An industrial data correction method and apparatus based on an improved LSTM neural network

Abstract

This invention discloses an industrial data correction method and apparatus based on an improved LSTM neural network, relating to the field of industrial process variable data correction technology. The invention uses an improved robust estimation function as the loss function of the LSTM neural network model to suppress the contamination of other normal data by measurement data containing significant errors. Simultaneously, it updates the kernel width parameter in the loss function and uses the obtained industrial evaporation time-series data as input to the LSTM neural network model, performing forward and backward propagation to update the network weights. This enables the improved LSTM neural network model to learn the trends of industrial evaporation time-series data, reducing the impact of random and significant errors on the measurement data, thereby accurately acquiring measurement data and correcting errors in industrial data.

Description

Technical Field

[0001] This invention relates to the field of industrial process variable data correction technology, and in particular to an industrial data correction method and apparatus based on an improved LSTM neural network. Background Technology

[0002] Alumina is a very important raw material in industrial production. Due to its high melting point and high hardness, it is widely used in traditional refractory materials, ceramics, metallurgy, chemicals and other fields. The evaporation process in alumina production is a key process used to extract and concentrate alumina from aluminum-containing solutions, and the quality of the evaporation products is the core concern of the alumina production evaporation process. During the evaporation process, excess water in the feed solution needs to be evaporated to obtain a high-quality sodium aluminate solution. Various measuring instruments are inevitably affected by the high temperature and strong alkaline environment on site, which causes random errors in the measurement data.

[0003] Furthermore, due to interference such as improper operation by on-site workers and instrument leakage, measurement data may contain significant errors. Limited by measurement costs and environmental constraints, and the inherent risks of measuring intermediate process parameters, the evaporation process only monitors the flow rate and concentration of the feed liquid at the inlet and outlet, leaving most intermediate parameters as unknown information. Inaccurate and incomplete production information affects the accuracy of industrial process modeling and optimized control. Therefore, in the alumina production evaporation process, correcting for random and significant errors, estimating unknown parameter data, and improving the accuracy and effectiveness of industrial data are crucial for stable industrial operation, efficient production, and green energy conservation.

[0004] Currently, the correction of errors in industrial data usually involves using simulation software and production parameters to predict the obtained flow and concentration information. However, since industrial data errors typically include both random and significant errors, the uncertainty in prediction is usually high, resulting in inaccurate measurement data and making it difficult to correct industrial data errors. Summary of the Invention

[0005] This invention provides an industrial data correction method and apparatus based on an improved LSTM neural network, which can solve the problem in the prior art that the measurement data is inaccurate and it is difficult to correct the errors in industrial data.

[0006] This invention provides an industrial data correction method based on an improved LSTM neural network, comprising the following steps:

[0007] Acquire industrial evaporation time-series data during industrial production processes;

[0008] An improved robust estimation function based on random and significant errors in industrial evaporation time series data is constructed.

[0009] The improved robust estimation function is used as the loss function of the LSTM neural network to construct the LSTM neural network model;

[0010] Industrial evaporation time series data is used as input to the LSTM neural network model for forward propagation. Based on the result of forward propagation and the deviation of industrial evaporation time series data containing random and significant errors, the loss function value of the current LSTM neural network model is obtained. Backpropagation is then performed based on the loss function value to iteratively update the network weights of the LSTM neural network model, so as to obtain an improved LSTM neural network model.

[0011] The improved LSTM neural network model is used to obtain the correction value of industrial evaporation time series data.

[0012] Preferably, the improved robust estimation function is:

[0013]

[0014] Where: e represents the weighted residual between the contamination value and the correction value; δ represents the kernel width of the loss function;

[0015]

[0016] Where: x m This indicates a contamination value containing gross errors; x r ε represents the correction value; ε represents the standard deviation of the variable.

[0017] Preferably, after the improved robust estimation function is obtained, the kernel width parameter within the improved robust estimation function is adjusted, including:

[0018] Improve the robust estimation function exp(-e) 4 / (2δ 2 )) is represented as exp(-F), and as F increases, the value of exp(-F) continuously decreases;

[0019] Set the F value to a threshold F thre At the start of backpropagation, the F value is set to F. ini And F ini <F thre As the number of training iterations continues to increase, the F-value changes from F0 to F1 ini The value continues to increase to F thre The value is:

[0020] F = e 4 / (2δ 2 ) = F ini +η*epoch

[0021] δ 2 =e 4 / 2(F ini +η*epoch)

[0022] Where: e represents the weighted residual between the contamination value and the correction value; δ represents the kernel width of the loss function; η represents the iteration step size; epoch represents the number of iterations;

[0023] The kernel width parameter δ is updated using the η*epoch factor.

