Method and device for dynamic pH regulation of autoclaved aerated concrete blocks
By using a backpropagation neural network prediction model and precise adjustment of multi-node control points, the lag problem of pH control in autoclaved aerated concrete production was solved, and dynamic control of slurry pH was achieved, ensuring stable product quality and efficient production.
Patent Information
- Application Number
- CN202610724131.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-25
- Publication Date
- 2026-06-30
Smart Images
Figure CN122299784A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of autoclaved aerated concrete (AAC) production technology, and particularly to a method and apparatus for dynamic pH control of AAC blocks. Background Technology
[0002] Autoclaved aerated concrete (AAC) is a lightweight, porous concrete made primarily from siliceous and calcareous materials, with the addition of aluminum powder as a foaming agent. It is produced through processes such as mixing, pouring, pre-curing, cutting, and autoclaving. The pH value of the slurry is a key parameter affecting the stability of foaming, the development of green body strength, and the final mechanical properties of the product. Aluminum powder reacts in an alkaline environment to produce hydrogen gas, thus forming a porous structure in the slurry. Simultaneously, the dissolution of active silica and aluminum components in the slurry, as well as the formation of hydration products, are highly dependent on a suitable pH environment. If the pH value is too low, the aluminum powder foaming rate decreases, the green body strength development is slow, and the porous structure deteriorates. If the pH value is too high, it may lead to excessive polymerization of the hydration gel, making it difficult to transform into well-crystallized tobermorite during autoclaving, thereby reducing the strength of the finished product.
[0003] Coal gasification fly ash (CGFA) is a solid waste generated during coal gasification. Rich in silicon and aluminum, it has the potential to replace cement or coal gasification fly ash in the production of autoclaved aerated concrete (AAC). However, CGFA contains unburned residual carbon and acidic oxygen-containing functional groups, making it acidic itself. Adding large amounts will significantly lower the pH value of the slurry. Currently, adjustments are often made by adding alkaline agents or alkaline aggregates, but these are mostly based on manual experience, leading to delays and inconsistencies. During the recycling of waste slurry, the continuous accumulation of acidic components from CGFA can cause significant fluctuations in the slurry pH value within a single day, making it difficult for technicians to respond accurately and promptly, resulting in unstable product quality.
[0004] Existing pH control methods are mostly based on simple feedback control or timed dosing, failing to establish predictive models between raw material ratios, slurry state and final product performance, and also unable to predict the pH value and finished product strength during casting, thus missing the best opportunity to make adjustments at the production front end.
[0005] In summary, to address the technical challenges in existing autoclaved aerated concrete (AAC) production, such as reliance on manual experience for pH control, lack of advanced prediction models, and poor multi-node coordination, a method and device for dynamic pH control of AAC blocks are proposed. Summary of the Invention
[0006] The main objective of this invention is to provide a method and apparatus for dynamic pH control of autoclaved aerated concrete blocks, so as to overcome the shortcomings of existing technologies, such as reliance on human experience, lack of advanced prediction models, and lack of multi-node coordination.
[0007] The main objective of this invention is to provide a method for dynamic pH control of autoclaved aerated concrete blocks, comprising the following steps:
[0008] Collect data on the pH value of the coal gasification fly ash slurry after slurry preparation and the amount of raw materials added: the raw materials include at least cement, lime, and coal gasification fly ash slurry;
[0009] The pH value of the coal gasification fly ash slurry and the amount of raw materials added are input into a pre-trained backpropagation neural network prediction model, and the predicted values are output; the predicted values include the pH prediction value of the casting slurry and the compressive strength prediction value of the product.
[0010] The predicted value is compared with a pre-set target range. If it exceeds the target range, the required dose of alkaline drug is calculated, and the drug is added at one or more control nodes.
[0011] After casting, the actual pH value and product compressive strength are collected and compared with the predicted value to obtain the error result. When the error result exceeds the set threshold, the online incremental update of the backpropagation neural network prediction model is triggered.
[0012] As a further preferred option, in the aforementioned method, before collecting data on the pH value of coal gasification fly ash slurry and the amount of raw materials added, it is also necessary to set up a first pH monitoring point and a second pH monitoring point on the production line.
