Temperature control method and system for sodium carbonate production process

Abstract

The invention relates to the technical field of industrial production temperature control, and discloses a temperature control method and system for a sodium carbonate production process. Comprising the steps of training a temperature strategy model and a temperature prediction model based on historical data; real-time data in the sodium carbonate production process are obtained, and related equipment is controlled to conduct temperature adjustment according to a temperature control strategy output by the temperature strategy model; the real-time production equipment temperature is compared with a temperature prediction curve output by the temperature prediction model, and whether the real-time production equipment temperature meets the temperature range requirement or not is judged; when an abnormal condition occurs, a temperature control strategy is automatically adjusted, and a new temperature prediction curve is generated; and if the abnormal condition is not solved within the preset adjustment times, triggering an early warning mechanism. Therefore, the problem that a traditional temperature control method lacks real-time data analysis and prediction capacity is effectively solved, and the intelligent and automatic temperature control capacity in the sodium carbonate production process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial production temperature control, and in particular to a temperature control method and system for a soda ash production process. Background Art

[0002] In the soda ash production process, temperature is a key factor affecting product quality and production efficiency. Traditional temperature control methods rely primarily on manual operation and empirical judgment, which is not suitable for complex and changing production environments. Soda ash production involves multiple variables, and even slight changes in these variables can affect production temperature, leading to fluctuations in product quality and reduced production efficiency.

[0003] Furthermore, traditional temperature control methods lack real-time data analysis and forecasting capabilities, making it difficult to respond promptly to temperature fluctuations during the production process. When production equipment temperatures deviate from the preset range, manual intervention and adjustment are often required. This not only increases labor costs but can also exacerbate production issues due to untimely or inappropriate adjustments. Therefore, achieving intelligent and automated temperature control in the soda ash production process has become a pressing issue. Summary of the Invention

[0004] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a temperature control method and system for a soda ash production process to solve the problems existing in the prior art.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a temperature control method for a soda ash production process, the method comprising: Step S100: Acquire first historical data in the soda ash production process, perform data preprocessing to obtain second historical data, and use the second historical data to train a temperature strategy model and a temperature prediction model; Step S200: Acquire real-time data during the soda ash production process, and control the temperature of the equipment in the soda ash production process according to the temperature control strategy output by the temperature strategy model; Step S300: Inputting real-time data and real-time temperature control strategy into the temperature prediction model to obtain a temperature prediction curve, comparing the real-time production equipment temperature with the temperature prediction curve to determine whether the real-time production equipment temperature meets the temperature range requirement; Step S400: When an abnormal situation occurs, automatically adjust the temperature control strategy and generate a new temperature prediction curve; Step S500: If the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism is triggered.

[0006] Preferably, in a possible implementation of the first aspect, step S100 further includes: Obtain the first historical data of the soda ash production process and transmit it to the control center; Performing data preprocessing on the first historical data, including data cleaning and data conversion, to obtain second historical data; Based on the second historical data, the temperature strategy model and the temperature prediction model are trained separately using the deep learning method.

[0007] Preferably, in a possible implementation of the first aspect, the first historical data includes historical temperature and precipitation data, historical production equipment temperature data, historical production raw material data, historical natural gas supply speed data, historical combustion air flow data, and historical cooling water flow data; the second historical data is the first historical data after cleaning and data conversion to meet the preset model training data format.

[0008] Preferably, in a possible implementation of the first aspect, the temperature strategy model outputs the temperature control strategy of the equipment within a preset control time during the soda ash production process, including controlling the natural gas supply rate, combustion air flow rate and cooling water flow rate; the temperature prediction model outputs a temperature prediction curve within the preset control time according to the temperature control strategy within the preset control time.

[0009] Preferably, in a possible implementation of the first aspect, step S200 further includes: The control system obtains real-time data from the soda ash production process, including real-time temperature and precipitation data, real-time production equipment temperature data, and real-time production raw material data. The real-time data is input into the temperature strategy model, and the temperature control strategy output by the temperature strategy model is operated to generate control instructions to control the equipment in the soda ash production process for temperature adjustment.

[0010] Preferably, in a possible implementation of the first aspect, the step S300 determines whether the real-time production equipment temperature meets the preset temperature range requirements, specifically by obtaining a temperature prediction curve, collecting real-time production equipment temperature data according to a preset time interval, and comparing it with the temperature prediction curve to determine whether the real-time production equipment temperature is within the preset upper and lower floating range of the temperature prediction curve at that time point.

[0011] Preferably, in a possible implementation of the first aspect, step S400 further includes: The abnormal situation includes that the real-time production equipment temperature is not within the preset upper and lower floating range of the temperature prediction curve or the predicted temperature change curve does not meet the preset target temperature of the soda ash production equipment; The automatic adjustment of the temperature control strategy, when an abnormal situation occurs, inputs the real-time temperature and precipitation data, the real-time production equipment temperature data and the real-time production raw material data into the temperature strategy model to adjust the temperature control strategy.

