Hot-blast stove gas consumption prediction method based on equipment state signal real-time correction

Abstract

The invention provides a hot blast stove gas consumption prediction method based on equipment state signal real-time correction, which comprises the following steps: reading hot blast stove gas consumption historical data with a time index and historical equipment state signal data in a database as initial data, and performing targeted processing for training a neural network prediction model; historical hot-blast stove production actual gas consumption data with time indexes near the current moment and hot-blast stove equipment state signal data are used as input data of the neural network prediction model, future gas consumption prediction is carried out, and a prediction result is subjected to targeted processing; marking a state transition point of the predicted hot-blast stove according to the change characteristics of the predicted data; receiving hot-blast stove state signal data with time indexes in real time, and marking actual hot-blast stove state transition points according to change characteristics of the signal data; and confirming the correlation between the predicted hot-blast stove state transition point and the actual hot-blast stove state transition point, and correcting the prediction result according to the correlation.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals. Background Art

[0002] Coal gas is an important secondary energy source in the steel production process. The full use of coal gas can not only reduce the external energy consumption of steel enterprises, but also avoid the negative impact of coal gas emissions on the environment. In order to better manage coal gas, the prediction of future consumption of coal gas equipment has become a key research direction for energy management and control in steel enterprises. As a major consumer of coal gas, the prediction of the gas consumption of hot blast furnaces is crucial in overall energy management and control. However, in the actual production process, due to the complexity and variability of the production site, the equipment rhythm of the hot blast furnace has great uncertainty, making it difficult for traditional prediction models to fully obtain the rhythm change law of the hot blast furnace from historical data, resulting in great difficulties in accurately predicting the gas consumption of the hot blast furnace.

[0003] The existing hot blast furnace gas consumption prediction technology only considers static factors such as historical data characteristics and production plans, so it has the defects of low prediction accuracy and prediction results deviating from the production rhythm. Summary of the invention

[0004] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method for predicting the gas consumption of a hot blast furnace based on real-time correction of equipment status signals. The gas consumption prediction results can be corrected by obtaining the real-time equipment status signals of the hot blast furnace, thereby avoiding the deviation of the prediction results from the actual production rhythm caused by the simple data prediction method, thereby improving the prediction accuracy of the hot blast furnace gas consumption, and providing reliable support for steel mills to grasp the production rhythm and conduct gas scheduling.

[0005] The technical means adopted by the present invention are as follows:

[0006] A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals, comprising:

[0007] S1. Read the hot blast furnace gas consumption historical data and historical equipment status signal data with time index in the database and use them as initial data;

[0008] S2, process the initial data in a targeted manner, and use the processed initial data to train the neural network prediction model;

[0009] S3, inputting the actual gas consumption data of the hot blast furnace production with time index near the current time and the hot blast furnace equipment status signal data into the neural network prediction model as input data of the neural network prediction model, predicting the future gas consumption, and performing targeted processing on the predicted future gas consumption, and predicting the hot blast furnace state transition point according to the change feature mark of the prediction data;

[0010] S4, receiving hot blast furnace state signal data with time index in real time, and marking the actual hot blast furnace state transition point according to the change characteristics of the signal data;

[0011] S5. Confirm the correlation between the predicted hot blast stove state transition point and the actual hot blast stove state transition point, and correct the prediction result according to the correlation.

[0012] Further, step S2 includes:

[0013] S21. Targeted processing of missing values:

[0014] According to the data characteristics of the periodic fluctuation of hot blast furnace gas consumption and the cliff-like change of consumption values ​​at the state change node, the method of filling the mean of the previous and next effective values ​​is adopted;

[0015] S22. Targeted processing of abnormal values:

[0016] The principle is to replace values ​​not close to zero with non-zero means and values ​​close to zero with zero values.

[0017] Furthermore, step S21 specifically includes:

[0018] S211. Get the time index of the missing value and record the value of this index as the missing value Q mv ;

[0019] S212, search forward for the nearest valid value index, and record the value of this index as the previous valid value Q pev , and record the difference between the two indexes as T p ;

[0020] S213, search backward for the nearest valid value index, and record the value of this index as the next valid value Q poev , and record the difference between the two indexes as T po ;

[0021] S214. When the current and the next valid values ​​are both zero or both non-zero, calculate the missing value Q mv ,as follows:

[0022]

[0023] S215. When one of the current and the next valid values ​​is zero and the other is non-zero, the following conditions apply:

[0024] When T p ≥T po When mv =Q pev ;

[0025] When T p <T po When mv =Q poev .

