Method for realizing automatic gas scheduling based on converter gas prediction model

Abstract

The invention provides a method for realizing automatic gas scheduling based on a converter gas prediction model, and the method comprises the steps: collecting production data in the production process of a converter, and carrying out the preprocessing operation of the production data, and obtaining the preprocessing data; performing data selection operation on the preprocessed data to obtain input data; predicting the input data through the trained neural network model to obtain a prediction result; and adjusting the consumption and storage of the gas according to the prediction result. According to the method, coal gas distribution and utilization are more actively and accurately carried out according to the potential supply and demand imbalance problem, the continuity and stability of the production process are guaranteed, and the energy utilization efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of industrial Internet and energy management technology, and in particular to a method for realizing automatic gas scheduling based on a converter gas prediction model. Background Art

[0002] In the steel production process, converter gas is a high-value energy produced as a byproduct of the steelmaking process. Its main components include carbon monoxide, carbon dioxide, nitrogen, etc., and it is an important fuel for heating and power generation within steel enterprises. Within steel enterprises, gas tanks are important equipment for storing converter gas, and the volatility of their storage capacity (i.e., changes in gas tank positions) directly reflects the balance of gas supply and demand. Among them, the amount of converter gas produced is affected by many factors, such as the type of raw materials, process parameters in the steelmaking process, and the operating status of the equipment. The volatility of these factors makes it difficult to accurately predict the output of gas. At the same time, the demand for gas consumption will also fluctuate due to adjustments to production plans, changes in equipment operating conditions, etc. At present, there are some data-driven scheduling systems, such as gas scheduling through simple statistical models or empirical rules.

[0003] However, the existing converter gas scheduling methods have several technical defects. First, the traditional manual experience scheduling method has low prediction accuracy for gas, which is difficult to adapt to these fluctuations, thus affecting production stability; second, the gas tank capacity is limited and is subject to strict operating procedures, and the traditional scheduling method lacks accurate prediction and dynamic scheduling of gas tank positions; finally, most gas scheduling systems rely on linear models or simple empirical algorithms, which are difficult to handle complex and variable data in gas production and consumption. Therefore, it is necessary to design a method for realizing automatic gas scheduling based on the converter gas prediction model. Summary of the invention

[0004] The purpose of the present invention is to provide a method for realizing automatic gas scheduling based on a converter gas prediction model, so as to more actively and accurately distribute and utilize gas according to potential supply and demand imbalance problems, ensure the continuity and stability of the production process, and improve energy utilization efficiency.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for realizing automatic gas scheduling based on a converter gas prediction model comprises the following steps:

[0007] Collecting production data during the converter production process, and performing preprocessing operations on the production data to obtain preprocessed data;

[0008] Performing data selection operation on the preprocessed data to obtain input data;

[0009] Use the trained neural network model to predict the input data and obtain the prediction results;

[0010] Adjust gas consumption and storage based on the forecast results.

[0011] Optionally, the production data includes: gas generation flow, gas consumption flow and gas tank storage capacity.

[0012] Optionally, the preprocessing operation includes: cleaning the production data, removing noise and outliers in the production data, and normalizing the production data.

[0013] Optionally, a data selection operation is performed on the preprocessed data to obtain input data, and the specific steps include:

[0014] Calculate the correlation of variables in production data, and screen according to the correlation to obtain screened data;

[0015] Perform feature extraction on the screened data to obtain variable features;

[0016] Select input data based on variable characteristics.

[0017] Optionally, the input data is predicted using a trained neural network model to obtain a prediction result. The specific steps include:

[0018] The historical data of converter production process is divided into training set and test set;

[0019] The BP neural network model is trained through the training set and back propagation algorithm to obtain a preliminary model;

[0020] Use the preliminary model to predict the test set, obtain preliminary results, and calculate the error between the preliminary results and the actual results;

[0021] The mean square error algorithm and the mean absolute error algorithm are used to analyze the error between the preliminary result and the actual result to obtain the analysis result;

[0022] Adjust the preliminary model according to the analysis results to obtain a trained neural network model;

[0023] Input the input data into the trained neural network model for prediction and obtain the prediction results.