[0013] The first pH monitoring point is located at the outlet of the coal gasification fly ash slurry mixing tank; the second pH monitoring point is located at the mold inlet before casting after the material mixing is completed.
[0014] The amount of raw materials added is collected through a weighing and metering system.
[0015] As a further preferred embodiment, in the aforementioned method, the backpropagation neural network prediction model has a three-layer structure;
[0016] The number of input layer nodes corresponds to the pH value and the quantity of raw materials collected from the coal gasification fly ash slurry. The number of output layer nodes is 2, corresponding to the casting pH value and compressive strength, respectively. The number of hidden layer nodes is 8, and the activation function is the Sigmoid function.
[0017] As a further preferred embodiment, in the aforementioned method, the backpropagation neural network prediction model adopts a staged incremental training approach, as follows:
[0018] The historical production dataset is divided into several subsets in ascending order of quantity, and then sequentially input into the backpropagation neural network prediction model for training. After each stage of training is completed, the weights and thresholds are saved as the initial parameters for the next stage of training.
[0019] As a further preferred embodiment, in the aforementioned method, the pre-set target range includes the target pH range of the casting slurry and the target compressive strength of the product;
[0020] The calculation of the required amount of alkaline drug is based on a nonlinear mapping relationship established using material balance, acid-base neutralization principles, and historical data.
[0021] The alkaline agent is any one of NaOH solution, Na2CO3 solution, or alkaline aggregate.
[0022] As a further preferred embodiment, in the aforementioned method, the target pH range of the casting slurry is 12.45~12.65, and the target compressive strength of the product is ≥3.5MPa.
[0023] As a further preferred embodiment, in the aforementioned method, the control nodes include at least a first control node for the coal gasification fly ash slurry preparation stage, a second control node for the circulating water stage, a third control node for the circulating material stage, and a fourth control node for the final mixing and stirring stage.
[0024] When performing the dosing action, alkaline agents are added first through the first control node, and then sequentially through the second, third and fourth control nodes, so that the predicted pH value of the casting slurry enters the target range.
[0025] As a further preferred embodiment, in the aforementioned method, the execution steps for triggering the online incremental update of the backpropagation neural network prediction model are as follows:
[0026] When several error results exceed a set threshold, an online incremental update of the backpropagation neural network prediction model is triggered. After triggering, new production data is added to the training set, and incremental training is performed based on the original weights of the backpropagation neural network prediction model to update the network parameters.
[0027] Based on the second main aspect of the present invention, a dynamic pH control device for autoclaved aerated concrete blocks is provided for implementing the aforementioned method, comprising an online pH monitoring unit, a weighing and metering unit, a data acquisition and transmission unit, an intelligent control unit, and a reagent dosing unit;
[0028] The pH online monitoring unit consists of at least a first pH meter located at the outlet of the coal gasification fly ash slurry preparation equipment and a second pH meter located at the inlet of the casting mold, and each pH meter is equipped with an automatic rinsing device;
[0029] Weighing and metering unit is used to measure the amount of each raw material input in real time;
[0030] The data acquisition and transmission unit is used to acquire data from the pH meter and the weighing unit and upload it to the storage platform;
[0031] The intelligent control unit has a built-in backpropagation neural network that receives data and outputs control commands.
[0032] The reagent dosing unit includes an alkaline reagent storage tank and a metering dosing device located at the coal gasification fly ash slurry preparation equipment, circulating water pipeline, circulating material pipeline and mixing equipment, and executes the dosing action according to the instructions of the intelligent control unit.
[0033] As a further preferred embodiment, in the aforementioned device, the alkaline reagent storage tanks respectively store NaOH solution, Na2CO3 solution, and alkaline aggregate;
[0034] The metering and dosing device can be either a frequency conversion metering pump or a screw feeder.
