An automatic annotation method and system for the production historical data of a blast furnace hot blast stove

Abstract

The present invention discloses an automatic annotation method and system for the production historical data of a blast furnace hot blast stove. First, zero-phase filtering is performed on the arch top temperature data to remove noise and eliminate data lag caused by filtering, which is convenient for determining the lag time in the next step. The lag time is determined according to the change curve of the arch top temperature with the valve position, and the arch top temperature data is shifted forward by the corresponding lag time for time alignment. Finally, the rising and falling states of the arch top temperature are annotated. By effectively preprocessing the data of the blast furnace hot blast stove, it is ensured that the data input into the model has good consistency and accuracy, thereby guaranteeing the accuracy, reliability, and stability of the subsequent model output. The problem of lag introduced by the filtering algorithm is solved, the problem of data annotation errors caused by data inertia and the problem of invalid annotation caused by insufficient in-depth combination with actual business are solved, and dynamic parameters are introduced to improve the adaptability of the annotation method.

Description

Technical Field

[0001] This application relates to the technical field of automatic annotation of blast furnace hot stove production historical data, and particularly to a method and system for automatic annotation of blast furnace hot stove production historical data. Background Art

[0002] Currently, for the production sites in domestic blast furnace ironmaking plants where flowmeter devices are prone to damage, flowmeters cannot be accurately calibrated, maintained, and serviced for a long time, and even there is no flowmeter, each automation control company has put forward its own solutions. Among them, the solution of predicting the output valve position based on the regression model of data analysis has been preliminarily applied. A key step in this is data preprocessing, where data is annotated to obtain the best stove burning data that can increase the dome temperature for model training. However, in the actual project usage process, the output of this solution is still unstable, such as inaccurate or jumping, and it cannot guarantee long-term operation. The reasons for this problem may be multi-factorial, but fundamentally, there are deficiencies in data preprocessing, the consistency of the input model data is poor, and there are many one-to-many situations (that is, one input data x corresponds to multiple outputs y), resulting in problems such as inaccurate output and jumping.

[0003] In response to the national call for carbon peak and carbon neutrality, more and more blast furnace ironmaking plants with flowmeters have carried out technological transformation and upgrading by purchasing optimized control systems to reduce energy consumption and pollutant emissions. The most crucial part of this is combustion optimization, which aims to achieve full combustion, energy conservation, and consumption reduction by optimizing the best ratio of gas and combustion-supporting gas. A key step in using data analysis technology to optimize the air-fuel ratio is also data preprocessing. Only by finding all the best stove burning data that can increase the dome temperature can we further analyze and find the optimal air-fuel ratio, and then guide the closed-loop control of the control system to achieve the optimization of the air-fuel ratio.

[0004] Taking the annotation of the dome temperature as an example, the existing processing method is to filter and smooth the historical dome temperature data, and then subtract the latter item from the former item. If the calculation result is greater than 0, it means the temperature rises; if it is less than 0, it means the temperature drops; if it is equal to 0, it means the temperature remains unchanged, so as to annotate the data.

[0005] The filtering operation of the existing processing method introduces new hysteresis. The vault temperature is time-series data, and its change is caused by the change in flow rate after the valve position operation or pressure change. Therefore, there must be a certain hysteresis in the vault temperature data relative to the valve position or pressure data. The existing processing method does not handle this problem, that is, it does not perform time alignment on the time-series data. There is a certain inertia in the vault temperature data, that is, at the end of the rise or fall of the vault temperature, in fact, the valve position or air-fuel ratio is already inappropriate, but due to inertia or the hysteresis of the response time, the vault temperature still maintains a small period of rise or fall. According to the existing processing method, incorrect labeling will occur, and at the same time, the data processing does not fully combine the actual situation of the project, and there are many invalid data. The data obtained by the existing preprocessing method will inevitably have many errors and invalid data, which will be catastrophic for model training. Summary of the Invention

[0006] Based on this, in view of the above technical problems, an automatic annotation method and system for the production historical data of a blast furnace hot blast stove are provided to solve the problem that the data obtained by the existing preprocessing method will inevitably have many errors and invalid data and introduce new hysteresis.

