Method for identifying state of stopper rod of continuous casting crystallizer

Abstract

The invention discloses a continuous casting crystallizer stopper rod state identification method, which comprises the following steps of: establishing a stopper rod state identification model by collecting operation data of a crystallizer liquid level control system in real time and measuring left-right and front-back offset distances of a stopper rod on a crystallizer; and the characteristic parameters of the stopper rod are calculated in real time through the stopper rod state recognition model, the state of the stopper rod is monitored, and a state prompt or alarm is given according to threshold value comparison. The method for identifying the state of the stopper rod based on the data is simple and rapid to implement, can quantitatively give the running state of the stopper rod, has the characteristics of good real-time performance and high reliability, and can provide a guidance function for field actual production and application.

Description

Technical Field

[0001] This invention relates to continuous casting crystallizer technology, and more specifically, to a method for identifying the state of stopper rods in a continuous casting crystallizer. Background Technology

[0002] Fluctuations in the liquid level of a continuous casting mold can lead to quality defects such as slag inclusions and cracks, and in severe cases, even molten steel leakage. Liquid level control in the mold typically employs a stopper rod mechanism. A servo motor drives the stopper rod up and down via a transmission mechanism, controlling the opening of the stopper rod and the top nozzle, thereby controlling the amount of molten steel flowing into the mold. Therefore, the stability and precision of the stopper rod's movement are crucial. However, in actual production, because both the stopper rod and the nozzle are immersed in the molten steel in the tundish, they cannot be directly observed, making it difficult to identify the stopper rod's status in real time. This includes the inertia of the stopper rod's movement, the precision of its movement, stopper rod head melting, stopper rod blockage, whether the stopper rod is aligned, and any shaking. Therefore, it is necessary to develop a new technology and method to provide real-time operational status information for the stopper rod system to guide production operations.

[0003] Currently, patent applications, such as Chinese patent application number 202311060532.X, propose a method and system for identifying and processing flocculent material shedding. This method collects relevant data on various process conditions during continuous casting production and uses this data to determine the current production status of the continuous casting machine. When the continuous casting machine is in normal production status, it calculates the fluctuation of the crystallizer level and the real-time change rate of the stopper rod opening. If the fluctuation of the crystallizer level and the real-time change rate of the stopper rod opening meet preset thresholds, it is considered that large pieces of flocculent material have shed. Although this method can monitor the state of flocculent material on the nozzle and stopper rod in real time, it does not quantitatively provide the action time and gain information of the stopper rod actuator. Furthermore, this method may have unstable identification results in continuous casting production processes with a large number of process disturbances.

[0004] For example, Chinese patent application number 201310104332.X proposes an online operation judgment system for a continuous casting machine crystallizer. This system collects process signals related to the crystallizer level control process by installing hydraulic cylinder displacement sensors, hydraulic cylinder pressure sensors, crystallizer cooling water temperature sensors, and crystallizer copper plate temperature sensors. The crystallizer operation status monitoring system receives process data from the crystallizer vibration PLC system and the crystallizer leakage prediction PLC system, processes the process signals, and uses intelligent algorithms to determine whether the crystallizer is in its optimal operating state. However, this method does not provide information on the action time and gain of the stopper rod actuator, nor does it quantitatively provide the operating status of the stopper rod actuator. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method for identifying the status of stopper rods in continuous casting molds. This data-based method for identifying stopper rod status is simple and quick to implement, can quantitatively provide the operating status of the stopper rods, and features good real-time performance and high reliability. It can provide guidance for actual production applications on-site.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for identifying the status of stopper rods in a continuous casting crystallizer:

[0008] A stopper state recognition model is established by collecting real-time operating data of the crystallizer liquid level control system and measuring the left-right and front-back offset distances of the stopper rod on the crystallizer.

[0009] The stopper's characteristic parameters are calculated in real time using the stopper's status recognition model, the stopper's status is monitored, and status prompts or alarms are given based on threshold comparisons.

[0010] Preferably, the operating data of the crystallizer level control system includes the set value and actual value of the stopper rod opening; and

[0011] The thickness, width, and drawing speed of the crystallizer.

[0012] Preferably, by collecting the set value and actual value of the stopper opening, the inertial time constant T and system gain A of the stopper are calculated using a recursive least squares algorithm. If the inertial time constant T and the system gain A exceed the threshold, an alarm is triggered for the inertia and execution accuracy of the stopper.

