Deep learning and physical model guide fused dense medium sorting density decision-making method
Through the integration of deep learning and physical models, combining attention mechanism and composite loss function, the problem that traditional sorting density prediction methods are complex and cannot meet real-time requirements is solved, and more accurate and stable sorting density prediction is achieved, improving selection Stability and efficiency of coal process.
Patent Information
- Application Number
- CN202411497946.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-10-25
AI Technical Summary
Traditional sorting density prediction methods are complex, time-consuming and cannot meet the real-time online prediction requirements of modern coal processing systems. Data-driven machine learning models are insufficient in stability and reliability in unknown or extreme situations, and cannot guarantee the physical rationality of the prediction results. .
Using the remediated sorting density decision-making method that integrates deep learning and physical model guidance, through data acquisition and preprocessing, constructing theoretical mapping models, deep neural network prediction models based on attention mechanisms, and defining compound loss functions, combining empirical loss and Physical loss, optimize model training to improve prediction accuracy and stability.
The prediction accuracy of sorting density is significantly improved, the generalization ability and stability of the model is enhanced, the prediction results are ensured in line with physical laws, the stability and efficiency of the coal preparation process are improved, and the interpretability and transparency of the model are enhanced.
Smart Images

Figure CN119049595A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of computer technology, and specifically relates to a heavy-medium separation density decision method guided by the fusion of deep learning and physical models. Background Art
[0002] In the process of coal sorting and processing, sorting density is a key indicator for evaluating coal sorting. Sorting density directly affects the quality of clean coal and the rational use of resources: too high sorting density may increase the impurity content in clean coal and reduce the economic value of coal; while too low sorting density may cause waste of high-quality coal resources. Therefore, accurate prediction of sorting density is of great significance for efficient coal sorting and processing.
[0003] Traditionally, the prediction methods of sorting density include empirical methods, experimental methods and numerical simulation methods. Although these methods can provide high-precision measurement results, the process is complicated and time-consuming, and cannot meet the needs of real-time online prediction in modern coal processing systems, resulting in measurement lags and affecting rapid decision-making in actual production.
[0004] In recent years, academia and industry have begun to focus on using machine learning technology to improve the accuracy of sorting density prediction. These methods analyze a large amount of historical data on coal preparation processes and use machine learning algorithms to make predictions in order to improve the accuracy of predictions. However, these data-driven machine learning models also have some problems. They rely too much on the quality of the data and cannot ensure the stability and reliability of the models when facing unknown or extreme situations. In addition, these models may not ensure the physical rationality of the prediction results. In fact, the coal preparation process is only one of the multiple factors that affect the ash content of clean coal products, and the quality of raw coal is the key factor in determining the quality of clean coal products. According to coal preparation theory, there is a theoretical connection between the quality of raw coal and sorting density. The present invention is based on this theoretical connection to guide the data-driven model to achieve a more stable and accurate sorting density prediction. Summary of the invention
[0005] In view of the above-mentioned technical problems existing in the prior art, the present invention proposes a heavy medium separation density decision method guided by deep learning and physical model fusion, which has a reasonable design, overcomes the shortcomings of the prior art, and has good results.
[0006] The technical solution of the present invention is as follows: A dense medium separation density decision method guided by deep learning and physical model fusion includes the following steps: Step 1: Data collection and preprocessing, building a sample data set; Step 2: construct a theoretical mapping model and calculate the theoretical sorting density; Step 3: Build a deep neural network prediction model based on the attention mechanism to predict the sorting density; Step 4: Define the composite loss function; Step 5: Divide the samples constructed in step 1 into training set and test set in proportion, and perform model training based on the training set; Step 6: Model evaluation and application.
