Control method and device of fly ash mineralization system

Abstract

The invention relates to the technical field of computers, in particular to a control method and device of a coal ash mineralization system. The method comprises the following steps: acquiring historical operation parameters of the fly ash mineralization system; wherein the historical operation parameters comprise the historical coal ash feeding rate, the historical leaching agent ratio, the historical leaching reaction temperature pressure and the historical stirring rotating speed of the leaching reaction kettle, the historical mineralization reaction temperature pressure of the mineralization reaction kettle and the historical carbon dioxide air inflow; inputting the historical operation parameters into a preset long-short-term memory network model to obtain a prediction result; wherein the prediction result comprises the calcium ion leaching rate, the carbon dioxide conversion rate and the calcium carbonate purity; inputting the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result comprises the current leaching agent ratio and the current mineralization reaction temperature and pressure; and controlling the fly ash mineralization system based on the adjustment result. Therefore, the product conversion rate of coal ash mineralization can be improved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a control method and apparatus for a fly ash mineralization system. Background Technology

[0002] Fly ash is a major industrial solid waste emitted by coal-fired power plants. It is rich in oxides such as calcium, silicon, and aluminum, and possesses properties similar to... The potential for mineralization reactions. Mineralization technology can realize the resource utilization of fly ash and... The permanent storage of calcium carbonate, while producing high-value-added calcium carbonate products, has significant environmental and economic benefits.

[0003] The conversion rate of products from existing fly ash mineralization is low: firstly, the leaching reaction parameters are poorly controlled, resulting in a low calcium ion leaching rate; secondly, the temperature and pressure conditions of the mineralization reaction are poorly matched, leading to a low carbon dioxide conversion rate and disordered product crystal forms.

[0004] Based on this, the present invention proposes a control method and apparatus for a fly ash mineralization system to solve the above-mentioned technical problems. Summary of the Invention

[0005] This invention describes a control method and apparatus for a fly ash mineralization system, which can improve the conversion rate of fly ash mineralization products.

[0006] According to a first aspect, the present invention provides a control method for a fly ash mineralization system. The method is applied to a controller of the fly ash mineralization system. The system includes the controller and, in sequence, a raw material storage unit, a fly ash grinding and classifying machine, a leaching reactor, a first solid-liquid separator, a carbon dioxide preheater, a mineralization reactor, a second solid-liquid separator, a product storage silo, and a tail gas recovery tank. The raw material storage unit includes a carbon dioxide storage tank, a fly ash storage tank, and a reagent storage tank. The controller is electrically connected to the fly ash grinding and classifying machine, the leaching reactor, and the mineralization reactor, respectively. The method includes: Obtain historical operating parameters of the fly ash mineralization system; wherein, the historical operating parameters include historical fly ash feed rate of the leaching reactor, historical leaching reagent ratio, historical leaching reaction temperature and pressure and historical stirring speed, and historical mineralization reaction temperature and pressure and historical carbon dioxide intake of the mineralization reactor; The historical operating parameters are input into a preset long short-term memory network model to obtain prediction results; wherein, the prediction results include calcium ion leaching rate, carbon dioxide conversion rate and calcium carbonate purity; The prediction results are input into a preset reinforcement learning optimization model to obtain the adjustment results; wherein, the adjustment results include the current leaching agent ratio and the current mineralization reaction temperature and pressure; The fly ash mineralization system is controlled based on the adjustment results.

[0007] According to a second aspect, the present application provides a control device of a fly ash mineralization system, the system comprising a controller and sequentially connected raw material storage unit, fly ash grinding classifier, leaching reactor, first solid-liquid classifier, carbon dioxide preheater, mineralization reactor, second solid-liquid classifier, product storage bin and tail gas recovery tank; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, and the device comprises: an acquisition unit configured to acquire historical operation parameters of the fly ash mineralization system; wherein the historical operation parameters comprise historical fly ash feeding rate, historical leaching reagent ratio, historical leaching reaction temperature and pressure and historical stirring speed of the leaching reactor, and historical mineralization reaction temperature and pressure and historical carbon dioxide inlet amount of the mineralization reactor; a first data processing unit configured to input the historical operation parameters into a preset long short-term memory network model to obtain a prediction result; wherein the prediction result comprises calcium ion leaching rate, carbon dioxide conversion rate and calcium carbonate purity; a second data processing unit configured to input the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result comprises current leaching reagent ratio and current mineralization reaction temperature and pressure; a third data processing unit configured to control the fly ash mineralization system based on the adjustment result.