[0013] By collecting the opening degree of the stopper rod and the thickness, width, and billet drawing speed of the crystallizer, a stopper rod status recognition model is established. By determining the amount of molten steel flowing into the crystallizer through the stopper rod, the flow coefficient λ of the stopper rod is calculated. If the flow coefficient λ deviates from the given range, it is considered that the stopper rod has melted or blocked, and a prompt or alarm is issued.

[0014] Preferably, the inertial time constant T and system gain A of the stopper are calculated as follows:

[0015] The actuator of the stopper rod is considered as a first-order inertial element, as shown in equation (1):

[0016]

[0017] The inertial time constant T and the system gain A are identified in real time using a recursive least squares algorithm, as shown in equation (2) below:

[0018]

[0019] In equation (2), θ k Let X be the solution of the recursive least squares algorithm at time k. k-1 Given all input data for the first k–1 time steps, x k Given the input data at time k, y k This is the output data at time k;

[0020] Discretize equation (1) using backward difference discretization:

[0021]

[0022] In equation (3), T s The sampling period is s, the Laplace transform factor is s, and the z transform factor is z.

[0023] Substituting equation (3) into equation (1) yields the pulse transfer function of the actuator of the stopper rod:

[0024]

[0025] Transform equation (4) into a difference equation:

[0026]

[0027] In equation (5), y(k) is the output of the difference equation at time k, and u(k) is the input of the difference equation at time k; y is the actual opening of the stopcock, and u is the set opening of the stopcock. Let:

[0028]

[0029] In equation (6), θ a and θ b The two solutions of recursive least squares at a certain time represent the coefficients of y(k-1) and u(k) in equation (5), respectively;

[0030] Combining equations (5) and (6), we can obtain the solution for the system gain A of the actuator of the stopper rod and the inertial time constant T of the stopper rod:

[0031]

[0032] In equation (7), θ1 and θ2 are obtained by the recursive least squares algorithm given in equation (2);

[0033] Given θ1, θ2 and sampling time T s The system gain A of the stopper rod actuator and the inertial time constant T of the stopper rod can be obtained according to equation (7).

[0034] Preferably, calculating the flow coefficient λ of the stopper rod specifically includes the following steps:

[0035] S1, Obtain the signal interference frequency of the stopper rod;

[0036] S2, perform real-time noise reduction on the opening signal of the stopper rod;

[0037] S3, Identify the flow coefficient λ of the stopper rod according to the flow balance equation.

[0038] Preferably, in step S1, the FFT algorithm is used to obtain the signal interference frequency, including the following steps:

[0039] S11, decomposition, decomposes the sequence x(n) into two subsequences with even-numbered terms and odd-numbered terms:

[0040] x(n)=x(2r)+x(2r+1)(8)

[0041] In equation (8), is an integer representing the index of even-numbered and odd-numbered terms;

[0042] S12, recursively calculates and performs DFT transformations on the even-numbered and odd-numbered subsequences of the decomposition, respectively, to obtain a DFT result of length N / 2. Let x... even (k) and x odd (k) are the DFT results for even-numbered and odd-numbered terms, respectively;

[0043] S13, Merge: Merge the DFT results of even-numbered and odd-numbered terms to obtain the complete DFT result.

[0044]

[0045] Preferably, in step S2, a notch filter is used to denoise the opening signal of the stopcock in real time.

[0046] The transfer function of the notch filter is:

[0047]

[0048] In equation (10), ξ1 and ξ2 are related to the parameters of the notch filter, including the center frequency w of the notch filter. n The depth of the trap and the bandwidth of the notch filter. b ;

[0049] Wherein, the center frequency w of the notch filter n The trap depth and the bandwidth w of the notch filter are obtained using the FFT algorithm in step S1. b Selection based on experience.