[0007] Furthermore, the specific process of step 1 is as follows: First, the key production process parameters of the heavy medium separation process and the floating and sinking data of raw coal are comprehensively collected through the production monitoring system of the coal preparation plant; the key production process parameters of the heavy medium separation process specifically include sampling date, sampling time, magnetic content, coal slime content, separation density, clean coal ash content, cyclone inlet pressure, and combined medium barrel liquid level; the floating and sinking data of raw coal specifically include sampling date, sampling time, density grade yield and density grade ash rate; Subsequently, the collected data were standardized and missing values were processed.
[0008] Furthermore, in step 2, according to the raw coal floating and sinking experimental data, a theoretical mapping relationship between the sorting density and the clean coal ash content is constructed; first, a discrete mapping relationship between the two is determined, and then the offline points are smoothed by an interpolation method to establish a continuous value theoretical mapping model between the sorting density and the clean coal ash content; the specific discrete mapping relationship is as follows: (1); in, The density level is The density range of is an integer and ; Sorting density for The upper bound of for The corresponding clean coal ash content; The density level zone number is The yield of The density level zone number is Density grade ash; is the density level area serial number index; is the maximum value function; The continuous value theoretical mapping model is used to calculate the theoretical sorting density value under the set target clean coal ash value. The theoretical sorting density value is one of the inputs of the subsequent deep neural network prediction model.
[0009] Furthermore, in step 3, the deep neural network prediction model based on the attention mechanism includes a one-dimensional convolutional neural network layer, a bidirectional long short-term memory network layer, and an attention mechanism layer; the input of the model includes the data preprocessed in step 1 and the theoretical sorting density value calculated in step 2, and the output of the model is the sorting density; Among them, the input of the one-dimensional convolutional neural network layer is the key production process parameters of coal preparation at different times. By performing convolution operations on the local areas of the key production process parameter data of coal preparation at different times, the corresponding feature maps are output to extract the effective short-term features of the key production process parameters and historical data; the calculation formula of the one-dimensional convolution is: (2); in, Represents the output value after convolution; are different elements in the input time series; , , are the weights of different convolution kernels respectively; The bidirectional long short-term memory network layer consists of two layers of long short-term memory networks in opposite directions, namely the forward layer and the backward layer, which are used to extract long-term dependent features in the time dimension; the bidirectional long short-term memory network layer includes an input layer, a forward layer, a backward layer and an output layer; the bidirectional long short-term memory network layer processes the input data from the forward and reverse directions of the sequence respectively, and generates a forward hidden state sequence and a reverse hidden state sequence respectively; for each time step, the forward and reverse hidden states are concatenated to generate the final hidden state, which is also used as the input of the attention mechanism layer; The forward process is as follows: (3); (4); (5); (6); (7); (8); in, for The state of the candidate unit is always positive; , They are time, The state of the forward storage unit at all times; , , Forward Process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the forward process respectively; Enter the state bias for the forward process; , , They are the weights of the forget gate, input gate, and output gate in the forward process respectively; Enter the state weights for the forward process; is the sigmoid function; is the tanh activation function; , They are Moment and The output value of the positive storage unit at all times; for The input value at the moment; The reverse process is: (9); (10); (11); (12); (13); (14); in, for Reverse candidate unit status at all times; , They are time, The state of the storage unit is reversed at all times; , , Reverse process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the reverse process respectively; Enter the state bias for the reverse process; , , They are the weights of the forget gate, input gate, and output gate in the reverse process respectively; Enter state weights for the reverse process; , They are Moment and The output value of the reverse storage unit at all times; Will and The bidirectional long short-term memory network layer is obtained by splicing Output vector at time ; During the calculation of the attention mechanism layer, The latent representation after dimension transformation is : (15); in, is the weight parameter of the attention mechanism layer; is the bias parameter of the attention mechanism layer; By calculation With context vector To measure the similarity between The importance of During the training process, it is randomly initialized and obtained through learning. Then, the importance of each input is normalized using the softmax function shown below. The specific formula is: (16); in, for The normalized attention value of For the bidirectional long short-term memory network layer Output vector at time instant; After dimension transformation The implicit representation of is an exponential function with base e; is the sum variable index of the moment; Finally, the output of the attention layer The weighted average of all outputs of the bidirectional long short-term memory network layer is obtained: (17); The output of the attention mechanism layer is transformed through the fully connected layer to finally obtain a one-dimensional sorting density.