[0008] In a third aspect, the embodiments of the present specification also provide an electronic device comprising a memory and a processor, the memory storing a computer program, and the processor executes the computer program to implement the method described in any of the embodiments of the present specification.

[0009] In a fourth aspect, the embodiments of the present specification also provide a control system of a fly ash mineralization system, the system comprising the controller and sequentially connected raw material storage unit, fly ash grinding classifier, leaching reactor, first solid-liquid classifier, carbon dioxide preheater, mineralization reactor, second solid-liquid classifier, product storage bin and tail gas recovery tank; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, and the controller is used to execute the method described in any of the embodiments of the present specification.

[0010] The control method and device of the fly ash mineralization system provided by the application obtain historical operation parameters of the fly ash mineralization system; wherein, the historical operation parameters include historical fly ash feeding rate (unit: kg / h, conventional adjustment range of 50-200 kg / h) of the leaching reaction kettle, historical leaching reagent ratio (volume ratio of hydrochloric acid concentration and alcohol amine type auxiliary agent), historical leaching reaction temperature and pressure (temperature: 30-80 DEG C, normal pressure working condition) and historical stirring speed (100-300 r / min) of the leaching reaction kettle, and historical mineralization reaction temperature and pressure of the mineralization reaction kettle and historical carbon dioxide gas inlet amount; then, the historical operation parameters are input into a preset long short-term memory network model to obtain a prediction result. The model captures the time sequence correlation and nonlinear mapping relationship between parameters, outputs a prediction result of a future specific period, and the prediction result includes calcium ion leaching rate (reflecting the dissolution efficiency of calcium element in fly ash), carbon dioxide conversion rate (reflecting the synergistic effect of carbon capture and mineralization reaction) and calcium carbonate purity (measuring the quality grade of the final product). Based on the prediction result of the preset long short-term memory network model, the prediction result is further input into a preset reinforcement learning optimization model to obtain an adjustment result; wherein, the adjustment result includes current leaching reagent ratio and current mineralization reaction temperature and pressure; finally, the fly ash mineralization system is controlled based on the adjustment result, so that the application can ensure that the raw material composition fluctuates or the equipment state changes, the system can still stably output high-purity calcium carbonate products, the carbon dioxide conversion efficiency and the calcium ion leaching rate are improved, and the product conversion rate of the fly ash mineralization is improved. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, brief descriptions will be given to the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0012] Figure 1 A flowchart of a control method of a fly ash mineralization system according to one embodiment is shown; Figure 2 A schematic block diagram of a control device of a fly ash mineralization system according to one embodiment is shown; Figure 3 A schematic block diagram of a control system of a fly ash mineralization system according to one embodiment is shown. DETAILED DESCRIPTION

[0013] The scheme provided by the application will be described below with reference to the drawings.

[0014] Figure 1A flowchart showing a control method of a fly ash mineralization system according to an embodiment. It can be understood that the method can be executed by any device, equipment, platform, cluster of equipment with computing and processing capabilities. The control method of the fly ash mineralization system, the method is applied to a controller of the fly ash mineralization system, the system comprising the controller and a raw material storage unit, a fly ash grinding classifier, a leaching reactor, a first solid-liquid classifier, a carbon dioxide preheater, a mineralization reactor, a second solid-liquid classifier, a product storage bin and a tail gas recovery tank connected in sequence; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, as shown in Figure 1 The method comprises: Step 100, obtaining historical operation parameters of the fly ash mineralization system; wherein the historical operation parameters include historical fly ash feeding rate of the leaching reactor, historical leaching reagent ratio, historical leaching reaction temperature and pressure and historical stirring speed and historical mineralization reaction temperature and pressure of the mineralization reactor, and historical carbon dioxide gas inlet amount; Step 102, inputting the historical operation parameters into a preset long short-term memory network model to obtain a prediction result; wherein the prediction result includes calcium ion leaching rate, carbon dioxide conversion rate and calcium carbonate purity; Step 104, inputting the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result includes current leaching reagent ratio and current mineralization reaction temperature and pressure; Step 106, controlling the fly ash mineralization system based on the adjustment result.