[0007] In a first aspect, an automatic annotation method for the production historical data of a blast furnace hot blast stove, the method includes:

[0008] Obtain target firing data, and perform forward filtering on the vault temperature of the firing data;

[0009] Arrange the target firing data in reverse order according to the time sequence to obtain reverse-order firing data, perform forward filtering on the vault temperature of the reverse-order firing data, and then arrange the vault temperature data in forward order according to the time;

[0010] Draw the relationship curves of the vault temperature with the gas valve position feedback, air valve position feedback, gas main pipe pressure, and air main pipe pressure measurement points. Find the parts where the vault temperature changes significantly with the valve position under the condition that the data of multiple other measurement points are basically unchanged. Determine the time intervals of each group of data according to a preset method, and obtain the hysteresis duration of the vault temperature relative to the gas valve position feedback by taking the average. Move the vault temperature data forward by the corresponding hysteresis duration according to the average time;

[0011] According to the target requirements, set the target window size, perform sliding window traversal annotation on the vault temperature to obtain the annotation result, and then use the target dynamic parameters according to the input service to further optimize the annotation result to obtain the optimized firing data.

[0012] In the above solution, optionally, the positive filtering of the vault temperature of the furnace data is specifically: using a Butterworth filter to perform positive filtering on the vault temperature of the target furnace data, where the target furnace data is a complete furnace data.

[0013] In the above solution, further optionally, the setting of the target window size according to the target requirement and the sliding window traversal annotation of the vault temperature to obtain the annotation result includes:

[0014] According to the actual situation, in response to the appropriate target window size set by the user, perform sliding window traversal annotation on the vault temperature to obtain the furnace data after the initial annotation.

[0015] In the above solution, further optionally, after performing sliding window traversal annotation on the vault temperature according to the actual situation in response to the appropriate target window size set by the user and obtaining the vault temperature data after the initial annotation, it further includes:

[0016] Dynamically constrain the quantiles of the vault temperature in the target furnace data and correct the annotation;

[0017] Dynamically constrain the average value of the vault temperature in the target furnace data and correct the annotation;

[0018] Dynamically constrain the quantiles of the furnace time in the target furnace data and correct the annotation to obtain the optimized furnace data.

[0019] In the second aspect, an automatic annotation system for the production historical data of a blast furnace hot stove, the system includes:

[0020] An acquisition module: used to acquire target furnace data and perform positive filtering on the vault temperature of the furnace data;

[0021] A filtering module: used to reverse the order of the target furnace data in chronological order to obtain reverse-order furnace data, perform positive filtering on the vault temperature of the reverse-order furnace data, and then arrange the vault temperature data in chronological order;

[0022] A processing module: used to draw the relationship curves of the vault temperature with the gas valve position feedback, air valve position feedback, gas main pipe pressure, and air main pipe pressure measurement points, find the parts where the vault temperature changes significantly with the valve position under the condition that the data of multiple other measurement points are basically unchanged, determine the time intervals of each group of data according to a preset method, obtain the lag duration of the vault temperature relative to the gas valve position feedback by taking the average value, and shift the vault temperature data forward by the corresponding lag duration according to the average time.

[0023] Annotation module: It is used to set the target window size according to the target requirements, perform sliding window traversal annotation on the vault temperature to obtain an annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized furnace burning data.

[0024] In the above solution, optionally, the forward filtering of the vault temperature of the furnace burning data is specifically: using a Butterworth filter to perform forward filtering on the vault temperature of the target furnace burning data, where the target furnace burning data is a complete furnace burning data.

[0025] In the above solution, further optionally, the setting of the target window size according to the target requirements and performing sliding window traversal annotation on the vault temperature to obtain an annotation result includes:

[0026] According to the actual situation, in response to the appropriate target window size set by the user, perform sliding window traversal annotation on the vault temperature to obtain the furnace burning data after the initial annotation.

[0027] In the above solution, further optionally, after obtaining the vault temperature data after the initial annotation by performing sliding window traversal annotation on the vault temperature in response to the appropriate target window size set by the user according to the actual situation, it further includes:

[0028] Dynamically constrain the quantiles of the vault temperature in the target furnace burning data and correct the annotation;

[0029] Dynamically constrain the average value of the vault temperature in the target furnace burning data and correct the annotation;

[0030] Dynamically constrain the quantiles of the furnace burning time in the target furnace burning data and correct the annotation to obtain the optimized furnace burning data.