[0012] The new temperature prediction curve is generated by inputting real-time temperature and precipitation data, real-time production equipment temperature data, real-time production raw material data and the adjusted temperature control strategy into the temperature prediction model to generate a new temperature prediction curve.

[0013] Preferably, in a possible implementation of the first aspect, step S500 further includes: When the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism is triggered. The early warning mechanism includes the control center generating early warning information, a pop-up prompt on the control center display screen, and sending the early warning information to the terminal device.

[0014] Preferably, in a possible implementation of the first aspect, if the abnormal situation is not resolved within a preset number of adjustments, that is, if the abnormal situation is adjusted for a preset number of adjustments and the abnormal situation occurs again within a preset warning time, the warning mechanism is triggered; the preset number of adjustments is at least once.

[0015] In a second aspect, the present invention provides a temperature control system for a soda ash production process, the system comprising: Model training module: obtains the first historical data of the soda ash production process, performs data preprocessing to obtain the second historical data, and uses the second historical data to train the temperature strategy model and the temperature prediction model; Temperature control module: This module obtains real-time data from the soda ash production process and uses the temperature control strategy output by the temperature strategy model to control the temperature of the equipment in the soda ash production process. Temperature judgment module: inputs real-time data and real-time temperature control strategy into the temperature prediction model to obtain the temperature prediction curve, compares the real-time production equipment temperature with the temperature prediction curve, and determines whether the real-time production equipment temperature meets the temperature range requirements; Temperature adjustment module: When an abnormal situation occurs, it automatically adjusts the temperature control strategy and generates a new temperature prediction curve; Early warning module: If the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism will be triggered.

[0016] The beneficial effects of the present invention are as follows: It utilizes deep learning technology to achieve intelligent, automated control, significantly improving the accuracy and real-time nature of temperature control while reducing manual intervention. It enables real-time monitoring and adjustment of production equipment temperatures, ensuring that temperatures remain within the optimal range, stabilizing product quality, and improving production efficiency. Furthermore, it provides automatic abnormality handling and early warning capabilities, reducing production risks, labor costs, and further lowering production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A flow chart of a temperature control method for a soda ash production process is provided for this application.

[0019] Figure 2 The present application provides a schematic structural diagram of a temperature control system for a soda ash production process.

[0020] Explanation of the accompanying drawings: 1-model training module, 2-temperature control module, 3-temperature judgment module, 4-temperature adjustment module, 5-early warning module. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] Example 1: Figure 1 As shown, the present invention provides a temperature control method for a soda ash production process, comprising: Step S100: Acquire first historical data in the soda ash production process, perform data preprocessing to obtain second historical data, and use the second historical data to train a temperature strategy model and a temperature prediction model; Specifically, step S100 of this application also includes: Step S110: Acquire first historical data during the soda ash production process and transmit it to the control center.

[0023] The control center is a core facility that integrates data collection, storage, processing, analysis, mining, and visualization, providing comprehensive, accurate, and timely data support and decision-making for the entire soda ash production process. The control center used in this example is equipped with a high-performance computing unit, a PLC communication interface, a database, a deep learning training environment, and a display.

[0024] In this embodiment, the ammonia-soda process is used to produce soda ash, using natural gas as fuel. The production process is temperature-controlled in four steps, including limestone calcination, brine ammonia absorption, carbonation, and sodium bicarbonate calcination. The first historical data contains historical data for the four steps: The limestone calcination step includes lime kiln target temperature, historical temperature and precipitation data, historical lime kiln temperature data, historical natural gas supply rate, historical combustion air flow rate, and historical limestone data; The brine ammonia absorption step includes the target temperature of the ammonia absorption tower, historical temperature and precipitation data, historical ammonia absorption tower temperature data, historical ammonia and brine flow data, and historical cooling water flow data; The carbonation reaction step includes the carbonation tower target temperature, historical air temperature and precipitation data, historical carbonation tower temperature data, historical ammonia brine and carbon dioxide concentration data, historical air intake and water outlet speed, and historical cooling water flow data; The sodium bicarbonate calcination step includes rotary kiln target temperature, historical air temperature and precipitation data, historical rotary kiln temperature data, historical natural gas supply rate, historical combustion air flow rate, and historical sodium bicarbonate data.