[0015] In the embodiment, the historical operation parameters of the fly ash mineralization system are acquired; wherein, the historical operation parameters include the historical fly ash feeding rate (unit: kg / h, conventional adjustment range of 50-200 kg / h) of the leaching reaction kettle, the historical leaching reagent ratio (volume ratio of hydrochloric acid concentration and alcohol amine type auxiliary agent), the historical leaching reaction temperature and pressure (temperature of 30-80 DEG C, normal pressure working condition) and the historical stirring speed (100-300 r / min) of the leaching reaction kettle, and the historical mineralization reaction temperature and pressure of the mineralization reaction kettle and the historical carbon dioxide feeding amount; then, the historical operation parameters are input into the preset long short-term memory network model to obtain a prediction result. The model captures the time sequence correlation and nonlinear mapping relationship between parameters, outputs the prediction result of a specific period in the future, and the prediction result includes the calcium ion leaching rate (reflecting the dissolution efficiency of calcium element in fly ash), the carbon dioxide conversion rate (reflecting the synergistic effect of carbon capture and mineralization reaction) and the calcium carbonate purity (measuring the quality grade of the final product). Based on the prediction result of the preset long short-term memory network model, the prediction result is further input into the preset reinforcement learning optimization model to obtain an adjustment result; wherein, the adjustment result includes the current leaching reagent ratio and the current mineralization reaction temperature and pressure; finally, the fly ash mineralization system is controlled based on the adjustment result, so that the fly ash mineralization system can still stably output high-purity calcium carbonate products while ensuring that the raw material composition fluctuates or the equipment state changes, and the carbon dioxide conversion efficiency and the calcium ion leaching rate are improved, thereby the product conversion rate of fly ash mineralization can be improved.

[0016] In the embodiment, the raw material storage unit serves as a front-end supply center and integrates a carbon dioxide storage tank (providing a mineralization reaction gas source), a fly ash storage tank (storing a solid-phase raw material) and a reagent storage tank (storing hydrochloric acid and alcohol amine type auxiliary agents required for leaching); the discharge end of the raw material storage unit is connected to a fly ash grinding and grading machine for fine pretreatment of fly ash; the ground fly ash is conveyed to a leaching reaction kettle to complete a calcium element dissolution reaction with leaching reagents supplied by the reagent storage tank; the slurry after the leaching reaction enters a first solid-liquid classifier to realize separation of leaching liquid and waste residue; the separated leaching liquid flows to a carbon dioxide preheater, and after being combined with pretreated carbon dioxide gas, enters a mineralization reaction kettle to complete a calcium carbonate generation reaction; the mineralization reaction product is separated by a second solid-liquid classifier, and the solid product is stored in a product storage bin, and the unreacted gas is recycled by a tail gas recovery tank.

[0017] In one embodiment of the present application, after the historical operation parameters are input into the preset long short-term memory network model to obtain the prediction result, the following steps are further included: When the calcium ion leaching rate is less than the first preset threshold value, the current grinding frequency, the current grading speed of the fly ash grinding and grading machine and the calcium ion leaching rate are input into the adaptive gradient correction model to obtain control parameters of the fly ash grinding and grading machine; wherein, the control parameters include the adjusted grinding frequency and the adjusted grading speed. The fly ash grinding and classifying machine is controlled based on the adjusted grinding frequency and the adjusted classifying rotating speed.

[0018] In the embodiment, when the calcium ion leaching rate prediction value output by the model is lower than the first preset threshold (such as 75%): first, the real-time running data of the fly ash grinding and classifying machine is collected, including the current grinding frequency (unit: Hz) and the current classifying rotating speed (unit: r / min), and the data is input into the preset adaptive gradient correction model together with the calcium ion leaching rate prediction value. The fly ash grinding and classifying machine is controlled based on the adjusted grinding frequency and the adjusted classifying rotating speed. By improving the grinding intensity and the classifying precision, the proportion of fine fly ash particles (the proportion of 80-120 mesh particles is increased by more than 10%) is increased, so that the reaction contact area with the leaching reagent is increased, and the calcium ion leaching rate is pushed back to the target interval.