[0024] Optionally, the prediction result is the 30-minute offset of the gas tank position; the calculation formula for the offset is: Q = GH t+30 -GH middle ; where Q is the offset, GH t+30 It is the gas tank value after 30 minutes, GH middle It is the middle counter value.

[0025] Optionally, the consumption and storage of gas are adjusted according to the prediction results. The specific steps include:

[0026] Calculate the boiler's adjustment rate to eliminate offset;

[0027] The regulating rate is compared with the maximum regulating rate of the boiler to obtain a comparison result, and the regulating rate is corrected once according to the comparison result to obtain a first corrected rate;

[0028] The first correction rate is corrected twice according to the first correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler, and the minimum converter gas consumption flow rate of the boiler to obtain a second correction rate;

[0029] Gas consumption and storage are adjusted based on the second modified rate.

[0030] Optionally, the specific calculation formula for the adjustment rate is: q t =Q / Δt; where q t is the adjustment rate, and Δt is the adjustment time.

[0031] Optionally, the regulating rate is compared with the maximum regulating rate of the boiler to obtain a comparison result, and the regulating rate is corrected once according to the comparison result to obtain a first corrected rate. The specific steps are:

[0032] When the regulating rate is greater than the maximum regulating rate of the boiler, the regulating rate is corrected to the maximum regulating rate of the boiler;

[0033] When the regulation rate is less than the negative maximum regulation rate of the boiler, the regulation rate is corrected to the negative maximum regulation rate of the boiler;

[0034] When the regulation rate is greater than or equal to the negative maximum regulation rate of the boiler and less than or equal to the maximum regulation rate of the boiler, the regulation rate remains unchanged.

[0035] Optionally, the first correction rate is corrected twice according to the first correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler and the minimum converter gas consumption flow rate of the boiler to obtain a second correction rate. The specific steps are: the first correction rate, the second correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler and the minimum converter gas consumption flow rate of the boiler are respectively recorded as q t2 ,q t3 ,q gl.0 ,q gl.max , and q gl.min ;

[0036] When gl,0 +q t2 >q gl,max When q t3=q gl,max -q gl,0 ;

[0037] When gl,0 +q t2 <q gl,min When q t3 =q gl,0 -q gl,min ;

[0038] When gl,min ≦q gl,0 +q t2 ≦q gl,max When q t3 =q t2 .

[0039] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: the method for realizing automatic gas scheduling based on the converter gas prediction model provided by the present invention comprises: collecting production data in the converter production process, and performing preprocessing operations on the production data to obtain preprocessed data; performing data selection operations on the preprocessed data to obtain input data; predicting the input data through a trained neural network model to obtain a prediction result; and adjusting the generation and consumption of gas according to the prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0041] Figure 1 A flow chart of a method for realizing automatic gas scheduling based on a converter gas prediction model according to an embodiment of the present invention;

[0042] Figure 2 A neural network model prediction flow chart of an embodiment of the present invention;

[0043] Figure 3 The figure is a gas regulation flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] like Figure 1 As shown, the embodiment of the present invention provides a method for realizing gas automatic scheduling based on a converter gas prediction model, comprising the following steps:

[0047] Step 100: collecting production data in the converter production process, and performing preprocessing operations on the production data to obtain preprocessed data.

[0048] Specifically, production data is obtained through PLC control systems, sensors, flow meters and other equipment. Production data includes: gas generation flow, gas consumption flow and gas tank storage capacity, as well as temperature, pressure, flow and other related parameters of the converter gas system.

[0049] Specifically, the preprocessing operation includes: cleaning the production data, removing noise and outliers in the production data, and normalizing the production data. The preprocessing operation ensures the accuracy and consistency of the data, and also enables data of different dimensions to be effectively analyzed in the same model.

