Splash state monitoring method and system based on audio

Abstract

The invention relates to an audio-based slag splashing state monitoring method and system, and the method comprises the steps: collecting real-time data needed by model calculation in a slag splashing process after the starting of the slag splashing of a converter; analyzing and processing the collected real-time data; setting a slag rising threshold value, observing and learning sound intensity time sequence characteristics during slag rising, establishing a slag rising point automatic detection model, and determining a slag rising point; establishing a slag splashing early warning point detection model, and determining an early warning point; and establishing a slag splashing endpoint detection model, determining a slag splashing endpoint, and realizing automatic forecasting of the slag splashing endpoint. According to the method, the audio signal in the slag splashing process of the converter is monitored in real time, the audio time sequence characteristics are extracted, and the detection of the slag starting point and the accurate forecasting of the slag splashing end point are realized in combination with a machine learning algorithm and an end point prediction technology. According to the method, the production efficiency in the slag splashing process of the converter is improved, the manual intervention cost is reduced, the negative influence on the service life of the converter due to overlong slag splashing time is avoided, and a powerful technical support is provided for continuous progress of the steelmaking industry.

Description

Technical Field

[0001] The invention relates to the technical field of converter steelmaking, and in particular to an audio-based slag splashing state monitoring method and system. Background Art

[0002] The slag splashing technology is a major advancement in converter protection technology. The use of converter slag splashing technology can not only effectively improve the protection effect of the furnace lining and increase the service life of the converter, but also significantly reduce the consumption of refractory materials, which is of great significance for improving the economy and reliability of steelmaking production.

[0003] At present, domestic and foreign steel enterprises have adopted a series of measures and means such as static control models to optimize the slag splashing furnace protection operation. However, these methods cannot achieve real-time monitoring of the entire slag splashing process and automatic converter slag splashing endpoint prediction. Traditionally, it is judged based on the personal experience of the operator, which increases the cost and labor intensity of manual intervention. According to the statistics of the slag splashing endpoint time of the steel plant converter, the proportion of furnaces with too long or too short slag splashing time due to improper operation is larger. On the one hand, too long slag splashing time will increase the erosion of the inner wall of the converter and shorten the service life of the converter. In order to extend the service life of the converter, a large number of tile repairs are required to maintain the furnace condition, which wastes a lot of manpower and reduces production efficiency; on the other hand, too short slag splashing time may lead to poor slag quality and affect the final quality of the product. According to field experience and industry knowledge, the slag starting time in the slag splashing process has a key influence on the slag splashing duration, and the change of sound can reflect the slag state in real time.

[0004] Therefore, how to realize automatic prediction of converter slag splashing endpoint becomes a key challenge of slag splashing furnace protection technology. Summary of the invention

[0005] In order to solve at least one of the above technical problems, the present invention provides an audio-based slag splashing state monitoring method, comprising: Collect historical data of the slag splashing process, including audio data, gun height data and slag adjusting agent data; and mark the slag starting point, warning point and slag splashing end point of the historical data; According to the slag starting point and previous historical data, a slag starting point detection model is established; According to the historical data from the slag starting point to the early warning point, a slag splashing early warning point detection model is established; According to the historical data from the early warning point to the slag splashing endpoint, a slag splashing endpoint detection model is established; Collect real-time data of the slag splashing process, including audio data, gun height data and slag adjusting agent data. First, input the slag starting point detection model to predict the slag starting point; then input the real-time data after the slag starting point into the slag splashing warning point detection model to predict the warning point; then input the real-time data after the warning point into the slag splashing endpoint detection model to predict the slag splashing endpoint.

[0006] Furthermore, collecting historical data of the slag splashing process also includes: generating a sound intensity change curve reflecting the slag splashing process according to the audio data, the gun height data and the slag adjusting agent data; specifically: Extract the time domain and frequency domain features of audio data, and select the audio channels that effectively describe the slag splashing process from multiple frequency bands; According to the selected audio channel value, combined with the data sampling rate, the average value of the sound intensity data per second is calculated as the instantaneous sound intensity value at the current moment. ; According to the gun height data and slag adjusting agent data, the gun position compensation value and slag adjusting agent compensation value are calculated to compensate for the instantaneous sound intensity value, and the compensated instantaneous sound intensity value is obtained. , to generate a sound intensity change curve.

