Exhaust gas prediction method and system

Abstract

This specification provides a method and system for predicting waste gas emissions. The method includes: acquiring process parameters of a drying furnace in lithium carbonate production, and waste gas data corresponding to the process parameters of the drying furnace; determining key parameters among the process parameters of the drying furnace; determining the mapping relationship between the key parameters and the waste gas data; and obtaining predicted waste gas data based on the mapping relationship and the current process parameters of the drying furnace. By employing the above technical solution, environmental pollution problems in the lithium carbonate production process can be reduced by predicting waste gas emissions.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of lithium carbonate production technology, and in particular to a waste gas prediction method and system. Background Technology

[0002] Lithium carbonate is an inorganic compound with wide applications in various fields. For example, in the battery industry, especially in lithium batteries, lithium carbonate is one of the key raw materials.

[0003] Currently, lithium carbonate is typically prepared by reacting lithium ore (such as spodumene, lepidolite, etc.) with carbonates (such as sodium carbonate or potassium carbonate). For example, lithium ore is crushed and ground, then reacted with carbonates at high temperatures to produce lithium carbonate.

[0004] However, there are some problems in the production process of lithium carbonate, such as the high dust content in the exhaust gas, which directly affects environmental pollution and production efficiency. Summary of the Invention

[0005] In view of this, embodiments of this specification provide a method and system for predicting exhaust gas emissions, which can reduce environmental pollution problems in the lithium carbonate production process by predicting exhaust gas emissions.

[0006] First, this specification provides an embodiment of a waste gas prediction method, including:

[0007] Obtain the process parameters of the drying furnace in lithium carbonate production, as well as the waste gas data corresponding to the process parameters of the drying furnace;

[0008] Determine the key parameters in the process parameters of the drying oven;

[0009] Determine the mapping relationship between the key parameters and the exhaust gas data;

[0010] Based on the mapping relationship and the current process parameters of the drying oven, predicted exhaust gas data is obtained.

[0011] Optionally, determining the key parameters in the process parameters of the drying oven includes:

[0012] Change any process parameter of the drying oven and obtain the corresponding actual exhaust gas data;

[0013] The actual waste gas data is compared with the waste gas data, and when it is determined that the difference in dust content between the actual waste gas data and the waste gas data is greater than a preset difference value, the changed process parameter is taken as the key parameter.

[0014] Optionally, determining the mapping relationship between the key parameters and the exhaust gas data includes:

[0015] A preset neural network model is used to determine the mapping relationship between the key parameters and the exhaust gas data.

[0016] Optionally, the neural network model includes: a graph neural network, a time series network, a graph convolutional network, and at least one fully connected layer;

[0017] The step of using a preset neural network model to determine the mapping relationship between the key parameters and the exhaust gas data includes:

[0018] The graph neural network is used to obtain the correlation between key parameters;

[0019] The time series model is used to obtain the dependency relationship of exhaust gas data over time;

[0020] The graph convolutional network is used to obtain the correlation between different production areas in the drying oven;

[0021] At least one fully connected layer is used to fuse the correlation between key parameters, the dependence of waste gas data on changes over time, and the correlation between different production areas in the drying oven, so as to obtain the mapping relationship between the key parameters and the waste gas data.

[0022] Optionally, the step of using the graph neural network to obtain the correlation between key parameters includes:

[0023] The graph neural network is used to determine the nodes corresponding to each key parameter and the boundary information between each node. The boundary information is used to characterize the correlation between each key parameter.

[0024] Optionally, the step of using the time series model to obtain the dependency relationship of exhaust gas data over time includes:

[0025] Obtain time-series data related to key parameters and exhaust gas data in the drying oven;

[0026] Extract key feature data from the time series data;

[0027] Based on the aforementioned key feature data, the dependence of exhaust gas data on changes over time is determined.

[0028] Optionally, the step of using the graph convolutional network to obtain the correlation between different production areas in the drying oven includes:

[0029] The drying oven is divided into production areas, and a corresponding area map is constructed.

[0030] Determine the weight information between each production region in the region map, whereby the weight information is used to characterize the correlation strength between the production regions;

[0031] Based on the weight information between each production area, the spatial feature representation corresponding to the drying oven is determined.

[0032] Optionally, the waste gas prediction method further includes adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted waste gas data.

[0033] Optionally, adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted exhaust gas data includes:

[0034] When the predicted exhaust gas data meets the production requirements, the current process parameters in the drying furnace during lithium carbonate production are used as the target process parameters; otherwise, the process parameters in the drying furnace during lithium carbonate production are adjusted until the predicted exhaust gas data meets the production requirements.

[0035] This specification also provides an exhaust gas prediction system, including:

[0036] The data acquisition unit is configured to acquire the process parameters of the drying furnace in lithium carbonate production, as well as the waste gas data corresponding to the process parameters of the drying furnace.