[0019] In an embodiment of the present application, the adaptive gradient correction model is constructed by the following functional formula:

[0020] In the formula, is the adjusted grinding frequency, is the preset upper limit of the grinding frequency, is the current grinding frequency, is the grinding frequency adjustment coefficient, is the leaching rate deviation gradient, is the first preset threshold, is the calcium ion leaching rate, is the extremely low leaching rate threshold, is the adjusted classifying rotating speed, is the preset upper limit of the rotating speed, is the current classifying rotating speed, is the classifying rotating speed adjustment coefficient.

[0021] In the embodiment, the adaptive gradient correction model generates targeted control parameters by fusing the leaching rate deviation gradient and the equipment running characteristics, wherein the adjustment amplitude adopts a nonlinear amplification algorithm, and dynamically increases with the deviation degree of the leaching rate and the threshold (for example, the adjustment amplitude is increased by 1.5 times for every 5% decrease in the leaching rate), and is strictly limited within the equipment safety threshold range (grinding frequency 30-50 Hz, classifying rotating speed 800-1500 r / min). The grinding frequency adjustment coefficient can be set to five hertz, and the classifying rotating speed adjustment coefficient can be set to two hundred revolutions per minute.

[0022] In an embodiment of the present application, the preset reinforcement learning optimization model is trained in the following manner: Obtain a historical data set; The initial reinforcement learning optimization model is trained according to a historical data set until a preset reward function converges to a preset interval, and the preset reinforcement learning optimization model is obtained. The preset reward function is constructed according to a calcium ion leaching rate function, a carbon dioxide conversion rate function, a calcium carbonate purity function, a constraint penalty term and a dynamic correction term.

[0023] In this embodiment, a historical data set is obtained, which contains the operation data of the fly ash mineralization system under different raw material characteristics (such as calcium-silicon content fluctuation) and operation parameter combinations. Each sample is associated with complete input parameters (such as leaching reagent ratio, mineralization temperature and pressure, etc.) and output indicators (such as calcium ion leaching rate, product purity, etc.). The initial reinforcement learning optimization model is trained according to the historical data set. During the training process, the model continuously learns the mapping relationship between parameter adjustment and system output, and constantly optimizes the decision strategy. The termination condition of the training is set as the convergence of the preset reward function to the target interval of 80-100. At this time, the model has stable optimization capability. The preset reward function is constructed according to the calcium ion leaching rate function (encouraging efficient dissolution), the carbon dioxide conversion rate function (strengthening carbon capture effect), the calcium carbonate purity function (ensuring product quality), the constraint penalty term and the dynamic correction term.

[0024] In an embodiment of the present application, the constraint penalty term includes a parameter fluctuation penalty function and an energy consumption penalty function. The preset reward function is constructed by the following formula:

[0025] In the formula, is the reward function value, is the first preset weight coefficient, is the calcium ion leaching rate function, is the carbon dioxide conversion rate function, is the second preset weight coefficient, is the third preset weight coefficient, is the calcium carbonate purity function, is the parameter fluctuation penalty function, is the energy consumption penalty function, is the dynamic correction term, is the weight of the parameter fluctuation penalty, is the weight of the energy consumption penalty, is the weight of the dynamic correction term.

[0026] In the present embodiment, the preset reward function creatively combines the three objective functions through dynamic weight coefficients, can adjust the priority of each objective according to different reaction stages (such as leaching stage, mineralization stage), and realizes the common optimization of efficient leaching, sufficient mineralization and high-quality products. The formula innovatively adds a parameter fluctuation penalty term and an energy consumption penalty term, which avoids the model output exceeding the safety or economic threshold through quantitative constraints; at the same time, a dynamic correction term is introduced, which adjusts the reward according to the system running trend (such as the change of reward value for several times in succession), guides the model to the long-term optimization direction, and solves the problem that the traditional reinforcement learning is easy to fall into short-term optimization. The first preset weight coefficient (high in the leaching stage, low in the mineralization stage), the second preset weight coefficient (high in the mineralization stage, low in the leaching stage).