[0025] In this embodiment, historical temperature and precipitation data are obtained from the meteorological department. The target temperature of the lime kiln is 1100°C, the target temperature of the ammonia absorption tower is 50°C, the target temperature of the carbonation tower is 25°C, and the target temperature of the rotary kiln is 175°C. The temperature sensors in the lime kiln, ammonia absorption tower, carbonation tower, and rotary kiln transmit the collected temperature data back to the control center every 5 minutes, that is, the historical temperature data is obtained from the control center; the historical natural gas supply rate in the lime kiln and rotary kiln, the historical combustion air flow rate, and the historical cooling water flow data in the ammonia absorption tower and carbonation tower are obtained through historical operation logs; historical limestone data, historical ammonia and brine flow data, historical ammonia brine and carbon dioxide concentration data, air intake and alkali output rate, and historical sodium bicarbonate data are also obtained through historical operation logs. All the first historical data obtained contain a timestamp.

[0026] The temperature sensor uses a thermocouple to collect device temperature. Thermocouples can maintain high-precision temperature measurements even in high-temperature environments. The temperature sensor is equipped with a ZigBee module. A gateway device, equipped with dual ZigBee and Ethernet communication interfaces, is used between the ZigBee module and the PLC, enabling simultaneous communication with both the ZigBee network and the PLC. After the thermocouple is integrated with the ZigBee module, it transmits temperature data to the gateway device via the ZigBee wireless network. Upon receiving the ZigBee signal, the gateway device converts it into an Ethernet data packet and sends it to the PLC via the Ethernet interface. Finally, the PLC transmits data to the control center via a data interface.

[0027] Step S120: performing data preprocessing on the first historical data, including data cleaning and data conversion, to obtain second historical data.

[0028] In this embodiment, Python's Pandas library is used to load the first historical data. First, data cleaning is performed. For each missing value, the average of the previous valid value and the next valid value is used to fill it. Check whether the data has exactly the same timestamp and historical data records. For redundant records, only one of them is retained. The interquartile range (IQR) method is used to determine whether it contains outliers or extreme values. The calculation formula is IQR = Q3 - Q1, where Q1 is the first quartile, which is the data point at the 25% position after the data is sorted, and Q3 is the third quartile, which is the data point at the 75% position after the data is sorted. The definition of an outlier is a data point lower than Q1 - 1.5 * IQR or higher than Q3 + 1.5 * IQR.

[0029] After data cleaning is completed, data conversion is performed to convert the first historical data into the second historical data. The second historical data is the first historical data after cleaning and data conversion, which meets the preset model training data format. In this embodiment, the unit of all temperature data is ℃, and it is retained to one decimal place. The unit of all flow data is , rounded to two decimal places. All precipitation data is in ml, rounded to two decimal places. All concentration data is in %, rounded to one decimal place. Limestone data includes limestone weight and limestone moisture content. The unit of limestone weight is kilograms, rounded to one decimal place. The unit of limestone moisture content is %, rounded to one decimal place. Sodium bicarbonate data includes sodium bicarbonate weight. The unit of sodium bicarbonate weight is kilograms, rounded to one decimal place. The second historical data is divided into training set, validation set, and test set in a ratio of 8:1:1.

[0030] Step S130: Using a deep learning method based on the second historical data, respectively train the temperature strategy model and the temperature prediction model.

[0031] A deep learning model is trained based on the second historical data. In this embodiment, different models are trained for different production steps, outputting temperature control strategies for the equipment within a preset control time for each production step, as well as temperature prediction curves within the preset control time. In this embodiment, the preset control time is 6 hours. This embodiment includes eight models, all using a PyTorch architecture. The model structures are described in detail using a limestone calcination temperature strategy model and a limestone calcination temperature prediction model as examples.

[0032] Limestone Calcination Temperature Strategy Model: The input layer defines a class called InputLayer, which inherits from PyTorch's nn.Module. The InputLayer initialization method __init__ accepts four parameters: num_temp_features, num_precip_features, num_lime_features, and num_stone_features, representing temperature-related features, precipitation-related features, lime kiln temperature features, and limestone features, respectively. For each feature type, a fully connected layer (nn.Linear) is created to map the input features to a higher-dimensional space.

[0033] In the forward method, three parameters are received: temp_precip, lime_temp, and stone_data, which correspond to the batch data of temperature and precipitation characteristics, lime kiln temperature characteristics, and limestone characteristics, respectively. Each parameter represents the corresponding feature data of all samples in a batch at the same time step.

[0034] Each type of feature data is first processed through its own fully connected layer and the ReLU activation function is applied to increase nonlinearity. The processed feature data is then concatenated along the feature dimension (i.e., the second dimension, dim=1) to form a comprehensive feature vector combined_features that contains information about all types of features.

[0035] The feature extraction and fusion layer defines the LSTMFeatureExtractor module, which extracts and fuses features of sequence data through the LSTM (Long Short-Term Memory) layer and inherits the nn.Module base class to build it.