[0031] In a third aspect, a computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0032] Obtain target furnace burning data and perform forward filtering on the vault temperature of the furnace burning data;

[0033] Arrange the target furnace burning data in reverse order according to time to obtain reverse-order furnace burning data, perform forward filtering on the vault temperature of the reverse-order furnace burning data, and then arrange the vault temperature data in forward order according to time;

[0034] Plot the relationship curves between the vault temperature and the feedback of the gas valve position, the feedback of the air valve position, the pressure of the gas main pipe, and the pressure of the air main pipe measurement points. Under the condition that the data of multiple other measurement points are basically unchanged, find the part where the vault temperature changes significantly with the valve position. Determine the time intervals of each group of data according to a preset method, obtain the lag duration of the vault temperature relative to the feedback of the gas valve position by taking the average value, and shift the vault temperature data forward by the corresponding lag duration according to the average value time;

[0035] According to the target requirements, set the target window size, perform sliding window traversal annotation on the vault temperature to obtain an annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized furnace burning data. Fourth aspect, a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0036] Obtain the target furnace burning data and perform forward filtering on the vault temperature of the furnace burning data;

[0037] Arrange the target furnace burning data in reverse order according to the time sequence to obtain reverse-order furnace burning data, perform forward filtering on the vault temperature of the reverse-order furnace burning data, and then arrange the vault temperature data in forward order according to the time;

[0038] Plot the relationship curves between the vault temperature and the feedback of the gas valve position, the feedback of the air valve position, the pressure of the gas main pipe, and the pressure of the air main pipe measurement points. Under the condition that the data of multiple other measurement points are basically unchanged, find the part where the vault temperature changes significantly with the valve position. Determine the time intervals of each group of data according to a preset method, obtain the lag duration of the vault temperature relative to the feedback of the gas valve position by taking the average value, and shift the vault temperature data forward by the corresponding lag duration according to the average value time;

[0039] According to the target requirements, set the target window size, perform sliding window traversal annotation on the vault temperature to obtain an annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized furnace burning data.

[0040] The present invention has at least the following beneficial effects:

[0041] Based on further analysis and research of the problems in the prior art, it is recognized that the data obtained by the existing preprocessing methods will inevitably have a large number of errors and invalid data, which will be catastrophic for model training. The present invention obtains target stove data and performs forward filtering on the vault temperature of the stove data; arranges the target stove data in reverse chronological order to obtain reverse-order stove data, performs forward filtering on the vault temperature of the reverse-order stove data, and then arranges the vault temperature data in forward chronological order; plots the relationship curves of the vault temperature with the gas valve position feedback, air valve position feedback, gas main pipe pressure, and air main pipe pressure measurement points, finds the parts where the vault temperature changes significantly with the valve position under the condition that the data of multiple other measurement points are basically unchanged, determines the time intervals of each group of data according to a preset method, obtains the lag duration of the vault temperature relative to the gas valve position feedback by taking the average value, and shifts the vault temperature data forward by the corresponding lag duration according to the average value time; according to the target requirements, sets the target window size, performs sliding window traversal annotation on the vault temperature to obtain an annotation result, and then further optimizes the annotation result using the target dynamic parameters according to the input service to obtain the optimized stove data.

[0042] By effectively preprocessing the data of the blast furnace hot blast stove, it is ensured that the data input into the model has good consistency and accuracy, thereby ensuring the accuracy, reliability, and stability of the subsequent model output, solving the problem of lag introduced by the filtering algorithm; solving the problem of lag in the vault temperature data itself; solving the problem of data annotation errors caused by data inertia and invalid annotation problems caused by insufficient combination with actual business, and introducing dynamic parameters to improve the adaptability of the annotation method. Brief Description of the Drawings

[0043] Figure 1 It is a schematic flow chart of the method for automatically annotating the production history data of the blast furnace hot blast stove provided by an embodiment of the present invention;

[0044] Figure 2 It is a schematic diagram for processing the filtering lag problem provided by an embodiment of the present invention;

[0045] Figure 3 It is a schematic diagram for processing the lag of the vault temperature data provided by an embodiment of the present invention;

[0046] Figure 4 It is a schematic diagram for annotating the vault temperature of the blast furnace hot blast stove provided by an embodiment of the present invention;