[0027] In an embodiment of the present application, the calcium ion leaching rate function, the carbon dioxide conversion rate function and the calcium carbonate purity function are constructed by the following formula:

[0028] In the formula, is a raw material calcium content correction coefficient, is the calcium ion leaching rate output by the prediction model in the training process, is the carbon dioxide conversion rate output by the prediction model in the training process, is the mineralization reaction duration, is the calcium carbonate purity output by the prediction model in the training process.

[0029] In the present embodiment, the raw material calcium content correction coefficient, the lower the raw material calcium content, the greater the coefficient, when the calcium carbonate purity output by the prediction model in the training process is greater than 95%, the reward increases faster, when the calcium carbonate purity output by the prediction model in the training process is greater than 92% and less than or equal to 95%, the penalty increases sharply when approaching the threshold, and when the calcium carbonate purity output by the prediction model in the training process is less than or equal to 92%, severe punishment.

[0030] In an embodiment of the present application, the constraint penalty term is constructed by the following formula:

[0031] In the formula, is the penalty weight of each parameter, is the deviation of leaching reaction pH from the target value, is the deviation of mineralization reaction pressure from the set value, is the deviation of mineralization reaction temperature from the set value, is the maximum allowed deviation of the i th parameter, is the current unit product energy consumption, is the baseline energy consumption.

[0032] In the embodiment, based on the ratio of unit product energy consumption (E, unit: kWh / ton of calcium carbonate) to reference energy consumption, an exponential penalty is adopted, when the current unit product energy consumption is less than or equal to the reference energy consumption, the penalty is 0, when the current unit product energy consumption is equal to 1.2 times of the reference energy consumption, the penalty is about equal to 10, when the current unit product energy consumption is greater than 1.5 times of the reference energy consumption, the penalty is about equal to 30.

[0033] In an embodiment of the present application, the dynamic correction term is constructed by the following formula:

[0034] In the formula, is the reward value at the current time, is the reward value at the last time, is the reward value at the time before the last time.

[0035] In the embodiment, if the reward value at the current time shows an upward trend for the first two times, the modified reward is increased, and if it shows a downward trend, the dynamic correction term is reduced.

[0036] In an embodiment of the present application, when the carbon dioxide conversion rate is less than the second preset threshold, the current heating power of the carbon dioxide preheater and the carbon dioxide conversion rate are input into the power compensation formula driven by the carbon dioxide conversion rate to obtain the preheater heating power after compensation; Based on the preheater heating power after compensation, the carbon dioxide preheater is controlled.

[0037] In the embodiment, when the carbon dioxide conversion rate monitoring value of the mineralization reaction kettle is lower than the second preset threshold (such as 85%), the current heating power (unit: kW) of the carbon dioxide preheater is first collected, and the real-time carbon dioxide conversion rate is extracted, the current heating power of the carbon dioxide preheater and the carbon dioxide conversion rate are input into the power compensation formula driven by the carbon dioxide conversion rate to obtain the preheater heating power after compensation.

[0038] In an embodiment of the present application, the power compensation formula driven by the carbon dioxide conversion rate is determined by the following formula:

[0039] In the formula, is the preheater heating power after compensation, is the current heating power, is the conversion rate compensation coefficient (0.3), is the carbon dioxide conversion rate, is the second preset threshold.

[0040] In the embodiment, For conversion rate bias amplification terms (such as carbon dioxide conversion rate equals 80 percent, amplification 1.65 times, carbon dioxide conversion rate equals 75 percent, amplification 2.72 times).

[0041] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in a different order than the order in which they are recited in the embodiments and still achieve the desired results. In addition, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous or necessary.