[0036] In the initialization method of LSTMFeatureExtractor, several key parameters are defined: input_size, hidden_size, num_layers, bidirectional, and dropout. These parameters are used to configure the input feature dimension of the LSTM layer, the size of the hidden layer, the number of LSTM layers, whether to use bidirectional LSTM, and whether to apply dropout between layers to prevent overfitting.

[0037] In the forward method, the input data x (of shape (batch, seq_len, input_size)) is passed to the LSTM layer. After processing, the LSTM layer returns two main outputs: outputs and (hn, cn). outputs contains the hidden state and cell state of the LSTM at each time step, while (hn, cn) contains the hidden state and cell state at the last time step. Finally, the forward method returns the processed hidden state hn as the feature representation of the entire sequence.

[0038] The output layer defines a PyTorch module named OutputLayer, which inherits from nn.Module. Its primary function is to predict the control strategy for the natural gas supply rate and combustion air flow rate based on the LSTM output. The __init__ method of OutputLayer accepts three parameters: hidden_size, num_time_steps, and num_outputs. The hidden_size is the size of the hidden layer of the LSTM, which determines the number of features input to the fully connected layer (nn.Linear). num_outputs is the number of neurons in the output layer, which represents the number of predicted values ​​for the natural gas supply rate and combustion air flow rate.

[0039] self.fc_gas and self.fc_air are two fully connected layers that are used to predict the natural gas supply rate and combustion air flow rate, respectively, from the hidden state h_n of an LSTM or other recurrent neural network layer. Both fully connected layers have the same number of input features as hidden_size and the same number of output features as num_outputs.

[0040] The forward method receives the final hidden state h_n of the LSTM or other recurrent neural network layer as input. This hidden state is then passed to self.fc_gas and self.fc_air to calculate the predicted values ​​for the natural gas supply rate and combustion air flow rate, respectively. These two predicted values ​​are then returned together.

[0041] The MSE loss function is used as the model's loss function. A PyTorch module named ModelLoss is defined, which inherits from nn.Module. In ModelLoss's __init__ method, basic initialization is performed by calling the parent class's __init__ method. It also instantiates a PyTorch built-in mean squared error (MSE) loss function object, self.mse_loss, for subsequent loss calculations.

[0042] In the forward method, two parameters are received: predictions and targets. predictions is a tuple containing two elements, namely the predicted value of the natural gas supply speed gas_speed_pred and the predicted value of the combustion air flow rate air_flow_pred. targets is also a tuple containing two elements, namely the true value of the natural gas supply speed gas_speed_target and the true value of the combustion air flow rate air_flow_target. The instantiated MSE loss function self.mse_loss is used to calculate the losses between the predicted values ​​and the true values ​​of the natural gas supply speed and the combustion air flow rate, namely gas_loss and air_loss. Finally, the two loss values ​​are added together to obtain the total loss value total_loss, which is returned.

[0043] Limestone Calcination Temperature Prediction Model: This model defines a class called LimeKilnTemperaturePredictor, which inherits from PyTorch's nn.Module. In the __init__ method of the LimeKilnTemperaturePredictor class, several key model parameters are defined: input_dim, the feature dimension of each time step (i.e., the number of features in the data input to the LSTM layer); hidden_dim, the hidden layer dimension of the LSTM layer, which determines the dimension of the LSTM unit's internal state; num_layers, the number of LSTM layers; multi-layer LSTMs can capture more complex sequential relationships; and output_dim, the output dimension, which refers to the dimension of the model's final output.

[0044] An LSTM layer is defined through nn.LSTM, whose input dimension is input_dim, hidden layer dimension is hidden_dim, number of layers is num_layers, and batch_first=True is set. The first dimension of the input data is the batch size. The LSTM layer is responsible for extracting useful features from the input time series data.

[0045] Through a fully connected layer nn.Linear, the features of the last time step output by LSTM are converted into the final prediction result, which is mapped to the output dimension through the fully connected layer, that is, the temperature prediction values ​​​​at several future time points are obtained.

[0046] The forward method defines the forward propagation of data through the model. First, the input data x (of shape (batch_size, seq_len, input_dim)) is fed into the LSTM layer, where seq_len is the length of the time series. After the LSTM layer processes the entire sequence, it extracts the output of the last time step as a summary of the entire sequence. This summary is fed into a fully connected layer to produce the final predictions, predictions, of shape (batch_size, output_dim), representing the predicted future temperature value for each sample. The MSE loss function is also used as the model's loss function.

[0047] Models are trained using the training set, and the test set is used to evaluate the model's generalization ability after training, that is, its performance on unseen data. The validation set tests the model at different stages of training to adjust the model's hyperparameters to avoid overfitting and ensure good performance on unseen data.