[0047] Figure 5 It is a second schematic flow chart of the method for automatically annotating the production history data of the blast furnace hot blast stove provided by an embodiment of the present invention;

[0048] Figure 6Schematic diagram of data sliding window annotation for blast furnace hot blast stove provided by an embodiment of the present invention;

[0049] Figure 7 Internal structure diagram of a computer device in an embodiment. Specific implementation manners

[0050] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] The automatic annotation method for blast furnace hot blast stove production historical data provided by the present application, as Figure 1 shown, includes the following steps:

[0052] Obtain target stove firing data, and perform forward filtering on the dome temperature of the stove firing data;

[0053] Arrange the target stove firing data in reverse chronological order to obtain reverse-ordered stove firing data, perform forward filtering on the dome temperature of the reverse-ordered stove firing data, and then arrange the dome temperature data in forward chronological order;

[0054] Draw the relationship curves between the dome temperature and the feedback of the gas valve position, the feedback of the air valve position, the pressure of the gas main pipe, and the pressure measuring point of the air main pipe. Find the parts where the dome temperature changes significantly with the valve position under the condition that the data of multiple other measuring points are basically unchanged. Determine the time intervals of each group of data according to a preset method, and obtain the lag duration of the dome temperature relative to the feedback of the gas valve position by taking the average. Move the dome temperature data forward by the corresponding lag duration according to the average time;

[0055] According to the target requirements, set the target window size, perform sliding window traversal annotation on the dome temperature to obtain an annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized stove firing data.

[0056] In an embodiment, the performing forward filtering on the dome temperature of the stove firing data specifically is: using a Butterworth filter to perform forward filtering on the dome temperature of the target stove firing data, where the target stove firing data is a complete stove firing data.

[0057] In an embodiment, the according to the target requirements, setting the target window size, and performing sliding window traversal annotation on the dome temperature to obtain an annotation result includes:

[0058] According to the actual situation, in response to the appropriate target window size set by the user, perform sliding window traversal annotation on the dome temperature to obtain the stove firing data after initial annotation.

[0059] In one embodiment, after obtaining the vault temperature data after initial annotation by performing a sliding window traversal annotation on the vault temperature according to the actual situation in response to a suitable target window size set by the user, the following steps are further included:

[0060] Dynamically constrain the quantiles of the vault temperature in the target stove firing data and correct the annotation;

[0061] Dynamically constrain the average value of the vault temperature in the target stove firing data and correct the annotation;

[0062] Dynamically constrain the quantiles of the stove firing time in the target stove firing data and correct the annotation to obtain optimized stove firing data.

[0063] In the above method for automatically annotating the production historical data of the blast furnace hot blast stove, effective data preprocessing is performed on the data of the blast furnace hot blast stove to ensure that the data input into the model has good consistency and accuracy, thereby ensuring the accuracy, reliability, and stability of the subsequent model output, solving the problem of lag introduced by the filtering algorithm; solving the problem of lag in the vault temperature data itself; solving the problems of data annotation errors caused by data inertia and invalid annotations caused by insufficient in-depth combination with actual business, and introducing dynamic parameters to improve the adaptability of the annotation method.

[0064] In one embodiment, as Figures 2 - 6 shown, the overall technical solution idea is as follows:

[0065] First, perform zero-phase filtering on the vault temperature data to remove noise and eliminate the data lag caused by filtering, facilitating the determination of the subsequent lag time; secondly, determine the lag time according to the change curve of the vault temperature with the valve position, and perform time alignment on the vault temperature data by shifting it forward by the corresponding lag time; finally, annotate the rising and falling states of the vault temperature.

[0066] In one embodiment, the implementation of the solution includes:

[0067] Zero-phase filtering: As Figure 2 shown in the schematic diagram for dealing with the filtering lag problem, generally, filtering operations on data or signals will produce a certain lag. The purpose of zero-phase filtering is to eliminate this lag. The specific operations taken include performing a forward filtering on the data once, and then using the same filtering method and parameters to perform a reverse filtering once (the black arrows in the figure represent the two filtering directions). The specific filtering parameters need to be adjusted according to the data situation.

[0068] Treatment of the lag of the vault temperature: The lag time of the vault temperature needs to be determined according to specific data. It can be determined by observing the position where the curve shows obvious fluctuations, such as Figure 3Schematic diagram. The black double arrow represents the lag time of the vault temperature. By counting multiple lag times in this way and taking the average value, the lag duration of the vault temperature can be obtained. Finally, the data is shifted accordingly to complete the time alignment operation.