[0042] According to another aspect, the present application provides a control device of a fly ash mineralization system. Figure 2 A schematic block diagram of a control device of a fly ash mineralization system according to an embodiment is shown. It can be understood that the device can be realized by any device, equipment, platform and cluster of equipment with computing and processing capabilities. The system includes a controller and a raw material storage unit, a fly ash grinding classifier, a leaching reactor, a first solid-liquid classifier, a carbon dioxide preheater, a mineralization reactor, a second solid-liquid classifier, a product storage bin and a tail gas recovery tank connected in turn; wherein the raw material storage unit includes a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, as shown in Figure 2 The device includes an acquisition unit 200, a first data processing unit 202, a second data processing unit 204 and a third data processing unit 206, as shown. The main functions of each component unit are as follows: The acquisition unit 200 is configured to acquire historical running parameters of the fly ash mineralization system; wherein the historical running parameters include historical fly ash feeding rate of the leaching reactor, historical leaching reagent ratio, historical leaching reaction temperature and pressure and historical stirring speed and historical mineralization reaction temperature and pressure of the carbon dioxide of the mineralization reactor; The first data processing unit 202 is configured to input the historical running parameters into a preset long short-term memory network model to obtain a prediction result; wherein the prediction result includes calcium ion leaching rate, carbon dioxide conversion rate and calcium carbonate purity; The second data processing unit 204 is configured to input the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result includes current leaching reagent ratio and current mineralization reaction temperature and pressure; The third data processing unit 206 is configured to control the fly ash mineralization system based on the adjustment result.

[0043] In an embodiment of the present application, the device further comprises a fourth data processing unit, which is configured to perform the following operations: When the calcium ion leaching rate is less than a first preset threshold, inputting the current grinding frequency, the current classification rotating speed of the fly ash grinding and classifying machine and the calcium ion leaching rate into a preset adaptive gradient correction model to obtain a control parameter of the fly ash grinding and classifying machine; wherein the control parameter comprises an adjusted grinding frequency and an adjusted classification rotating speed. Based on the adjusted grinding frequency and the adjusted classification rotating speed, controlling the fly ash grinding and classifying machine.

[0044] In an embodiment of the present application, the adaptive gradient correction model is constructed by the following functional formula:

[0045] In the formula, is the adjusted grinding frequency, is a preset upper limit of the grinding frequency, is the current grinding frequency, is a grinding frequency adjustment coefficient, is a leaching rate deviation gradient, is the first preset threshold, is the calcium ion leaching rate, is an extremely low leaching rate threshold, is the adjusted classification rotating speed, is a preset upper limit of the rotating speed, is the current classification rotating speed, is a classification rotating speed adjustment coefficient.

[0046] In an embodiment of the present application, the device further comprises a fifth data processing unit, which is configured to perform the following operations: obtaining a historical data set; training an initial reinforcement learning optimization model according to the historical data set until a preset reward function converges to a preset interval to stop training, to obtain the preset reinforcement learning optimization model; wherein the preset reward function is constructed according to a calcium ion leaching rate function, a carbon dioxide conversion rate function, a calcium carbonate purity function, a constraint penalty term and a dynamic correction term.

[0047] In an embodiment of the present application, the constraint penalty term comprises a parameter fluctuation penalty function and an energy consumption penalty function; and the preset reward function is constructed by the following formula:

[0048] In the formula, is a reward function value, is a first preset weight coefficient, is the calcium ion leaching rate function, is the carbon dioxide conversion rate function, is a second preset weight coefficient, is a third preset weight coefficient, is the calcium carbonate purity function, is a parameter fluctuation penalty function, is an energy consumption penalty function, is a dynamic correction term, is a weight of the parameter fluctuation penalty, is a weight of the energy consumption penalty, is a weight of the dynamic correction term.

[0049] In an embodiment of the present application, the calcium ion leaching rate function, the carbon dioxide conversion rate function and the calcium carbonate purity function are constructed by the following formula:

[0050] In the formula, is a raw material calcium content correction coefficient, is the calcium ion leaching rate output by the prediction model in the training process, is the carbon dioxide conversion rate output by the prediction model in the training process, is the duration of the mineralization reaction, is the calcium carbonate purity output by the prediction model in the training process.

[0051] In an embodiment of the present application, the constraint penalty term is constructed by the following formula:

[0052] In the formula, is the reward value at the current moment, is the reward value at the last moment, is the reward value at the moment before the last moment, is the penalty weight of each parameter, is the deviation of the leaching reaction pH from the target value, is the deviation of the mineralization reaction pressure from the set value, is the deviation of the mineralization reaction temperature from the set value, is the maximum allowed deviation of the i-th parameter, is the current unit product energy consumption, is the reference energy consumption.