[0037] The processing unit is configured to determine key parameters in the process parameters of the drying oven, determine the mapping relationship between the key parameters and the exhaust gas data, and obtain predicted exhaust gas data based on the mapping relationship and the current process parameters of the drying oven.

[0038] The waste gas prediction scheme adopted in the embodiments of this specification can determine the mapping relationship between key parameters and waste gas data by identifying the key parameters in the process parameters of the drying furnace and the waste gas data corresponding to the process parameters of the drying furnace. Since the waste gas data is generated in the production of lithium carbonate, and the mapping relationship can reflect the influence of the process parameters in the drying furnace on the generated waste gas data, the predicted waste gas data can be obtained based on the mapping relationship and the current process parameters in the drying furnace. It can then be determined whether the waste gas data meets the environmental protection requirements, thereby reducing the environmental pollution problem in the production process of lithium carbonate. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of this specification, the drawings used in the description of the embodiments of this specification or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart of an exhaust gas prediction method according to an embodiment of this specification is shown;

[0041] Figure 2 A flowchart illustrating a method for determining key parameters in the process parameters of a drying oven, as described in an embodiment of this specification, is shown.

[0042] Figure 3 This specification illustrates a flowchart of an embodiment for determining the mapping relationship between key parameters and exhaust gas data.

[0043] Figure 4 A schematic diagram of the structure of an exhaust gas prediction system according to an embodiment of this specification is shown. Detailed Implementation

[0044] As described in the background section, there are some problems in the production process of lithium carbonate, such as the high dust content in the generated waste gas, which directly affects environmental pollution and production efficiency.

[0045] Currently, the traditional method is to predict dust content based on simple statistical models, but this method ignores the complex relationship between dust content and the process parameters of the drying oven.

[0046] To address the aforementioned technical problems, this specification provides an exhaust gas prediction scheme. By determining the key parameters in the process parameters of the drying furnace and the exhaust gas data corresponding to the process parameters of the drying furnace, the mapping relationship between the key parameters and the exhaust gas data can be determined. Since the exhaust gas data is generated during lithium carbonate production, the predicted exhaust gas data can be obtained based on the mapping relationship and the current process parameters in the drying furnace. This allows for a determination of whether the exhaust gas data meets environmental protection requirements, thereby reducing environmental pollution problems in the lithium carbonate production process.

[0047] To enable those skilled in the art to better understand and implement the embodiments of this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings.

[0048] Reference Figure 1 The flowchart shown here illustrates a method for predicting exhaust gas emissions, which may specifically include the following steps:

[0049] S11, Obtain the process parameters of the drying furnace in lithium carbonate production, and the waste gas data corresponding to the process parameters of the drying furnace.

[0050] Specifically, in the process of lithium carbonate production, different process parameters are set in the drying furnace, resulting in different waste gas data. Therefore, waste gas data under the corresponding process parameters can be obtained as reference data to predict the waste gas data generated in subsequent lithium carbonate production processes.

[0051] In this embodiment, sensors (e.g., temperature sensors, humidity sensors, etc.) can be used to acquire process parameters and exhaust gas data in the drying oven.

[0052] In some examples, exhaust gas data may include dust content.

[0053] In some examples, process parameters may include temperature, humidity, drying rate, production time, etc.

[0054] S12, determine the key parameters in the process parameters of the drying oven.

[0055] Specifically, in the lithium carbonate production process, different process parameters have varying degrees of impact on the generated waste gas. For example, some process parameters have a relatively small impact on waste gas when they change within a controllable range; while for other process parameters, even small changes can significantly affect the content of the generated waste gas. In these cases, these process parameters can be considered as critical parameters.

[0056] In short, the key parameters in this invention can refer to process parameters that have a significant impact on exhaust gas data.

[0057] It should be noted that the terms "smaller" and "larger" in the embodiments of this specification are relative concepts and do not limit specific values. They are used to describe the degree of influence of process parameters on exhaust gas data, especially dust content.

[0058] S13, determine the mapping relationship between the key parameters and the exhaust gas data.

[0059] Specifically, after determining the key parameters, the mapping relationship between the two can be determined by using a preset processing method based on the acquired exhaust gas data.

[0060] S14. Based on the mapping relationship and the current process parameters of the drying oven, the predicted exhaust gas data is obtained.

[0061] Specifically, the mapping relationship can characterize the influence of process parameters in the drying oven on the generated waste gas data. Then, based on the current process parameters in the drying oven, the predicted waste gas data corresponding to the current process parameters in the drying oven can be obtained.