[0048] The remaining six models are the brine ammonia absorption temperature strategy model, the brine ammonia absorption temperature prediction model, the carbonation reaction temperature strategy model, the carbonation reaction temperature prediction model, the sodium bicarbonate calcination temperature strategy model, and the sodium bicarbonate calcination temperature prediction model. These models are similar in structure to the limestone calcination temperature strategy model and the limestone calcination temperature prediction model, so they are not described in detail here.

[0049] Step S200: acquiring real-time data of the soda ash production process, and controlling the temperature of the equipment in the soda ash production process according to the temperature control strategy output by the temperature strategy model.

[0050] Specifically, the control system receives real-time sensor data based on the temperature control strategy output by the deep learning model. The PLC, serving as the core control unit, processes the data according to the preset temperature control strategy and generates control instructions. These instructions are executed by devices such as electric control valves to adjust relevant parameters (including cooling water flow, natural gas supply rate, and combustion air flow rate). Sensors continue to monitor the adjusted parameters and feed real-time data back to the PLC, forming a closed-loop control system to ensure the stability and accuracy of the production process.

[0051] Temperature regulation of equipment in the soda ash production process is divided into four steps: limestone calcination, brine ammonia absorption, carbonation, and sodium bicarbonate calcination. This example uses the limestone calcination and brine ammonia absorption steps as examples. Temperature regulation in the carbonation and brine ammonia absorption steps is similar, both utilizing cooling water flow rate for temperature control. Temperature regulation in the sodium bicarbonate and limestone calcinations is similar, utilizing adjustments to the natural gas supply rate and combustion air flow rate for temperature control.

[0052] Limestone calcination steps: Load the pre-trained limestone calcination temperature strategy model in the control center, obtain real-time temperature and precipitation data from the meteorological department, obtain real-time lime kiln temperature data through the temperature sensor, and obtain the used limestone data from the control center. The limestone data includes limestone weight and moisture content.

[0053] Real-time data is input into the limestone calcination temperature strategy model, which outputs a temperature control strategy, specifically involving the control scheme of the natural gas supply rate and combustion air flow rate during the calcination process.

[0054] The natural gas supply rate is related to combustion efficiency. When the natural gas supply rate is moderate, the fuel and air are fully mixed, achieving complete combustion, thereby improving combustion efficiency. The natural gas supply rate also affects flame temperature. Given a fixed fuel supply, increasing the natural gas supply rate can increase flame temperature because more fuel burns in a shorter period of time, releasing more heat. The combustion air flow rate affects the degree of mixing between natural gas and air. An appropriate air flow rate helps the fuel and air mix thoroughly, forming a uniform mixture, thereby improving combustion efficiency.

[0055] The control system adjusts the natural gas supply rate and combustion air flow rate during the limestone calcination process in real time based on the temperature control strategy output by the model. Natural gas flow rate sensors and combustion air flow rate sensors are used in the gas supply pipeline to measure and transmit flow rate data in real time. Electric regulating valves are installed in the natural gas pipelines to adjust the valve opening based on control signals, thereby controlling the natural gas supply rate. Electric regulating valves are also installed in the combustion air pipelines to adjust the combustion air flow rate.

[0056] The control center uses a programmable logic controller as the core control unit, which receives data from sensors and performs calculations and processing according to preset control logic and algorithms; it is equipped with a human-machine interface to set process parameters, monitor the production process, receive alarm information, and interact with the PLC for data.

[0057] The PLC outputs the natural gas supply rate and combustion air flow rate, sending control instructions to the electric control valve. Sensors collect the natural gas and combustion air flow rates in real time and transmit the data to the PLC. The PLC analyzes and compares the received data to determine the deviation between the current state and the target state. Based on the magnitude and direction of the deviation, the PLC adjusts the control instructions, gradually narrowing the deviation by adjusting the opening of the natural gas and combustion air control valves, bringing the actual parameters close to or reaching the set values. The sensors continue to monitor the adjusted parameters and transmit the updated data to the PLC. The PLC makes further adjustments based on the updated data, forming a closed loop to ensure the stability and accuracy of the limestone calcination process.

[0058] Brine ammonia absorption steps: load the pre-trained brine ammonia absorption temperature strategy model in the control center, obtain real-time temperature and precipitation data from the meteorological department, obtain real-time ammonia absorption tower temperature data through the temperature sensor, and obtain ammonia and brine flow data from the control center.

[0059] The real-time temperature and precipitation data, real-time ammonia absorption tower temperature data, and ammonia and brine flow data are input into the brine ammonia absorption temperature strategy model, and the temperature control strategy during the brine ammonia absorption process is output, that is, the cooling water flow control strategy during the entire limestone calcination process.

