, including molten salt temperature, flow rate, pressure, heat exchanger-related data, and external data such as ambient temperature and solar radiation intensity;

[0031] The collected molten salt flow rate data Q(t) is measured by an electromagnetic flowmeter with a measurement accuracy of ±0.5%. Its measurement principle is based on Faraday's law of electromagnetic induction. The relationship between the induced electromotive force E, the molten salt flow velocity v, the pipe diameter D, and the magnetic induction intensity B is E = B·v·D. The molten salt flow rate is calculated by measuring the induced electromotive force.

[0032] In addition to the basic data, collect the heat exchange area of the heat exchanger, the changes in the heat transfer coefficient, and the molten salt composition data. For the heat exchanger, record in detail its plate material, thickness, corrugation shape, and size, as these parameters affect fluid flow and heat transfer efficiency. For the molten salt composition, regularly detect the impurity content because impurities will change the thermophysical properties of the molten salt, thereby affecting the system performance.

[0033] Calibrate the electromagnetic flowmeter regularly, at least once every six months. The calibration uses a standard flow source. Connect the electromagnetic flowmeter to the standard flow source, compare the measured value with the actual flow value, and perform error correction. During installation, use an electromagnetic interference detector to evaluate the electromagnetic environment at the installation location, select the location with the least interference for installation, and use a metal shielding tube to shield the electromagnetic flowmeter to reduce electromagnetic interference.

[0034] S2: Preprocess the collected data, calculate the mean, standard deviation, remove outliers and normalize;

[0035] When calculating the data mean and standard deviation, the number of data samples n is determined according to the system operation stability. When the system runs stably, n takes the value of the past 7 consecutive days, with 24 data points collected every day, that is, n = 7×24; if the system runs unstably, then n is the past 3 consecutive days, with 1 data point collected every hour, that is, n = 3×24.

[0036] When the system is unstable, use a moving window to calculate the mean and standard deviation, and set the window duration to 3 days. After each new data collection, recalculate the mean and standard deviation of the hourly data within the window to reflect the system operation changes in a timely manner.

[0037] The mean calculation formula for the first window is: The mean calculation formula for K moving windows is: The size of the moving window is m.

[0038] Use the Isolation Forest algorithm to assist in detecting outliers. Compare the outliers detected based on the mean and standard deviation with the results of the Isolation Forest algorithm. If both determine it as an outlier, then determine that data as an outlier and remove it to improve the accuracy of outlier detection.

[0039] First, train the Isolation Forest model using normal data to obtain the model's parameters. Then, input the data points to be detected into the model and calculate their anomaly scores. When the anomaly score exceeds the pre-set threshold of 0.5, the data point is determined to be an outlier. Compare the detection results of the Isolation Forest algorithm with the outliers detected based on the mean and standard deviation. If both determine it to be an outlier, then confirm that the data is an outlier and remove it, thereby improving the accuracy of outlier detection and providing a more reliable data basis for subsequent data processing and system performance analysis.

[0040] S3: Build a system performance analysis model, establish an energy balance model based on the principles of energy conservation and heat transfer, and combine with machine learning algorithms to build a performance prediction model;

[0041] In the energy balance model, the logarithmic mean temperature difference ΔT hx (t) in the heat transfer formula Q m (t) is calculated by the formula:

[0042] where ΔT1(t) is the temperature difference between the hot fluid inlet and the cold fluid outlet of the heat exchanger, and ΔT2(t) is the temperature difference between the hot fluid outlet and the cold fluid inlet.

[0043] The machine learning algorithm uses a backpropagation neural network to build a performance prediction model; this network includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is determined according to the dimension of the input data, and the number of nodes N h in the hidden layer is calculated through the empirical formula

[0044] where N i is the number of nodes in the input layer, N o is the number of nodes in the output layer, and a is a constant between 1 and 10; during network training, the mean squared error MSE is used as the loss function, where i is the number of training samples, y i is the true value, and is the predicted value.

[0045] Consider heat loss from the pipeline and throttling losses of the valve, and improve the energy balance model. For heat loss from the pipeline, based on the thermal conductivity of the pipeline insulation material, pipe diameter, length, and ambient temperature, use the heat conduction formula (where Q is the heat dissipation, λ is the thermal conductivity, A is the pipeline surface area, ΔT is the temperature difference between the inside and outside of the pipeline, and L is the pipeline length) to calculate. The throttling loss of the valve is calculated from the pressure difference and flow rate before and after the valve.