[0050] Preferably, the center frequency w of the notch filter n The trap depth and the bandwidth w of the notch filterb The relationship with ξ1 and ξ2 is as follows:

[0051]

[0052] Preferably, in step S3, the crystallizer is treated as a typical integral element, as shown in equation (12) below:

[0053]

[0054] In equation (12), Q in Q is the inflow rate of the crystallizer. out A is the outflow rate of the crystallizer. m The cross-sectional area of ​​the crystallizer;

[0055] If the liquid level in the crystallizer is stable at this time, then:

[0056]

[0057] The flow balance equation of the crystallizer is then approximately:

[0058] Q in ≈Q out (14)

[0059] in:

[0060]

[0061] Combining equations (14) and (15), the identification formula for the flow coefficient λ of the stopper rod is:

[0062]

[0063] In equation (16), th is the thickness of the crystallizer, w is the width of the crystallizer, v is the drawing speed, and h is the opening of the stopper rod.

[0064] The present invention provides a method for identifying the status of a stopper rod in a continuous casting mold. This method establishes a stopper rod status identification model by real-time acquisition of data from the mold's liquid level control system and measurement of the left-right and front-back offset distances of the stopper rod. It then calculates the stopper rod's characteristic parameters in real time, monitors the stopper rod status, and provides status prompts and alarms based on threshold comparisons to guide production control operations. Furthermore, it offers the following advantages:

[0065] 1) This invention is simple, quick, and easy to deploy. Using this invention, only on-site process data needs to be acquired and combined with the identification algorithm to output the stopper rod status identification result;

[0066] 2) This invention can quantitatively provide the operating state of the stopper rod, namely the inertial time T of the stopper rod actuator, the gain A of the stopper rod actuator, and the flow coefficient λ of the stopper rod. For example, in Example 1, the inertial time T of the stopper rod actuator is calculated to be 0.047 seconds, the gain A of the stopper rod actuator is 1.000, and the flow coefficient λ of the stopper rod is 0.32;

[0067] 3) This invention has good real-time performance and high reliability. The recursive least squares algorithm and the flow balance equation identification algorithm have fast calculation speed and can calculate the stopper rod operating parameters based on the real-time process data collected during on-site production;

[0068] 4) This invention has high identification accuracy. First, the goodness of fit of the identification results of the recursive least squares algorithm is measured by the R² coefficient of determination, and the calculated R² coefficients are all above 0.8. Second, the fast Fourier transform combined with notch filter technology is used to perform real-time noise reduction on the stopcock signal, which reduces the error of the flow coefficient identification results.

[0069] 5) This invention provides status prompts and alarms based on the stopper rod's operating parameters. When the identified parameters exceed a certain threshold, it indicates that the stopper rod actuator is in an abnormal operating state. At this time, corresponding status prompts and alarms are given to guide production control operations. Attached Figure Description

[0070] Figure 1 This is a schematic diagram illustrating the implementation of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0071] Figure 2 This is a schematic diagram of the frequency characteristics of the notch filter in the continuous casting crystallizer stopper rod state identification method of the present invention;

[0072] Figure 3 This is a schematic diagram of the ibaAnalyzer software used to collect PLC data in Embodiment 1 of the continuous casting crystallizer stopper rod status identification method of the present invention;

[0073] Figure 4 This is a schematic diagram of the identification results using the recursive least squares algorithm in Embodiment 1 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0074] Figure 5 This is a schematic diagram comparing the estimated opening degree with the actual opening degree in Embodiment 1 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0075] Figure 6 yes Figure 5 A magnified view of a portion of the image;

[0076] Figure 7 This is a schematic diagram of the flow coefficient identified without denoising the stopper rod signal in Embodiment 1 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0077] Figure 8 This is a schematic diagram of the Fourier transform result of the stopper rod signal in Embodiment 1 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0078] Figure 9 This is a schematic diagram of the time-domain results of filtering out the bulging signal in the stopper opening using a notch filter in Embodiment 1 of the continuous casting crystallizer stopper state identification method of the present invention.

[0079] Figure 10 This is a schematic diagram of the frequency domain results of filtering out the bulging signal in the stopper opening using a notch filter in Embodiment 1 of the continuous casting crystallizer stopper state identification method of the present invention.