[0060] By controlling the cooling water flow rate, the brine temperature within the ammonia absorption tower can be effectively regulated and maintained, facilitating the dissolution and absorption of ammonia in the brine. Keeping the brine at a lower temperature accelerates the diffusion of ammonia molecules into the brine, thereby improving ammonia absorption efficiency. If the brine temperature is too high, it will not only affect the ammonia absorption effect, but may also cause certain components in the brine to undergo undesirable chemical reactions or crystallize, clogging equipment and pipelines.

[0061] The control system adjusts the cooling water flow rate during the ammonia absorption process based on the temperature control strategy for the output brine. Flow sensors are installed on the cooling water pipelines to monitor the cooling water flow rate in real time, providing real-time data to the control system. The PLC receives signals from the flow sensors and is responsible for issuing and executing commands. Electric regulating valves are installed on the cooling water pipelines to automatically adjust the valve opening according to the controller's instructions, thereby controlling the cooling water flow rate.

[0062] A flow sensor monitors the cooling water flow in real time and transmits the data to the PLC. The PLC compares the received flow data with the output of the brine ammonia absorption temperature strategy model to determine whether the cooling water flow needs to be adjusted. If adjustment is necessary, the PLC calculates the adjustment amount and generates control instructions.

[0063] The PLC sends control commands to the electric control valve, which automatically adjusts the valve opening according to the commands, thereby changing the cooling water flow rate. During the adjustment process, the flow sensor continues to monitor the cooling water flow rate and feeds real-time data back to the controller. The PLC evaluates the adjustment results based on this feedback and makes further adjustments as needed.

[0064] Step S300: input the real-time data and the real-time temperature control strategy into the temperature prediction model to obtain a temperature prediction curve, compare the real-time production equipment temperature with the temperature prediction curve, and determine whether the real-time production equipment temperature meets the temperature range requirements.

[0065] Specifically, determining whether the real-time production equipment temperature meets a preset temperature range requirement includes obtaining a temperature prediction curve, collecting real-time production equipment temperature data at preset time intervals, comparing it with the temperature prediction curve, and determining whether the real-time production equipment temperature is within a preset upper and lower floating range of the temperature prediction curve at that time point. This embodiment includes four stages: a limestone calcination stage, a brine ammonia absorption stage, a carbonation reaction stage, and a sodium bicarbonate calcination stage.

[0066] S310 limestone calcination phase: Before production begins, real-time temperature and precipitation data are obtained from the meteorological department. A temperature sensor is used to obtain the real-time temperature of the lime kiln. The control center also collects the weight and moisture content of the limestone, as well as the natural gas supply rate and combustion air flow rate output by the limestone calcination temperature strategy model. This data is input into the limestone calcination temperature prediction model to generate a lime kiln temperature prediction curve for the next six hours. During production, the lime kiln temperature is collected every five minutes, and the recorded temperature value is then compared with the prediction curve to determine whether it fluctuates within a range of 25°C above or below the prediction curve. Under normal circumstances, the temperature management strategy and temperature prediction curve for limestone calcination are updated hourly, meaning the relevant models reanalyze the data and adjust the strategy and prediction curve every hour.

[0067] During the S320 brine ammonia absorption phase, before production begins, real-time meteorological data is obtained from the meteorological department. A temperature sensor measures the real-time temperature of the ammonia absorption tower. Ammonia and brine flow rates, along with the cooling water flow rate output from the brine ammonia absorption temperature strategy model, are collected from the control center. This input data is fed into the brine ammonia absorption temperature prediction model to generate a six-hour ammonia absorption tower temperature forecast curve. During production, the ammonia absorption tower temperature is recorded every five minutes and compared to the forecast curve to confirm that it is within a predicted range of 5°C. Unless there are any abnormalities, the brine ammonia absorption temperature control strategy and temperature forecast curve are also updated hourly to ensure precise control.

[0068] S330 Carbonation Reaction Stage: Before production begins, meteorological data, the carbonation tower's real-time temperature, the concentrations of ammoniacal brine and carbon dioxide, the air inlet and outlet rates, and the cooling water flow rate output by the carbonation reaction temperature strategy model are collected. This data is input into the carbonation reaction temperature prediction model to generate a temperature forecast curve for the next six hours. During production, the carbonation tower temperature is monitored every five minutes and compared to the forecast curve to verify that it is within the predicted range of plus or minus 5°C. Unless otherwise specified, the carbonation reaction temperature control strategy and forecast curve are updated hourly.

[0069] During the S340 sodium bicarbonate calcination phase, before production begins, meteorological data, the current rotary kiln temperature, the weight of sodium bicarbonate, and the natural gas and air flow rates output by the sodium bicarbonate calcination strategy model are collected. This data is fed into the sodium bicarbonate calcination temperature prediction model to generate a six-hour rotary kiln temperature forecast curve. During production, the rotary kiln temperature is recorded every five minutes and compared to the forecast curve to confirm that it is within a predicted range of 10°C. Under normal circumstances, the sodium bicarbonate calcination temperature control strategy and temperature forecast curve are also updated hourly.