[0035] Compared with existing technologies, this invention provides a method and apparatus for dynamic pH control of autoclaved aerated concrete blocks. First, this invention adopts a backpropagation neural network prediction model. Before material mixing and stirring, based on only four input parameters—pH value of gasified fly ash slurry, cement dosage, lime dosage, and coal gasification fly ash slurry dosage—it can output predicted values of the casting slurry pH and the block compressive strength. This allows the control decision to be completed before casting, providing sufficient response time for the addition of alkaline agents. It solves the problem of adjustment lag in manual experience-based adjustment and realizes predictive control and proactive adjustment.
[0036] Secondly, this invention sets up pH monitoring points and reagent dosing points at four process nodes: the outlet of the coal gasification fly ash slurry mixing tank, the circulating water pipeline, the circulating material reuse pipeline, and the material mixing tank. The metering pumps or screw feeders at each node independently execute the dosing action according to the calculation results in the intelligent control unit. This increases the pH value of the coal gasification fly ash slurry from the source, inhibits the accumulation of acidic substances in the circulating water and circulating materials in the middle, and fine-tunes the mixed slurry at the end, forming a coordinated pH control chain throughout the process to prevent the continuous enrichment of acidic components in the waste slurry circulation.
[0037] Meanwhile, the backpropagation neural network model established by adopting a phased incremental training method is deployed in the intelligent control unit. When the error between the actual value and the predicted value continuously exceeds the set threshold, the system automatically reads the latest production data to perform online incremental learning on the model, updates the network weights and thresholds, so that the prediction model can continuously adapt to changes in the source of coal gasification fly ash, grade fluctuations and operating condition drift, and maintain high prediction accuracy in the long term.
[0038] Finally, this invention ensures smooth gas generation of aluminum powder and normal development of green body strength through precise dynamic pH control, enabling autoclaved aerated concrete blocks with a high proportion of acidic coal gasification fly ash to meet national standards. It ensures product qualification rate while disposing of a large amount of industrial solid waste, and reduces labor intensity and operating costs. It realizes the automation and intelligence of pH control, reduces reliance on manual experience, and avoids waste of reagents through precise dosing. Attached Figure Description
[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, obtaining other drawings based on these drawings without creative effort still falls within the scope of the present invention.
[0040] Figure 1 The following is a flowchart illustrating the execution of a method for dynamic pH control of autoclaved aerated concrete blocks according to an embodiment of the present invention.
[0041] Figure 2 The diagram shows a backpropagation neural network model training flowchart for a method of dynamic pH control of autoclaved aerated concrete blocks according to an embodiment of the present invention. Detailed Implementation
[0042] The preferred embodiments of the present invention will be described in detail below to provide a clearer understanding of the purpose, features, and advantages of the invention. It should be understood that the following embodiments are not intended to limit the scope of the invention, but are merely illustrative of the essential spirit of the technical solution of the invention.
[0043] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments in order to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the art will recognize that embodiments may be practiced without one or more of these specific details. In other instances, well-known techniques associated with the invention may not have been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.
[0044] Throughout this specification, references to "an embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Therefore, the appearance of "in an embodiment" or "an embodiment" in various places throughout the specification does not necessarily refer to the same embodiment. Furthermore, a particular feature, structure, or characteristic may be combined in any manner in one or more embodiments.
[0045] Combination Figure 1As shown, in one feasible embodiment, the present invention provides a method for dynamic pH control of autoclaved aerated concrete blocks, comprising the following steps S1-S4:
[0046] S1, Collect data on the pH value of the coal gasification fly ash slurry after slurry preparation and the amount of raw materials added: the raw materials include at least cement, lime, and coal gasification fly ash slurry;
[0047] S2, input the pH value of the coal gasification fly ash slurry and the amount of raw materials added into the pre-trained backpropagation neural network prediction model, and output the predicted values; the predicted values include the pH prediction value of the casting slurry and the compressive strength prediction value of the product.
[0048] S3. Compare the predicted value with the preset target range. If it exceeds the target range, calculate the amount of alkaline drug that needs to be added and perform the drug addition action at one or more control nodes.
[0049] S4. After casting, the actual pH value and product compressive strength are collected and compared with the predicted value to obtain the error result. When the error result exceeds the set threshold, the online incremental update of the backpropagation neural network prediction model is triggered.