[0080] Figure 11 This is a schematic diagram of the flow coefficient identification result after processing the stopper rod signal in Embodiment 1 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0081] Figure 12 This is a schematic diagram of the ibaAnalyzer software used to collect PLC data in Embodiment 2 of the continuous casting crystallizer stopper rod status identification method of the present invention;

[0082] Figure 13 This is a schematic diagram of the identification results using the recursive least squares algorithm in Embodiment 2 of the continuous casting crystallizer stopper state identification method of the present invention;

[0083] Figure 14 This is a schematic diagram comparing the estimated opening degree with the actual opening degree in Embodiment 2 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0084] Figure 15 yes Figure 14 A magnified view of a portion of the image;

[0085] Figure 16 This is a schematic diagram of the Fourier transform result of the stopper rod signal in Embodiment 2 of the continuous casting crystallizer stopper rod state identification method of the present invention;

[0086] Figure 17 This is a schematic diagram of the time-domain results of filtering out the bulging signal in the stopper opening using a notch filter in Embodiment 2 of the continuous casting crystallizer stopper state identification method of the present invention.

[0087] Figure 18 This is a schematic diagram of the frequency domain results of filtering out the bulging signal in the stopper opening using a notch filter in Embodiment 2 of the continuous casting crystallizer stopper state identification method of the present invention.

[0088] Figure 19 This is a schematic diagram of the flow coefficient identification results in Embodiment 2 of the continuous casting crystallizer stopper rod state identification method of the present invention. Detailed Implementation

[0089] To better understand the above-mentioned technical solutions of the present invention, the technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0090] Combination Figure 1 As shown, the present invention provides a method for identifying the state of a stopper rod in a continuous casting crystallizer:

[0091] By real-time acquisition of operating data from the crystallizer level control system and measurement of the lateral and longitudinal offset distances of the stopper rods on the crystallizer, a stopper rod status recognition model is established. This model is used to calculate the characteristic parameters of the stopper rods in real time, enabling status monitoring and providing status prompts or alarms based on threshold comparisons. Specifically, this includes:

[0092] (1) Collect the set value and actual value of the stopper opening, treat the stopper's actuator as a first-order inertial element, and use the recursive least squares algorithm to calculate the stopper's inertial time constant T and system gain A in a timely manner. If the inertial time constant T and system gain A exceed the given threshold, an alarm is triggered for the stopper's inertia and execution accuracy.

[0093] (2) Collect data such as stopper opening, crystallizer width, crystallizer thickness, and casting speed, and establish a crystallizer flow balance model. By determining the amount of molten steel flowing into the crystallizer from the stopper, the flow coefficient λ of the stopper is calculated. If the flow coefficient λ deviates from the given reasonable range, it indicates that the stopper has melted or blocked, and an alarm is given. However, due to interference in the actual production process, such as unstable bulging, there is a certain amount of noise in the production data. When identifying the flow coefficient based on the data, it is necessary to filter out the noise contained in the data. Considering the real-time requirements of parameter identification, this invention also uses a notch filter to filter out the noise signal in the data.

[0094] (1.1) Identification of the inertial time constant T and system gain A of the stopper rod

[0095] This invention collects real-time data on the actual opening degree and the set opening degree of the stopper rod during the operation of the crystallizer control system. Through mechanistic model analysis of the stopper rod's servo mechanism, the stopper rod's actuator is treated as a first-order inertial element, as shown in equation (1):

[0096]

[0097] The inertial time constant T and system gain A are identified in real time using a recursive least squares algorithm, as shown in equation (2) below:

[0098]

[0099] In equation (2), θ k Let X be the solution of the recursive least squares algorithm at time k. k-1 Given all input data for the first k–1 time steps, xk Given the input data at time k, y k This is the output data at time k;

[0100] Compared with the traditional least squares algorithm, the recursive least squares algorithm has the advantages of high real-time performance, strong adaptability and high memory efficiency.

[0101] To facilitate parameter identification of the stopcock actuator using the recursive least squares algorithm, equation (1) is discretized using backward difference discretization:

[0102]

[0103] In equation (3), T s The sampling period is s, the Laplace transform factor is s, and the z transform factor is z.

[0104] Substituting equation (3) into equation (1) yields the pulse transfer function of the stopper rod actuator:

[0105]

[0106] Transform equation (4) into a difference equation:

[0107]

[0108] In equation (5), y(k) is the output of the difference equation at time k, and u(k) is the input of the difference equation at time k; y is the actual opening of the stopcock, and u is the set opening of the stopcock. Let:

[0109]

[0110] In equation (6), θ a and θ b The two solutions of recursive least squares at a certain time represent the coefficients of y(k-1) and u(k) in equation (5), respectively;

[0111] Combining equations (5) and (6), we can obtain the solutions for the system gain A of the stopper rod actuator and the inertial time constant T of the stopper rod:

[0112]

[0113] In equation (7), θ1 and θ2 are obtained by the recursive least squares algorithm given in equation (2);

[0114] Given θ1, θ2 and sampling time T s The system gain A of the stopper rod actuator and the inertial time constant T of the stopper rod can be obtained according to equation (7).