[0007] Furthermore, the gun position compensation value , through the formula ) calculation, where is the gun position compensation coefficient, is the instantaneous gun height, the initial value of the gun position ; Slag adjustment agent compensation value , through the formula Calculate, where is the compensation coefficient of slag conditioning agent, Δm(t) is the instantaneous change of slag adjusting agent, is the real-time correction coefficient, Collect the actual amount of steel added during tapping; the instantaneous sound intensity value after compensation , through the formula calculate.

[0008] Furthermore, the instantaneous sound intensity after compensation is calculated to be After that, it also includes: S141: Set the moving window to calculate the instantaneous sound intensity value The sliding average ; S142: Set the exponentially weighted moving average weighting factor ewmaAlpha to assign the instantaneous sound intensity value at the current moment Higher weight; S143: Calculate the final instantaneous sound intensity value according to the instantaneous sound intensity value, the sliding average value and the exponentially weighted moving average weighting factor , and regenerate the sound intensity change curve.

[0009] Furthermore, the final instantaneous sound intensity value , through the formula calculate.

[0010] Furthermore, the steps of establishing a slag point detection model include: According to the slag starting point and previous historical data, the slag starting parameter thresholds of different furnace slag starting points are counted, including any one or more of: change in gun height, sound intensity and sudden change, trend slope range of sound intensity curve, movement of frequency peak, amplitude change of sound waveform in time domain, and slag starting duration; A slag point detection model is constructed with slag parameter threshold and historical data as input and slag point as output, which is used to learn the first sound intensity time series characteristics; Through multi-furnace learning based on historical data, the parameter settings of the slag point detection model are fine-tuned to obtain the trained slag point detection model, and the slag point is automatically predicted.

[0011] Furthermore, the step of establishing a slag splash warning point detection model includes: The instantaneous sound intensity values ​​from the slag starting point to the warning point of different furnaces are counted to construct a time series data set of instantaneous sound intensity; A supervised deep learning time series prediction model is constructed, the instantaneous sound intensity time series dataset is input into the time series prediction model, the second audio time series features are learned, and the trained time series prediction model is obtained as a slag splash warning point detection model to automatically predict the warning point.

[0012] Furthermore, the step of establishing a slag splashing endpoint detection model includes: According to the historical data from the early warning point to the slag splashing endpoint, the endpoint parameter thresholds of different furnaces are counted, including: the intensity and increase and decrease speed of the slag splashing endpoint sound, the trend slope range of the sound intensity curve, the frequency distribution and the movement of the peak value, the amplitude change and energy distribution of the sound waveform in the time domain, and any one or more of the slag splashing duration; A slag splashing endpoint detection model is constructed with endpoint parameter threshold and historical data as input and slag splashing endpoint as output, which is used to learn the third tone intensity time series characteristics; The historical data is divided into different slag splashing modes, different slag splashing modes are learned, and the parameter settings of the slag splashing endpoint inspection model are fine-tuned to obtain the trained slag splashing endpoint detection model to automatically predict the slag splashing endpoint.

[0013] On the other hand, the present invention also provides an audio-based slag splashing state monitoring system, including: an audio data acquisition device, a multi-band audio analyzer, a PLC data acquisition device and an audio slag-making industrial computer, which is used to implement any of the above-mentioned slag splashing state monitoring methods.

[0014] Furthermore, the audio data acquisition device includes an audio sensor, which is arranged on the outside of the converter fire retaining wall and is used to collect audio signals of nitrogen impacting the slag during slag splashing.

