Low-temperature steam generation system capable of intelligently regulating and controlling gas production rate and regulating and controlling method

Abstract

The invention relates to a low-temperature steam generation system capable of intelligently regulating and controlling the gas production rate and a regulating and controlling method, and aims to solve the technical problem that the steam production rate is unstable when a coal-fired power plant fluctuates. The intelligent control system inputs parameters related to boiler combustion, an intelligent neural network prediction model is obtained through analysis and training, the output unit outputs the group number of heat exchanger modules according to boiler loads and liquid level values of a first liquid level meter and a second liquid level meter, and the group number of the input heat exchanger modules, the liquid level of a steam drum or the liquid level of a heat exchanger are regulated and controlled. According to the method, the boiler load can be matched with the group number of the heat exchanger modules put into operation, and the problem of steam parameter lag caused by boiler load change of a traditional system is effectively solved, so that stable and reliable operation of the multi-effect evaporation system is realized, and collaborative coupling of desulfurization wastewater low-energy-consumption treatment and unit load is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of desulfurization and environmental protection of coal-fired power plants, and in particular relates to a low-temperature steam generation system and a control method capable of intelligently controlling gas production. Background Art

[0002] The existing desulfurization wastewater from coal-fired power plants is mainly treated by a multi-effect evaporation concentration process, and its evaporation heat source comes from the waste heat of the flue gas. Specifically, a heat exchanger is installed in the flue between the dust collector and the first-effect heater, and the high-temperature flue gas of 110-140°C is used to heat the circulating water in the heat exchanger to generate steam, which is then used to heat the desulfurization wastewater. Since the steam output directly affects the operating effect of the entire multi-effect evaporation concentration system, the stability of steam output is crucial. However, due to the influence of many factors such as power supply and demand balance, energy-saving and emission reduction policies, regional power grid dispatching, and new energy policies, the load of coal-fired power plants is in a dynamic fluctuation state. Load changes cause fluctuations in flue gas volume and flue gas temperature, which in turn affects the steam output of the heat exchanger, resulting in unstable steam output.

[0003] Traditional flue heat exchangers have the following main problems:

[0004] 1. The contact area between the flue heat exchanger and the circulating water is small, the heat exchanger is large, the cost is high, and the installation is inconvenient;

[0005] 2. When the load of coal-fired power plants fluctuates, the steam production is unstable and cannot be adjusted efficiently and quickly;

[0006] 3. Low degree of automation control;

[0007] 4. Affect the efficient and stable operation of the subsequent multiple-effect evaporation concentration process. Summary of the invention

[0008] The purpose of the present invention is to solve the above-mentioned problem and to provide a low-temperature steam generation system and a control method capable of intelligently controlling the gas production.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0010] A low-temperature steam generation system capable of intelligently regulating gas production, comprising a steam drum, a steam pipe, a first-effect heater, a condensate tank, a water inlet pipe, a vacuum pump, a feed water pump and a sewage tank, wherein the steam drum is connected to the inlet of the first-effect heater through the steam pipe, the outlet of the first-effect heater is connected to the inlet of the condensate tank, the bottom outlet of the condensate tank is connected to the steam drum through the water inlet pipe, the feed water pump is provided on the water inlet pipe, the top outlet of the condensate tank is connected to an exhaust pipe, the exhaust pipe is provided with a vacuum pump, and the other end of the exhaust pipe extends to the atmosphere;

[0011] It also includes several groups of heat exchanger modules and intelligent control systems connected in parallel. The heat exchanger module includes a heat exchanger body, several finned heat exchange tubes, an ascending tube, a downcomer, a connecting valve and a three-way switching valve. The upper header on one side of the heat exchanger body is connected to the lower end of the ascending tube, the upper end of the ascending tube is connected to the steam drum, the upper end of the downcomer is connected to the steam drum, and the lower end is connected to the lower header on the other side of the heat exchanger body. The two ends of the finned heat exchange tube are respectively connected to the upper header and the lower header. The bottoms of the several heat exchanger bodies are connected by a connecting valve. The three-way switching valve is arranged at the upper end of the ascending tube, and one end of the three-way switching valve leads to the steam drum, and the other end leads to the atmosphere. A control valve is arranged on the downcomer, a first liquid level gauge is arranged on the heat exchanger body, and a second liquid level gauge is arranged on the steam drum. The heat exchanger body is connected to the sewage tank through a drain pipe, and a drain valve is arranged on the drain pipe;