[0026] Further, step S22 specifically includes:

[0027] S221. According to the hot blast stove data characteristics, obtain the non-zero mean value Q of the initial data nzm ,as follows:

[0028]

[0029] Q i is the non-zero value in the initial data, and n is the number of non-zero values;

[0030] S222. Obtain the time index of the abnormal value and record the value of this index as the abnormal value Q av ;

[0031] S223. According to the data characteristics of the hot blast furnace gas consumption, the principle of replacing non-zero values ​​with non-zero mean values ​​and replacing near-zero values ​​with zero values ​​is adopted:

[0032] When the values ​​on both sides of the outlier are greater than zero: Q av =Q nzm ;

[0033] When the values ​​at both ends of the outlier are equal to zero: Q av =0;

[0034] When the value on one side of the outlier is equal to zero and the value on the other side is greater than zero:

[0035] If the conditions are met: Q av <Q nzm -Q dv , then Q av =0;

[0036] If the conditions are met: Q nzm +Q dv <Q av , then Q av =Q nzm ;

[0037] In the formula, Q dvis the maximum value allowed for the outlier to deviate. According to the three sigma criterion, it can be set to [σ, 3σ], where σ is the standard deviation of the non-zero values ​​in the initial data.

[0038] Further, step S3 specifically includes:

[0039] S31. Conventional hot air furnaces include but are not limited to two equipment states: furnace burning and air supply:

[0040] When the hot blast furnace is in the burning state, it consumes gas, and the gas consumption data value is greater than 0;

[0041] When the hot air furnace is in the air supply state, no gas is consumed, and the gas consumption data value is equal to 0;

[0042] According to the changing characteristics of gas consumption of hot blast furnace, when the hot blast furnace changes from the burning state to the air supply state, the gas consumption data will suddenly change from a non-zero value to 0;

[0043] When the hot air furnace changes from the air supply state to the furnace burning state, the gas consumption data will suddenly change from 0 to a non-zero value;

[0044] S32, predicting the hot blast stove state transition point according to the change characteristic mark of the prediction data:

[0045] When the gas consumption data values ​​for more than three consecutive time steps are all 0, the time index of the point where the first gas consumption data value is 0 is marked as the air supply switching point t w ;

[0046] When the gas consumption data values ​​for more than three consecutive time steps are all non-zero values, the time index of the point where the first gas consumption data value is non-zero is marked as the furnace switching point t b .

[0047] Further, step S4 specifically includes:

[0048] S41. In actual production, when the hot air furnace is in the burning state, the burning signal data value will be displayed as 1, and when the burning is stopped, the burning signal data value will be displayed as 0;

[0049] S42. At the same time, when the hot air furnace is in the air supply state, the air supply signal data value will be displayed as 1, and when the air supply is stopped, the air supply signal data value will be displayed as 0;

[0050] S43, when the value of the furnace burning status signal received in real time changes from 0 to 1, mark this time index point as the furnace burning transition point T b ;

[0051] S44, when the real-time received air supply status signal data changes from 0 to 1, mark this time index point as the air supply switching point Tw .

[0052] Furthermore, step S5 specifically includes:

[0053] S51, every time the air supply switching point T w When it appears, calculate the air supply switching point T w The last air supply switching point t w The time difference Δt w :

[0054] Δt w =t aw -T w

[0055] Where, t aw is the most recent air supply switching point t w ;

[0056] S52, every time the furnace switching point T b When it appears, calculate the furnace switching point T b The last furnace switching point t b The difference Δt b :

[0057] Δt b =t ab -T b

[0058] Where, t ab The most recent furnace switching point t b ;

[0059] S53, every time the air supply state conversion time difference is obtained, the furnace conversion point t b The time index of all future forecast values ​​is modified as follows:

[0060]

[0061] Where, t ac-wi is the corrected air supply switching point t w The time index of the i-th time step after t wi is the air supply switching point before correction t w The time index of the i-th time step after t wi That is t w , when i=n, t wi The time index of the last predicted value at the current moment;

[0062] S54, every time the furnace state conversion time difference is obtained, the nearest furnace conversion point t b The time indexes of all predicted values ​​at the beginning are modified as follows:

[0063]

[0064] Where, t ac-bi is the corrected furnace switching point t b The time index of the i-th time step after t bi is the furnace switching point before correction b The time index of the i-th time step after t bi That is t b , when i=n, t bi The time index of the last predicted value at the current moment.