[0015] The present invention provides an audio-based slag splashing status monitoring method and system, which can realize the whole process monitoring and endpoint prediction of the slag splashing process, so as to solve the technical problems that the slag splashing time is too long or too short, affecting the life of the converter and the production efficiency, the slag splashing effect is poor, and the cost of manual intervention is high, which is caused by excessive reliance on the experience and intuition of operators in the slag splashing furnace protection process. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of an embodiment of a method for monitoring slag splashing status based on audio according to the present invention; Figure 2 It is a structural schematic diagram of an embodiment of a prediction model of a slag splashing state monitoring method based on audio in the present invention; Figure 3 The present invention is a schematic structural diagram of an embodiment of an audio-based slag splashing status monitoring system of the present invention. DETAILED DESCRIPTION

[0017] 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.

[0018] It should be noted that if the embodiments of the present invention involve directional indications, such as up, down, left, right, front, back, etc., then the directional indication is only used to explain the relative position relationship, movement status, etc. between the components in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly. In addition, if the embodiments of the present invention involve descriptions of "first, second", "S1, S2", "step one, step two", etc., then such descriptions are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of indicated technical features or indicating the execution order of the method, etc. Those skilled in the art can understand that anything that does not violate the gist of the invention under the technical concept of the invention should be included in the protection scope of the present invention.

[0019] like Figure 1 -2, the present invention provides a method for monitoring slag splashing status based on audio, comprising: S1: Collect historical data of the slag splashing process, including audio data, gun height data and slag adjustment agent data; and mark the slag starting point of the historical data , early warning points and splashing end point ; Specifically, for step S1, Figure 3As shown, a slag splashing state monitoring system based on audio is provided, which may optionally include: an audio data acquisition device A, a multi-band audio analyzer B, a PLC data acquisition device C and an audio slag-melting industrial computer D.

[0020] The audio data acquisition device A may include audio sensors, sound collection tubes, etc. The intelligent audio sensor is installed on the outside of the converter fire wall, and the sound collection tube in the intelligent purge protection box collects audio signals during slag splashing, such as the audio signals emitted by nitrogen impacting the slag. The multi-band audio analyzer B is responsible for pre-processing the collected original audio signals, converting the signals into channel data of multiple frequency bands, reducing noise interference, and thus extracting useful slag splashing information. More specifically, the audio data may include audio intensity values, and the optional unit sampling rate is 8192. Through the audio acquisition device, all audio signals of nitrogen impacting the slag during slag splashing are collected, and the multi-band audio analyzer is used to obtain the original audio intensity values ​​after frequency division in real time. The center frequencies of different audio intensity values ​​include: 107HZ, 130HZ, 155HZ, 190HZ, 225HZ, and 267HZ.

[0021] The PLC data acquisition device C may include a liquid level meter, a weight sensor, etc., which is responsible for accurately collecting and transmitting the gun height data and slag adjusting agent data in the converter, ensuring the real-time monitoring and accurate recording of these key process parameters, and providing data support for the control and optimization of the slag splashing process. More specifically, the gun height data may be the gun position height, that is, the height difference between the oxygen gun and the furnace bottom; the slag adjusting agent data may include the type of slag adjusting agent, the addition time and weight of each slag adjusting agent, etc.; for example, the types of slag adjusting agents include: limestone, dolomite, iron oxide, lime, return ore, and iron powder briquette.

[0022] The audio slag-removing industrial computer D is the core of the entire system. It is responsible for real-time monitoring and analysis of audio and other data, implementing the monitoring method, displaying the slag-initiating point detected later, and automatically predicting the early warning point and the slag-splattering end point based on the analysis results. Through the intelligent processing and analysis of the industrial computer, the operator can more intuitively understand the status of the slag-splattering process and make more accurate operational decisions.

[0023] More specifically, for historical data, it is possible to observe and analyze historical data of at least 100 furnaces covering all shifts, and mark the actual slagging time based on the changing characteristics of audio, gun height and other data at each stage, combined with the experience of on-site operators, flame images, etc. The slag starting point; the time when the gun position reaches the lowest point at the end of slag splashing, that is, the start time of slag splashing warning , is the warning point; mark the time corresponding to the end point of slag splashing , is the slag splashing endpoint; after marking, the historical data is divided into three parts.