[0012] The intelligent control system includes an input unit, an analysis unit and an output unit;

[0013] The input unit is used to input parameters related to boiler combustion, including boiler load, coal feed rate, primary air volume, secondary air volume, flue gas temperature and flue gas volume;

[0014] The analysis unit includes a data processing unit and a model training unit. The data processing unit is used to process the data of the input unit; the model training unit is used to construct an intelligent neural network prediction model, and train the intelligent neural network prediction model according to the data processed by the data processing unit to obtain a trained intelligent neural network prediction model;

[0015] The output unit is used to output the number of heat exchanger modules, the liquid level values ​​of the first liquid level gauge and the second liquid level gauge through an intelligent neural network prediction model according to the boiler load;

[0016] The intelligent control system is electrically connected to the connecting valve, the control valve, the three-way switching valve, the first liquid level gauge, the second liquid level gauge, the vacuum pump, the water supply pump and the drain valve respectively.

[0017] Furthermore, the outer side walls of the tubes of the finned heat exchange tubes are evenly provided with spiral fins, and the inner side walls are evenly provided with protrusions.

[0018] Furthermore, the protrusion is in any one of a sawtooth strip shape, a spherical protrusion shape or a smooth strip shape.

[0019] Furthermore, the upper edge of the riser is at the same horizontal position as the center line of the steam drum.

[0020] Furthermore, the steam drum is provided with a pressure monitor and a temperature monitor, the steam pipe is provided with a steam flow meter, and the pressure monitor, temperature monitor and steam flow meter are electrically connected to the intelligent control system respectively.

[0021] A method for intelligently controlling gas production of a low-temperature steam generation system, characterized in that it comprises the following steps:

[0022] Step 1) Determine the model input:

[0023] Select key indicators closely related to boiler combustion and import them into the input unit of the intelligent control system, including boiler coal feed x1, primary air volume x2, secondary air volume x3, flue gas temperature x4 and flue gas volume x5, to form an input vector x = [x1, x2, x3, x4, x5], which fully reflects the real-time working condition of the boiler;

[0024] Step 2) Data preprocessing:

[0025] The data processing unit performs 0-1 standardization processing on numerical data such as coal feed rate x1, primary air volume x2 and secondary air volume x3 to reduce the impact of dimensional differences on the model;

[0026] For data with time series characteristics, such as flue gas temperature x4 and flue gas volume x5, the sliding window method is used for feature extraction. The window w is set to 5, and the data of w consecutive time steps are extracted as a feature vector to capture its dynamic change trend;

[0027] Step 3) Build the intelligent neural network prediction model structure:

[0028] Step 3.1) In the model training unit, a multi-layer perceptron (MLP) is used to build a neural network structure with one hidden layer, and the number of neurons in the hidden layer n is set to 64;

[0029] Step 3.2) The activation function is set to ReLU (f(x) = max(0, x)) to improve the model convergence speed and avoid gradient vanishing;

[0030] Step 3.3) The model learning rate α is initially set to 0.001, and the learning rate decay strategy is adopted during the training process. The final learning rate is set to 0.0005;

[0031] Step 3.4) The number of iterations T of the model is set to 800, and the early stopping method is introduced to monitor the validation set loss. When the validation set loss does not decrease for 15 consecutive iterations, the training is terminated early to prevent overfitting.

[0032] Step 4) Model training and optimization:

[0033] Based on massive big data statistics and simulation of data samples under different load conditions, the backpropagation algorithm is used to continuously optimize the weights;

[0034] Collect real-time system operation data every 15 minutes, divide the new data into training set and validation set at a ratio of 80%:20%, retrain the model, and continuously optimize the weight matrix;

[0035] Step 5) Model performance evaluation:

[0036] According to the mean square error (MSE) formula Evaluate the effect of model performance improvement to ensure steady growth in system accuracy and sensitivity;

[0037] Step 6) Model output and application:

[0038] The model output is the number of heat exchanger module groups required to be invested or the liquid level values ​​of the first liquid level gauge and the second liquid level gauge;

[0039] Step 7) Heat exchanger module or liquid level control

[0040] When the boiler load is between 50% and 69%, a total of 8 modules are put into use in the heat exchanger. The intelligent control system adjusts the liquid level in the drum by controlling the start and stop of the feedwater pump to match the output value of the first or second liquid level gauge.