[0065] Compared with the prior art, the present invention has the following advantages:

[0066] 1. The present invention provides a method for predicting the gas consumption of a hot blast furnace based on real-time correction of equipment status signals. According to the periodic and cliff-like change characteristics of the gas consumption data and signal data of the hot blast furnace, targeted data processing and signal processing methods are adopted, which not only improves the validity of the initial data, but also fully retains the change characteristics of the data, reduces the training time of the prediction model, and improves the accuracy of the model parameters.

[0067] 2. The present invention provides a method for predicting the gas consumption of a hot blast furnace based on real-time correction of equipment status signals. The method analyzes the characteristics of the predicted gas consumption data to mark the predicted equipment status transition points, and analyzes the characteristics of the equipment signal data in real-time production to mark the actual equipment status transition points. The predicted gas consumption data of the hot blast furnace is then corrected according to the correlation between the predicted equipment status transition points and the actual equipment status transition points. The method cleverly combines the production process characteristics with the intelligent algorithm to improve the prediction accuracy.

[0068] 3. The present invention provides a method for predicting the gas consumption of a hot blast furnace based on real-time correction of equipment status signals. Due to its high prediction accuracy and no production rhythm deviation problem, it realizes accurate prediction of gas consumption and provides reliable support for steel mills to master production rhythm and conduct gas scheduling.

[0069] Based on the above reasons, the present invention can be widely promoted in the fields of gas consumption prediction and the like. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] 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 or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are 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.

[0071] Figure 1 The figure is a flow chart of the method of the present invention.

[0072] Figure 2 This is the effect of the prediction result before correction provided by the embodiment of the present invention.

[0073] Figure 3 This is the corrected effect of the prediction result provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0074] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0075] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0076] The prediction accuracy of the hot blast stove gas consumption is affected by many factors, including the production rhythm change of the hot blast stove, the processing of the data used for training and prediction, etc. The processing effect of each factor will affect the prediction accuracy of the hot blast stove gas consumption. Therefore, in order to make a higher-precision prediction of the hot blast stove gas consumption, an embodiment of the present invention provides a hot blast stove gas consumption prediction method based on real-time correction of the device status signal.

[0077] like Figure 1 As shown, the present invention provides a method for predicting gas consumption of a hot blast furnace based on real-time correction of equipment status signals, comprising:

[0078] S1. Read the hot blast furnace gas consumption historical data and historical equipment status signal data with time index in the database and use them as initial data;

[0079] In this embodiment, the length of reading the hot blast stove gas consumption data and the equipment status signal data is 20,000 time steps, and each time step is 1 minute. However, in some other embodiments, the time length and time step have no fixed range, and the specific values ​​should be determined based on actual conditions, such as data quality, predicted business requirements, and server computing capabilities, and are not limited here.

[0080] S2, process the initial data in a targeted manner, and use the processed initial data to train the neural network prediction model;

[0081] S3, inputting the actual gas consumption data of the hot blast furnace production with time index near the current time and the hot blast furnace equipment status signal data into the neural network prediction model as input data of the neural network prediction model, predicting the future gas consumption, and processing the predicted future gas consumption in a targeted manner, and predicting the hot blast furnace state transition point according to the change characteristics of the prediction data;

[0082] In this embodiment, the neural network prediction model used is the LSTM algorithm, but in other embodiments, other algorithms that can achieve the same purpose can also be used, which is not limited here.

[0083] S4, receiving hot blast furnace state signal data with time index in real time, and marking the actual hot blast furnace state transition point according to the change characteristics of the signal data;

[0084] S5. Confirm the correlation between the predicted hot blast stove state transition point and the actual hot blast stove state transition point, and correct the prediction result according to the correlation.

[0085] In specific implementation, as a preferred embodiment of the present invention, step S2 includes:

[0086] S21. Targeted processing of missing values:

[0087] According to the data characteristics of the periodic fluctuation of hot blast furnace gas consumption and the cliff-like change of consumption values ​​at the state change node, the method of filling the mean of the previous and next effective values ​​is adopted;

[0088] S22. Targeted processing of outliers (values ​​that satisfy a certain correlation with non-zero mean and zero value are outliers):

[0089] The principle of replacing non-zero values ​​with non-zero means and replacing near-zero values ​​with zero values ​​is adopted;

[0090] S23. After outlier processing, the processed initial data is used to train a neural network prediction model.