[0010] Furthermore, in step 4, the composite loss function combines the empirical loss and the physical loss. For the empirical loss part, the mean square error is used as the measurement standard to measure the difference between the predicted value and the actual value of the sorting density. The calculation formula is: (18); in, is the mean square error value; is the total number of samples; Representative The actual value of the sorting density of each sample; It is The predicted value of sorting density for each sample; In the physical loss part, additional constraints based on the physical process of coal preparation are introduced. By perturbing the process parameters and raw coal floating and sinking data in the sample except for the sorting density in a targeted manner and comparing the prediction results before and after the disturbance, the physical loss is established to reflect the physical influence of the disturbed parameters on the sorting density. The physical loss part can evaluate the actual influence of the disturbance on the predicted value of the sorting density. The physical loss function is defined as: (19); in, is the physical loss function; and They are the sorting density values predicted by the deep neural network based on the attention mechanism for the input data before and after the disturbance; is the ReLU function; Finally, the composite loss function of the attention-based deep neural network prediction model is set to the weighted sum of empirical loss and physical loss.
[0011] Furthermore, in step 5, during the training process, the empirical loss and the physical loss are optimized simultaneously; the model hyperparameters are set: the learning rate is 0.001; the optimizer is the Adam algorithm; the batch size is 32; the number of training rounds is 50; and an early stopping mechanism is introduced to monitor the training process. Once the model performance no longer improves, the training is stopped to prevent the model from overfitting.
[0012] Furthermore, in step 6, after the model training is completed, the trained model is evaluated using the test set, and the performance of the model is measured by comparing the difference between the predicted value and the actual value of the sorting density, calculating the mean square error, relative absolute error and relative absolute error key indicators; at the same time, the difference between the predicted value and the actual value is intuitively displayed through a chart; after the evaluation is completed, the optimal model in the evaluation process is selected and applied to actual production to realize the prediction of the sorting density and assist engineers in making decisions.
[0013] The beneficial technical effects brought about by the present invention are as follows.
[0014] 1. The present invention significantly improves the prediction accuracy of sorting density by combining the machine learning model of neural network with the theoretical model based on physical guidance. This method can not only more accurately predict the sorting density under complex working conditions, but also enhance the generalization ability of the model by introducing the prior knowledge of the physical model, so that it can better adapt to different production conditions. Such optimization helps to improve the stability and efficiency of the coal preparation process, especially when data is scarce or working conditions vary greatly, the model can still provide reliable prediction results.
[0015] 2. The present invention constructs a composite loss function by introducing theoretical prediction values and physical losses, so that the prediction results of the model are not only based on data, but also follow physical laws. This method improves the interpretability of the model, making the prediction process more transparent, and making it easier for engineers to understand and trust the prediction basis of the model. In actual operation, this transparency and interpretability are crucial for engineers to make decisions and adjustments, and help improve the intelligence level and operation accuracy of the entire coal preparation process. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is the overall architecture diagram of the heavy medium separation density decision method guided by deep learning and physical model in the present invention.
[0017] Figure 2 This is a structural diagram of the deep neural network prediction model based on the attention mechanism of the present invention.
[0018] Figure 3 It is a schematic diagram of the one-dimensional convolution process in the model of the present invention.
[0019] Figure 4 This is the LSTM network architecture diagram in the model of the present invention.