[0041] When the boiler load is greater than 70%, the intelligent control system controls the number of heat exchanger modules involved in the operation by controlling the opening and closing of the connecting valve and the control valve and regulating the three-way switching valve;

[0042] Furthermore, in step 3.4), L2 regularization is introduced, the parameter value is set to 0.0005, constraints are imposed on the weight vector, and the optimization objective function is Where m is the number of samples, y i is the true label, is the predicted value, and N is the number of weights to prevent the model from overfitting.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention adopts several groups of heat exchanger modules and intelligent control systems arranged in parallel. Through the intelligent neural network prediction model, the boiler load is matched with the number of heat exchanger modules put into operation. When the boiler load increases, the system will predict and start the appropriate number of heat exchanger modules to meet the steam demand in time, effectively solving the problem of steam parameter lag caused by boiler load changes in the traditional system, thereby achieving stable and reliable operation of the multi-effect evaporation system and realizing the coordinated coupling of low-energy consumption treatment of desulfurization wastewater and unit load;

[0045] 2. The inner wall of the finned heat exchange tube of the present invention is evenly distributed with a number of serrated strips, spherical protrusions or smooth strip-shaped protrusions. The design of the protrusions greatly increases the contact area between the finned heat exchange tube and the circulating water, reduces the volume of the heat exchanger body, reduces the equipment cost, and improves the heat transfer coefficient of the heat exchanger body. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0047] Figure 2 It is a schematic diagram of the heat exchanger module structure of the present invention;

[0048] Figure 3 Schematic diagram of the connection structure of the three-way switching valve in the heat exchanger module of the present invention;

[0049] Figure 4 It is a schematic diagram of the structure of the fin heat exchange tube of the present invention;

[0050] Figure 5 This is a schematic diagram of a serrated structure in which the inner wall protrusions of the fin heat exchange tube of the present invention are serrated;

[0051] Figure 6 This is a schematic diagram of a spherical protrusion structure of the inner wall protrusion of the fin heat exchange tube of the present invention;

[0052] Figure 7 This is a schematic diagram of a fin heat exchange tube of the present invention having a smooth strip-shaped protrusion on the inner wall;

[0053] Figure 8 It is a schematic diagram of the structure of the intelligent control system of the present invention;

[0054] In the figure: 1. heat exchanger module; 2. connecting valve; 3. riser; 4. downcomer; 5. control valve; 6. steam drum; 7. three-way switching valve; 8. steam pipe; 9. single-effect heater; 10. condensate tank; 11. water inlet pipe; 12. first liquid level gauge; 13. second liquid level gauge; 14. pressure monitor; 15. temperature monitor; 16. steam flow meter; 17. heat exchanger body; 18. finned heat exchange tube; 19. fin; 20. tube; 21. protrusion; 22. vacuum pump; 23. feed water pump; 24. sewage tank; 25. drain valve. DETAILED DESCRIPTION

[0055] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0056] like Figure 1-7As shown, a low-temperature steam generation system capable of intelligently regulating gas production includes a steam drum 6, a steam pipe 8, a first-effect heater 9, a condensed water tank 10, a water inlet pipe 11, a vacuum pump 22, a feed water pump 23 and a sewage tank 24, wherein the steam drum 6 is connected to the inlet of the first-effect heater 9 through the steam pipe 8, and a steam flow meter 16 is provided on the steam pipe 8, which can monitor the steam production in real time, and the outlet of the first-effect heater 9 is connected to the inlet of the condensed water tank 10, and the bottom outlet of the condensed water tank 10 is connected to the steam drum 6 through the water inlet pipe 11, and a feed water pump 23 is provided on the water inlet pipe 11, and the top outlet of the condensed water tank 10 is connected to an exhaust pipe, and a vacuum pump 22 is provided on the exhaust pipe, and the other end of the exhaust pipe extends to the atmosphere side;