[0091] In specific implementation, as a preferred embodiment of the present invention, step S21 specifically includes:

[0092] S211. Get the time index of the missing value and record the value of this index as the missing value Q mv ;

[0093] S212, search forward for the nearest valid value index, and record the value of this index as the previous valid value Q pev , and record the difference between the two indexes as T p ;

[0094] S213, search backward for the nearest valid value index, and record the value of this index as the next valid value Q poev , and record the difference between the two indexes as T po ;

[0095] S214. When the current and the next valid values ​​are both zero or both non-zero, calculate the missing value Q mv ,as follows:

[0096]

[0097] S215. When one of the current and the next valid values ​​is zero and the other is non-zero, the following conditions apply:

[0098] When T p ≥T po When mv =Q pev ;

[0099] When T p <T po When mv =Q poev .

[0100] In specific implementation, as a preferred embodiment of the present invention, step S22 specifically includes:

[0101] S221. According to the hot blast stove data characteristics, obtain the non-zero mean value Q of the initial data nzm ,as follows:

[0102]

[0103] Q i is the non-zero value in the initial data, and n is the number of non-zero values;

[0104] S222. Obtain the time index of the abnormal value and record the value of this index as the abnormal value Q av ;

[0105] S223. According to the data characteristics of the hot blast furnace gas consumption, the principle of replacing non-zero values ​​with non-zero mean values ​​and replacing near-zero values ​​with zero values ​​is adopted:

[0106] When the values ​​on both sides of the outlier are greater than zero: Q av =Q nzm ;

[0107] When the values ​​at both ends of the outlier are equal to zero: Q av =0;

[0108] When the value on one side of the outlier is equal to zero and the value on the other side is greater than zero:

[0109] If the conditions are met: Q av <Q nzm -Q dv , then Q av =0;

[0110] If the conditions are met: Q nzm +Q dv <Q av , then Q av =Q nzm ;

[0111] In the formula, Q dv is the maximum value allowed for the outlier to deviate. According to the three sigma criterion, it can be set to [σ, 3σ], where σ is the standard deviation of the non-zero values ​​in the initial data.

[0112] In this embodiment, the number of non-zero values ​​n is determined according to the specific number of non-zero values ​​in the given 20,000 historical gas consumption data and signal data, and is not limited here; the maximum value Q of the abnormal value allowed to deviate in this embodiment is dv It is preferably 2σ, which is substituted into the formula as the basis for selecting outliers and processing the data. The maximum value Q allowed for the outlier to deviate is dv The selection can be made based on actual data characteristics and is not limited here.

[0113] In specific implementation, as a preferred embodiment of the present invention, step S3 specifically includes:

[0114] S31. Conventional hot air furnaces include but are not limited to two equipment states: furnace burning and air supply:

[0115] When the hot blast furnace is in the burning state, it consumes gas, and the gas consumption data value is greater than 0;

[0116] When the hot air furnace is in the air supply state, no gas is consumed, and the gas consumption data value is equal to 0;

[0117] According to the changing characteristics of gas consumption of hot blast furnace, when the hot blast furnace changes from the burning state to the air supply state, the gas consumption data will suddenly change from a non-zero value to 0;

[0118] When the hot air furnace changes from the air supply state to the furnace burning state, the gas consumption data will suddenly change from 0 to a non-zero value;

[0119] S32, predicting the hot blast stove state transition point according to the change characteristic mark of the prediction data:

[0120] When the gas consumption data values ​​for more than three consecutive time steps are all 0, the time index of the point where the first gas consumption data value is 0 is marked as the air supply switching point t w ;

[0121] When the gas consumption data values ​​for more than three consecutive time steps are all non-zero values, the time index of the point where the first gas consumption data value is non-zero is marked as the furnace switching point t b .

[0122] In specific implementation, as a preferred embodiment of the present invention, step S4 specifically includes:

[0123] S41. In actual production, when the hot air furnace is in the burning state, the burning signal data value will be displayed as 1, and when the burning is stopped, the burning signal data value will be displayed as 0;

[0124] S42. At the same time, when the hot air furnace is in the air supply state, the air supply signal data value will be displayed as 1, and when the air supply is stopped, the air supply signal data value will be displayed as 0;

[0125] S43, when the value of the furnace burning status signal received in real time changes from 0 to 1, mark this time index point as the furnace burning transition point T b ;

[0126] S44, when the real-time received air supply status signal data changes from 0 to 1, mark this time index point as the air supply switching point T w .