[0069] Annotation of vault temperature: It is agreed that when the air-fuel ratio is appropriate and the vault temperature rises, Label = 1; when the air-fuel ratio is inappropriate and the vault temperature drops, Label = -1; when it is impossible to fully determine whether the air-fuel ratio is appropriate, Label = 0 is marked at that position.

[0070] Taking the firing data of each furnace as a unit, the data is traversed using a sliding window method. As Figure 6 shown, a window of size 5 is used to traverse and annotate the data in sequence. The specific operations are as follows:

[0071] Taking the current data as the center of the window, it extends equidistantly forward and backward until the window size is 5;

[0072] Judging the data within the window, if all are in ascending order, the data at the current position is marked as 1; if all are in descending order, the data at the current position is marked as -1; if there are both ascending and descending or the window length is insufficient, it is marked as 0. This can avoid incorrect annotation due to the inertia of temperature data.

[0073] To further improve the self-adaptability of this method, reduce the negative impact of inappropriate parameter settings on the data annotation results, and achieve dynamic constraints. Based on the previous step, the 0.5 quantiles "p" and "n" of all data marked as 1 and -1 are calculated. The data marked as 1 but with a vault temperature less than "p" is re-marked as 0, and the data marked as -1 but with a vault temperature greater than "n" is re-marked as 0;

[0074] Combined with the actual business situation, for all data currently marked as 1, all those with a vault temperature less than the average temperature are marked as 0;

[0075] Combined with the actual firing of the hot blast stove, the availability of the data in the first small period and near the end of the firing is relatively weak, which is not conducive to the subsequent training of the model. However, the firing time of each furnace is also not fixed in actual production. Therefore, the data before the 0.15 quantile of the firing time and the data after the 0.95 quantile of the firing time are re-marked as 0; The finally marked data is as Figure 4 shown.

[0076] In one embodiment, the zero-phase filtering: The vault temperature of a complete firing data is filtered forward using a Butterworth filter, then the data is arranged in reverse order by time, filtered using the same filtering method and parameters, and then the vault temperature data is arranged in forward order by time, thus completing the zero-phase filtering and solving the problem of lag introduced by the filtering algorithm;

[0077] In one embodiment, for the vault temperature lag processing: draw curves of the vault temperature against measuring points such as the gas valve position feedback, air valve position feedback, main gas pipe pressure, and main air pipe pressure. Find the parts where the vault temperature changes significantly with the valve position under the condition that the data of multiple other measuring points are basically unchanged. Determine the time intervals for each group according to the above method, and calculate the average value to obtain the lag duration of the vault temperature relative to the gas valve position feedback. Finally, shift the vault temperature data forward by the corresponding lag duration in time to complete the processing of the vault temperature lag, achieving data alignment and solving the lag problem existing in the time series data itself.

[0078] This embodiment fully considers the characteristics of industrial time-series data and proposes a preprocessing process and method for solving the data lag problem. Moreover, by deeply integrating data preprocessing with actual production, a data automatic annotation method is proposed, further improving the accuracy, effectiveness, and consistency of the data preprocessing results.

[0079] This embodiment solves the problem of lag introduced by the filtering algorithm, solves the lag problem existing in the time series data itself, solves the problem of incorrect data annotation caused by data inertia and the problem of ineffective annotation caused by insufficient in-depth combination with actual business, and introduces dynamic parameters to improve the adaptability of the annotation method.

[0080] In one embodiment, for the vault temperature annotation: according to the actual situation and the implementation method of the above solution, set an appropriate window size and perform sliding window traversal annotation on the vault temperature to solve the problem of incorrect annotation caused by data inertia. Then, in combination with the actual business, use dynamic parameters to further optimize the annotation results: change the data marked as 1 but with the vault temperature less than "p" to 0, change the data marked as -1 but with the vault temperature greater than "n" to 0, change the data marked as 1 but with the vault temperature less than the average vault temperature to 0, change the data marked as -1 but with the vault temperature higher than a certain threshold to 0, and also change the data at the initial and final stages of the furnace burning time to 0 to ensure good consistency of the data.