[0057] The system also includes 4-6 groups of heat exchanger modules 1 connected in parallel and an intelligent control system. Each heat exchanger module 1 can be controlled separately. The heat exchanger module 1 includes a heat exchanger body 17, a plurality of finned heat exchange tubes 18, an ascending tube 3, a descending tube 4, a connecting valve 2 and a three-way switching valve 7. The upper header on one side of the heat exchanger body 17 is connected to the lower end of the ascending tube 3, the upper end of the ascending tube 3 is connected to the steam drum 6, and its upper edge is at the same horizontal position as the center line of the steam drum 6. The upper end of the descending tube 4 is connected to the steam drum 6. The lower end is connected to the lower header on the other side of the heat exchanger body 17, and the two ends of the finned heat exchange tube 18 are respectively connected to the upper header and the lower header. The outer wall of the tube 20 of the finned heat exchange tube 18 is evenly provided with spiral fins 19, and the inner wall is evenly provided with serrated protrusions 21. Taking the φ45 tube 20 as an example, after adding the serrated protrusions 21 on the inner wall, the contact area between the heat exchanger side and the circulating water can be increased by about 3.85 times, which greatly improves the overall heat transfer coefficient of the heat exchanger without increasing the water side resistance of the heat exchanger.

[0058] The bottoms of the 4-6 heat exchanger bodies 17 are connected by a connecting valve 2, the three-way switching valve 7 is arranged at the upper end of the riser 3, and one end of the three-way switching valve 7 leads to the steam drum 6, and the other end leads to the atmosphere. The downcomer 4 is provided with a control valve 5, the heat exchanger body 17 is provided with a first liquid level gauge 12, the steam drum 6 is provided with a second liquid level gauge 13, a pressure monitor 14 and a temperature monitor 15, the pressure monitor 14, the temperature monitor 15 and the steam flow meter 16 are electrically connected to the intelligent control system respectively, the pressure monitor 14 and the temperature monitor 15 can check the pressure and temperature in the steam drum 6 in real time, the heat exchanger body 17 is connected to the sewage tank 24 through a drain pipe, and the drain pipe is provided with a drain valve 25.

[0059] like Figure 8 As shown, the intelligent control system includes an input unit, an analysis unit and an output unit;

[0060] The input unit is used to input parameters related to boiler combustion, including boiler load, coal feed rate, primary air volume, secondary air volume, flue gas temperature and flue gas volume;

[0061] The analysis unit includes a data processing unit and a model training unit. The data processing unit is used to process the data of the input unit; the model training unit is used to construct an intelligent neural network prediction model, and train the intelligent neural network prediction model according to the data processed by the data processing unit to obtain a trained intelligent neural network prediction model;

[0062] The output unit is used to output the number of groups of the heat exchanger modules 1, the liquid level values ​​of the first liquid level gauge 12 and the second liquid level gauge 13 according to the boiler load through the intelligent neural network prediction model;

[0063] The intelligent control system is electrically connected to the connecting valve 2 , the control valve 5 , the three-way switching valve 7 , the first liquid level gauge 12 , the second liquid level gauge 13 , the vacuum pump 22 , the water supply pump 23 and the drain valve 25 , respectively.

[0064] The protrusion 21 may be in a spherical protrusion shape or a smooth strip shape.

[0065] During operation, a negative pressure environment of the system is first established through the vacuum pump 22, and the system pressure is controlled according to the required saturated steam temperature. The circulating water in the heat exchanger body 17 is heated by the flue gas and enters the steam drum 6 through the riser 3. Part of the water becomes saturated steam and enters the first-effect heater 9 through the steam pipe 8 to heat the desulfurization wastewater inside. After the heating is completed, the steam becomes condensed water and enters the condensed water tank 10. The condensed water is then transported to the steam drum 6 along the water inlet pipe 11 through the feed water pump 23 to replenish the water. The water in the steam drum 6 enters the heat exchanger body 17 through the downcomer 4 to be heated again, and the cycle repeats.