[0024] Preferably, determining the forward propagation of the LSTM neural network model includes:

[0025] Based on the input and output dimensions of industrial evaporation time series data in the LSTM neural network model, the input dimension, hidden layer dimension, output dimension, LSTM layers, number of LSTM layers, and linear layers of the LSTM neural network model are defined.

[0026] Based on the correction target of industrial evaporation time series data in the LSTM neural network model, the forward propagation mode of the LSTM neural network model is defined as follows: the input industrial evaporation time series data is processed through the LSTM layer and then processed through the linear layer to obtain the correction value.

[0027] Preferably, the iterative update of the LSTM neural network model weights includes:

[0028] After obtaining the loss function value of the LSTM neural network model, the loss function value is used as the input of the LSTM neural network model for backpropagation. The improved robust estimation function iteratively updates the network weights of the LSTM neural network model, and the weight parameters are updated using the gradient descent method. The update method is as follows:

[0029]

[0030] Among them: W ih and b ih W represents the weight and bias parameters between the input state and the hidden state, respectively; hh and b hh α and Ln represent the weight and bias parameters between hidden states, respectively; α represents the fixed learning rate, and α>0; Ln represents the loss function.

[0031] Preferably, the correction value for acquiring the industrial evaporation time series data includes:

[0032] An improved LSTM neural network model is used to obtain the correction adjustment value of industrial evaporation time series data. Based on this correction adjustment value, the industrial evaporation time series data is inversely normalized to obtain the corrected value. The inverse normalization method is as follows:

[0033] x r =x nr *(x maxm -x minm )+x minm

[0034] Where: x r This represents the final corrected value; x nr This represents the normalized correction value; This represents the maximum value containing a significant error; This represents the minimum value containing a significant error.

[0035] This invention also provides an industrial data correction device based on an improved LSTM neural network, comprising:

[0036] The data module is used to acquire industrial evaporation time-series data during industrial production processes;

[0037] An improved robust estimation function based on random and significant errors in industrial evaporation time series data is constructed.

[0038] The model module is used to construct an LSTM neural network model by using the improved robust estimation function as the loss function of the LSTM neural network.

[0039] Industrial evaporation time series data is used as input to the LSTM neural network model for forward propagation. Based on the result of forward propagation and the deviation of industrial evaporation time series data containing random and significant errors, the loss function value of the current LSTM neural network model is obtained. Backpropagation is then performed based on the loss function value to iteratively update the network weights of the LSTM neural network model, so as to obtain an improved LSTM neural network model.

[0040] The correction module is used to obtain correction values ​​for industrial evaporation time series data using an improved LSTM neural network model.

[0041] This invention provides an industrial data correction method and apparatus based on an improved LSTM neural network. Compared with the prior art, its advantages are as follows:

[0042] This invention uses an improved robust estimation function as the loss function of an LSTM neural network model. The loss function is used to determine the forward propagation process of the constructed LSTM neural network model, suppressing the influence of significant and random errors in the industrial evaporation time-series data on normal data during processing. Then, the industrial evaporation time-series data is used as input to the LSTM neural network model, and forward and backward propagation are performed to update the network weights. This allows the improved LSTM neural network model to learn the trends in the industrial evaporation time-series data, reducing the impact of random and significant errors on the measurement data, thereby accurately acquiring measurement data and correcting errors in the industrial data. Attached Figure Description

[0043] Figure 1 A schematic diagram of the overall process of an industrial data correction method based on an improved LSTM neural network provided in an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of a robust correction structure for industrial evaporation data based on an improved LSTM network, provided as an embodiment of the present invention, for an industrial data correction method based on an improved LSTM neural network.