[0070] Step S400: When an abnormal situation occurs, the temperature control strategy is automatically adjusted and a new temperature prediction curve is generated.

[0071] Abnormal situations include that the real-time production equipment temperature is not within the preset upper and lower floating range of the temperature prediction curve or the predicted temperature change curve does not meet the preset target temperature of the soda ash production equipment.

[0072] Specifically, in the limestone calcination step, when the real-time temperature of the lime kiln at a certain moment does not meet the range of 25°C above and below the temperature change curve, the control center will infer the limestone calcination temperature strategy model based on the current real-time data, and adjust the equipment through the control center based on the inferred temperature control strategy, and generate a new temperature change prediction curve through the limestone calcination temperature prediction model according to the new temperature control strategy.

[0073] If the temperature change prediction curve at a certain moment does not meet the target temperature, that is, during the limestone calcination process, the temperature change prediction curve shows a temperature outside the range of 1075°C to 1125°C, and the limestone calcination has already begun but has not yet been completed, it indicates that if the current temperature control strategy is continued, the lime kiln will not meet the target temperature. The control center will infer the limestone calcination temperature strategy model based on the current real-time data, adjust the inferred temperature control strategy through the control center, and generate a new temperature change prediction curve based on the new temperature control strategy using the limestone calcination temperature prediction model.

[0074] Similarly, in steps such as brine ammonia absorption, carbonation reaction, and sodium bicarbonate calcination, when it is detected that the real-time temperature deviates from the allowable range of the predicted temperature change curve of each step (up or down 5°C, up or down 5°C, up or down 10°C, respectively) or the predicted curve indicates that the future temperature will deviate from the target range (45°C to 55°C, 20°C to 30°C, and 165°C to 185°C, respectively), the control center will make inference adjustments based on the corresponding temperature strategy model and update the temperature prediction curve in real time to ensure the stability and efficiency of the production process.

[0075] Step S500: If the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism is triggered.

[0076] Specifically, if an abnormality is not resolved within a preset number of adjustments, an early warning mechanism is triggered. This mechanism involves the control center generating an early warning message, displaying a pop-up window on the control center's display screen, and sending the early warning message to the terminal device. If an abnormality is not resolved within the preset number of adjustments, i.e., if the abnormality is adjusted for the preset number of adjustments and then recurs within the preset early warning time, the early warning mechanism is triggered. In this embodiment, the preset number of adjustments is one, and the preset early warning time is one hour.

[0077] In this embodiment, when the real-time temperature of the device deviates from the temperature change curve or the temperature change curve does not meet the target temperature, the temperature control strategy is automatically adjusted and the temperature change prediction curve is updated. If the real-time temperature deviates from the temperature change curve or the temperature change curve does not meet the target temperature again within one hour, the early warning mechanism is triggered.

[0078] A warning prompt is displayed on the control center screen. Clicking the warning prompt includes detailed data on temperature changes during the operation. If the warning prompt displayed on the screen is not processed for more than 5 minutes, an alert is sent to the terminal device. The warning is sent to the terminal device via text message to the manager's mobile phone. The SMS sending platform uses Alibaba Cloud SMS Service. By calling the API interface, the warning information is formatted as SMS content and a list of manager mobile phone numbers to receive the SMS is specified. Alibaba Cloud SMS Service will return the sending results, including a list of mobile phone numbers where the message was successfully sent and a list of mobile phone numbers where the message was failed. The system will handle the returned results accordingly, including resending failed SMS messages and recording the sending log.

[0079] Example 2: Figure 2 As shown, the present invention provides a temperature control system for a soda ash production process, including 1-model training module, 2-temperature control module, 3-temperature judgment module, 4-temperature adjustment module, 5-early warning module, wherein: Model training module: obtains the first historical data of the soda ash production process, performs data preprocessing to obtain the second historical data, and uses the second historical data to train the temperature strategy model and the temperature prediction model; Temperature control module: This module obtains real-time data from the soda ash production process and uses the temperature control strategy output by the temperature strategy model to control the temperature of the equipment in the soda ash production process. Temperature judgment module: inputs real-time data and real-time temperature control strategy into the temperature prediction model to obtain the temperature prediction curve, compares the real-time production equipment temperature with the temperature prediction curve, and determines whether the real-time production equipment temperature meets the temperature range requirements; Temperature adjustment module: When an abnormal situation occurs, it automatically adjusts the temperature control strategy and generates a new temperature prediction curve; Early warning module: If the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism will be triggered.