[0066] An intelligent control method for low-temperature steam generation gas production includes the following steps:

[0067] Step 1) Determine the model input:

[0068] Select key indicators closely related to boiler combustion and import them into the input unit of the intelligent control system, including boiler coal feed x1, primary air volume x2, secondary air volume x3, flue gas temperature x4 and flue gas volume x5, to form an input vector x = [x1, x2, x3, x4, x5], which fully reflects the real-time working condition of the boiler;

[0069] Step 2) Data preprocessing:

[0070] The data processing unit performs 0-1 standardization processing on numerical data such as coal feed rate x1, primary air volume x2 and secondary air volume x3 to reduce the impact of dimensional differences on the model;

[0071] For data with time series characteristics, such as flue gas temperature x4 and flue gas volume x5, the sliding window method is used for feature extraction. The window w is set to 5, and the data of w consecutive time steps are extracted as a feature vector to capture its dynamic change trend;

[0072] Step 3) Build the intelligent neural network prediction model structure:

[0073] Step 3.1) In the model training unit, a multi-layer perceptron (MLP) is used to build a neural network structure with one hidden layer, and the number of neurons in the hidden layer n is set to 64;

[0074] Step 3.2) The activation function is set to ReLU (f(x) = max(0, x)) to improve the model convergence speed and avoid gradient vanishing;

[0075] Step 3.3) The model learning rate α is initially set to 0.001, and the learning rate decay strategy is adopted during the training process. The final learning rate is set to 0.0005;

[0076] Step 3.4) The number of iterations T of the model is set to 800, and the early stopping method is introduced to monitor the validation set loss. When the validation set loss does not decrease for 15 consecutive iterations, the training is terminated early to prevent overfitting. In addition, L2 regularization is introduced, and the parameter value is set to 0.0005 to impose constraints on the weight vector. The optimization objective function is: Where m is the number of samples, y i is the true label, is the predicted value, N is the number of weights to prevent the model from overfitting;

[0077] Step 4) Model training and optimization:

[0078] Based on massive big data statistics and simulation of data samples under different load conditions, the backpropagation algorithm is used to continuously optimize the weights;

[0079] Collect real-time system operation data every 15 minutes, divide the new data into training set and validation set at a ratio of 80%:20%, retrain the model, and continuously optimize the weight matrix;

[0080] Step 5) Model performance evaluation:

[0081] According to the mean square error (MSE) formula Evaluate the effect of model performance improvement to ensure steady growth in system accuracy and sensitivity;

[0082] Step 6) Model output and application:

[0083] The model output is the number of heat exchanger modules 1 required to be put into operation or the liquid level values ​​of the first liquid level gauge 12 and the second liquid level gauge 13;

[0084] Step 7) Heat exchanger module or liquid level control

[0085] The number of heat exchanger modules 1 is determined according to the actual project parameters. Taking two heat exchangers, each heat exchanger is divided into four modules, a total of eight modules as an example, the control process is as follows:

[0086] When the boiler load is between 50% and 69%, a total of 8 modules are put into the heat exchanger. The intelligent control system adjusts the liquid level in the drum 6 by controlling the start and stop of the feed water pump 23 to match the output value of the first liquid level gauge 12 or the second liquid level gauge 13. The specific corresponding liquid levels are shown in Table 2.

[0087] When the boiler load is greater than or equal to 70%, the intelligent control system controls the number of heat exchanger modules involved in the operation by controlling the opening and closing of the connecting valve 2 and the control valve 5 and regulating the three-way switching valve 7, as follows:

[0088] Under the premise that the boiler unit load is greater than 70%, when the unit load is high, the flue gas temperature is high and the flue gas volume is large, the intelligent control system will close one or several groups of heat exchanger modules 1: close the control valve 5 on the corresponding downcomer 4, switch the three-way switching valve 7 on the riser 3 to the atmosphere side, and the steam generated in the heat exchanger module 1 will be discharged into the atmosphere, and the module will stop operating;

[0089] When the unit load is low, the flue gas temperature is low and the flue gas volume is small, the module is put back into use: the control valve 5 on the corresponding downcomer 4 is opened, the three-way switching valve 7 on the corresponding riser 3 is switched to the drum 6 side, the steam generated in the heat exchanger module 1 is discharged into the drum 6, and the module is put into operation in the system;

[0090] Since the boiler load is constantly fluctuating, the intelligent neural network prediction model is used to fit the boiler load to the heat exchanger module 1 put into operation to ensure stable steam production when the load fluctuates.