[0062] Using the waste gas prediction method in the above example, since the waste gas data is generated in the production of lithium carbonate, and the mapping relationship can reflect the influence of the process parameters in the drying furnace on the generated waste gas data, the predicted waste gas data can be obtained based on the mapping relationship and the current process parameters in the drying furnace. This allows us to determine whether the waste gas data meets environmental protection requirements, thereby reducing environmental pollution problems in the production process of lithium carbonate.

[0063] To enable those skilled in the art to better understand and implement the embodiments of this specification, the following describes in detail the concept, scheme, principle, and advantages of the embodiments of this specification in conjunction with the accompanying drawings and specific examples.

[0064] In some embodiments of this specification, the inventors conducted extensive experiments and research on factors affecting exhaust gas data and found that by changing the process parameters under the same conditions in the same drying oven, the key factors affecting the dust content in exhaust gas data can be determined.

[0065] As an example, see Figure 2 The flowchart shown in this specification illustrates a method for determining key parameters in the process parameters of a drying oven, as illustrated in the embodiment. Figure 2 As shown, the following steps can be performed:

[0066] S21, change any process parameter of the drying oven and obtain the corresponding actual exhaust gas data.

[0067] Specifically, for lithium carbonate production processes, multiple process parameters are often set in the drying furnace. By changing any one of these process parameters and obtaining actual waste gas data, the degree of influence of that process parameter on the waste gas data can be determined.

[0068] It should be noted that when changing any process parameter in the drying oven, the changed process parameter should be able to meet the requirements of lithium carbonate production, so as to avoid the inability to obtain accurate actual waste gas data due to excessive changes in process parameters, and thus the inability to determine key parameters.

[0069] S22, compare the actual waste gas data with the waste gas data, and when it is determined that the difference in dust content between the actual waste gas data and the waste gas data is greater than a preset content difference value, the changed process parameter is taken as a key parameter.

[0070] Specifically, by comparing the actual exhaust gas data obtained in step S21 with the historical exhaust gas data obtained without changing any process parameters, the difference in dust content between the two can be obtained. The specific value of the dust content difference reflects the degree of influence of the change in process parameters on the exhaust gas data. Therefore, when it is determined that the difference in dust content between the actual exhaust gas data and the historical exhaust gas data is greater than the preset difference value, it indicates that the change in this process parameter has a significant impact on the exhaust gas data. In this case, the changed process parameter can be regarded as a key parameter.

[0071] By comparing the actual exhaust gas data when any process parameter changes with the exhaust gas data when the process parameter remains unchanged, the difference in dust content can be obtained. Since the difference in dust content reflects the degree of impact of changes in process parameters on exhaust gas data, the accuracy of the obtained key parameters can be improved, thereby enhancing the accuracy of the mapping relationship.

[0072] In some embodiments of this specification, when key parameters are obtained, a correspondence between them and exhaust gas data can be established.

[0073] As an optional example, a preset neural network model can be used to determine the mapping relationship between the key parameters and the exhaust gas data. The preset neural network model can be obtained based on the key parameters in the process parameters of the drying oven and the corresponding exhaust gas data.

[0074] Because neural network models have better generalization ability and versatility, when using neural network models to determine the mapping relationship between key parameters and exhaust gas data, the accuracy of the mapping relationship can be improved, thereby obtaining more accurate predicted exhaust gas data, such as more accurate dust content.

[0075] In some examples in this manual, neural network models can be obtained in the following ways:

[0076] First, determine the type of neural network model.

[0077] As an alternative example, the neural network model could be a Regional Temporal Graph Convolutional Network (Reg T-GCN), a special form of Decomposed Graph Neural Network (DGNN) suitable for handling temporal and spatial data associations.

[0078] In this embodiment of the invention, the regional temporal graph neural network may include: a graph neural network, a time series network, a graph convolutional network, and at least one fully connected layer.

[0079] Next, the neural network model is trained, specifically including:

[0080] 1) Data preparation

[0081] Prepare a dataset for training, for example, collect historical exhaust gas data, including dust content, temperature, humidity, drying rate, etc., where temperature, humidity, drying rate, etc. can be used as feature data, and dust content as the target variable.

[0082] Before using a dataset, it can be cleaned and preprocessed to remove outliers and missing values. Afterward, the data can be standardized to ensure that the data in the dataset are of the same magnitude. Ensuring data quality requires necessary cleaning and preprocessing.

[0083] 2) Choosing a loss function

[0084] The loss function can quantify the difference between the predicted result and the actual result, so an appropriate loss function can be selected according to the type of dataset.

[0085] For example, mean squared error (MSE) is typically used for continuous prediction tasks; cross-entropy loss can be used for classification tasks.

[0086] 3) Select an optimization algorithm

[0087] Choose an appropriate optimization algorithm to tune the model parameters, such as Adam, stochastic gradient descent (SGD), etc.

[0088] It should be noted that when selecting an optimization algorithm, the convergence speed and stability of the algorithm need to be considered.