[0046] S4: Determine the key optimization parameters and their ranges, including molten salt physical properties, structural parameters of the heat storage tank and heat exchanger, and operating parameters, and set the target values and constraints according to the design objectives and costs;

[0047] Specific heat capacity C of molten salt p The optimized target value is determined according to different temperature ranges; in the range of 300–400 °C, the target value is 1.5 - 1.8 kJ / (kg·K); in the range of 400 - 500 °C, the target value is 1.8 - 2.0 kJ / (kg·K), and its value needs to meet the requirements of the overall energy balance equation of the system for the heat storage capacity.

[0048] S5: Using simulation software, perform simulation calculations based on the analysis model and parameter range, considering different operating conditions factors, and find the optimal parameter combination by comparing indicators and using optimization algorithms;

[0049] When using ANSYS Fluent for simulation, the finite volume method is used to discretize the control equations; taking a two-dimensional steady-state heat transfer problem as an example, for the heat conduction equation The discretized equation is a p T p = ∑ n a n T n where a p and a n are discretization coefficients, T p is the central node temperature, and T n is the adjacent node temperature. The temperature distribution is obtained by iteratively solving this discretized equation.

[0050] When using the genetic algorithm to find the optimal parameter combination, binary coding is used for parameter coding. The parameter value range is [x min , x max , and the coding length is l. Then the formula for converting the binary coding value b corresponding to the parameter x into the actual value is

[0051] For the selection operation of the genetic algorithm, the roulette wheel selection method is used. The probability f i that an individual is selected is the fitness value of individual i, and N is the population size. The genetic algorithm introduces an elite retention strategy, and the 5% individuals with the highest fitness in each generation are directly retained and enter the next generation. Combining with the particle swarm optimization algorithm, use its global search ability in the initial stage of the genetic algorithm to quickly locate the optimal solution region, and use the selection, crossover, and mutation operations of the genetic algorithm for local fine search in the later stage to improve the optimization efficiency.

[0052] Try various discretization methods such as the finite element method and the finite difference method. For the same simulation operating condition, perform simulations using different methods respectively, compare the errors between the simulation results and the actual test data, and select the discretization method with the smallest error for system simulation.

[0053] S6: Modify or adjust the system according to the optimal parameter combination, and build a real-time monitoring system to compare the actual and simulation data;

[0054] The temperature sensor of the real-time monitoring system uses a K-type thermocouple with a measurement accuracy of ±1°C. Its thermoelectromotive force E and the temperatures T and T of the measurement end and the reference end O have the relationship where S(T) is the Seebeck coefficient; the pressure sensor uses a strain gauge pressure sensor, and its output voltage U O has the relationship with the pressure P as U O = kP, where k is the sensor sensitivity.

[0055] S7: If there is a deviation, analyze the reasons, dynamically adjust and optimize the parameters and operation strategies, continuously collect data for performance analysis and optimization to continuously improve the system performance;

[0056] When the actual heat exchange efficiency η a is lower than the simulated and optimized value η s , the method for judging whether the heat exchanger is fouled is: calculate the fouling thermal resistance

[0057] K a is the actual overall heat transfer coefficient, and K d is the designed overall heat transfer coefficient; if R f > 0.0001 m 2 ·K / W, it is determined that the heat exchanger is fouled, and the cleaning cycle needs to be adjusted. The new cleaning cycle t n is calculated according to the fouling rate r, R f，max is the maximum allowable fouling thermal resistance. Optimize the fouling countermeasures; consider on-line cleaning when the heat exchanger is fouled and select ultrasonic cleaning according to the fouling type. Establish a fouling prediction model for the heat exchanger, predict the fouling trend based on the system operation data, and adjust the operation strategy in advance.

Claims

1. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system, characterized in that, It includes the following steps: S1: Collect system operation data and equipment parameters, covering molten salt temperature, flow rate, pressure, data related to heat exchangers, as well as external data such as ambient temperature and solar radiation intensity; S2: Preprocess the collected data, calculate the mean value, standard deviation, remove outliers and normalize; S3: Build a system performance analysis model, establish an energy balance model based on the principles of energy conservation and heat transfer, and build a performance prediction model in combination with machine learning algorithms; S4: Determine the key optimization parameters and their ranges, including molten salt physical properties, heat storage tank and heat exchanger structure parameters, operation parameters, and set the target values and constraints according to the design objectives and costs; S5: Use simulation software to perform simulation calculations based on the analysis model and parameter ranges, consider different working conditions factors, and find the optimal parameter combination by comparing indicators and using optimization algorithms; S6: Modify or adjust the system according to the optimal parameter combination, build a real-time monitoring system to compare the actual and simulated data; S7: If there are deviations, analyze the reasons, dynamically adjust the optimization parameters and operation strategies, continuously collect data for performance analysis and optimization to achieve continuous improvement of system performance.

2. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In the step S1, the collected molten salt flow rate data Q(t) is measured by an electromagnetic flowmeter with a measurement accuracy of ±0.5%. Its measurement principle is based on Faraday's law of electromagnetic induction. The relationship between the induced electromotive force E, the molten salt flow velocity v, the pipe diameter D, and the magnetic induction intensity B is E = B·v·D. The molten salt flow rate is calculated by measuring the induced electromotive force 3. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In S2, when calculating the mean value and standard deviation of the data, the number of data samples n is determined according to the system operation stability. When the system operates stably, n takes the value of the past 7 consecutive days, with 24 data points collected every day, that is, n = 7×24; If the system operates unstably, then n is the past 3 consecutive days, with 1 data point collected every hour, that is, n = 3×24.

4. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In S3, the logarithmic mean temperature difference ΔT hx (t) in the heat transfer formula Q of the heat exchanger in the energy balance model m (t) is calculated by the formula: Where ΔT1(t) is the temperature difference between the hot fluid inlet and the cold fluid outlet of the heat exchanger, and ΔT2(t) is the temperature difference between the hot fluid outlet and the cold fluid inlet.

5. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In S3, the machine learning algorithm uses a backpropagation neural network to build a performance prediction model; the network includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is determined according to the dimension of the input data, and the number of nodes in the hidden layer is N h through an empirical formula Calculation, N i is the number of nodes in the input layer, N o is the number of nodes in the output layer, a is a constant between 1 and 10; the mean squared error MSE is used as the loss function during network training, is the number of training samples, y i is the true value, is the predicted value.

6. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that In S4, the specific heat capacity C of the molten salt p The optimized target value is determined according to different temperature ranges; in the range of 300–400 °C, the target value is 1.5-1.8 kJ / (kg·K); in the range of 400-500 °C, the target value is 1.8-2.0 kJ / (kg·K), and its value needs to meet the requirements of the overall energy balance equation of the system for the heat storage capacity.

7. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that In S5, when using ANSYS Fluent for simulation, the finite volume method is used to discretize the control equations; Taking the two-dimensional steady-state heat transfer problem as an example, for the heat conduction equation The discretized equation is a p T p = ∑ n a n T n , a p and a n are the discretization coefficients, T p is the temperature of the central node, T n is the temperature of the adjacent node, and the temperature distribution is obtained by iteratively solving the discretized equation.

8. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that In S5, when using the genetic algorithm to find the optimal parameter combination, binary coding is used for parameter coding, the parameter value range is [x min , x max , and the coding length is l. Then the formula for converting the binary coding value b corresponding to the parameter x into the actual value is The selection operation of the genetic algorithm uses the roulette wheel selection method, and the probability that an individual is selected f i is the fitness value of individual i, and N is the population size.

9. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In S6, the temperature sensor of the real-time monitoring system uses a K-type thermocouple with a measurement accuracy of ±1°C. Its thermoelectromotive force E and the temperatures T and T of the measurement end and the reference end O have the relationship of where S(T) is the Seebeck coefficient; the pressure sensor uses a strain gauge pressure sensor, and the relationship between its output voltage U O and the pressure P is U O = kP, where k is the sensor sensitivity.

10. A method for enhancing the performance of an efficient molten salt thermal storage and heat exchange system according to claim 1, characterized in that, In the S7, when the actual heat exchange efficiency η a is lower than the simulated and optimized value η s , the method for determining whether the heat exchanger is fouled is: calculate the fouling heat resistance K a is the actual overall heat transfer coefficient, K d is the designed overall heat transfer coefficient; if R f > 0.0001 m 2 ·K / W, it is determined that the heat exchanger is fouled and the cleaning cycle needs to be adjusted. The new cleaning cycle t n is calculated according to the fouling rate r R f，max is the maximum allowable fouling thermal resistance.

Images