[0091] Through the intelligent neural network prediction model, the boiler load is fitted with the number of heat exchanger modules put into operation, which solves the problems of large lag and delay of low-temperature steam parameters caused by boiler load changes and difficulty in reliable operation of multi-effect evaporation system, and realizes the coordinated coupling of low-energy consumption treatment of desulfurization wastewater and wide load of the unit (35-100%).

[0092] The corresponding relationship between the load and the number of heat exchanger groups in the actual operation of the system and the liquid level of the steam drum and heat exchanger body is shown in Table 1, and the specific water level values ​​are shown in Table 2.

[0093] Table 1 Correspondence between load, number of heat exchanger groups, and liquid level of drum and heat exchanger body

[0094] load 50％ 60％ 70％ 75％ 80％ 90％ 100％ Number of input groups 8 groups 8 groups 8 groups 7 groups 6 groups 5 groups 4 groups Water Level High Water Level Medium water level Normal water level Normal water level Normal water level Normal water level Normal water level

[0095] Table 2 Different liquid level values ​​during system operation

[0096] High Water Level 1 / 2 level of the second level gauge on the drum Medium water level Full liquid level of the first liquid level gauge on the heat exchanger body Normal water level 7 / 8 liquid level of the first liquid level gauge on the heat exchanger body

Claims

1. A low-temperature steam generation system capable of intelligently regulating gas production, comprising a steam drum (6), a steam pipe (8), a first-effect heater (9), a condensate tank (10), a water inlet pipe (11), a vacuum pump (22), a feed water pump (23) and a sewage tank (24), wherein the steam drum (6) is connected to the inlet of the first-effect heater (9) through the steam pipe (8), the outlet of the first-effect heater (9) is connected to the inlet of the condensate tank (10), the bottom outlet of the condensate tank (10) is connected to the steam drum (6) through the water inlet pipe (11), the feed water pump (23) is provided on the water inlet pipe (11), the top outlet of the condensate tank (10) is connected to an exhaust pipe, the exhaust pipe is provided with a vacuum pump (22), and the other end extends to the atmosphere; It is characterized in that The invention also comprises a plurality of sets of heat exchanger modules (1) connected in parallel and an intelligent control system, wherein the heat exchanger module (1) comprises a heat exchanger body (17), a plurality of finned heat exchange tubes (18), an ascending tube (3), a descending tube (4), a connecting valve (2) and a three-way switching valve (7), wherein an upper header on one side of the heat exchanger body (17) is connected to the lower end of the ascending tube (3), the upper end of the ascending tube (3) is connected to the steam drum (6), the upper end of the descending tube (4) is connected to the steam drum (6), and the lower end is connected to a lower header on the other side of the heat exchanger body (17), and the two ends of the finned heat exchange tube (18) are connected to the steam drum (6). The heat exchanger bodies (17) are connected to the upper header and the lower header respectively. The bottoms of the plurality of heat exchanger bodies (17) are connected via a connecting valve (2). The three-way switching valve (7) is arranged at the upper end of the riser (3), and one end of the three-way switching valve (7) leads to the steam drum (6), and the other end leads to the atmosphere. The downcomer (4) is provided with a control valve (5). The heat exchanger body (17) is provided with a first liquid level gauge (12), and the steam drum (6) is provided with a second liquid level gauge (13). The heat exchanger body (17) is connected to a sewage tank (24) via a discharge pipe, and the discharge pipe is provided with a discharge valve (25); The intelligent control system includes an input unit, an analysis unit and an output unit; The input unit is used to input parameters related to boiler combustion, including boiler load, coal feed rate, primary air volume, secondary air volume, flue gas temperature and flue gas volume; The analysis unit includes a data processing unit and a model training unit. The data processing unit is used to process the data of the input unit; the model training unit is used to construct an intelligent neural network prediction model, and train the intelligent neural network prediction model according to the data processed by the data processing unit to obtain a trained intelligent neural network prediction model; The output unit is used to output the number of groups of heat exchanger modules (1), the liquid level values ​​of the first liquid level meter (12) and the second liquid level meter (13) according to the boiler load through an intelligent neural network prediction model; The intelligent control system is electrically connected to the connecting valve (2), the control valve (5), the three-way switching valve (7), the first liquid level gauge (12), the second liquid level gauge (13), the vacuum pump (22), the water supply pump (23) and the drain valve (25), respectively.