[0020] Figure 5 This is a diagram of the bidirectional LSTM network architecture in the model of the present invention. DETAILED DESCRIPTION
[0021] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments: A density decision method for heavy-medium separation guided by deep learning and physical model is as follows: Figure 1 As shown, the following steps are included: Step 1: Data collection and preprocessing to build a sample data set; the specific process is as follows: First, the key production process parameters of the heavy medium separation process and the floating and sinking data of raw coal are comprehensively collected through the production monitoring system of the coal preparation plant. The key production process parameters of the heavy medium separation process include sampling date, sampling time, magnetic content, coal slime content, separation density, clean coal ash content, cyclone inlet pressure, and combined medium barrel liquid level. The floating and sinking data of raw coal include sampling date, sampling time, density grade yield, and density grade ash rate.
[0022] Subsequently, the collected data are carefully standardized and missing value processed to ensure the integrity and consistency of the data and provide high-quality input for subsequent neural network models.
[0023] The data in the embodiment of the present invention are collected from a coal preparation plant under Shandong Energy Group, including key production process parameters of heavy medium separation process and floating and sinking characteristics of raw coal. Specifically, as shown in Table 1, the key production process parameters of heavy medium separation process cover multiple dimensions such as sampling date, sampling time, magnetic content, coal slime content, separation density, cyclone inlet pressure and combined medium barrel liquid level, and these data are continuously sampled at intervals of 1 minute.
[0024] Table 1 Example of heavy medium separation process parameters .
[0025] At the same time, as shown in Table 2, the raw coal floating and sinking data also includes information such as sampling date, sampling time, density level, yield and density level ash. In the daily production of the coal preparation plant, the operation is divided into three 8-hour shifts. Each shift usually conducts 2 to 3 raw coal floating and sinking experiments according to production needs. The time of the floating and sinking experiment is uncertain, and the duration of the heavy medium separation process is about 4 minutes. In the present invention, samples are constructed based on the data of each raw coal floating and sinking experiment to form a sample data set. The output of the sample is the sorting density corresponding to the time of clean coal collection, and the input is the raw coal floating and sinking experiment data 4 minutes ago and the process parameters in the past 4 minutes. Through this method, a total of 5290 samples were generated, laying a reliable data foundation for subsequent multi-model training and testing.
[0026] Table 2 Example table of raw coal floating and sinking data .
[0027] Step 2, theoretical sorting density calculation: Using the raw coal floating and sinking experimental data, a theoretical mapping model is constructed to formalize the relationship between sorting density and clean coal ash content. First, the present invention establishes a discrete mapping relationship between sorting density and clean coal ash content, and further smoothes the discrete data points through interpolation technology to establish a continuous value theoretical mapping model. The specific discrete mapping relationship is as follows: (1); in, The density level is The density range of is an integer and ; Sorting density for The upper bound of for The corresponding clean coal ash content; The density level zone number is The yield of The density level zone number is Density grade ash; is the density level area serial number index; is the maximum value function; Taking the raw coal floating and sinking data at 6:20 on June 1 in Table 2 as an example, when the theoretical sorting density is 1.4, the calculation formula for clean coal ash is (50.17*4.83+15.57*7.01) / (50.17+15.57). By calculating the clean coal ash under different theoretical sorting densities and interpolating the results, a continuous value theoretical mapping model between sorting density and clean coal ash can be established. Using the continuous value theoretical mapping model (referred to as the theoretical model), the theoretical sorting density value under the set target clean coal ash value can be calculated. This continuous value theoretical sorting density value (i.e., theoretical prediction value) is not only one of the inputs of the subsequent deep neural network prediction model, but also crucial to improving the prediction accuracy of the model.
[0028] Step 3: Construct a deep neural network prediction model based on the attention mechanism to predict the final sorting density. The present invention designs a deep neural network prediction model based on the attention mechanism (referred to as the neural network model), which is a deep learning model that integrates the convolutional neural network CNN and the bidirectional long short-term memory network BiLSTM. The model architecture diagram is shown in the figure below. Figure 2 As shown in the figure, it includes a one-dimensional convolutional neural network layer (1D CNN layer), a bidirectional long short-term memory network layer (BiLSTM layer), and an attention mechanism layer. The input of the model includes the preprocessed data in step 1 and the theoretical sorting density value calculated in step 2. The output of the model is the final sorting density.