[0115] (1.2) Identification of the flow coefficient λ of the stopper rod

[0116] This invention identifies the stopper rod flow coefficient by real-time acquisition of process data from the crystallizer control system. The main data collected includes: crystallizer thickness, crystallizer width, billet drawing speed, and stopper rod opening. Considering the interference from non-stopper rod flow characteristics in the process data, such as unstable bulging, noise reduction processing of the process signal is necessary. Specifically, the following steps are included:

[0117] S1, Obtain the signal interference frequency of the stopcock.

[0118] In the field of signal processing, the most commonly used method for obtaining signal frequency is the Fourier transform. The purpose of the Fourier transform is to transform a signal in the time domain into the frequency domain. The idea is to approximate the original signal with a series of sinusoidal signals to obtain the frequency components of the signal. The most commonly used and most efficient Fourier transform algorithm is the Fast Fourier Transform (FFT) algorithm.

[0119] The FFT algorithm is an improved algorithm of the Discrete Fourier Transform (DFT). Given a discrete-time signal X(n) of length N, the DFT transform is performed on it, and its expression is:

[0120]

[0121] in

[0122] The FFT algorithm accelerates computation by decomposing the DFT into multiple smaller DFT calculations. Assuming N is a power of 2, the input sequence can be divided into two subsequences, and their DFTs can be performed separately. Then, by combining the frequency domain results of these subsequences, the frequency domain representation of the entire input sequence is obtained. This includes the following steps:

[0123] S11, decomposition, decomposes the sequence x(n) into two subsequences with even-numbered terms and odd-numbered terms:

[0124] x(n)=x(2r)+x(2r+1)(8)

[0125] In equation (8), is an integer representing the index of even-numbered and odd-numbered terms;

[0126] S12, recursively calculates and performs DFT transformations on the even-numbered and odd-numbered subsequences of the decomposition, respectively, to obtain a DFT result of length N / 2. Let x... even (k) and x odd (k) are the DFT results for even-numbered and odd-numbered terms, respectively;

[0127] S13, Merge: Merge the DFT results of even-numbered and odd-numbered terms to obtain the complete DFT result.

[0128]

[0129] This allows the FFT algorithm to obtain the signal frequency components, which significantly improves computational performance compared to the DFT algorithm.

[0130] S2 performs real-time noise reduction on the opening signal of the stopper rod.

[0131] A notch filter is a filter used to filter out signal components at specific frequencies. It is widely used in signal processing and communication fields, and is often used to remove interference or noise at specific frequencies.

[0132] The transfer function of the notch filter is:

[0133]

[0134] In equation (10), ξ1 and ξ2 are related to the parameters of the notch filter, including the center frequency w of the notch filter. n The depth of the trap and the bandwidth of the notch filter. b ;

[0135] Among them, the center frequency w of the notch filter n The trap depth and notch filter bandwidth w are obtained using the FFT algorithm in step S1. b Selection based on experience.

[0136] The center frequency w of the notch filter n Trap depth, notch filter bandwidth w b The relationship with ξ1 and ξ2 is as follows:

[0137]

[0138] Combination Figure 2 As shown, the frequency characteristics of the notch filter indicate that the amplitude of a signal located at the center frequency will be attenuated by a factor of depth, and its phase angle will have a certain lag / lead at the center frequency; the amplitude of a signal far from the center frequency will hardly be attenuated, and its phase angle will have almost no lag / lead.

[0139] S3, Identify the flow coefficient λ of the stopper rod according to the flow balance equation.

[0140] The liquid level in the crystallizer is determined by both the input and output flow rates of the crystallizer. The crystallizer is considered as a typical integral element, as shown in equation (12) below:

[0141]

[0142] In equation (12), Q in Q is the inflow rate of the crystallizer. out A is the outflow rate of the crystallizer. mThe cross-sectional area of ​​the crystallizer;

[0143] If the liquid level in the crystallizer is stable at this time, then:

[0144]

[0145] The flow balance equation for the crystallizer is then approximated as:

[0146] Q in ≈Q out (14)

[0147] in:

[0148]

[0149] Combining equations (14) and (15), the identification formula for the flow coefficient λ of the stopper rod is:

[0150]

[0151] In equation (16), th is the thickness of the crystallizer, w is the width of the crystallizer, v is the drawing speed, and h is the opening of the stopper rod.