[0081] It should be understood that although Figure 1 the steps in the flowchart Figure 1At least some of the steps may include multiple steps or multiple stages, which do not necessarily need to be executed and completed at the same moment, but can be executed at different moments, and the execution order of these steps or stages does not necessarily need to be sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0082] In one embodiment, an automatic annotation system for blast furnace hot stove production historical data is provided, including the following program modules:

[0083] Acquisition module: used to acquire target firing data and perform forward filtering on the vault temperature of the firing data;

[0084] Filtering module: used to reverse the order of the target firing data according to the time sequence to obtain reverse-order firing data, perform forward filtering on the vault temperature of the reverse-order firing data, and then arrange the vault temperature data in the forward order of time;

[0085] Processing module: used to draw the relationship curves between the vault temperature and the gas valve position feedback, air valve position feedback, gas main pipe pressure, and air main pipe pressure measurement points, find the parts where the vault temperature changes significantly with the valve position under the condition that the data of multiple other measurement points are basically unchanged, determine the time intervals of each group of data according to a preset method, obtain the lag duration of the vault temperature relative to the gas valve position feedback by taking the average value, and shift the vault temperature data forward by the corresponding lag duration according to the average time;

[0086] Annotation module: used to set a target window size according to the target requirements, perform sliding window traversal annotation on the vault temperature to obtain an annotation result, and then further optimize the annotation result using target dynamic parameters according to the input service to obtain optimized firing data.

[0087] In one embodiment, the forward filtering of the vault temperature of the firing data is specifically: using a Butterworth filter to perform forward filtering on the vault temperature of the target firing data, where the target firing data is a complete firing data.

[0088] In one embodiment, the setting of the target window size according to the target requirements and the performing of sliding window traversal annotation on the vault temperature to obtain an annotation result include:

[0089] According to the actual situation, in response to the appropriate target window size set by the user, perform sliding window traversal annotation on the vault temperature to obtain the firing data after initial annotation.

[0090] In one embodiment, after obtaining the vault temperature data after the initial annotation by performing a sliding window traversal annotation on the vault temperature according to the actual situation in response to a suitable target window size set by the user, the following steps are further included:

[0091] Dynamically constrain the quantiles of the vault temperature in the target furnace operation data and correct the annotation;

[0092] Dynamically constrain the average value of the vault temperature in the target furnace operation data and correct the annotation;

[0093] Dynamically constrain the quantiles of the furnace operation time in the target furnace operation data and correct the annotation to obtain optimized furnace operation data.

[0094] For the specific limitations of the automatic annotation system for blast furnace hot stove production historical data, reference can be made to the limitations of the automatic annotation method for blast furnace hot stove production historical data described above, which will not be elaborated here. Each module in the above automatic annotation system for blast furnace hot stove production historical data can be implemented in whole or in part by software, hardware, and their combinations. The above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form so that the processor can call and execute the operations corresponding to the above modules.

[0095] In one embodiment, a computer device is provided. This computer device can be a terminal, and its internal structure diagram can be as shown in Figure 7 The figure shows. The computer device includes a processor, a memory, a communication interface, a display screen, and an input system connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WI FI, a carrier network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it realizes an automatic annotation method for blast furnace hot stove production historical data. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input system of the computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad set on the shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0096] Those skilled in the art can understand that Figure 7The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0097] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, which involves all or part of the processes in the method of the above embodiment.

[0098] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which involves all or part of the processes in the method of the above embodiment.

[0099] Those of ordinary skill in the art can understand that all or part of the processes in the above embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0100] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0101] The above-described embodiments only represent several implementation manners of this application, and their descriptions are relatively specific and detailed, but they should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of the patent of this application should be subject to the appended claims.