[0045] Figure 3 A schematic diagram comparing industrial variable measurement values ​​and data correction values ​​for an industrial data correction method based on an improved LSTM neural network provided in an embodiment of the present invention;

[0046] Figure 4 A schematic diagram comparing the influence functions of the improved robust estimation function, the Welsh function, and the Correntropy robust estimation function of an industrial data correction method based on an improved LSTM neural network provided in an embodiment of the present invention.

[0047] Figure 5 A schematic diagram comparing the standard deviations of different variables calculated by an improved robust data correction method based on an improved LSTM neural network and a correction method based on Welsh and Correntropy robust estimation functions, provided in an embodiment of the present invention.

[0048] Figure 6 This diagram illustrates a comparison between the improved robust data correction method based on an improved LSTM neural network and the data standard deviations calculated by LSTM, NN, and NRE-NN methods, as provided in this embodiment of the invention. Detailed Implementation

[0049] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0050] See Figure 1 This invention provides an industrial data correction method based on an improved LSTM neural network, comprising the following steps:

[0051] Step 1: To avoid significant errors contaminating measurement data containing only random errors, a data correction model based on the Novel Robust Estimator (NRE) is constructed to suppress the effects of significant error propagation.

[0052] Step 2: In order to improve the solution accuracy of the data correction model, the improved robust estimation function is used as the loss function of the LSTM neural network, so as to effectively combine data correction and deep learning.

[0053] Step 3: Train and tune the LSTM neural network based on the improved robust estimation function to learn the trend of evaporation data and improve the robustness of the measurement data to significant errors.

[0054] Step 4: Validate the correction results of the LSTM neural network based on the improved robust estimation function using evaluation metrics to eliminate significant errors, reduce random errors, and improve the validity and reliability of the measurement data.

[0055] Specifically:

[0056] Step 1: First, determine the mathematical expression of the improved robust estimation function, as shown in the formula below, and calculate its partial derivative with respect to the bias. The influence function is obtained, as shown in the formula below. Its convergence is analyzed to effectively eliminate significant errors in evaporation measurement data and improve the robustness of the measurement data. The formulas for each part are as follows:

[0057]

[0058] Where: x m This indicates a contamination value containing gross errors; x r ε represents the correction value; e represents the standard deviation of the variable; e represents the weighted residual between the contamination value and the correction value; δ represents the kernel width of the loss function.

[0059] Step 2: Using the improved robust estimation function as the loss function of the LSTM neural network, it can automatically learn the trend of the data to process industrial evaporation time series data in industrial production; the LSTM network includes an input gate I for reading data into the unit. t Output gate O from the unit output entry t Forget Gate F for resetting unit content t Memory cell C used to record additional information t and candidate memory and hidden state H t The calculation formula is:

[0060] I t =σ(X) t W xi +H t-1 W hi +b i )

[0061] F t =σ(X) t W xf +H t-1 W hf +b f )

[0062] O t =σ(X) t W xo +H t+1 W ho +b o )

[0063]

[0064] H t =O t ☉tanh(C t )

[0065] Where: σ represents a fully connected layer with an activation function, X t H represents the input data at the current moment. t-1 W represents the hidden state of the previous time step. xi 、W hi 、W xf 、W hf 、W xo 、W ho 、W xc 、W hc These represent the weights of the current input, the current hidden state, the relationship between the input gate and the forget gate, the relationship between the hidden state and the forget gate, the relationship between the input and the output, the relationship between the hidden state and the output, the relationship between the input and the memory cell, and the relationship between the hidden state and the memory cell, respectively. i 、b f 、bo 、b c represents the biases of the input gate, forget gate, output gate, and memory cell, respectively, and 'e' represents element-wise multiplication. The input gate, forget gate, output gate, and candidate memory cells are all obtained by the combined action of the current input and the hidden state of the previous time step.

[0066] Because of I t With F t After passing through a fully connected layer with an activation function, the value is between 0 and 1. When the content of the cell from the previous time step is discarded, F... t If I is set to 0, and the content of the candidate memory cell at the current moment needs to be retained, then... t The value is set to 1 to determine the current content of the memory cell. However, the content of the memory cell does not necessarily need to be passed to the hidden state, so the output gate controls whether any content needs to be passed to the hidden state.