[0053] In an embodiment of the present application, the dynamic correction term is constructed by the following formula:

[0054] wherein, is a reward value of a current moment, is a reward value of a previous moment, is a reward value of a moment before the previous moment.

[0055] In an embodiment of the present application, the device further comprises a sixth data processing unit, which is configured to perform the following operation: when the carbon dioxide conversion rate is less than the second preset threshold value, inputting the current heating power of the carbon dioxide preheater and the carbon dioxide conversion rate into a carbon dioxide conversion rate driven power compensation formula to obtain a compensated preheater heating power; controlling the carbon dioxide preheater based on the compensated preheater heating power.

[0056] In an embodiment of the present application, the carbon dioxide conversion rate driven power compensation formula is determined by the following formula:

[0057] wherein, is the compensated preheater heating power, is the current heating power, is a conversion rate compensation coefficient (0.3), is the carbon dioxide conversion rate, is the second preset threshold value.

[0058] According to another aspect of the embodiments, Figure 3 a schematic block diagram of a control system of a fly ash mineralization system according to an embodiment is shown, the system comprising the controller and a raw material storage unit, a leaching reaction kettle, a first solid-liquid classifier, a mineralization reaction kettle, a second solid-liquid classifier and a product storage bin connected in sequence; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reaction kettle and the mineralization reaction kettle respectively, and the controller is configured to cause the computer to execute the executable code to implement the method as described in combination with Figure 1 the described method.

[0059] According to still another aspect of the embodiments, an electronic device is also provided, comprising a memory and a processor, the memory having stored therein executable code, and the processor being configured to implement the method as described in combination with Figure 1 the described method.

[0060] The various embodiments of the present application are described in a progressive manner, and the same or similar parts among the various embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, the device embodiments are described simply because they are basically similar to the method embodiments, and the relevant parts can be referred to the description of the method embodiments.

[0061] Those skilled in the art can understand that the functions described in the above one or more examples can be implemented by hardware, software, firmware or any combination thereof. When implemented by software, the functions can be stored in a computer readable medium or transmitted as one or more instructions or codes on a computer readable medium.

[0062] The above detailed description sets forth the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above detailed description is only a specific implementation of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solutions of the present application shall be included in the protection scope of the present application.

Claims

1. A control method of a fly ash mineralization system, characterized by, The method is applied to a controller of a fly ash mineralization system, the system comprising the controller and a raw material storage unit, a fly ash grinding classifier, a leaching reactor, a first solid-liquid classifier, a carbon dioxide preheater, a mineralization reactor, a second solid-liquid classifier, a product storage bin and a tail gas recovery tank connected in sequence; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, and the method comprises: obtaining historical operation parameters of the fly ash mineralization system; wherein the historical operation parameters comprise a historical fly ash feeding rate of the leaching reactor, a historical leaching reagent ratio, a historical leaching reaction temperature and pressure and a historical stirring speed, and a historical mineralization reaction temperature and pressure of the mineralization reactor, and a historical carbon dioxide gas inlet amount; inputting the historical operation parameters into a preset long short-term memory network model to obtain a prediction result; wherein the prediction result comprises a calcium ion leaching rate, a carbon dioxide conversion rate and a calcium carbonate purity; inputting the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result comprises a current leaching reagent ratio and a current mineralization reaction temperature and pressure; controlling the fly ash mineralization system based on the adjustment result.

2. The method of claim 1, wherein, After the historical operation parameters are input into the preset long short-term memory network model to obtain the prediction result, the method further comprises: when the calcium ion leaching rate is less than a first preset threshold, inputting a current grinding frequency and a current classification speed of the fly ash grinding classifier and the calcium ion leaching rate into an adaptive gradient correction model to obtain a control parameter of the fly ash grinding classifier; wherein the control parameter comprises an adjusted grinding frequency and an adjusted classification speed; controlling the fly ash grinding classifier based on the adjusted grinding frequency and the adjusted classification speed.