Claims

1. An automatic annotation method for the production historical data of a blast furnace hot blast stove, characterized in that, the method includes: Obtain the target firing data and perform forward filtering on the arch top temperature of the firing data; Arrange the target firing data in reverse chronological order to obtain reverse-order firing data, perform forward filtering on the arch top temperature of the reverse-order firing data, and then arrange the arch top temperature data in forward chronological order; Draw the relationship curves of the arch top temperature with the gas valve position feedback, air valve position feedback, gas main pipe pressure, and air main pipe pressure measurement points. Find the part where the arch top temperature changes with the gas valve position under the condition that the data of multiple other measurement points remain unchanged. Determine the time interval of the part where the arch top temperature changes with the gas valve position according to a preset method. Obtain the lag duration by calculating the mean value of multiple time intervals, and shift the arch top temperature data forward by the corresponding lag duration according to the mean time; According to the target requirements, set the target window size, perform sliding window traversal annotation on the arch top temperature to obtain the annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized firing data; Specifically, according to the actual situation, in response to the target window size set by the user, perform sliding window traversal annotation on the arch top temperature to obtain the firing data after the initial annotation. Judge the data within the window. If all are in ascending order, label the current position data as 1. If all are in descending order, label the current position data as -1. If there are both ascending and descending or the window length is insufficient, label it as 0; and calculate the 0.5 quantiles "p" and "n" of all the data labeled as 1 and labeled as -1. Among them, it is agreed that when the air-fuel ratio is correct and the arch top temperature rises, label = 1; when the air-fuel ratio is incorrect and the arch top temperature drops, label = -1; and label = 0 for the positions where the air-fuel ratio cannot be completely determined; The specific optimization process includes: changing the data labeled as 1 but with an arch top temperature less than "p" to 0, changing the data labeled as -1 but with an arch top temperature greater than "n" to 0, changing the data labeled as 1 but with an arch top temperature less than the average arch top temperature to 0, changing the data labeled as -1 but with an arch top temperature higher than a certain threshold to 0, and also changing the data at the beginning and end of the firing time to 0.

2. The method according to claim 1, characterized in that, The forward filtering of the arch top temperature of the firing data is specifically: using a Butterworth filter to perform forward filtering on the arch top temperature of the target firing data, where the target firing data is a complete firing data.

3. An automatic annotation system for the production historical data of a blast furnace hot blast stove, characterized in that, the system includes: An acquisition module: used to acquire the target firing data and perform forward filtering on the arch top temperature of the firing data; A filtering module: used to arrange the target firing data in reverse chronological order to obtain reverse-order firing data, perform forward filtering on the arch top temperature of the reverse-order firing data, and then arrange the arch top temperature data in forward chronological order; Processing module: It is used to draw the relationship curves of the vault temperature with the feedback of the gas valve position, the feedback of the air valve position, the pressure of the gas main pipe, and the pressure measurement points of the air main pipe, find the part where the vault temperature changes with the gas valve position under the condition that the data of multiple other measurement points remain unchanged, determine the time interval of the part where the vault temperature changes with the gas valve position according to a preset method, obtain the lag duration by calculating the mean value of multiple time intervals, and shift the vault temperature data forward by the corresponding lag duration according to the mean time. Annotation module: It is used to set the target window size according to the target requirements, perform sliding window traversal annotation on the vault temperature to obtain the annotation result, and then further optimize the annotation result using the target dynamic parameters according to the input service to obtain the optimized furnace burning data. Specifically, according to the actual situation, in response to the target window size set by the user, perform sliding window traversal annotation on the vault temperature to obtain the furnace burning data after the initial annotation. Judge the data within the window. If all are in ascending order, label the data at the current position as 1. If all are in descending order, label the data at the current position as -1. If there are both ascending and descending or the window length is insufficient, label it as 0. And calculate the 0.5 quantiles "p" and "n" of all the data labeled as 1 and -1. Here, it is agreed that when the air-fuel ratio is correct and the vault temperature rises, Label = 1; when the air-fuel ratio is incorrect and the vault temperature drops, Label = -1; and the position where the air-fuel ratio cannot be completely determined is labeled as Label = 0. The specific optimization process includes: changing the data labeled as 1 but with the vault temperature less than "p" to 0, changing the data labeled as -1 but with the vault temperature greater than "n" to 0, changing the data labeled as 1 but with the vault temperature less than the average vault temperature to 0, changing the data labeled as -1 but with the vault temperature higher than a certain threshold to 0, and also changing the data at the beginning and end of the furnace burning time to 0.

4. The system according to claim 3, wherein, The forward filtering of the vault temperature of the furnace burning data is specifically: using a Butterworth filter to perform forward filtering on the vault temperature of the target furnace burning data, where the target furnace burning data is a complete furnace burning data.

5. A computer device, including a memory and a processor, the memory stores a computer program, wherein, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 2.

6. A computer-readable storage medium, on which a computer program is stored, wherein, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 2.

Images