[0152] To ensure the accuracy of flow coefficient identification, the opening h of the stopper rod generally needs to be denoised using the notch filter involved in step S2.

[0153] Combining steps S1 to S3, the dominant frequency of the interference component in the stopper signal is determined according to step S1. Then, the notch filter designed in step S2 is used to notch filter the bulging component in the stopper signal to remove it. Finally, the flow coefficient identification formula (16) proposed in step S3 is used to identify the flow coefficient of the stopper. Compared with the traditional flow coefficient identification method, the present invention will improve the accuracy of the flow coefficient identification of the stopper to a certain extent.

[0154] Example 1

[0155] In this embodiment 1, ibaAnalyzer software is used to collect on-site PLC data, mainly including: set opening degree, actual opening degree, crystallizer width, crystallizer thickness, and billet pulling speed, such as... Figure 3 As shown.

[0156] (1) Identify the system gain A and inertial time constant T of the stopper rod actuator based on field data.

[0157] use Figure 3 The actual opening data and opening data of the stopper rod are used to identify the inertial time and gain of the stopper rod mechanism through a recursive least squares algorithm. Assume the stopper rod actuator is a first-order inertial element:

[0158]

[0159] In equation (3.1), A is the system gain of the stopper rod actuator, and T is the inertial time constant of the stopper rod actuator. The identification results are as follows: Figure 4 As shown in the figure. From the identification results, it can be obtained that the system gain A of the stopper rod actuator is 1.000 and the inertial time constant T is 0.047s. Therefore, the transfer function of the stopper rod actuator is shown in equation (3.2).

[0160]

[0161] The transfer function identified using the set opening data is used to obtain the estimated actual opening of the stopcock. The estimated opening is then compared and verified with the actual opening. Figure 5 and Figure 6 As shown in the curve, it can be seen from the recursive least squares identification algorithm that the transfer function fits the actual opening degree and the set opening degree data very well. This is a result intuitively perceived from the graphical curve. Next, the coefficient of determination R is calculated. 2 To quantify the fit of this transfer function model, R 2 Ri is a coefficient used to measure the goodness of fit of a model, with a maximum value of 1. 2 The closer the value is to 1, the better the model fits. Coefficient of Determination R0 2 The calculation definition is:

[0162]

[0163] In equation (3.3), y i For the sake of truth, For predicted values, Calculate R0 between the estimated opening and the actual opening, taking the mean of the true values. 2 The calculated value is 0.84735, which is close to 1, indicating that the transfer function identified by the recursive least squares algorithm fits the data well.

[0164] (2) Identifying the stopper rod flow coefficient λ based on field data

[0165] according to Figure 3 The data is identified using the flow balance equation. If the interference components in the stopper rod are not processed, the identified flow coefficient will be as follows: Figure 7 As shown, the identified flow coefficient varies from 0.26 to 0.36, with a large deviation, which will affect the subsequent threshold judgment. Therefore, it is necessary to process the interference in the stopcock signal.

[0166] S1, Perform a Fourier transform on the stopcock signal to obtain the interference component.

[0167] The Fourier transform result of the stopper signal is as follows Figure 8 As shown in the figure, the stopper rod signal contains a component with a frequency of 0.38Hz and an amplitude of 0.92mm (bulging component) and a component with a frequency of 4.16Hz and an amplitude of 0.29Hz (jittering component). The jittering component is inherent to the stopper rod mechanism, while the bulging component is transmitted from the liquid level to the stopper rod through the feedback mechanism of the crystallizer liquid level control system. The bulging component should be removed; therefore, a notch filter is designed in step S2 to filter out the bulging component.

[0168] S2, designed with a swirl remover to eliminate bulging portion.

[0169] According to step S1, the frequency of the notch filter is 0.38Hz. Therefore, a notch filter with a center frequency of 0.38Hz, a bandwidth of 0.1Hz, and a trap depth of 0.01 is designed, and its transfer function is as follows:

[0170]

[0171] Signal processing results are as follows Figure 9 and Figure 10 As shown, the 0.38Hz stopper bulge component is filtered out.