[0089] 4) Model Training

[0090] The preprocessed dataset is input into the model, and during training, changes in the loss function and other performance metrics are monitored.

[0091] In some examples, the model can capture the correspondence between dust content and key parameters by acquiring the correlations between various parameters of the lithium carbonate production drying furnace; time series data related to the lithium carbonate production drying furnace, such as historical temperature, humidity, drying speed and dust content records; and spatial correlations between different areas of the lithium carbonate production drying furnace.

[0092] Optionally, the following steps may also be performed during training:

[0093] 5) Hyperparameter tuning

[0094] Model performance can be optimized by adjusting the model's hyperparameters (such as learning rate, batch size, number of network layers, etc.).

[0095] For example, grid search, random search, or Bayesian optimization methods can be used to find the optimal combination of hyperparameters.

[0096] 6) Cross-validation

[0097] Cross-validation is used to evaluate the model's generalization ability to ensure that the model has stable performance on different subsets of data.

[0098] 7) Model Evaluation and Validation

[0099] The model was evaluated using an independent test dataset to assess its accuracy, predictive ability, and robustness.

[0100] In some examples, these test datasets may be obtained during the actual production process of lithium carbonate.

[0101] 8) Model optimization and fine-tuning

[0102] The model will be further optimized and fine-tuned based on the evaluation results.

[0103] For example, regularization techniques such as Dropout and L1 / L2 regularization can be used to prevent overfitting.

[0104] Through the above steps, we can train and optimize the Reg T-GCN model, enabling it to accurately predict the dust content in the exhaust gas from lithium carbonate production drying furnaces. This process requires meticulous adjustments and repeated validations to ensure the model's reliability and effectiveness in practical applications.

[0105] In the embodiments of this specification, to further improve the reliability and effectiveness of the obtained model, it may also include: model validation, including: evaluating the model using a validation dataset; checking the model's generalization ability and prediction accuracy; and adjusting the model as needed.

[0106] Specifically, the neural network model (Reg T-GCN model) in this embodiment of the invention can be trained in the following manner:

[0107] A1): Define the RegT-GCN network structure

[0108] A11): Determine the nodes of the graph neural network.

[0109] The nodes represent various parameters of the lithium carbonate production drying furnace, such as temperature, humidity, drying speed, and dust content.

[0110] Choosing appropriate parameters as nodes is crucial, as these parameters should be the key factors affecting the dust content in exhaust gas.

[0111] A12) Determine the edges between nodes in a graph neural network.

[0112] Edges represent the relationships between nodes, which may include physical relationships or statistical correlations.

[0113] For example, there may be a physical relationship between temperature and humidity, while drying speed and dust content may have a statistical correlation.

[0114] A13): Design the structure of a graph neural network

[0115] The structure of the entire graph is designed based on the relationship between nodes and edges. The structural design should take into account the actual operation process and physical laws in the production process.

[0116] A14) Characteristic representation of the graph

[0117] Assign features to each node. These features may include historical data, real-time data, etc., and these features should be able to fully reflect the state and changes of each parameter.

[0118] In some other examples, it may also include:

[0119] A15) Integrate other data sources

[0120] Consider integrating environmental data or other relevant data into graph neural networks, which can help the model gain a more comprehensive understanding of production processes and environmental changes.

[0121] A2) Time dimension processing

[0122] A21) Collection and processing of time series data

[0123] Collect time-series data related to the drying furnace during the lithium carbonate production process, such as historical temperature, humidity, drying speed, and dust content records, and organize the data to ensure the continuity and integrity of the time series.

[0124] Identification of inter-dependencies (A22)

[0125] Analyze the time dependence of various parameters, such as how dust content changes with time and other factors.

[0126] Use statistical methods, such as autocorrelation and cross-correlation functions, to identify and quantify these time dependencies.

[0127] A23) Feature Extraction of Time Series

[0128] Using time series analysis techniques, such as Fourier transform or wavelet transform, key features of time series data are extracted, and it is determined which features are most critical for dust content prediction.

[0129] A24): Construction of Time Series Models

[0130] Recurrent Neural Networks (RNNs) or Long Short-Term Memory Networks (LSTMs) can be used to handle time dependencies.

[0131] Among them, RNN or LSTM can effectively capture long-term and short-term dependencies in time series data.

[0132] A25) Training and Validation of Time Series Models

[0133] The time series model was trained using historical data to validate its predictive ability and ensure its accuracy and stability at different time scales.

[0134] In some optional examples, the following may also be included:

[0135] A26) Integrating Time Series Models and Graph Neural Networks

[0136] Ensure that the output of the time series model is effectively integrated with the spatial features in the graph neural network.

[0137] Through these steps, the Reg T-GCN model can not only capture the spatial correlations between parameters of lithium carbonate production drying furnaces, but also effectively process time-series data, thereby more accurately predicting the dust content in exhaust gas. This time-dimensional processing provides the model with a more comprehensive data perspective, enhancing the accuracy and reliability of predictions.