[0024] Preferably, step S1 further includes: generating a sound intensity change curve reflecting the slag splashing process according to the audio data, the gun height data and the slag adjusting agent data; Specifically, for the subsequent steps S2-S4, in order to better learn the audio timing characteristics, step S1 can also optionally perform one or more processing processes such as frequency division data analysis, instantaneous sound intensity noise reduction processing, curve dynamic compensation processing, and sound intensity curve weighted processing on the acquired original audio data based on the audio data, gun height data, and slag adjustment agent data, so as to more effectively and accurately describe the audio changes.

[0025] Preferably, it includes: S11: Extract the time domain and frequency domain characteristics of the audio data, and select the audio channels that effectively describe the slag splashing process from multiple frequency bands; specifically, the frequency division data analysis function needs to observe the time domain and frequency domain characteristics of the sound intensity data in the slag splashing stage, and select the audio channels that effectively and accurately describe the slag splashing process from multiple frequency bands. The data of these channels will serve as the basis for drawing the sound intensity curve. Specifically, when the slag starts to splash, it is affected by the slag temperature and the final slag composition, and the sound is relatively soft. As the high-pressure nitrogen cools the slag, the gun position drops, the slag aggregates, and the thickness decreases, the sound in the furnace will gradually increase to the maximum.

[0026] S12: According to the selected audio channel value and the data sampling rate, the average value of the sound intensity data per second is calculated as the instantaneous sound intensity value at the current moment. Specifically, the instantaneous sound intensity noise reduction processing function is to reduce the impact of background noise on the signal and enhance the stability of the signal.

[0027] S13: Calculate the gun position compensation value based on the gun height data and slag adjustment agent data and slag adjustment agent compensation value , to compensate for the instantaneous sound intensity The instantaneous sound intensity after compensation is , to generate a sound intensity change curve.

[0028] Specifically, the curve dynamic compensation function needs to calculate the gun position compensation value in real time according to the change of gun position during slag splashing and the time and weight of slag adjusting agent added. and slag adjustment agent compensation value Optionally, on the one hand, the real-time gun position compensation value is calculated by the formula ) calculation, where is the gun position compensation coefficient, is the instantaneous gun height, the initial value of the gun position , according to the three sub-mode setting strategies of furnace bottom height (low furnace bottom, normal furnace bottom and high furnace bottom), the real-time slag adjustment agent compensation value is adjusted by formula Calculate, where is the compensation coefficient of slag conditioning agent, Δm(t) The instantaneous change of slag adjusting agent. For the newly added slag adjusting agent modification function in the steel-making process, the actual addition amount during steel-making is collected. , and set the corresponding real-time correction coefficient , the correction coefficient is related to the type of slag conditioning agent, specific furnace conditions, environment, etc. According to the above formula, the instantaneous sound intensity after compensation is .

[0029] More preferably, the instantaneous sound intensity value after compensation obtained in step S13 is After that, it also includes: S141: Set the moving window to calculate the instantaneous sound intensity value The sliding average ; S142: Set the exponentially weighted moving average weighting factor ewmaAlpha to assign the instantaneous sound intensity value at the current moment Higher weight; S143: Calculate the final instantaneous sound intensity value according to the instantaneous sound intensity value, the sliding average value and the exponentially weighted moving average weighting factor , and regenerate the sound intensity change curve.

[0030] Specifically, in order to further utilize the sound intensity time series data, extract its short-term and long-term trend characteristics, which helps to smooth the random fluctuations of the slag splashing sound intensity in the short term and help identify potential changes in the sound intensity trend, so as to meet the trend curve analysis needs of audio-guided slag splashing furnace protection. Optionally, first, set a moving window size between 5 and 10 and calculate the sliding average of the instantaneous sound intensity. ; Secondly, set the exponentially weighted moving average weighting factor ewmaAlpha in the range of 0.8 to 0.9 to give the current moment sound intensity data a higher weight; finally, according to the formula , calculate the final instantaneous sound intensity value, which is used to describe the audio trend curve of slag splashing.