[0050] This invention provides a method and apparatus for dynamic pH control of autoclaved aerated concrete (AAC) blocks, particularly for AAC block production lines that incorporate coal gasification fly ash. The method implements multi-node monitoring in the stages of fly ash slurry preparation, circulating water reuse, circulating material utilization, and pre-casting stages. A backpropagation neural network model is used to predict pH value and product compressive strength in advance, and the amount of alkaline agent added is automatically adjusted accordingly to stabilize the pH value of the slurry during casting within the target range, ensuring consistent product quality.
[0051] (1) Data acquisition and preprocessing
[0052] Several monitoring points are set up on the autoclaved aerated concrete production line: the first pH monitoring point is located at the outlet of the coal gasification fly ash slurry mixing tank, which is used to collect the pH value of the coal gasification fly ash slurry; the second pH monitoring point is located at the mold inlet before pouring after the material is mixed and stirred, which is used to collect the pH value of the pouring slurry.
[0053] Simultaneously, the weighing and metering system collects the amounts of cement, lime, coal gasification fly ash slurry, and coal gasification fly ash (or fixed values) added for each batch. The collected data is uploaded to a cloud platform or local industrial control computer for storage via a data acquisition module.
[0054] The collected data is normalized to map each parameter value to the [0,1] interval, eliminating the influence of dimensions and accelerating the convergence speed of the subsequent model.
[0055] The normalization formula is:
[0056] ;
[0057] in, This represents the normalized value. This represents the original collected data. , These represent the minimum and maximum values of the parameter in the historical dataset, respectively.
[0058] (2) Establish a backpropagation neural network prediction model (BP neural network prediction model)
[0059] A three-layer backpropagation (BP) neural network was constructed, with the following structure: an input layer with 4 nodes, corresponding to the pH value of coal gasification fly ash slurry, cement dosage, lime dosage, and coal gasification fly ash slurry dosage, respectively; a hidden layer with 8 nodes; and an output layer with 2 nodes, corresponding to the predicted pH value of the castable slurry and the predicted compressive strength value of autoclaved aerated concrete blocks, respectively. The sigmoid function was used as the activation function.
[0060] The BP neural network prediction model is trained using historical production data.
[0061] The training method adopts incremental training: the total dataset is divided into several stages in an increasing manner, such as 120 groups, 220 groups, 320 groups, and 420 groups. After the training of each stage is completed, the weights and thresholds are saved as the initial parameters for the next stage of training.
[0062] The training objective is to minimize the mean squared error (MSE) and the mean absolute error percentage (MAPE).
[0063] After the model is trained, it receives input parameters online in real time and outputs the predicted values of casting pH and compressive strength corresponding to the current mix ratio.
[0064] The output of the BP neural network prediction model is placed before the actual mixing and pouring of materials, transforming the traditional post-event feedback regulation into pre-event feedback regulation. By learning the nonlinear mapping relationship between the pH value of coal gasification fly ash slurry, cement dosage, lime dosage, coal gasification fly ash slurry dosage and the actual pH and finished product degree after mixing in historical data, the pH value of the slurry at the pouring time under the current mix ratio can be directly calculated without actual mixing reaction.
[0065] The BP neural network prediction model implicitly fits the complex and diverse buffer system in the actual slurry and the perturbation of local pH by aluminum powder gas generation through the nonlinear transformation of the 8 nodes of the hidden layer.
[0066] In traditional methods, the actual pH value can only be measured by a pH meter after the slurry is mixed and stirred. If the pH is found to be too low, an alkali agent is added. At this time, the aluminum powder has already started to generate gas, and pH adjustment cannot reverse the poor pore structure that has already been formed. However, this invention predicts the problem and completes the adjustment before casting. The adjusted slurry is already in the target pH range when it enters the mold, and the aluminum powder can complete the gas generation in the optimal alkaline environment.
[0067] (3) Regulation decision-making and implementation
[0068] The target pH range for the casting slurry is set at 12.45~12.65 and the target compressive strength of the product is ≥3.5MPa.