[0138] A3) Region Dimension Processing

[0139] This step focuses on capturing the spatial correlations between different areas of the drying furnace during lithium carbonate production to improve the accuracy of the predictive model. The following is a detailed description of this step:

[0140] A31) Identifying Spatial Correlation

[0141] The spatial relationships between different zones within the lithium carbonate production drying furnace were analyzed, including the mutual influence between parameters such as temperature, humidity, and dust content in different zones.

[0142] For example, environmental conditions in some areas may directly affect neighboring areas.

[0143] A32) Construct a spatial relationship diagram

[0144] Based on the identified spatial correlations, a graph representation is constructed, where nodes represent different regions, edges represent spatial correlations between regions, and the weights of the edges are determined to reflect the strength of the correlations between regions.

[0145] A33) Applications of graph convolutional networks

[0146] Graph Convolutional Networks (GCNs) are used to process spatial relational graphs in order to capture complex spatial relationships.

[0147] A34) Feature Extraction and Fusion

[0148] Features, such as the statistical characteristics of environmental parameters, are extracted from each regional node to form a comprehensive spatial feature representation.

[0149] A35) Integration of Regional and Temporal Characteristics

[0150] By processing spatial features and the temporal dimension, and integrating the features, the model can be ensured to consider the influence of both time and spatial dimensions simultaneously.

[0151] A36) Model Training and Optimization

[0152] The model is trained using real data to optimize the representation of spatial relationships, and the model parameters are tuned to ensure accurate capture of spatial relationships.

[0153] Through these steps, the Reg T-GCN model can effectively handle the spatial dimension and capture the complex relationships between different regions within the lithium carbonate production drying furnace. This handling of the spatial dimension allows the model to take into account the spatial heterogeneity of the production process when predicting the dust content in the exhaust gas, thereby improving the accuracy and reliability of the prediction.

[0154] In some embodiments, deep learning techniques such as multilayer perceptrons (MLPs) can be used to fuse the results of temporal and spatial processing.

[0155] A4) Output Layer Design

[0156] In this step, we will focus on the design of the output layer in the Reg T-GCN model. The output layer is the final part of the model and is responsible for transforming the features learned from the previous layers (temporal and region feature processing layers) into the final prediction result. The following is a detailed description of the sub-steps:

[0157] A41): Output Layer Design

[0158] Define the output target, that is, clarify the prediction target of the model, such as predicting the dust content in the exhaust gas of the lithium carbonate production drying furnace, and determine the structure and function of the output layer based on the prediction target.

[0159] A42) Design the output layer structure

[0160] The output layer can be a simple fully connected layer or it can contain multiple layers, depending on the complexity of the problem and the required accuracy.

[0161] For example, a linear output unit is typically used for continuous prediction targets (such as dust content); while a softmax function is used for classification tasks.

[0162] A43) Selection of activation function

[0163] For regression tasks (such as dust content prediction), activation functions are usually not required, or linear activation functions are used.

[0164] For classification tasks, an appropriate non-linear activation function, such as softmax, can be selected.

[0165] A44) Optimize the output layer

[0166] Adjust the parameters of the output layer, such as weights and biases, to optimize prediction performance.

[0167] Use appropriate loss functions, such as mean squared error (MSE) for regression tasks and cross-entropy loss for classification tasks.

[0168] A45) Verification and Adjustment of the Output Layer

[0169] Use the validation dataset to test the performance of the output layer.

[0170] Based on the validation results, adjust the structure or parameters of the output layer to improve prediction accuracy.

[0171] A46) Considerations for Ensemble Learning

[0172] In some complex scenarios, ensemble learning methods can be considered to improve model performance.

[0173] For example, the accuracy and robustness of the overall prediction can be improved by combining the outputs of multiple models.

[0174] By following the steps above, an effective Reg T-GCN model can be constructed to predict the dust content in the exhaust gas from the drying furnace in lithium carbonate production, while also considering the complex temporal and spatial relationships. This method not only provides accurate prediction results but also helps us to gain a deeper understanding of the various dynamic changes in the production process.

[0175] In some embodiments, when a trained neural network model is obtained, the mapping relationship between key parameters and exhaust gas data can be determined based on the preset neural network model.

[0176] As an optional example, see Figure 3 The flowchart shown in this specification illustrates a method for determining the mapping relationship between key parameters and exhaust gas data, as illustrated in the embodiments of this specification. Figure 3 As shown, it includes:

[0177] S31, The graph neural network is used to obtain the correlation between the key parameters.

[0178] Specifically, a graph neural network can be used to process each key parameter as a node. By analyzing the nodes, the correlation between each key parameter can be determined.