[0172] S3, using the processed stopper rod signal for flow coefficient identification.

[0173] After filtering out the bulging component in the stopcock signal using the notch filter in step S2, the flow coefficient identified by it is as follows: Figure 11 As shown. (Through) Figure 11 It can be seen that after removing the bulging component from the stopper rod signal, the fluctuation amplitude of the flow coefficient becomes smaller and the trend becomes stronger, resulting in a more accurate result.

[0174] Example 2

[0175] In this embodiment 2, ibaAnalyzer software is used to collect PLC data from other dates on-site. This data mainly includes: set opening degree, actual opening degree, crystallizer width, crystallizer thickness, and billet pulling speed. Figure 12 As shown.

[0176] (1) Identify the system gain A and inertial time constant T of the stopper rod actuator based on field data.

[0177] use Figure 12 The actual opening data and the opening data of the stopper rod are used to identify the inertial time constant T and system gain A of the stopper rod mechanism through a recursive least squares algorithm. The results are as follows: Figure 13 As shown in the figure. The identification results show that the system gain A of the stopcock actuator is 0.999, and the inertial time constant T is 0.156s. Therefore, the identified transfer function model is shown in equation (4.1).

[0178]

[0179] Using the set aperture data to drive the identification of the transfer function model (4.1), the estimated aperture data is obtained. The estimated aperture is then compared with the actual aperture data, such as... Figure 14 and Figure 15 As shown. Calculate the coefficient of determination R of the model. 2 =0.97707, indicating that the identified transfer function fits the original data well.

[0180] However, the inertial time constant T of the stopper rod was identified as 0.156s, which is too long. At this time, the stopper rod mechanism is in an abnormal state and an alarm should be issued.

[0181] (2) Identifying the stopper rod flow coefficient λ based on field data

[0182] according to Figure 12 The data is identified using the flow balance equation. The identification steps are as follows:

[0183] S1, Perform a Fourier transform on the stopcock signal to obtain the interference component.

[0184] The Fourier transform result of the stopper rod signal is as follows Figure 15 As shown. From Figure 16 As can be seen, there are two interference components in the stopcock signal: one with a frequency of 0.073Hz and an amplitude of 0.88, and the other with a frequency of 3.82Hz and an amplitude of 0.078. The 0.073Hz interference is a bulging disturbance, and the 3.82Hz interference is a stick jitter signal, which is an inherent signal of the stopcock. A notch filter is needed to filter out the bulging component.

[0185] S2, designed with a swirl remover to eliminate bulging portion.

[0186] According to step S1, the bulging component in the stopcock signal is 0.073Hz. Therefore, a notch filter with a center frequency of 0.073Hz, a trap depth of 0.01, and a bandwidth of 0.05Hz is designed, and its transfer function is:

[0187]

[0188] Processing with a notch filter in equation (4.2) Figure 12 The results of the middle plug signal are as follows Figure 17 and Figure 18 As shown in the notch filter results, the bulging component in the stopcock signal is removed.

[0189] S3, using the processed stopper rod signal for flow coefficient identification.

[0190] The flow coefficient is identified using the stopper rod signal processed in step S2, such as... Figure 19 As shown in the figure, the identification results show that after removing the bulging signal component of the stopper rod, the trend of the identified flow coefficient is more obvious, and the fluctuation of the flow coefficient becomes smaller. However, the flow coefficient identified using this set of data is too large, and the flow coefficient value keeps increasing, indicating that the stopper rod may have been eroded, and the erosion is intensifying over time, so an alarm should be issued.

[0191] In summary, the method for identifying the state of plug rods in continuous casting molds that are rejected according to the present invention mainly includes two aspects:

[0192] (1) Collect the set value and actual value of the stopper rod opening. Treat the stopper rod actuator as a first-order inertial element and use the recursive least squares algorithm to calculate the inertial time constant T and system gain A of the stopper rod system in a timely manner. If the time constant T and gain A exceed the given threshold, an alarm is triggered regarding the inertia and execution accuracy of the stopper rod. The transfer function model identified by the recursive least squares algorithm has good fitting performance, and the determination coefficients calculated in the two embodiments are... The values ​​are all above 0.8.