[0050] Step 200: Perform a data selection operation on the pre-processed data to obtain input data.

[0051] It should be noted that in the process of selecting input data, the main basis is to ensure that the selected data can fully and accurately reflect the key factors affecting the change of converter gas tank position, so as to improve the accuracy and robustness of the prediction model. The specific selection steps are as follows:

[0052] First, determine the variables in the production data that can affect the size of the converter gas tank, including:

[0053] Gas production is the core factor affecting the gas tank position. This parameter directly determines the filling rate of the gas tank and is therefore the key input data of the model.

[0054] Gas tank pressure is also a core factor affecting the gas tank position, and is positively correlated with the tank position.

[0055] Gas consumption. The gas consumption in different production links will affect the storage level of the gas tank. Therefore, this parameter is also an important factor affecting the changes in gas tank positions.

[0056] The initial position of the gas tank determines the basic level of the gas tank at the beginning of scheduling and has a direct impact on subsequent predictions.

[0057] Process parameters are parameters such as the type of raw materials, furnace temperature, furnace pressure, etc. used by gas consumers. These parameters will indirectly affect the gas generation and consumption flow. Combining these process parameters with the main parameters such as gas generation flow, gas consumption flow, gas tank pressure, and gas tank initial position can capture more details of changes in the production process.

[0058] Equipment status parameters, the operating status of the equipment (such as whether it is running under optimal conditions) will affect the efficiency of gas generation and consumption. Introducing these parameters can improve the adaptability of the model to actual production conditions.

[0059] Historical data trends, time series features in historical data (such as changing trends and periodic fluctuations in gas cabinet positions) can capture some potential rules and be used to improve the predictive ability of the model.

[0060] Secondly, the correlation of each variable in the determined production data is calculated, and the variables with high correlation are selected to obtain the screening data. By calculating and screening the correlation, the parameters with large noise are eliminated, so that the variables input into the model have good stability and robustness.

[0061] Then, the principal component analysis method is used to reduce the dimension and extract features of the screened data to obtain the variable characteristics of each variable in the screened data, thereby simplifying the model input, avoiding the interference of redundant variables, and improving the prediction accuracy of the model.

[0062] Finally, the input data of the neural network model are selected according to the variable characteristics. The specific input data selected include: gas generation flow, residual oxygen content in gas, carbon monoxide content in gas, gas consumption flow of converter, gas consumption flow and load of generator sets in converter gas system, pressure of converter gas tank and pressure of gas main network in converter gas system.

[0063] Step 300: Use the trained neural network model to predict the input data and obtain the prediction result. Figure 2 As shown, including:

[0064] Step 301: Divide the historical data of the converter production process into a training set and a test set. The specific steps are: extract 4,000 groups of samples from the historical data of the converter production process, use the first 2,000 groups of samples as the training set of the model, and use the last 2,000 groups of samples as the test set of the model. This division ensures that the model can fully learn the characteristics of the data during training and verify its prediction ability during testing.

[0065] Step 302: Use the training set to train the BP neural network model, and adjust the weights and biases of the model through the back propagation algorithm, thereby optimizing the model parameters.

[0066] Specifically, the BP neural network model is a multi-layer feedforward neural network model trained based on the error back propagation algorithm. It performs self-learning through the gradient descent method, can realize nonlinear mapping between input and output, is used to solve various problems such as classification and regression, and can learn complex functional relationships through a large amount of training data. The structure of the BP neural network model includes an input layer, a hidden layer (can have multiple) and an output layer. Each layer is composed of multiple neurons, which are connected by weights and transmit information to each other under the action of an activation function. In the forward propagation process, the input data is passed to the hidden layer after being weighted and processed by the activation function, and finally to the output layer. If the predicted value of the output layer does not match the expected value, it will enter the error back propagation process, and the weights between the layers will be continuously adjusted by the gradient descent method to reduce the error.

[0067] Step 303: The BP neural network model is evaluated and predicted using the test set, and the error between the predicted result and the actual result is calculated.