[0179] In the embodiments of this specification, it may specifically include: using the graph neural network to determine the nodes corresponding to each key parameter, as well as the boundary information between each node, wherein the boundary information is used to characterize the correlation between each key parameter.

[0180] In this context, nodes represent various parameters of the lithium carbonate production drying furnace, such as temperature, humidity, drying speed, and dust content; edges represent the relationships between nodes, which may include physical relationships or statistical correlations. For example, there may be a physical relationship between temperature and humidity, while drying speed and dust content may have a statistical correlation.

[0181] S32, using the time series model, obtain the dependency relationship of the exhaust gas data over time.

[0182] Specifically, by incorporating time series analysis, long-term and short-term dependencies in time series data can be effectively captured, that is, the dependencies of exhaust gas data over time.

[0183] In the embodiments of this specification, the process may specifically include: acquiring time-series data related to key parameters and exhaust gas data in the drying oven; extracting key feature data from the time-series data; and determining the dependency relationship of exhaust gas data over time based on the key feature data.

[0184] S33, using the graph convolutional network, obtain the correlation between different production areas in the drying oven.

[0185] Specifically, due to different regional locations, even with the same process parameters, the production processes in different production areas of the drying oven will vary. These differences will cause fluctuations between regions. Therefore, the correlation between different production areas can be obtained to determine the impact of spatial heterogeneity in the production process on the exhaust gas data.

[0186] S34. At least one fully connected layer is used to fuse the correlation between key parameters, the dependence of waste gas data on changes over time, and the correlation between different production areas in the drying oven, so as to obtain the mapping relationship between the key parameters and the waste gas data.

[0187] Specifically, through steps S31 to S33, the temporal and spatial relationship between key parameters and exhaust gas data can be obtained. Then, through at least one fully connected layer, the correspondence between key parameters and exhaust gas data can be captured, that is, the mapping relationship between key parameters and exhaust gas data can be determined.

[0188] It should be noted that for steps S31 to S34, the specific prediction process can be found in the description of the model training process. This is because the training process involves further analyzing the data to capture the mapping relationship, while the prediction process follows the reasoning process fixed in the training process, parsing the input data to obtain the corresponding output results.

[0189] By adopting the method in the above example and comprehensively considering the changes in the drying oven over time and space, it is possible to more accurately reflect various dynamic changes in the production process, thereby improving the accuracy of the obtained mapping relationship.

[0190] By determining the mapping relationship between key parameters and exhaust gas data, predicted exhaust gas data generated during the lithium carbonate production process can be applied. Based on this predicted exhaust gas data, process parameters in the drying furnace during lithium carbonate production can be adjusted to reduce dust emissions.

[0191] As an optional example, if it is determined that the predicted exhaust gas data meets the production requirements, the current process parameters in the drying furnace during lithium carbonate production are used as the target process parameters; otherwise, the process parameters in the drying furnace during lithium carbonate production are adjusted until the predicted exhaust gas data meets the production requirements.

[0192] Specifically, if the predicted exhaust gas data meets the production requirements, it means that the current exhaust gas emissions can balance production efficiency and environmental protection requirements as shown in the diagram, and the current process parameters can be used as the target process parameters; if the predicted exhaust gas data does not meet the production requirements, the process parameters in the drying furnace in lithium carbonate production can be adjusted until the predicted exhaust gas data that meets the production requirements is obtained.

[0193] In practical applications, the lithium carbonate production process can be carried out using the following steps:

[0194] The trained model is used to make predictions on real-time data, including: collecting real-time operating data of the drying oven; inputting the data into the model to obtain predicted dust content; and adjusting production parameters and optimizing waste gas treatment based on the prediction results.

[0195] Optionally, it may also include: results analysis and application, including: analyzing the prediction results to determine whether measures need to be taken to adjust the production process; optimizing production parameters based on the prediction results, such as adjusting the drying speed to reduce dust emissions.

[0196] Furthermore, it may also include: continuous monitoring and optimization, including: continuously monitoring the predictive performance of the model to ensure its accuracy and stability; and further optimizing and adjusting the model based on feedback from practical applications.

[0197] Using the exhaust gas prediction method in the above example can effectively reduce exhaust gas emissions.

[0198] As an example, dust emissions from drying ovens may increase in high humidity environments. By using the Reg T-GCN model, the dust content in such environments can be predicted in real time, and drying parameters can be adjusted based on the prediction results to reduce dust emissions.

[0199] As another example, dust emissions can vary significantly at different drying speeds. By using the Reg T-GCN model, dust content at different drying speeds can be predicted, providing operators with guidance to select the optimal drying rate.

[0200] As another example, while pursuing high production efficiency, environmental protection requirements, especially the control of dust emissions, need to be considered. By using the Reg T-GCN model, dust levels under different production conditions can be predicted, helping management to formulate strategies that balance production efficiency and environmental protection.