[0031] S2: Establish a slag point detection model based on the slag point and previous historical data; Specifically, it is possible to choose to establish a slag point detection model by learning the time series characteristics of sound intensity within the period of time through a neural network method based on the slag point and previous historical data; the specific model structure can adopt any method of neural network, such as convolutional neural network, etc.

[0032] Preferably, step S2 comprises: S21: According to the slag starting point and the previous historical data, the threshold values ​​of slag starting parameters at different furnaces are counted, including but not limited to: any one or more of the change in gun height, the magnitude and mutation of sound intensity, the trend slope range of sound intensity curve, the movement of frequency peak, the amplitude change of sound waveform in the time domain, and the slag starting time; S22: construct a slag starting point detection model with slag starting parameter threshold and historical data as input and slag starting point as output, so as to learn the first sound intensity time series feature; S23: Perform multi-furnace learning through historical data, fine-tune the parameter settings of the slag initiation point detection model, obtain the trained slag initiation point detection model, and automatically predict the slag initiation point.

[0033] Preferably, step S2 further includes screening the historical data according to the slag parameter threshold, for example eliminating abnormal heats including slag overoxidation, and then performing multi-heat learning.

[0034] S3: Establish a slag splash warning point detection model based on the historical data from the slag starting point to the warning point; Specifically, it is possible to choose to establish a slag splash warning point detection model based on historical data from the slag initiation point to the warning point, and similarly learn the time series characteristics of sound intensity during this period through a neural network method; the specific model structure can adopt any method of a neural network.

[0035] Preferably, step S3 includes: S31: Count the instantaneous sound intensity values ​​from the slag starting point to the warning point of different furnaces, and construct an instantaneous sound intensity time series data set; S32: construct a supervised deep learning time series prediction model, input the instantaneous sound intensity time series data set into the time series prediction model, learn the second audio time series features, obtain the trained time series prediction model as a slag splash warning point detection model, and automatically predict the warning point; Specifically, you can choose to observe and analyze at least 100 furnaces covering all shifts, and mark the time when the gun position reaches the lowest range at the end of slag splashing. , as the start time of slag splashing warning. To the warning point The instantaneous sound intensity value during this period of time As a time series dataset for training the slag splash warning point detection model, a supervised deep learning time series prediction method is designed to automatically learn the corresponding audio time series features. , and extract the audio timing features when the slag splashing ends, and dynamically detect the start signal of the slag splashing warning.

[0036] S4: Establish a slag splashing endpoint detection model based on historical data from the warning point to the slag splashing endpoint; Specifically, similar to steps S2 and S3, it is possible to optionally establish a slag splashing endpoint detection model by learning the time series characteristics of sound intensity within this period of time based on the historical data from the warning point to the slag splashing endpoint through a neural network method; the specific model structure can adopt any method of neural network, such as convolutional neural network, etc.

[0037] Preferably, step S4 comprises: S41: Based on the historical data from the warning point to the slag splashing endpoint, the endpoint parameter thresholds of different furnaces are counted, including but not limited to: any one or more of the sound intensity and increase and decrease speed of the slag splashing endpoint, the trend slope range of the sound intensity curve, the frequency distribution and the movement of the peak value, the amplitude change and energy distribution of the sound waveform in the time domain, and the slag splashing duration; S42: constructing a slag splashing endpoint detection model with endpoint parameter threshold and historical data as input and slag splashing endpoint as output, for learning the third tone intensity time series feature; S43: Divide the historical data into different slag splashing modes, conduct learning of different slag splashing modes, fine-tune the parameter settings of the slag splashing endpoint inspection model, obtain the trained slag splashing endpoint additional measurement model, and automatically predict the slag splashing endpoint.