[0067] Step 3: Based on the established LSTM neural network with an improved robust estimation function, the training and parameter tuning of the LSTM network model includes the following steps:

[0068] S1: Initialize the parameters of the improved robust estimation function and set the dynamic adjustment rules for the kernel width; the kernel width of the loss function is an adjustable parameter, and the specific parameter tuning process is as follows:

[0069] exp(-e 4 / (2δ 2 The expression exp(-F) is used to express that as F increases, the value of exp(-F) initially decreases rapidly, and then the decreasing trend slows down. Once the decreasing trend slows down, the effect of increasing the value of F on the function also decreases. Let the value of F at this point be a threshold F. thre When starting backpropagation, first set the F value to F. ini F ini <F thre This results in a large gradient for the model during training; as the number of training iterations increases, the gradient gradually decreases, and the suppression of significant errors gradually increases, while the F-value also gradually increases until it reaches F0. thre At this point, as the F-value continues to increase, the suppression effect on significant errors no longer changes significantly. thre Different values ​​are taken depending on the confidence level; from the above principle, F = e 4 / (2δ 2 ) = F ini +η*epoch, therefore δ is derived 2 =e 4 / 2(F ini +η*epoch), the kernel width δ is adjusted by the η*epoch factor, where η is the iteration step size and epoch is the number of iterations.

[0070] S2: Based on the data input and output dimensions and the goal of data correction robustness, define the input dimension, hidden layer dimension, output dimension, and number of LSTM layers of the LSTM network; define the LSTM layers and linear layers and initialize the weights and biases; define the network's forward propagation method, that is, the forward propagation process that processes the input data through the LSTM layers, then through the linear layers, and finally obtains the corrected value, such as... Figure 2 As shown.

[0071] Based on the actual situation in industrial production processes, the LSTM network has 35 input and output dimensions, 100 hidden layers, and 5 LSTM layers. The network selects the Adam optimizer and sets the learning rate to 0.001. During training, the learning rate scheduler is used to dynamically adjust the learning rate to improve training performance. The network uses the Xavier initialization method to initialize the weights and biases. This initialization method helps the model converge better and helps avoid gradient explosion or gradient vanishing.

[0072] S3: Normalize the industrial evaporation data and use the normalized data as input to the LSTM network model for forward propagation; calculate the current network-coordinated loss function value based on the deviation between the forward propagation output and the data containing random and significant errors; perform backpropagation based on the loss function value of the improved robust estimation function, and iterate the weight updates of the LSTM network according to the robust loss function; update the network weights and optimize the parameters using gradient descent, the formula of which is:

[0073]

[0074] Among them: W ih and b ih These are the weight parameters and bias parameters between the input and the hidden state, respectively; W hh and b hh These are the weight parameters and bias parameters between hidden states, respectively; α represents a fixed learning rate, and α>0; Ln represents the loss function; the gradient space gradually decreases by subtracting the weighted partial derivative of the loss function with respect to the current parameter, until the optimal solution is found.

[0075] S4: Obtain the average loss value during the epoch iteration, and use the network with updated weights to proceed to step S3 until the iteration count is complete.

[0076] S5: Using the improved LSTM neural network model after training, obtain the standardized evaporation data correction value. Then, inversely normalize the obtained normalized data using the following formula to obtain the data correction value; the inverse normalization formula is:

[0077]

[0078] Where: x r This represents the final corrected value; x nr This represents the normalized correction value; This represents the maximum value containing a significant error; This represents the minimum value containing a significant error.

[0079] Step 4: For the constructed LSTM network based on the improved robust estimation function, compare the standard deviation of the measured values ​​and the corrected values ​​of the evaporation data in the alumina production evaporation process to measure the data correction effect and verify whether the correction results have detected and eliminated significant errors.

[0080] The specific experiment is as follows:

[0081] The alumina production evaporation process involves 35 variables, with 800 experimental samples for each variable. To analyze the robustness of data correction based on an improved robust estimation function embedded with an LSTM neural network, 200 sets of measured values ​​and corrected values ​​were compared as follows: Figure 3 As shown, from Figure 3 It can be clearly observed that all significant errors in the measured values ​​containing significant errors were successfully identified and effectively corrected after network training. The overall trend of the evaporation data is more stable after correction, and the interference of significant errors is also well suppressed. As shown in Table 1, 1 to 35 represent the raw liquid flow rate, feed liquid temperature, steam temperature, and condensate temperature in each effect evaporator and each stage of flash evaporator, respectively. From the comparison of the standard deviation between the measured values ​​and the corrected values, it can be seen that the standard deviation calculated by the corrected values ​​is significantly lower than that calculated by the measured values, with the lowest reduction being 0.41 (20%) and the highest reduction being 3.28 (87%).