[0127] In this embodiment, the furnace switching point T b and air supply switching point T w The time index point may be any time during the operation of the device. You can select it according to actual needs. There is no restriction here.

[0128] In specific implementation, as a preferred embodiment of the present invention, step S5 specifically includes:

[0129] S51, every time the air supply switching point T w When it appears, calculate the air supply switching point T w The last air supply switching point t w The time difference Δt w :

[0130] Δt w =taw -T w

[0131] Where, t aw is the most recent air supply switching point t w ;

[0132] S52, every time the furnace switching point T b When it appears, calculate the furnace switching point T b The last furnace switching point t b The difference Δt b :

[0133] Δt b =t ab -T b

[0134] Where, t ab The most recent furnace switching point t b ;

[0135] S53, every time the air supply state conversion time difference is obtained, the furnace conversion point t b The time index of all future forecast values ​​is modified as follows:

[0136]

[0137] Where, t ac-wi is the corrected air supply switching point t w The time index of the i-th time step after t wi is the air supply switching point before correction t w The time index of the i-th time step after t wi That is t w , when i=n, t wi The time index of the last predicted value at the current moment;

[0138] S54, every time the furnace state conversion time difference is obtained, the nearest furnace conversion point t b The time indexes of all predicted values ​​at the beginning are modified as follows:

[0139]

[0140] Where, t ac-bi is the corrected furnace switching point t b The time index of the i-th time step after t bi is the furnace switching point before correction b The time index of the i-th time step after t bi That is t b , when i=n, tbi The time index of the last predicted value at the current moment.

[0141] In this embodiment, the predicted future step size is set to 480 time steps, and the time step size is 1 minute. In other embodiments, the predicted step size and the actual step size can be set according to actual needs, and there is no restriction here; the furnace switching point t b and air supply switching point t w The time index point may be any time during the operation of the equipment. It can be selected according to actual needs and is not limited here. The value of n is based on the furnace conversion point t b and air supply switching point t w The intermediate step length between the time index and the time index of the last predicted value is not restricted here.

[0142] Figure 2 To correct the effect before the prediction result, Figure 3 To illustrate the effect of the correction of the prediction results, it can be seen from the comparison results before and after that this method can more accurately predict the gas consumption of the hot blast furnace, especially when the equipment state is converted, the prediction accuracy is significantly improved, making the predicted equipment production rhythm more consistent with the actual equipment production rhythm, and achieving real-time correction of the hot blast furnace gas consumption according to the equipment status signal.

[0143] In summary, the method of the present invention can correct the prediction result in combination with the real-time status signal of the hot blast stove, overcomes the shortcomings of poor prediction accuracy and deviation between the predicted production rhythm and the actual production rhythm caused by simple data prediction, and improves the prediction accuracy of the hot blast stove gas consumption.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals, characterized in that: include: S1. Read the hot blast furnace gas consumption historical data and historical equipment status signal data with time index in the database and use them as initial data; S2, process the initial data in a targeted manner, and use the processed initial data to train the neural network prediction model; S3, inputting the actual gas consumption data of the hot blast furnace production with time index near the current time and the hot blast furnace equipment status signal data into the neural network prediction model as input data of the neural network prediction model, predicting the future gas consumption, and processing the predicted future gas consumption in a targeted manner, and predicting the hot blast furnace state transition point according to the change characteristics of the prediction data; S4, receiving hot blast furnace state signal data with time index in real time, and marking the actual hot blast furnace state transition point according to the change characteristics of the signal data; S5. Confirm the correlation between the predicted hot blast stove state transition point and the actual hot blast stove state transition point, and correct the prediction result according to the correlation.

2. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S2 comprises: S21. Targeted processing of missing values: According to the data characteristics of the periodic fluctuation of hot blast furnace gas consumption and the cliff-like change of consumption values ​​at the state change node, the method of filling the mean of the previous and next effective values ​​is adopted; S22. Targeted processing of abnormal values: The principle is to replace values ​​not close to zero with non-zero means and values ​​close to zero with zero values.

3. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S21 specifically includes: S211. Get the time index of the missing value and record the value of this index as the missing value Q mv ; S212, search forward for the nearest valid value index, and record the value of this index as the previous valid value Q pev , and record the difference between the two indexes as T p ; S213, search backward for the nearest valid value index, and record the value of this index as the next valid value Q poev , and record the difference between the two indexes as T po ; S214. When the current and the next valid values ​​are both zero or both non-zero, calculate the missing value Q mv ,as follows: S215. When one of the current and the next valid values ​​is zero and the other is non-zero, the following conditions apply: When T p ≥T po When mv =Q pev ; When T p <T po When mv =Q poev .

4. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S22 specifically includes: S221. According to the hot blast stove data characteristics, obtain the non-zero mean value Q of the initial data nzm ,as follows: Q i is the non-zero value in the initial data, and n is the number of non-zero values; S222. Obtain the time index of the abnormal value and record the value of this index as the abnormal value Q av ; S223. According to the data characteristics of the hot blast furnace gas consumption, the principle of replacing non-zero values ​​with non-zero mean values ​​and replacing near-zero values ​​with zero values ​​is adopted: When the values ​​on both sides of the outlier are greater than zero: Q av =Q nzm ; When the values ​​at both ends of the outlier are equal to zero: Q av =0; When the value on one side of the outlier is equal to zero and the value on the other side is greater than zero: If the conditions are met: Q av <Q nzm -Q dv , then Q av =0; If the conditions are met: Q nzm +Q dv <Q av , then Q av =Q nzm ; In the formula, Q dv is the maximum value allowed for the outlier to deviate. According to the three sigma criterion, it can be set to [σ, 3σ], where σ is the standard deviation of the non-zero values ​​in the initial data.

5. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S3 specifically includes: S31. Conventional hot air furnaces include but are not limited to two equipment states: furnace burning and air supply: When the hot blast furnace is in the burning state, it consumes gas, and the gas consumption data value is greater than 0; When the hot air furnace is in the air supply state, no gas is consumed, and the gas consumption data value is equal to 0; According to the changing characteristics of gas consumption of hot blast furnace, when the hot blast furnace changes from the burning state to the air supply state, the gas consumption data will suddenly change from a non-zero value to 0; When the hot air furnace changes from the air supply state to the furnace burning state, the gas consumption data will suddenly change from 0 to a non-zero value; S32, predicting the hot blast stove state transition point according to the change characteristic mark of the prediction data: When the gas consumption data values ​​for more than three consecutive time steps are all 0, the time index of the point where the first gas consumption data value is 0 is marked as the air supply switching point t w ; When the gas consumption data values ​​for more than three consecutive time steps are all non-zero values, the time index of the point where the first gas consumption data value is non-zero is marked as the furnace switching point t b .

6. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S4 specifically includes: S41. In actual production, when the hot air furnace is in the burning state, the burning signal data value will be displayed as 1, and when the burning is stopped, the burning signal data value will be displayed as 0; S42. At the same time, when the hot air furnace is in the air supply state, the air supply signal data value will be displayed as 1, and when the air supply is stopped, the air supply signal data value will be displayed as 0; S43, when the value of the furnace burning status signal received in real time changes from 0 to 1, mark this time index point as the furnace burning transition point T b ; S44, when the real-time received air supply status signal data changes from 0 to 1, mark this time index point as the air supply switching point T w .

7. A method for predicting gas consumption of a hot blast stove based on real-time correction of equipment status signals according to claim 1, characterized in that: Step S5 specifically includes: S51, every time the air supply switching point T w When it appears, calculate the air supply switching point T w The last air supply switching point t w The time difference Δt w : Δt=tT w aw w Where, t aw is the most recent air supply switching point t w ; S52, every time the furnace switching point T b When it appears, calculate the furnace switching point T b The last furnace switching point t b The difference Δt b : Δt b =t ab -T b Where, t ab The most recent furnace switching point t b ; S53, every time the air supply state conversion time difference is obtained, the furnace conversion point t b The time index of all future forecast values ​​is modified as follows: In the formula, t ac-wi is the corrected air supply switching point t w The time index of the i-th time step after t wi is the air supply switching point before correction t w The time index of the i-th time step after t wi That is t w , when i=n, t wi The time index of the last predicted value at the current moment; S54, whenever the furnace state conversion time difference is obtained, the nearest furnace conversion point t b The time indexes of all predicted values ​​at the beginning are modified as follows: In the formula, t ac-bi is the corrected furnace switching point t b The time index of the i-th time step after t bi is the furnace switching point before correction b The time index of the i-th time step after t bi That is t b , when i=n, t bi The time index of the last predicted value at the current moment.

Images