[0080] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A temperature control method for a soda ash production process, characterized in that: include: Step S100: acquiring first historical data in the soda ash production process, performing data preprocessing to obtain second historical data, and using the second historical data to train a temperature strategy model and a temperature prediction model; Step S200: acquiring real-time data in the soda ash production process, and controlling the temperature of the equipment in the soda ash production process according to the temperature control strategy output by the temperature strategy model; Step S300: input the real-time data and the real-time temperature control strategy into the temperature prediction model to obtain a temperature prediction curve, compare the real-time production equipment temperature with the temperature prediction curve, and determine whether the real-time production equipment temperature meets the temperature range requirement; Step S400: When an abnormal situation occurs, the temperature control strategy is automatically adjusted and a new temperature prediction curve is generated; Step S500: If the abnormal situation is not resolved within the preset adjustment times, the early warning mechanism is triggered.

2. The method according to claim 1, characterized in that: The step S100 further comprises: Obtain the first historical data of the soda ash production process and transmit it to the control center; Performing data preprocessing on the first historical data, including data cleaning and data conversion, to obtain second historical data; Based on the second historical data, the temperature strategy model and the temperature prediction model are trained respectively using a deep learning method.

3. The method according to claim 2, characterized in that: The first historical data includes historical temperature and precipitation data, historical production equipment temperature data, historical production raw material data, historical natural gas supply speed data, historical combustion air flow data, and historical cooling water flow data; the second historical data is the first historical data after cleaning and data conversion to meet the preset model training data format.

4. The method according to claim 2, characterized in that: The temperature strategy model outputs the temperature control strategy of the equipment within the preset control time in the soda ash production process, including controlling the natural gas supply speed, combustion air flow rate and cooling water flow rate; the temperature prediction model outputs the temperature prediction curve within the preset control time according to the temperature control strategy within the preset control time.

5. The method according to claim 1, characterized in that: The step S200 further comprises: The control system obtains real-time data in the soda ash production process, including real-time temperature and precipitation data, real-time production equipment temperature data, and real-time production raw material data, inputs the real-time data into the temperature strategy model, performs operations according to the temperature control strategy output by the temperature strategy model, and generates control instructions to control the equipment in the soda ash production process to adjust the temperature.

6. The method according to claim 1, characterized in that: In the step S300, it is determined whether the real-time production equipment temperature meets the preset temperature range requirement, specifically: obtaining a temperature prediction curve, collecting real-time production equipment temperature data according to a preset time interval, comparing it with the temperature prediction curve, and determining whether the real-time production equipment temperature is within the preset upper and lower floating range of the temperature prediction curve at that time point.

7. The method according to claim 1, characterized in that: The step S400 further comprises: The abnormal situation includes that the real-time production equipment temperature is not within the preset upper and lower floating range of the temperature prediction curve or the predicted temperature change curve does not meet the preset target temperature of the soda ash production equipment; The temperature control strategy is automatically adjusted. When an abnormal situation occurs, the real-time temperature and precipitation data, the real-time production equipment temperature data, and the real-time production raw material data are input into the temperature strategy model to adjust the temperature control strategy. The new temperature prediction curve is generated by inputting real-time temperature and precipitation data, real-time production equipment temperature data, real-time production raw material data and adjusted temperature control strategy into the temperature prediction model to generate a new temperature prediction curve.

8. The method according to claim 1, characterized in that: The step S500 further comprises: When the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism is triggered. The early warning mechanism includes the control center generating early warning information, a pop-up prompt on the control center display screen, and sending the early warning information to the terminal device.

9. The method according to claim 8, characterized in that If the abnormal situation is not resolved within the preset adjustment times, that is, the abnormal situation is adjusted for the preset adjustment times, and the abnormal situation occurs again within the preset warning time, the warning mechanism is triggered; The preset adjustment times is at least once.

10. A temperature control system for a soda ash production process, characterized in that: include: Model training module: obtains the first historical data in the soda ash production process, performs data preprocessing to obtain the second historical data, and uses the second historical data to train the temperature strategy model and the temperature prediction model; Temperature control module: obtains real-time data in the soda ash production process, and controls the temperature of the equipment in the soda ash production process according to the temperature control strategy output by the temperature strategy model; Temperature judgment module: inputs real-time data and real-time temperature control strategy into the temperature prediction model to obtain the temperature prediction curve, compares the real-time production equipment temperature with the temperature prediction curve, and determines whether the real-time production equipment temperature meets the temperature range requirements; Temperature adjustment module: when abnormal conditions occur, the temperature control strategy is automatically adjusted and a new temperature prediction curve is generated; Early warning module: If the abnormal situation is not resolved within the preset number of adjustments, the early warning mechanism will be triggered.

Images