[0029] The input of the convolutional neural network layer is the key production process parameters of coal preparation at different times; Convolutional Neural Network Layers: Two-dimensional convolution is mainly used for image feature extraction, while one-dimensional convolution is mainly used for time series feature extraction. Figure 3 A schematic diagram of one-dimensional convolution is shown, and its calculation formula is: (2); in, Represents the output value after convolution, which participates in the subsequent calculation process as the hidden feature of the convolved area in the time series; are different elements in the input time series; , , are the different weights of the convolution kernel. For other elements in the time series containing different timestamps, define and For other elements at different timestamps in the time series, are the corresponding output values after convolution of these elements, and the calculation method is the same as formula (2).
[0030] The convolutional neural network layer extracts short-term features in the feature dimension (production process parameters) by performing convolution operations on local areas of key production process parameters of coal preparation at different times.
[0031] Bidirectional LSTM layer: LSTM and bidirectional LSTM are specific types of RNN networks used to model sequence data. The structure of LSTM is as follows Figure 4 As shown, it consists of a forget gate, an input gate, and an output gate.
[0032] The forward propagation process equation is as follows: (3); (4); (5); (6); (7); (8); in, for The state of the candidate unit is always positive; , They are time, The state of the forward storage unit at all times; , , Forward Process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the forward process respectively; Enter the state bias for the forward process; , , They are the weights of the forget gate, input gate, and output gate in the forward process respectively; Enter the state weights for the forward process; is the sigmoid function; is the tanh activation function; , They are Moment and The output value of the positive storage unit at the moment, After concatenation, it is used as the input of the attention mechanism layer; for The input value at the moment; The bidirectional LSTM used in the present invention is composed of two layers of LSTM in opposite directions, namely the forward layer and the backward layer, which are used to extract long-term dependent features in the time dimension. The structure of the bidirectional LSTM is as follows: Figure 5 As shown in the figure, it includes input layer, forward layer, backward layer and output layer. The bidirectional LSTM extracts the features of the input time series from the forward and reverse directions, and generates the forward hidden state sequence and the reverse hidden state sequence respectively. For each time step, the forward and reverse hidden states are concatenated to form the final output of the bidirectional LSTM layer, which also serves as the input of the attention mechanism layer. Among them, represents the input sequence, represents the hidden state of the forward layer, represents the hidden state of the reverse layer, Represents the final output of the bidirectional LSTM layer.
[0033] The reverse process is: (9); (10); (11); (12); (13); (14); in, for Reverse candidate unit status at all times; , They are time, The state of the storage unit is reversed at all times; , , Reverse process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the reverse process respectively; Enter the state bias for the reverse process; , , They are the weights of the forget gate, input gate, and output gate in the reverse process respectively; Enter state weights for the reverse process; , They are Moment and The output value of the reverse storage unit at all times; Will and The bidirectional long short-term memory network layer is obtained by splicing Output vector at time ; Attention mechanism layer: In order to further enhance the feature representation and enable the model to dynamically focus on the key features in the input sequence, the present invention introduces an attention mechanism to assign different weights to the output of the bidirectional LSTM to amplify the key production process parameters related to the sorting density prediction.
[0034] During the calculation of the attention mechanism layer, The latent representation after dimension transformation is : (15); in, is the weight parameter of the attention mechanism layer; is the bias parameter of the attention mechanism layer; By calculation With context vector To measure the similarity between The importance of During the training process, it is randomly initialized and obtained through learning. Then, the importance of each input is normalized using the softmax function shown below. The specific formula is: (16); in, for The normalized attention value of For the bidirectional long short-term memory network layer Output vector at time instant; After dimension transformation The implicit representation of is an exponential function with base e; is the sum variable index of the moment; Finally, the output of the attention layer The weighted average of all outputs of the bidirectional long short-term memory network layer is obtained: (17); The output of the attention mechanism layer is transformed through the fully connected layer to finally obtain a one-dimensional sorting density.