[0038] Specifically, real-time collection from The instantaneous sound intensity value from the beginning to the current moment is used to extract the sound intensity characteristics under different slag splashing modes (lowering furnace bottom, rising furnace bottom, protecting the molten pool). Then, the endpoint threshold is set based on the on-site operation experience and the audio change characteristics at the end of slag splashing. The endpoint threshold that needs to be fine-tuned for different converters is finally predicted based on the sound intensity time series characteristics and the satisfaction of the endpoint threshold. .

[0039] S5: Collect real-time data of the slag splashing process, including audio data, gun height data and slag adjusting agent data. First, input the slag starting point detection model to predict the slag starting point; then input the real-time data after the slag starting point into the slag splashing early warning point detection model to predict the early warning point; then input the real-time data after the early warning point into the slag splashing endpoint detection model to predict the slag splashing endpoint.

[0040] Specifically, real-time monitoring is performed to collect actual data of the slag splashing process. First, the previously trained slag initiation point detection model is input, combined with the slag initiation threshold, and the slag initiation point is predicted. Then, the subsequent data is input into the previously trained slag splashing early warning point detection model. After the early warning point is predicted, the subsequent data is input into the previously trained slag splashing endpoint detection model, combined with the endpoint threshold, to predict the slag splashing endpoint.

[0041] In this embodiment, a slag splashing state monitoring method based on audio of the present invention is provided, which divides converter monitoring into three stages: slag starting stage, early warning stage and endpoint stage, mainly including slag starting point detection model, slag splashing early warning point detection model and slag splashing endpoint detection model, and also involves the original audio corresponding to the slag splashing time, gun height, limestone, dolomite, iron oxide scale, lime, return ore, iron powder press ball adding time and weight and other key process parameters. 1. Slag starting point is a key turning point in the slag splashing process, and its accurate detection is crucial to ensure the accuracy of slag splashing endpoint prediction. Because before the slag initiation point, the audio data contains more noise and interference, and the extraction of audio timing features is more difficult. After the slag initiation point, the quality of the audio timing features is significantly improved, becoming clearer and more useful; setting up a slag initiation point detection model before this helps to obtain more accurate audio timing features, reduce data noise interference, and improve the quality and accuracy of feature extraction. The optimization of these features helps to improve the prediction accuracy and robustness of the entire slag splashing endpoint prediction model, thereby providing operators with more precise operating guidance, thereby optimizing the slag splashing process and improving production efficiency; in the preferred scheme, the precise detection of the slag initiation point and the analysis of the audio timing features, the slag splashing sound intensity change curve can more accurately reflect the state of the slag splashing process, quickly respond to changes in the slag splashing process, improve the stability and production efficiency of the slag splashing process, reduce manual observation and judgment, and thus improve the accuracy of the endpoint prediction. 2. The warning point is the time point when the gun position reaches the lowest range at the end of slag splashing. After that, the gun position will not change much. Before the endpoint prediction, adding a warning point and constructing a slag splashing warning detection model will help reduce the misjudgment of the endpoint during the gun lowering process and enhance the robustness of the model; 3. The slag splashing endpoint is the slag splashing termination point. Analyzing the audio timing characteristics after the warning point, establishing a slag splashing endpoint detection model, and determining the slag splashing endpoint can improve the hit rate of the slag splashing endpoint prediction and avoid unnecessary impact and damage to the furnace lining, thereby extending the service life of the furnace lining and reducing labor costs and operational risks.