[0201] Through the steps and examples above, Reg T-GCN can provide accurate prediction results and data support for decision-making in the production process.

[0202] It is understood that the above description provides multiple exhaust gas prediction methods and embodiments of the present specification. The optional methods described in each embodiment can be combined and cross-referenced without conflict, thereby extending to a variety of possible embodiments. These can all be considered as embodiments disclosed and made public in this specification.

[0203] The specification also provides an exhaust gas prediction system corresponding to the above-described exhaust gas prediction method, which will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0204] It should be noted that the exhaust gas prediction system described below can be considered as a functional module required to implement the exhaust gas prediction method provided in this specification; the content of the exhaust gas prediction system described below can be referred to in correspondence with the content of the exhaust gas prediction method described above.

[0205] Reference Figure 4 The exhaust gas prediction system shown in this embodiment of the specification may include, in some embodiments of the specification, the exhaust gas prediction system as follows:

[0206] The data acquisition unit 110 is configured to acquire the process parameters of the drying furnace in lithium carbonate production, as well as the waste gas data corresponding to the process parameters in the drying furnace.

[0207] The processing unit 120 is configured to determine key parameters in the process parameters of the drying oven, determine the mapping relationship between the key parameters and the exhaust gas data, and obtain predicted exhaust gas data based on the mapping relationship and the current process parameters of the drying oven.

[0208] The specific processes of the data acquisition unit 110 and the processing unit 120 can be found in the foregoing content and will not be described in detail here.

[0209] Since the waste gas data is generated during lithium carbonate production, and the mapping relationship can reflect the impact of process parameters in the drying furnace on the generated waste gas data, the waste gas data can be predicted based on the mapping relationship and the current process parameters in the drying furnace. This allows us to determine whether the waste gas data meets environmental protection requirements, thereby reducing environmental pollution problems in the lithium carbonate production process.

[0210] This specification also provides an electronic device for predicting exhaust gas data, such as dust content, including a memory and a processor, wherein the memory is adapted to store one or more computer instructions, and the processor, when executing the computer instructions, performs the steps of the exhaust gas prediction method described in any of the foregoing embodiments.

[0211] In practice, electronic devices may also include expansion interfaces suitable for connecting with other devices to enable data interaction.

[0212] Specifically, electronic devices can be general-purpose or special-purpose computer equipment, or more specifically, servers or computer terminals, such as personal computer equipment, portable terminal equipment, etc.

[0213] In practice, the memory, processor, and expansion interface can be connected via a bus.

[0214] In specific implementations, the processor can be implemented by a processing chip such as a central processing unit (CPU) or a field programmable gate array (FPGA), or by an application specific integrated circuit (ASIC) or one or more integrated circuits configured to implement the embodiments of this specification.

[0215] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device.

[0216] This specification also provides a computer program product, including a computer program and / or instructions, which, when executed by a processor, perform the steps of the exhaust gas prediction method described in any of the foregoing embodiments.

[0217] This specification also provides a computer-readable storage medium storing computer instructions that, when executed, can perform the steps of the exhaust gas prediction method described in any of the foregoing embodiments. The computer-readable storage medium can be any suitable readable storage medium, such as an optical disc, a hard disk drive, or a solid-state drive. The instructions stored on the computer-readable storage medium execute the steps of the exhaust gas prediction method described in any of the foregoing embodiments, which will not be described further here.

[0218] The computer-readable storage medium may include, for example, any suitable type of memory cell, memory device, memory article, memory medium, storage device, storage article, storage medium and / or storage cell, such as memory, removable or non-removable medium, erasable or non-erasable medium, writable or rewritable medium, digital or analog medium, hard disk, floppy disk, optical disc read-only memory (CD-ROM), recordable optical disc (CD-R), rewritable optical disc (CD-RW), optical disc, magnetic medium, magneto-optical medium, removable memory card or disk, various types of digital universal optical disc (DVD), magnetic tape, cassette tape, etc.

[0219] Computer instructions may include any suitable type of code implemented using any appropriate high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc.