[0035] Step 4, define a composite loss function; in order to improve the prediction accuracy of the model, the present invention designs a composite loss function, which cleverly integrates two key components: empirical loss and physical loss.
[0036] For the empirical loss part, the mean square error (MSE) was selected as the measurement standard to measure the difference between the predicted value of sorting density and the actual value of sorting density. The calculation formula is: (18); in, is the mean square error value; is the total number of samples; Representative The actual value of the sorting density of each sample; It is The predicted value of sorting density for each sample.
[0037] In the physical loss part, additional constraints based on the physical process of coal preparation are innovatively introduced. Specifically, by making targeted disturbances to the process parameters and raw coal floating and sinking data in the sample except for the sorting density, and comparing the prediction results before and after the disturbance, physical losses are established to reflect the physical impact of the disturbed parameters on the sorting density; the physical loss part can evaluate the actual impact of these disturbances on the predicted value of the sorting density. The physical loss function is defined as: (19); in, is the physical loss function; and They are the sorting density values predicted by the deep neural network based on the attention mechanism for the input data before and after the disturbance; It is the ReLU function. The ReLU function ensures that the loss is calculated only when there is a deviation between the predicted value and the actual value, so that the model focuses more on reducing the prediction error.
[0038] Finally, the composite loss function (also called final loss) of the deep neural network prediction model based on the attention mechanism is set to the weighted sum of the empirical loss and the physical loss, which not only ensures that the model has a good fit to the historical data in a statistical sense, but also ensures that the model prediction results conform to the actual laws of the physical process. Through this innovative composite loss function design, the model of the present invention can achieve higher accuracy and robustness when predicting the sorting density in the coal preparation process.
[0039] Step 5: Divide the samples constructed in step 1 into training sets and test sets in proportion, and perform model training based on the training set. During the training process, the loss function optimizes both the empirical loss and the physical loss to ensure that the model's predictions are both accurate and in line with physical laws, thereby improving the model's generalization ability and prediction accuracy.
[0040] In an embodiment of the present invention, the deep neural network prediction model based on the attention mechanism is trained using the historical big data of coal preparation production. Specifically, the samples constructed in step 1 are divided into a training set and a test set in a ratio of 8:2. During the training process, a composite loss function is specially designed, which combines the empirical loss and the physical loss. The purpose of this is to allow the model to follow the basic laws of the physical process while performing data-driven predictions, and it is a physical model. In addition, the hyperparameters of the model are carefully adjusted to further optimize the performance: the learning rate is 0.001; the optimizer is the Adam algorithm; the batch size is 32; and the number of training rounds is 50. At the same time, an early stopping mechanism is introduced to monitor the training process. Once the model performance is no longer improved, the training is stopped to prevent the model from overfitting.
[0041] Step 6: Model evaluation and application: After the model training is completed, the trained model is evaluated with the test set. By comparing the difference between the predicted value and the actual value of the sorting density, key indicators such as mean square error (MSE), relative absolute error (RSE) and relative absolute error (RAE) are calculated to measure the performance of the model. At the same time, the difference between the predicted and actual values is intuitively displayed through charts to quickly evaluate the model effect. After the evaluation is completed, the optimal model in the evaluation process is selected and applied to actual production to achieve accurate prediction of the sorting density, thereby improving the stability and efficiency of the coal preparation process, optimizing the production process, and assisting engineers in making decisions.
[0042] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.
Claims
1. A dense medium separation density decision method guided by deep learning and physical model, characterized in that: The steps include: Step 1: Data collection and preprocessing, building a sample data set; Step 2: construct a theoretical mapping model and calculate the theoretical sorting density; Step 3: Build a deep neural network prediction model based on attention mechanism to predict sorting density; Step 4: Define the composite loss function; Step 5: Divide the samples constructed in step 1 into training set and test set in proportion, and perform model training based on the training set; Step 6: Model evaluation and application.