[0042] In summary, the present invention provides a method and device for automatically predicting the end point of slag initiation and slag splashing in a converter based on audio timing characteristics, thereby realizing full-process monitoring and end point prediction of the slag splashing process, so as to solve the technical problems of over-reliance on the experience and intuition of operators in the slag splashing furnace protection process, resulting in too long or too short slag splashing time, affecting the life of the converter and production efficiency, poor slag splashing effect, and high cost of manual intervention. It is worth noting that the first key to the present invention is to divide the slag splashing process into three stages. The first stage is to determine the slag starting point to set aside the audio data with large noise interference and difficult to extract features; the second stage is to determine the warning point to give early warning and reduce key misjudgments; the third stage is to determine the slag splashing endpoint to determine the end of slag splashing and avoid unnecessary impact and damage to the furnace lining, thereby extending the service life of the furnace lining and reducing labor costs and operational risks; the second key is how to process the audio data to further facilitate the subsequent extraction of audio time series features; as for the specific model structure, any other learning algorithm, such as random forest, support vector machine, neural network to train the model, use deep learning technology, such as convolutional neural network, recurrent neural network to process time series data, extract the complex nonlinear relationship between the features, and learn the first, second, and third audio data features.

[0043] In the preferred scheme, step S1 provides several preferred embodiments for generating a sound intensity variation curve, which can facilitate the subsequent steps S2-S4 to better learn the audio timing characteristics; step S2, according to the actual situation of slag initiation, has some parameter thresholds that meet the slag initiation point, and proposes a slag initiation point detection model that combines the slag initiation threshold to learn the first audio timing characteristics, which is more in line with reality; step S3, since the indicators in this time period do not change much, only the audio timing characteristics can be considered. Therefore, in order to simplify the model structure, only the instantaneous sound intensity value of this time period is counted, the time series prediction model is trained, and the second audio timing characteristics are learned, and the To the slag splashing warning point detection model; step S4, according to the actual situation of the slag splashing endpoint, it has a parameter threshold that meets the endpoint, and proposes a slag splashing endpoint detection model that combines the slag starting endpoint to learn the third audio timing feature. More importantly, due to the significant differences in the characteristics of different slag splashing modes, the historical data of different modes are trained separately at this time to extract the timing features of different modes in a targeted manner. The prediction results are more accurate, which can greatly avoid endpoint prediction errors and the like, further improve the efficiency and effectiveness of slag splashing operations, reduce material consumption, shorten the smelting cycle, and achieve the goal of reducing costs and increasing efficiency.

[0044] On the other hand, the present invention also provides an audio-based slag splashing state monitoring system, such as Figure 3 As shown, it includes: an audio data acquisition device A, a multi-band audio analyzer B, a PLC data acquisition device C and an audio slag-splatting industrial computer D, which are used to implement the above-mentioned audio-based slag splashing state monitoring method.

[0045] The above-mentioned slag splashing state monitoring system is created based on the above-mentioned slag splashing state monitoring method, and its technical function and beneficial effects are no longer repeated here. The various technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the various technical features in the above-mentioned embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0046] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for monitoring slag splashing status based on audio, characterized in that: include: Collect historical data of the slag splashing process, including audio data, gun height data and slag conditioning agent data; And mark the slag starting point, warning point and slag splashing end point of the historical data; According to the slag starting point and previous historical data, a slag starting point detection model is established; According to the historical data from the slag starting point to the early warning point, a slag splashing early warning point detection model is established; According to the historical data from the early warning point to the slag splashing endpoint, a slag splashing endpoint detection model is established; Collect real-time data of the slag splashing process, including audio data, gun height data and slag adjusting agent data. First, input the slag starting point detection model to predict the slag starting point; then input the real-time data after the slag starting point into the slag splashing early warning point detection model to predict the early warning point; then input the real-time data after the early warning point into the slag splashing endpoint detection model to predict the slag splashing endpoint.

2. The slag splashing state monitoring method according to claim 1, characterized in that: Collecting historical data of the slag splashing process also includes: generating a sound intensity change curve reflecting the slag splashing process based on audio data, gun height data and slag adjusting agent data; specifically: Extract the time domain and frequency domain features of audio data, and select the audio channels that effectively describe the slag splashing process from multiple frequency bands; According to the selected audio channel value, combined with the data sampling rate, the average value of the sound intensity data per second is calculated as the instantaneous sound intensity value at the current moment. ; According to the gun height data and slag adjusting agent data, the gun position compensation value and slag adjusting agent compensation value are calculated to compensate for the instantaneous sound intensity value, and the compensated instantaneous sound intensity value is obtained. , to generate a sound intensity change curve.