[0220] While the embodiments disclosed in this specification are as described above, the present invention is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for predicting waste gas emissions, characterized in that, include: Obtain the process parameters of the drying furnace in lithium carbonate production, as well as the waste gas data corresponding to the process parameters of the drying furnace; Determine the key parameters in the process parameters of the drying oven; Determine the mapping relationship between the key parameters and the exhaust gas data; Based on the mapping relationship and the current process parameters of the drying oven, predicted exhaust gas data is obtained; Determining the mapping relationship between the key parameters and the exhaust gas data includes: A preset neural network model is used to determine the mapping relationship between the key parameters and the exhaust gas data; The neural network model includes: a graph neural network, a time series network, a graph convolutional network, and at least one fully connected layer; The step of using a preset neural network model to determine the mapping relationship between the key parameters and the exhaust gas data includes: The graph neural network is used to obtain the correlation between key parameters; The time series network is used to obtain the dependency relationship of exhaust gas data over time; The graph convolutional network is used to obtain the correlation between different production areas in the drying oven; At least one fully connected layer is used to fuse the correlation between key parameters, the dependence of waste gas data on changes over time, and the correlation between different production areas in the drying oven, so as to obtain the mapping relationship between the key parameters and the waste gas data. The step of using the graph neural network to obtain the correlation between key parameters includes: The graph neural network is used to determine the nodes corresponding to each key parameter and the boundary information between each node. The boundary information is used to characterize the correlation between each key parameter. The step of using the time series network to obtain the dependency relationship of exhaust gas data over time includes: Obtain time-series data related to key parameters and exhaust gas data in the drying oven; Extract key feature data from the time series data; Based on the aforementioned key feature data, the dependence of exhaust gas data on changes over time is determined; The step of using the graph convolutional network to obtain the correlation between different production areas in the drying oven includes: The drying oven is divided into production areas, and a corresponding area map is constructed. Determine the weight information between each production region in the region map, whereby the weight information is used to characterize the correlation strength between the production regions; Based on the weight information between each production area, the spatial feature representation corresponding to the drying oven is determined; The key parameters in determining the process parameters of the drying oven include: Change any process parameter of the drying oven and obtain the corresponding actual exhaust gas data; The actual waste gas data is compared with the waste gas data, and when it is determined that the difference in dust content between the actual waste gas data and the waste gas data is greater than the preset content difference value, the changed process parameter is taken as the key parameter. The waste gas prediction method further includes: adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted waste gas data; The step of adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted waste gas data includes: When the predicted exhaust gas data meets the production requirements, the current process parameters in the drying furnace during lithium carbonate production are used as the target process parameters; otherwise, the process parameters in the drying furnace during lithium carbonate production are adjusted until the predicted exhaust gas data meets the production requirements.

2. A waste gas prediction system, characterized in that, include: The data acquisition unit is configured to acquire the process parameters of the drying furnace in lithium carbonate production, as well as the waste gas data corresponding to the process parameters of the drying furnace. The processing unit is configured to determine key parameters in the process parameters of the drying oven, determine the mapping relationship between the key parameters and the exhaust gas data, and obtain predicted exhaust gas data based on the mapping relationship and the current process parameters of the drying oven. Determining the mapping relationship between the key parameters and the exhaust gas data includes: A preset neural network model is used to determine the mapping relationship between the key parameters and the exhaust gas data; The neural network model includes: a graph neural network, a time series network, a graph convolutional network, and at least one fully connected layer; The step of using a preset neural network model to determine the mapping relationship between the key parameters and the exhaust gas data includes: The graph neural network is used to obtain the correlation between key parameters; The time series network is used to obtain the dependency relationship of exhaust gas data over time; The graph convolutional network is used to obtain the correlation between different production areas in the drying oven; At least one fully connected layer is used to fuse the correlation between key parameters, the dependence of waste gas data on changes over time, and the correlation between different production areas in the drying oven, so as to obtain the mapping relationship between the key parameters and the waste gas data. The step of using the graph neural network to obtain the correlation between key parameters includes: The graph neural network is used to determine the nodes corresponding to each key parameter and the boundary information between each node. The boundary information is used to characterize the correlation between each key parameter. The step of using the time series network to obtain the dependency relationship of exhaust gas data over time includes: Obtain time-series data related to key parameters and exhaust gas data in the drying oven; Extract key feature data from the time series data; Based on the aforementioned key feature data, the dependence of exhaust gas data on changes over time is determined; The step of using the graph convolutional network to obtain the correlation between different production areas in the drying oven includes: The drying oven is divided into production areas, and a corresponding area map is constructed. Determine the weight information between each production region in the region map, whereby the weight information is used to characterize the correlation strength between the production regions; Based on the weight information between each production area, the spatial feature representation corresponding to the drying oven is determined; The key parameters in determining the process parameters of the drying oven include: Change any process parameter of the drying oven and obtain the corresponding actual exhaust gas data; The actual waste gas data is compared with the waste gas data, and when it is determined that the difference in dust content between the actual waste gas data and the waste gas data is greater than the preset content difference value, the changed process parameter is taken as the key parameter. It also includes: adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted waste gas data; The step of adjusting the process parameters in the drying furnace during lithium carbonate production based on the predicted waste gas data includes: When the predicted exhaust gas data meets the production requirements, the current process parameters in the drying furnace during lithium carbonate production are used as the target process parameters; otherwise, the process parameters in the drying furnace during lithium carbonate production are adjusted until the predicted exhaust gas data meets the production requirements.

Images