2. The dense medium separation density decision method guided by deep learning and physical model according to claim 1 is characterized in that: The specific process of step 1 is as follows: First, the key production process parameters of the heavy medium separation process and the floating and sinking data of raw coal are comprehensively collected through the production monitoring system of the coal preparation plant; the key production process parameters of the heavy medium separation process specifically include sampling date, sampling time, magnetic content, coal slime content, separation density, clean coal ash content, cyclone inlet pressure, and combined medium barrel liquid level; the floating and sinking data of raw coal specifically include sampling date, sampling time, density grade yield and density grade ash rate; Subsequently, the collected data were standardized and missing values were processed.
3. The dense-medium separation density decision method guided by deep learning and physical model according to claim 1 is characterized in that: In step 2, a theoretical mapping relationship between sorting density and clean coal ash content is constructed based on the raw coal floating and sinking experimental data; first, a discrete mapping relationship between the two is determined, and then the offline points are smoothed by an interpolation method to establish a continuous value theoretical mapping model between sorting density and clean coal ash content; the specific discrete mapping relationship is as follows: (1); in, The density level is The density range of is an integer and ; Sorting density for The upper bound of for The corresponding clean coal ash content; The density level zone number is The yield of The density level zone number is Density grade ash; is the density level area serial number index; is the maximum value function; The continuous value theoretical mapping model is used to calculate the theoretical sorting density value under the set target clean coal ash value. The theoretical sorting density value is one of the inputs of the subsequent deep neural network prediction model.
4. The dense-medium separation density decision method guided by deep learning and physical model according to claim 1 is characterized in that: In the step 3, the deep neural network prediction model based on the attention mechanism includes a one-dimensional convolutional neural network layer, a bidirectional long short-term memory network layer, and an attention mechanism layer; the input of the model includes the data preprocessed in step 1 and the theoretical sorting density value calculated in step 2, and the output of the model is the sorting density; Among them, the input of the one-dimensional convolutional neural network layer is the key production process parameters of coal preparation at different times. By performing convolution operations on the local areas of the key production process parameter data of coal preparation at different times, the corresponding feature maps are output to extract the effective short-term features of the key production process parameters and historical data; the calculation formula of the one-dimensional convolution is: (2); in, Represents the output value after convolution; are different elements in the input time series; , , are the weights of different convolution kernels respectively; The bidirectional long short-term memory network layer consists of two layers of long short-term memory networks in opposite directions, namely the forward layer and the backward layer, which are used to extract long-term dependent features in the time dimension; the bidirectional long short-term memory network layer includes an input layer, a forward layer, a backward layer and an output layer; the bidirectional long short-term memory network layer processes the input data from the forward and reverse directions of the sequence respectively, and generates a forward hidden state sequence and a reverse hidden state sequence respectively; for each time step, the forward and reverse hidden states are concatenated to generate the final hidden state, which is also used as the input of the attention mechanism layer; The forward process is as follows: (3); (4); (5); (6); (7); (8); in, for The state of the candidate unit is always positive; , They are time, The state of the forward storage unit at all times; , , Forward Process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the forward process respectively; Enter the state bias for the forward process; , , They are the weights of the forget gate, input gate, and output gate in the forward process respectively; Enter the state weights for the forward process; is the sigmoid function; is the tanh activation function; , They are Moment and The output value of the positive storage unit at all times; for The input value at the moment; The reverse process is: (9); (10); (11); (12); (13); (14); in, for Reverse candidate unit status at all times; , They are time, The state of the storage unit is reversed at all times; , , Reverse process The forget gate, input gate, and output gate at each moment; , , They are the biases corresponding to the forget gate, input gate, and output gate in the reverse process respectively; Enter the state bias for the reverse process; , , They are the weights of the forget gate, input gate, and output gate in the reverse process respectively; Enter state weights for the reverse process; , They are Moment and The output value of the reverse storage unit at all times; Will and The bidirectional long short-term memory network layer is obtained by splicing Output vector at time ; During the calculation of the attention mechanism layer, The latent representation after dimension transformation is : (15); in, is the weight parameter of the attention mechanism layer; is the bias parameter of the attention mechanism layer; By calculation With context vector To measure the similarity between The importance of During the training process, it is randomly initialized and obtained through learning. Then, the importance of each input is normalized using the softmax function shown below. The specific formula is: (16); in, for The normalized attention value of For the bidirectional long short-term memory network layer Output vector at time instant; After dimension transformation The implicit representation of is an exponential function with base e; is the sum variable index of the moment; Finally, the output of the attention layer The weighted average of all outputs of the bidirectional long short-term memory network layer is obtained: (17); The output of the attention mechanism layer is transformed through the fully connected layer to finally obtain a one-dimensional sorting density.