[0068] Step 304: Analyze the error between the preliminary result and the actual result using the mean square error algorithm and the mean absolute error algorithm to obtain an analysis result.

[0069] Step 305: Adjust the preliminary model according to the analysis results (such as increasing the number of hidden layer nodes, adjusting the learning rate, etc.) to obtain a trained neural network model.

[0070] Step 306: Input the input data into the trained neural network model for prediction to obtain a prediction result.

[0071] Specifically, the prediction result is the 30-minute offset of the gas tank position; the calculation formula of the offset is: Q = GH t+30 -GH middle ; where Q is the offset, GH t+30 It is the gas tank value after 30 minutes, GH middle It is the middle counter value.

[0072] Step 400: Adjust gas consumption and storage according to the prediction results.

[0073] Specifically, when it is predicted that the gas tank level will approach the upper limit in 30 minutes, priority is given to reducing gas output or increasing gas consumption to avoid gas tank level over-limit; when it is predicted that the gas tank level will approach the lower limit in 30 minutes, gas production is increased or gas consumption is reduced to ensure a stable gas supply. By dynamically adjusting the gas flow, the storage volume in the gas tank is maintained within a reasonable range, preventing energy waste or production interruptions.

[0074] It should be noted that the specific steps for adjusting the generation and consumption of gas according to the prediction results are as follows: Figure 3 As shown, including:

[0075] Step 401: Calculate the boiler offset elimination adjustment rate. The specific calculation formula is: q t =Q / Δt; where q t is the adjustment rate, and Δt is the adjustment time.

[0076] Step 402: Determine the adjustment rate q t Whether it exceeds the maximum adjustment rate q allowed by the boiler max , and make a first correction to the adjustment rate according to the judgment result to obtain a first correction rate. The specific steps are:

[0077] When t >q max When q t Corrected to the maximum adjustment rate, so that q t2 =q max ;

[0078] When t <-q max When q t Corrected to the negative maximum regulation rate, so that q t2 =—q max ;

[0079] When -q max ≦q t ≦q max When q t2 =q t .

[0080] Step 403: Adjust the converter gas consumption flow rate for the first time according to the adjustment rate after the first correction, and determine whether the adjusted converter gas consumption flow rate is within the maximum converter gas consumption flow rate q of the boiler. gl.max Minimum consumption flow rate of converter gas with boiler q gl.min The first adjustment rate is corrected for the second time according to the judgment result. The specific steps are: the first correction rate, the second correction rate and the current converter gas consumption flow are respectively recorded as q t2 ,q t3 and q gl.0 .

[0081] When gl,0 +q t2 >q gl,max When q t3 =q gl,max -q gl,0 ;

[0082] When gl,0 +q t2 <q gl,min When q t3 =q gl,0 -q gl,min ;

[0083] When gl,min ≦q gl,0 +q t2 ≦q gl,max When q t3 =q t2 .

[0084] Step 404: Adjust gas consumption and storage according to the second correction rate.

[0085] The beneficial effects of the present invention are as follows:

[0086] 1) Realizes real-time monitoring and automatic dispatching of converter gas, ensuring the continuity of production;

[0087] 2) It realizes efficient use of energy and effectively avoids waste of resources or production interruption caused by too high or too low gas tank level;

[0088] 3) Training the BP neural network model and using the trained model for prediction greatly improves the prediction accuracy and efficiency.

[0089] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0090] The present invention uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for realizing automatic gas scheduling based on converter gas prediction model, characterized in that: The steps include: Collecting production data during the converter production process, and performing preprocessing operations on the production data to obtain preprocessed data; Performing a data selection operation on the preprocessed data to obtain input data; Predicting the input data using a trained neural network model to obtain a prediction result; The consumption and storage of gas are adjusted according to the prediction results.

2. The method for realizing automatic gas scheduling based on converter gas prediction model according to claim 1, characterized in that: The production data includes: the gas generation flow, the gas consumption flow and the storage capacity of the gas tank.