[0082] Table 1. Comparison of standard deviations between measured and corrected data for the alumina production evaporation process.

[0083] variable 1 2 3 4 5 6 7 8 9 10 11 12 Measured values 3.77 2.05 3.74 1.65 1.37 1.65 1.64 1.69 1.68 1.81 1.77 1.89 Correction value 0.49 1.64 1.54 1.04 0.56 0.19 0.41 0.25 0.55 0.21 0.63 0.21 variable 13 14 15 16 17 18 19 20 21 22 23 24 Measured values 1.89 2.41 1.57 1.76 1.63 1.58 1.62 1.78 1.85 1.62 1.72 1.87 Correction value 0.42 0.77 0.11 0.04 0.30 0.41 0.63 0.60 0.52 0.35 0.13 0.37 variable 25 26 27 28 29 30 31 32 33 34 35 Measured values 1.5 1.76 1.68 1.69 1.71 1.76 1.70 1.55 1.69 1.88 1.64 Correction value 0.04 0.43 0.33 0.52 0.78 0.29 0.13 0.36 0.37 0.34 0.035

[0084] Based on the LSTM network, the loss function is replaced by the Welsh and Correntropy robust estimation functions, respectively, as follows: Figure 4 As shown, a comparative experiment was conducted, selecting five variables: variable 1, variable 22, variable 23, variable 24, and variable 25. The standard deviation of the corrected data is as follows: Figure 5 As shown, variable 1 contains significant error, while the other four variables contain only random error; Figure 5 It can be seen that the LSTM network based on the improved robust estimation function has good robustness to data containing significant errors, and can eliminate significant errors and reduce the impact of random errors.

[0085] Data correction was performed on the alumina production evaporation process data using NRE-LSTM, LSTM, NN, and NRE-NN, respectively, and the standard deviation comparison chart is shown below. Figure 6 As shown, the standard deviation of the correction results based on NRE-LSTM is smaller than that of the three comparison methods. The NN network using the improved robust estimation function has a better effect on suppressing random errors than the NN network without the improved robust estimation function, while the LSTM network obviously has a better effect on suppressing errors than the NN network.

[0086] This invention also provides an industrial data correction device based on an improved LSTM neural network, comprising:

[0087] The data module is used to acquire industrial evaporation time-series data during industrial production processes.

[0088] An improved robust estimation function based on random and significant errors in industrial evaporation time series data is constructed.

[0089] The model module is used to construct an LSTM neural network model by using the improved robust estimation function as the loss function of the LSTM neural network.

[0090] Industrial evaporation time series data is used as input to the LSTM neural network model for forward propagation. Based on the result of forward propagation and the deviation of the industrial evaporation time series data containing random and significant errors, the loss function value of the current LSTM neural network model is obtained. Backpropagation is then performed based on the loss function value to iteratively update the network weights of the LSTM neural network model, thereby obtaining an improved LSTM neural network model.

[0091] The correction module is used to obtain correction values ​​for industrial evaporation time series data using an improved LSTM neural network model.