5. The dense medium separation density decision method guided by deep learning and physical model according to claim 1 is characterized in that: In step 4, the composite loss function combines the empirical loss and the physical loss. For the empirical loss part, the mean square error is used as the measurement standard to measure the difference between the predicted value and the actual value of the sorting density. The calculation formula is: (18); in, is the mean square error value; is the total number of samples; Representative The actual value of the sorting density of each sample; It is The predicted value of sorting density for each sample; In the physical loss part, additional constraints based on the physical process of coal preparation are introduced. By perturbing the process parameters and raw coal floating and sinking data in the sample except for the sorting density in a targeted manner and comparing the prediction results before and after the disturbance, the physical loss is established to reflect the physical influence of the disturbed parameters on the sorting density. The physical loss part can evaluate the actual influence of the disturbance on the predicted value of the sorting density. The physical loss function is defined as: (19); in, is the physical loss function; and They are the sorting density values predicted by the deep neural network based on the attention mechanism for the input data before and after the disturbance; is the ReLU function; Finally, the composite loss function of the attention-based deep neural network prediction model is set to the weighted sum of empirical loss and physical loss.
6. The dense-medium separation density decision method guided by deep learning and physical model fusion according to claim 1 is characterized in that: In step 5, during the training process, the empirical loss and the physical loss are optimized simultaneously; the model hyperparameters are set: the learning rate is 0.001; the optimizer is the Adam algorithm; the batch size is 32; the number of training rounds is 50; and an early stopping mechanism is introduced to monitor the training process. Once the model performance no longer improves, the training is stopped to prevent the model from overfitting.
7. The dense medium separation density decision method guided by deep learning and physical model according to claim 1 is characterized in that: In step 6, after the model training is completed, the trained model is evaluated using the test set, and the performance of the model is measured by comparing the difference between the predicted value and the actual value of the sorting density, calculating the mean square error, relative absolute error and relative absolute error key indicators; at the same time, the difference between the predicted value and the actual value is intuitively displayed through a chart; after the evaluation is completed, the optimal model in the evaluation process is selected and applied to actual production to realize the prediction of the sorting density and assist engineers in making decisions.
Citation Information
Patent Citations
Intelligent control system and method for dense medium separation process
CN110605178A
Blast furnace thermal load abnormal state monitoring method based on attention mechanism
CN114015825A
Time projection-based method for predicting control variables in dense medium coal preparation process
CN114074022A
Multi-step prediction method for clean coal quality in dense medium coal separation process based on LSTM (Long Short Term Memory)
CN114330930A
Process industry product quality short-term prediction method based on deep learning
CN116011672A
Cited By
Dam deformation prediction method based on embedded feature selection
CN119538126A
Edge gateway-based electricity room environment regulation and control method and device
CN119781337A
Coal dressing full-process monitoring decision-making method and system based on Internet of Things sensing
CN120046873A
Coal dressing decision-making method and system based on multi-source data integration
CN120180050A
Coal preparation decision-making method and system based on multi-source data integration
CN120180050B