3. The slag splashing state monitoring method according to claim 2, characterized in that: Gun position compensation value , through the formula ) calculation, where is the gun position compensation coefficient, is the instantaneous gun height, the initial value of the gun position ; Slag adjustment agent compensation value , through the formula Calculate, where is the compensation coefficient of slag conditioning agent, Δm(t) is the instantaneous change of slag adjusting agent, is the real-time correction coefficient, Collect the actual amount of steel added during tapping; the instantaneous sound intensity value after compensation , through the formula calculate.

4. The slag splashing state monitoring method according to claim 2, characterized in that: The instantaneous sound intensity after compensation is calculated as After that, it also includes: S141: Set the moving window to calculate the instantaneous sound intensity value The sliding average ; S142: Set the exponentially weighted moving average weighting factor ewmaAlpha to assign the instantaneous sound intensity value at the current moment Higher weight; S143: Calculate the final instantaneous sound intensity value according to the instantaneous sound intensity value, the sliding average value and the exponentially weighted moving average weighting factor , and regenerate the sound intensity change curve.

5. The slag splashing state monitoring method according to claim 4, characterized in that: Final instantaneous sound intensity , through the formula calculate.

6. The slag splashing state monitoring method according to claim 1, characterized in that: The steps to establish a slag point detection model include: According to the slag starting point and previous historical data, the slag starting parameter thresholds of different furnace slag starting points are counted, including any one or more of: change in gun height, sound intensity and sudden change, trend slope range of sound intensity curve, movement of frequency peak, amplitude change of sound waveform in time domain, and slag starting duration; A slag point detection model is constructed with slag parameter threshold and historical data as input and slag point as output, which is used to learn the first sound intensity time series characteristics; Through multi-furnace learning based on historical data, the parameter settings of the slag point detection model are fine-tuned to obtain the trained slag point detection model, and the slag point is automatically predicted.

7. The slag splashing state monitoring method according to claim 1, characterized in that: The steps of establishing a slag splash warning point detection model include: The instantaneous sound intensity values ​​from the slag starting point to the warning point of different furnaces are counted to construct a time series data set of instantaneous sound intensity; A supervised deep learning time series prediction model is constructed, the instantaneous sound intensity time series dataset is input into the time series prediction model, the second audio time series features are learned, and the trained time series prediction model is obtained as a slag splash warning point detection model to automatically predict the warning point.

8. The slag splashing state monitoring method according to any one of claims 1 to 7, characterized in that: The steps of establishing the slag splash endpoint detection model include: According to the historical data from the early warning point to the slag splashing endpoint, the endpoint parameter thresholds of different furnaces are counted, including: the intensity and increase and decrease speed of the slag splashing endpoint sound, the trend slope range of the sound intensity curve, the frequency distribution and the movement of the peak value, the amplitude change and energy distribution of the sound waveform in the time domain, and any one or more of the slag splashing duration; A slag splashing endpoint detection model is constructed with endpoint parameter threshold and historical data as input and slag splashing endpoint as output, which is used to learn the third tone intensity time series characteristics; The historical data is divided into different slag splashing modes, different slag splashing modes are learned, and the parameter settings of the slag splashing endpoint inspection model are fine-tuned to obtain the trained slag splashing endpoint detection model to automatically predict the slag splashing endpoint.

9. An audio-based slag splashing status monitoring system, characterized in that: include: An audio data acquisition device, a multi-band audio analyzer, a PLC data acquisition device and an audio slag-removing industrial computer are used to implement the slag splashing state monitoring method described in any one of claims 1-8.

10. The slag splashing state monitoring system according to claim 9, characterized in that: The audio data acquisition device comprises an audio sensor which is arranged on the outer side of the converter fire retaining wall and is used for collecting the audio signal of nitrogen impacting the slag during slag splashing.

Images