[0092] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. An industrial data correction method based on an improved LSTM neural network, characterized in that, Includes the following steps: Acquire industrial evaporation time-series data during industrial production processes; An improved robust estimation function based on random and significant errors in industrial evaporation time series data is constructed. The improved robust estimation function is used as the loss function of the LSTM neural network to construct the LSTM neural network model; Industrial evaporation time series data is used as input to the LSTM neural network model for forward propagation. Based on the result of forward propagation and the deviation of industrial evaporation time series data containing random and significant errors, the loss function value of the current LSTM neural network model is obtained. Backpropagation is then performed based on the loss function value to iteratively update the network weights of the LSTM neural network model, so as to obtain an improved LSTM neural network model. The improved LSTM neural network model was used to obtain the correction values ​​of industrial evaporation time series data; The improved robust estimation function is: in: This represents the weighted residual between the pollution value and the correction value; This represents the kernel width of the loss function; in: This indicates a contamination value containing gross errors; Indicates the correction value; The standard deviation of a variable; After the improved robust estimation function is obtained, the kernel width parameter within the improved robust estimation function is adjusted, including: Improve the robust estimation function Represented as ,along with Increase, The value continues to decrease; Will The value is set to a threshold. At the start of backpropagation, Value set to ,and As the number of training sessions continues to increase, Value by The value continued to increase to The value is: in: This represents the weighted residual between the pollution value and the correction value; This represents the kernel width of the loss function; Indicates the iteration step size; Indicates the number of iterations; pass Factor for kernel width parameter Update; The correction values ​​for acquiring industrial evaporation time-series data include: An improved LSTM neural network model is used to obtain the correction adjustment value of industrial evaporation time series data. Based on this correction adjustment value, the industrial evaporation time series data is inversely normalized to obtain the corrected value. The inverse normalization method is as follows: in: This represents the final corrected value; This represents the normalized correction value; This represents the maximum value containing a significant error. This represents the minimum value containing a significant error.

2. The industrial data correction method based on an improved LSTM neural network according to claim 1, characterized in that, Determining the forward propagation of the LSTM neural network model includes: Based on the input and output dimensions of industrial evaporation time series data in the LSTM neural network model, the input dimension, hidden layer dimension, output dimension, LSTM layers, number of LSTM layers, and linear layers of the LSTM neural network model are defined. Based on the correction target of industrial evaporation time series data in the LSTM neural network model, the forward propagation mode of the LSTM neural network model is defined as follows: the input industrial evaporation time series data is processed through the LSTM layer and then processed through the linear layer to obtain the correction value.

3. The industrial data correction method based on an improved LSTM neural network according to claim 1, characterized in that, The iterative update of the LSTM neural network model weights includes: After obtaining the loss function value of the LSTM neural network model, the loss function value is used as the input of the LSTM neural network model for backpropagation. The improved robust estimation function iteratively updates the network weights of the LSTM neural network model, and the weight parameters are updated using the gradient descent method. The update method is as follows: in: and These represent the weight parameters and bias parameters between the input state and the hidden state, respectively. and These represent the weight parameters and bias parameters between hidden states, respectively; Indicates a fixed learning rate, and >0; This represents the loss function.

4. An industrial data correction device based on an improved LSTM neural network, characterized in that, include: The data module is used to acquire industrial evaporation time-series data during industrial production processes; An improved robust estimation function based on random and significant errors in industrial evaporation time series data is constructed. The model module is used to construct an LSTM neural network model by using the improved robust estimation function as the loss function of the LSTM neural network. Industrial evaporation time series data is used as input to the LSTM neural network model for forward propagation. Based on the result of forward propagation and the deviation of industrial evaporation time series data containing random and significant errors, the loss function value of the current LSTM neural network model is obtained. Backpropagation is then performed based on the loss function value to iteratively update the network weights of the LSTM neural network model, so as to obtain an improved LSTM neural network model. The correction module is used to obtain the correction value of industrial evaporation time series data using the improved LSTM neural network model; The improved robust estimation function is: in: This represents the weighted residual between the pollution value and the correction value; This represents the kernel width of the loss function; in: This indicates a contamination value containing gross errors; Indicates the correction value; The standard deviation of a variable; After the improved robust estimation function is obtained, the kernel width parameter within the improved robust estimation function is adjusted, including: Improve the robust estimation function Represented as ,along with Increase, The value continues to decrease; Will The value is set to a threshold. At the start of backpropagation, Value set to ,and As the number of training sessions continues to increase, Value by The value continued to increase to The value is: in: This represents the weighted residual between the pollution value and the correction value; This represents the kernel width of the loss function; Indicates the iteration step size; Indicates the number of iterations; pass Factor for kernel width parameter Update; The correction values ​​for acquiring industrial evaporation time-series data include: An improved LSTM neural network model is used to obtain the correction adjustment value of industrial evaporation time series data. Based on this correction adjustment value, the industrial evaporation time series data is inversely normalized to obtain the corrected value. The inverse normalization method is as follows: in: This represents the final corrected value; This represents the normalized correction value; This represents the maximum value containing a significant error. This represents the minimum value containing a significant error.

Images