3. The method for realizing automatic gas scheduling based on converter gas prediction model according to claim 1, characterized in that: The preprocessing operation includes: cleaning the production data, removing noise and abnormal values ​​in the production data, and normalizing the production data.

4. The method for realizing automatic gas scheduling based on converter gas prediction model according to claim 1, characterized in that: Performing a data selection operation on the preprocessed data to obtain input data, the specific steps include: Calculating the correlation of variables in the production data, and screening according to the correlation to obtain screening data; Performing feature extraction operation on the screening data to obtain variable features; The input data is selected according to the variable characteristics.

5. The method for realizing automatic gas scheduling based on converter gas prediction model according to claim 1, characterized in that: The input data is predicted by the trained neural network model to obtain a prediction result, and the specific steps include: Dividing the historical data of the converter production process into a training set and a test set; The BP neural network model is trained by using the training set and the back propagation algorithm to obtain a preliminary model; Predicting the test set using the preliminary model to obtain preliminary results, and calculating the error between the preliminary results and the actual results; Analyze the error between the preliminary result and the actual result using a mean square error algorithm and a mean absolute error algorithm to obtain an analysis result; Adjusting the preliminary model according to the analysis results to obtain the trained neural network model; The input data is input into the trained neural network model for prediction to obtain the prediction result.

6. The method for realizing automatic gas dispatching based on converter gas prediction model according to claim 1, characterized in that: The prediction result is the 30-minute offset of the gas tank position; the calculation formula of the offset is: Q = GH t+30 -GH middle ; Where Q is the offset, GH t+30 It is the gas tank value after 30 minutes, GH middle It is the middle counter value.

7. The method for realizing automatic gas dispatching based on converter gas prediction model according to claim 6, characterized in that: The consumption and storage of gas are adjusted according to the prediction results, and the specific steps include: calculating a rate of regulation of the boiler to eliminate the offset; The regulating rate is compared with the maximum regulating rate of the boiler to obtain a comparison result, and the regulating rate is corrected once according to the comparison result to obtain a first corrected rate; Performing secondary correction on the first correction rate according to the first correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler, and the minimum converter gas consumption flow rate of the boiler to obtain a second correction rate; The consumption and storage of the gas are adjusted according to the second correction rate.

8. The method for realizing automatic gas dispatching based on converter gas prediction model according to claim 7, characterized in that: The specific calculation formula of the adjustment rate is: t =Q / Δt; where q t is the adjustment rate, and Δt is the adjustment time.

9. The method for realizing automatic gas dispatching based on converter gas prediction model according to claim 7, characterized in that: The regulating rate is compared with the maximum regulating rate of the boiler to obtain a comparison result, and the regulating rate is corrected once according to the comparison result to obtain a first corrected rate. The specific steps are: When the regulation rate is greater than the maximum regulation rate of the boiler, the regulation rate is corrected to the maximum regulation rate of the boiler; When the regulation rate is less than the negative maximum regulation rate of the boiler, the regulation rate is corrected to the negative maximum regulation rate of the boiler; When the regulation rate is greater than or equal to the negative maximum regulation rate of the boiler and less than or equal to the maximum regulation rate of the boiler, the regulation rate remains unchanged.

10. The method for realizing automatic gas dispatching based on converter gas prediction model according to claim 7, characterized in that: The first correction rate is corrected twice according to the first correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler and the minimum converter gas consumption flow rate of the boiler to obtain a second correction rate. The specific steps are: the first correction rate, the second correction rate, the gas consumption flow rate, the maximum converter gas consumption flow rate of the boiler and the minimum converter gas consumption flow rate of the boiler are respectively recorded as q t2 ,q t3 ,q gl，0 ,q gl.max , and q gl.min ; When gl,0 +q t2 >q gl,max When q t3 =q gl,max -q gl,0 ; When gl,0 +q t2 <q gl,min When q t3 =q gl,0 -q gl,min ; When gl,min ≦q gl,0 +q t2 ≦q gl,max When q t3 =q t2 .

Images