[0069] When the predicted casting pH value is lower than the target lower limit, calculate the required dosage of alkaline reagent. Alkaline reagents include strong alkaline solutions, such as NaOH solution, Na2CO3 solution, or alkaline aggregates, such as lime, carbide slag, etc., which can be added at multiple control points.
[0070] First control node: Located in the coal gasification fly ash slurry mixing tank, alkaline agents or alkaline aggregates are added to the coal gasification fly ash slurry to directly increase the pH value of the coal gasification fly ash slurry.
[0071] The second control node is located in the circulating water pipeline. It adds strong alkali agents to the recycled process water to adjust the alkalinity of the circulating water.
[0072] The third control node is located in the circulating material reuse pipeline. Alkaline agents are added to the circulating material to prevent the accumulation of acidic substances from causing a decrease in the overall slurry pH.
[0073] The fourth control node is located in the material mixing tank, where strong alkali agents are directly added to the final mixed slurry as a final fine-tuning method.
[0074] The dosage of the reagent at each control node is calculated based on a preset algorithm. A shallow neural network is trained based on neural network data. Based on the actual input, such as the difference between the currently predicted pouring pH value and the target pH value, the slurry volume, and the pH of the fly ash slurry, the required dosage of alkaline reagent is directly output and sent to the control unit.
[0075] Specifically, based on the principles of material balance and acid-base neutralization, and combined with historical production data, a nonlinear mapping relationship is established. The control unit receives the output of the prediction model and the control decision, and automatically executes the dosing action through a variable frequency metering pump or screw feeder.
[0076] (4) Feedback correction and model update
[0077] We continuously collect actual casting pH values and product compressive strength data, compare them with predicted values, and calculate the prediction error.
[0078] When the error continuously exceeds the set threshold, new data is added to the training set to perform online incremental learning on the BP neural network, update the network weights and thresholds, and enable the model to adapt to fluctuations in raw material characteristics and changes in production conditions.
[0079] Furthermore, the present invention also provides a device for dynamically controlling the pH value of autoclaved aerated concrete slurry for implementing the above method, comprising:
[0080] It includes an online pH monitoring unit, a weighing and metering unit, a data acquisition and transmission unit, an intelligent control unit, and a reagent dosing unit;
[0081] The pH online monitoring unit consists of at least a first pH meter located at the outlet of the coal gasification fly ash slurry preparation equipment and a second pH meter located at the inlet of the casting mold, and each pH meter is equipped with an automatic rinsing device.
[0082] Both the first and second pH meters are industrial online pH meters with automatic temperature compensation. The probe is made of alkali-resistant polytetrafluoroethylene. The front end of the probe is equipped with an automatic flushing device controlled by a solenoid valve. The flushing water pressure is 0.3~0.5MPa, and the duration of each flush is 3~4 seconds.
[0083] The automatic flushing device consists of a solenoid valve, a flushing water pipe, and a nozzle. The nozzle is aligned with the probe surface, and the inlet of the solenoid valve is connected to the industrial water pipe within the factory area.
[0084] The weighing and metering unit is used to measure the input amount of each raw material in real time. It consists of a resistance strain gauge load cell and a transmitter. The transmitter converts the weight signal into a current signal.
[0085] The range of the sensor at the cement silo outlet is 0~500kg, the range of the sensor at the lime silo outlet is 0~300kg, and the range of the sensor at the coal gasification fly ash slurry storage tank outlet is 0~5000kg. The accuracy class of the sensor is C3.
[0086] The data acquisition and transmission unit is used to acquire data from the pH meter and the weighing unit and upload it to the storage platform.
[0087] The data acquisition and transmission unit includes an RS485 data communication module. The RS485 data communication module adopts the Modbus RTU communication protocol, with a baud rate of 9600bps, 8 data bits, 1 stop bit, and no parity.
[0088] The intelligent control unit has a built-in backpropagation neural network prediction model, which receives data and outputs control commands.
[0089] The intelligent unit is a Siemens S7-1200 series PLC or an Advantech IPC-610L industrial computer, equipped with an RS485 communication interface and an Ethernet interface.