2. A low-temperature steam generation system capable of intelligently regulating gas production according to claim 1, characterized in that: The outer side walls of the tubes (20) of the finned heat exchange tubes (18) are evenly provided with spiral fins (19), and the inner side walls are evenly provided with protrusions (21).

3. A low-temperature steam generation system capable of intelligently regulating gas production according to claim 2, characterized in that: The protrusion (21) is in any one of a sawtooth strip shape, a spherical protrusion shape or a smooth strip shape.

4. A low-temperature steam generation system capable of intelligently regulating gas production according to claim 2, characterized in that: The upper edge of the riser (3) is at the same horizontal position as the center line of the steam drum (6).

5. The low-temperature steam generation system capable of intelligently regulating gas production according to claim 2, characterized in that: The steam drum (6) is also provided with a pressure monitor (14) and a temperature monitor (15), and the steam pipe (8) is provided with a steam flow meter (16). The pressure monitor (14), the temperature monitor (15) and the steam flow meter (16) are respectively electrically connected to the intelligent control system.

6. An intelligent control method using a low-temperature steam generation system capable of intelligently controlling gas production as claimed in any one of claims 2 to 5, characterized in that: The steps include: Step 1) Determine the model input: The key indicators closely related to boiler combustion are selected and imported into the input unit of the intelligent control system, including boiler coal feed x1, primary air volume x2, secondary air volume x3, flue gas temperature x4 and flue gas volume x5, forming an input vector Comprehensively reflect the real-time working conditions of the boiler; Step 2) Data preprocessing: The data processing unit performs 0-1 standardization processing on numerical data such as coal feed rate x1, primary air volume x2 and secondary air volume x3 to reduce the impact of dimensional differences on the model; For data with time series characteristics, such as flue gas temperature x4 and flue gas volume x5, the sliding window method is used for feature extraction. The window w is set to 5, and the data of w consecutive time steps are extracted as a feature vector to capture its dynamic change trend; Step 3) Build the intelligent neural network prediction model structure: Step 3.1) In the model training unit, a multi-layer perceptron (MLP) is used to build a neural network structure with one hidden layer, and the number of neurons in the hidden layer n is set to 64; Step 3.2) The activation function is set to ReLU (f(x) = max(0, x)) to improve the model convergence speed and avoid gradient vanishing; Step 3.3) The model learning rate α is initially set to 0.001, and the learning rate decay strategy is adopted during the training process. The final learning rate is set to 0.0005; Step 3.4) The number of iterations T of the model is set to 800, and the early stopping method is introduced to monitor the validation set loss. When the validation set loss does not decrease for 15 consecutive iterations, the training is terminated early to prevent overfitting. Step 4) Model training and optimization: Based on massive big data statistics and simulation of data samples under different load conditions, the backpropagation algorithm is used to continuously optimize the weights; Collect real-time system operation data every 15 minutes, divide the new data into training set and validation set at a ratio of 80%:20%, retrain the model, and continuously optimize the weight matrix; Step 5) Model performance evaluation: According to the mean square error (MSE) formula Evaluate the effect of model performance improvement to ensure steady growth in system accuracy and sensitivity; Step 6) Model output and application: The model output is the number of heat exchanger modules (1) that need to be put into operation or the liquid level values ​​of the first liquid level gauge (12) and the second liquid level gauge (13); Step 7) Heat exchanger module or liquid level control When the boiler load is between 50% and 69%, a total of 8 modules are put into operation in the heat exchanger, and the intelligent control system adjusts the liquid level in the drum (6) by controlling the start and stop of the feed water pump (23) to match the output value of the first liquid level meter (12) or the second liquid level meter (13); When the boiler load is greater than 70%, the intelligent control system controls the number of heat exchanger modules involved in the operation by controlling the opening and closing of the connecting valve (2) and the control valve (5) and regulating the three-way switching valve (7).

7. A control method using a low-temperature steam generation system capable of intelligently controlling gas production as claimed in claim 6, characterized in that: In step 3.4), L2 regularization is introduced, the parameter value is set to 0.0005, constraints are imposed on the weight vector, and the optimization objective function is: Where m is the number of samples, y i is the true label, is the predicted value, and N is the number of weights to prevent the model from overfitting.

Images