3. The method of claim 2, wherein, The adaptive gradient correction model is constructed by the following functional formula: wherein, is the adjusted grinding frequency, is a preset upper limit of the grinding frequency, is the current grinding frequency, is a grinding frequency adjustment coefficient, is a leaching rate deviation gradient, is the first preset threshold value, is the calcium ion leaching rate, is an extremely low leaching rate threshold value, is the adjusted classification rotational speed, is a preset upper limit of the rotational speed, is the current classification rotational speed, is a classification rotational speed adjustment coefficient.

4. The method of claim 1, wherein, The preset reinforcement learning optimization model is obtained by training in the following manner: obtaining a historical data set; training an initial reinforcement learning optimization model according to the historical data set until a preset reward function converges to a preset interval to stop training, thereby obtaining the preset reinforcement learning optimization model; wherein the preset reward function is constructed according to a calcium ion leaching rate function, a carbon dioxide conversion rate function, a calcium carbonate purity function, a constraint penalty term and a dynamic correction term.

5. The method of claim 4, wherein, The constraint penalty term comprises a parameter fluctuation penalty function and an energy consumption penalty function; The preset reward function is constructed by the following formula: In the formula, is a reward function value, is a first preset weight coefficient, is the calcium ion leaching rate function, is the carbon dioxide conversion rate function, is a second preset weight coefficient, is a third preset weight coefficient, is the calcium carbonate purity function, is a parameter fluctuation penalty function, is an energy consumption penalty function, is a dynamic correction term, is a weight of the parameter fluctuation penalty, is a weight of the energy consumption penalty, is a weight of the dynamic correction term.

6. The method of claim 5, wherein, The calcium ion leaching rate function, the carbon dioxide conversion rate function and the calcium carbonate purity function are constructed by the following formula: wherein is a raw material calcium content correction factor, is the calcium ion leaching rate output by the prediction model during the training process, is the carbon dioxide conversion rate output by the prediction model during the training process, is the duration of the mineralization reaction, is the calcium carbonate purity output by the prediction model during the training process.

7. The method of claim 5, wherein, The constraint penalty term and the dynamic correction term are constructed by the following formula: In the formula, is a reward value at a current time, is a reward value at a previous time, is a reward value at a time two steps before, is a penalty weight of each parameter, is a deviation of leaching reaction pH from a target value, is a deviation of mineralization reaction pressure from a set value, is a deviation of mineralization reaction temperature from a set value, is a maximum allowable deviation of the i-th parameter, is a current unit product energy consumption, is a reference energy consumption.

8. A control device of a fly ash mineralization system, characterized by, The system comprises a controller and sequentially connected raw material storage unit, fly ash grinding classifier, leaching reactor, first solid-liquid classifier, carbon dioxide preheater, mineralization reactor, second solid-liquid classifier, product storage bin and tail gas recovery tank; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, and the device comprises: An acquisition unit is configured to acquire historical operation parameters of the fly ash mineralization system; wherein the historical operation parameters include historical fly ash feeding rate of the leaching reactor, historical leaching reagent ratio, historical leaching reaction temperature and pressure and historical stirring speed, and historical mineralization reaction temperature and pressure of the mineralization reactor, and historical carbon dioxide inlet amount; A first data processing unit is configured to input the historical operation parameters into a preset long short-term memory network model to obtain a prediction result; wherein the prediction result includes calcium ion leaching rate, carbon dioxide conversion rate and calcium carbonate purity; A second data processing unit is configured to input the prediction result into a preset reinforcement learning optimization model to obtain an adjustment result; wherein the adjustment result includes current leaching reagent ratio and current mineralization reaction temperature and pressure; A third data processing unit is configured to control the fly ash mineralization system based on the adjustment result.

9. An electronic device, comprising: The system comprises the controller and sequentially connected raw material storage unit, fly ash grinding classifier, leaching reactor, first solid-liquid classifier, carbon dioxide preheater, mineralization reactor, second solid-liquid classifier, product storage bin and tail gas recovery tank; wherein the raw material storage unit comprises a carbon dioxide storage tank, a fly ash storage tank and a reagent storage tank, the controller is electrically connected with the fly ash grinding classifier, the leaching reactor and the mineralization reactor respectively, and the controller is used for executing the method according to any one of claims 1-7.

10. A control system for a fly ash mineralization system, characterized in that, ​

Images