[0090] The reagent dosing unit includes an alkaline reagent storage tank and a metering dosing device located at the coal gasification fly ash slurry preparation equipment, circulating water pipeline, circulating material pipeline and mixing equipment, and executes the dosing action according to the instructions of the intelligent control unit.
[0091] Among them, the alkaline reagent storage tanks used to store NaOH solution are made of polyethylene and have a volume of 1~2m³; the alkaline reagent storage tanks used to store Na2CO3 solution are made of polyethylene and have a volume of 0.5~1m³; and the alkaline reagent storage tanks used to store alkaline aggregates such as lime powder are made of carbon steel and have a volume of 0.5~1m³. An arch-breaking device is provided at the bottom.
[0092] The metering and dosing device is a variable frequency metering pump or a screw feeder. The variable frequency metering pump can be a diaphragm metering pump with a flow rate adjustment range of 0~50L / h and the pump head material is polytetrafluoroethylene.
[0093] The screw feeder can be a volumetric feeder driven by a variable frequency motor, with a conveying capacity of 0~100kg / h.
[0094] In one feasible embodiment, the invention will be further described through the following examples.
[0095] This embodiment takes an autoclaved aerated concrete (AAC) block production line with an annual output of 200,000 cubic meters as an example. This production line uses coal gasification fly ash as the main siliceous material, with some coal gasification fly ash added to replace cement. Dry-process microencapsulated modified coal gasification fly ash is used as the cementitious component. The specific steps are as follows:
[0096] 1. Equipment installation and data acquisition
[0097] A first pH meter is installed on the outlet pipe of the coal gasification fly ash slurry mixing tank, and an automatic flushing device is configured. After each measurement is completed, the solenoid valve opens to flush the probe with clean water.
[0098] Install a second pH meter at the outlet of the mixing tank or the inlet of the casting mold, and also equip it with an automatic rinsing device.
[0099] Weighing and metering units are installed at the discharge ports of cement silos, lime silos, coal gasification fly ash slurry storage tanks, and coal gasification fly ash storage tanks to collect the quality of each batch of materials in real time.
[0100] All pH meters and metering devices are connected to the data acquisition module via RS485 data cables, which then uploads the data to the cloud platform for storage. The intelligent control unit reads the data from the cloud platform and executes subsequent algorithms.
[0101] 2. Establish a BP neural network prediction model
[0102] Combination Figure 2 As shown, the structure of the BP neural network consists of four nodes in the input layer: X1 = pH value of coal gasification fly ash slurry, X2 = cement mass (kg), X3 = lime mass (kg), and X4 = coal gasification fly ash slurry mass (kg) (the coal gasification fly ash is a fixed value of 50 kg and is not used as input).
[0103] The hidden layer has 8 nodes, and the output layer has 2 nodes: Y1 = predicted pH value of the casting slurry, Y2 = predicted compressive strength value of the product.
[0104] The activation functions for both the hidden and output layers are Sigmoid functions.
[0105] 420 sets of valid production data were selected from the historical database of the production line, of which 340 sets were used as the training set and 80 sets were used as the test set.
[0106] The training method adopts incremental training, which feeds the training set into the network in four stages in the order of 120 sets, 220 sets, 320 sets and 420 sets. After training 4000 times in each stage, the weights and thresholds are saved, and the next stage continues to train based on this.
[0107] The learning rate was set to 0.2 during training, with the goal of minimizing the mean squared error (MSE) and mean absolute error percentage (MAPE).
[0108] After incremental training, the final model achieved a casting pH prediction MSE of 0.0022 and a MAPE of 0.26% on the test set; and a compressive strength prediction MSE of 0.0201 and a MAE of 0.1173 MPa, with prediction accuracy meeting industrial requirements.
[0109] 3. Online control process
[0110] The target pH range for the casting slurry is set at 12.45~12.65, and the target compressive strength of the product is ≥3.5MPa.
[0111] During normal production, the intelligent control unit receives the current mix ratio data and the pH value of the coal gasification fly ash slurry for each batch, calls the BP neural network prediction model to perform forward calculations, and outputs the predicted pH value and the predicted compressive strength value for casting.
[0112] When the predicted pH value is below 12.45, the control unit calculates the required amount of OH⁻ moles to be added based on the built-in acid-base balance and material balance algorithm, and allocates the task according to the priority of each control node and the allowable dosage:
[0113] The first control point: Add NaOH solution (30% concentration) to the coal gasification fly ash slurry mixing tank through a metering pump to increase the pH value of the coal gasification fly ash slurry. This method can suppress the adverse effects of acidic fly ash at the source.
[0114] Second control point: When the low pH value of coal gasification fly ash slurry is caused by the accumulation of acidity in the circulating water, add an appropriate amount of alkaline aggregate, such as lime powder, into the circulating water pipeline to adjust the alkalinity of the circulating water before using it for slurry preparation.
[0115] Third control node: When the acidity of the circulating material (cutting waste return) is detected to be strong, Na2CO3 solution is sprayed into the circulating material pipeline to neutralize the residual acidic functional groups and prevent acidic substances from accumulating in the waste slurry.
[0116] Fourth adjustment point: If the predicted pouring pH is still slightly low after the above measures, add a small amount of NaOH solution to the mixing tank for final fine adjustment so that the actual pouring pH value enters the target range.
[0117] For example, when the pH value of a batch of coal gasification fly ash slurry drops to 11.79, and the cement is 170 kg, the lime is 279 kg, and the coal gasification fly ash slurry is 3680 kg, the model predicts that the pouring pH will be 12.38.
[0118] The control unit calculates that OH needs to be added. - Approximately 0.15 mol is converted into approximately 20 g of 30% NaOH solution, and is preferentially added through the first control node.
[0119] After adjustment, the actual casting pH rose to 12.49, and the product compressive strength reached 5.0 MPa, which meets the requirements.
[0120] Specifically, this invention utilizes spatially distributed dosing to achieve a graded treatment process that includes alkali extraction at the source, neutralization during the process, and fine-tuning at the end, thus avoiding drastic pH fluctuations caused by large-dose dosing at a single node.
[0121] Meanwhile, different alkaline agents can be used for different nodes, taking advantage of their respective reaction kinetic differences. For example, NaOH reacts quickly and is highly alkaline, making it suitable for rapid adjustment at the source. Lime powder reacts slowly but has a slow-release effect, making it suitable for the continuous stability of circulating water. Na2CO3 solution has a large buffer capacity, making it suitable for treating complex organic acids.
[0122] Multi-node synergistic dosing reduces the consumption of alkaline agents per cubic meter, while avoiding the problem of excessive polymerization of hydrated gels caused by single-node dosing.
[0123] 4. Online model updates
[0124] After every 50 batches of production, the newly obtained actual pH value and compressive strength are compared with the predicted pH value and compressive strength value of the casting. If the average absolute error increases by more than 20% compared to the previous value, the system automatically triggers a model update: the most recent 200 sets of data are added to the training set, the original weights and thresholds are retained as initial values, incremental training is performed again, and the network parameters are fine-tuned so that the model can continuously track the changes in raw material characteristics and process conditions.
[0125] After continuous operation and verification, the range of compressive strength fluctuation of the production line products was reduced from 3.6~5.2MPa to 4.5~5.2MPa after adopting the method and device of this invention. The rate of superior products increased by about 15%, the consumption of alkaline agents per cubic meter of blocks was reduced by about 12%, and the amount of coal gasification fly ash consumed was stabilized at more than 40,000 tons / year, achieving good technical and economic results.
[0126] The technical terms, principles, or means related to the technical solutions of the present invention mentioned in the above embodiments, which are not described in detail above, are all well-known technologies or common practices that are known to those skilled in the art.
[0127] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.
Claims
1. A method for dynamic pH control of autoclaved aerated concrete blocks, characterized in that, Includes the following steps: Collect data on the pH value of the coal gasification fly ash slurry after slurry preparation and the amount of raw materials added; the raw materials include at least cement, lime, and coal gasification fly ash slurry. The pH value of the coal gasification fly ash slurry and the amount of raw materials added are input into a pre-trained backpropagation neural network prediction model, and the predicted values are output; the predicted values include the pH prediction value of the casting slurry and the compressive strength prediction value of the product. The predicted value is compared with a pre-set target range. If it exceeds the target range, the required dose of alkaline drug is calculated, and the drug is added at one or more control nodes. After casting, the actual pH value and product compressive strength are collected and compared with the predicted value to obtain the error result. When the error result exceeds the set threshold, the online incremental update of the backpropagation neural network prediction model is triggered.
2. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 1, characterized in that, Before collecting data on the pH value of coal gasification fly ash slurry and the amount of raw materials added, it is necessary to set up a first pH monitoring point and a second pH monitoring point on the production line. The first pH monitoring point is located at the outlet of the coal gasification fly ash slurry mixing tank; the second pH monitoring point is located at the mold inlet before casting after the material mixing is completed. The amount of raw materials added is collected through a weighing and metering system.
3. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 1, characterized in that, The backpropagation neural network prediction model has a three-layer structure; The number of input layer nodes corresponds to the pH value and the quantity of raw materials collected from the coal gasification fly ash slurry. The number of output layer nodes is 2, corresponding to the casting pH value and compressive strength, respectively. The number of hidden layer nodes is 8, and the activation function is the Sigmoid function.
4. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 3, characterized in that, The backpropagation neural network prediction model adopts a phased incremental training method, as follows: The historical production dataset is divided into several subsets in ascending order of quantity, and then sequentially input into the backpropagation neural network prediction model for training. After each stage of training is completed, the weights and thresholds are saved as the initial parameters for the next stage of training.
5. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 1, characterized in that, The pre-set target ranges include the target pH range of the casting slurry and the target compressive strength of the product; The calculation of the required amount of alkaline drug is based on a nonlinear mapping relationship established using material balance, acid-base neutralization principles, and historical data. The alkaline agent is any one of NaOH solution, Na2CO3 solution, or alkaline aggregate.
6. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 5, characterized in that, The target pH range of the casting slurry is 12.45~12.65, and the target compressive strength of the product is ≥3.5MPa.
7. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 1, characterized in that, The control nodes include at least the first control node of the coal gasification fly ash slurry preparation stage, the second control node of the circulating water stage, the third control node of the circulating material stage, and the fourth control node of the final mixing and stirring stage. When performing the dosing action, alkaline agents are added first through the first control node, and then sequentially through the second, third and fourth control nodes, so that the predicted pH value of the casting slurry enters the target range.
8. The method for dynamic pH control of autoclaved aerated concrete blocks according to claim 1, characterized in that, The execution steps for triggering the online incremental update of the backpropagation neural network prediction model are as follows: When several error results exceed the set threshold, the online incremental update of the backpropagation neural network prediction model is triggered. After triggering, new production data is added to the training set, and incremental training is performed on the basis of the original backpropagation neural network prediction model weights to update the network parameters.
9. A dynamic pH control device for autoclaved aerated concrete blocks used to implement the methods described in claims 1-8, characterized in that, It includes an online pH monitoring unit, a weighing and metering unit, a data acquisition and transmission unit, an intelligent control unit, and a reagent dosing unit; The pH online monitoring unit consists of at least a first pH meter located at the outlet of the coal gasification fly ash slurry preparation equipment and a second pH meter located at the inlet of the casting mold, and each pH meter is equipped with an automatic rinsing device; Weighing and metering unit is used to measure the amount of each raw material input in real time; The data acquisition and transmission unit is used to acquire data from the pH meter and the weighing unit and upload it to the storage platform; The intelligent control unit has a built-in backpropagation neural network that receives data and outputs control commands. The reagent dosing unit includes an alkaline reagent storage tank and a metering dosing device located at the coal gasification fly ash slurry preparation equipment, circulating water pipeline, circulating material pipeline and mixing equipment, for performing dosing actions according to the instructions of the intelligent control unit.
10. A method for dynamic pH control of autoclaved aerated concrete blocks according to claim 9, characterized in that, The alkaline reagent storage tanks respectively store NaOH solution, Na2CO3 solution, and alkaline aggregates; The metering and dosing device can be either a frequency conversion metering pump or a screw feeder.