[0193] (2) Data such as stopper opening, mold width, mold thickness, and casting speed are collected, and the bulging component in the stopper signal is filtered out using a notch filter. A mold flow balance model is established, and the flow coefficient λ of the stopper is calculated by determining the amount of molten steel flowing into the mold from the stopper. If the flow coefficient deviates from the given reasonable range, it indicates that the stopper is melting or blocked, and an alarm is triggered. Compared with the flow coefficient identification of the stopper signal without processing, the flow coefficient identification using the notch filter results in smaller amplitude changes and a more obvious trend, thus better reflecting the actual flow coefficient.

[0194] Those skilled in the art should recognize that the above embodiments are merely illustrative of the present invention and are not intended to limit the present invention. Any variations or modifications to the above embodiments that are within the spirit and essence of the present invention will fall within the scope of the claims of the present invention.

Claims

1. A method for identifying the state of a stopper rod in a continuous casting crystallizer, characterized in that: A stopper state recognition model is established by collecting real-time operating data of the crystallizer liquid level control system and measuring the left-right and front-back offset distances of the stopper rod on the crystallizer. The stopper's characteristic parameters are calculated in real time using the stopper's status recognition model, the stopper's status is monitored, and status prompts or alarms are given based on threshold comparisons.

2. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 1, characterized in that: The operating data of the crystallizer level control system includes the set value and actual value of the stopper rod opening; and The thickness, width, and drawing speed of the crystallizer.

3. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 2, characterized in that: By collecting the set value and actual value of the stopper opening, the inertial time constant T and system gain A of the stopper are calculated using the recursive least squares algorithm. If the inertial time constant T and the system gain A exceed the threshold, an alarm is triggered for the inertia and execution accuracy of the stopper. By collecting the opening degree of the stopper rod and the thickness, width, and billet drawing speed of the crystallizer, a stopper rod status recognition model is established. By determining the amount of molten steel flowing into the crystallizer through the stopper rod, the flow coefficient λ of the stopper rod is calculated. If the flow coefficient λ deviates from the given range, it indicates that the stopper rod has melted or blocked, and a prompt or alarm is issued.

4. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 3, characterized in that, The inertial time constant T and system gain A of the stopper rod are calculated as follows: The actuator of the stopper rod is considered as a first-order inertial element, as shown in equation (1): The inertial time constant T and the system gain A are identified in real time using a recursive least squares algorithm, as shown in equation (2) below: In equation (2), θ k Let X be the solution of the recursive least squares algorithm at time k. k-1 Given all input data for the first k–1 time steps, x k Given the input data at time k, y k This is the output data at time k; Discretize equation (1) using backward difference discretization: In equation (3), T s s is the sampling period, s is the Laplace transform factor, and z is the z-transform factor; Substituting equation (3) into equation (1) yields the pulse transfer function of the actuator of the stopper rod: Transform equation (4) into a difference equation: In equation (5), y(k) is the output of the difference equation at time k, and u(k) is the input of the difference equation at time k; y is the actual opening of the stopcock, and u is the set opening of the stopcock. Let: In equation (6), θ a and θ b The two solutions of recursive least squares at a certain time represent the coefficients of y(k-1) and u(k) in equation (5), respectively; Combining equations (5) and (6), we can obtain the solution for the system gain A of the actuator of the stopper rod and the inertial time constant T of the stopper rod: In equation (7), θ1 and θ2 are obtained by the recursive least squares algorithm given in equation (2); Given θ1, θ2 and sampling time T s The system gain A of the stopper rod actuator and the inertial time constant T of the stopper rod can be obtained according to equation (7).

5. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 3, characterized in that, Calculating the flow coefficient λ of the stopper rod specifically includes the following steps: S1, Obtain the signal interference frequency of the stopper rod; S2, perform real-time noise reduction on the opening signal of the stopper rod; S3, Identify the flow coefficient λ of the stopper rod according to the flow balance equation.

6. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 5, characterized in that: In step S1, the FFT algorithm is used to obtain the signal interference frequency, which includes the following steps: S11, decomposition, decomposes the sequence x(n) into two subsequences with even-numbered terms and odd-numbered terms: x(n)=x(2r)+x(2r+1)(8) In equation (8), is an integer representing the index of even-numbered and odd-numbered terms; S12, recursively calculates and performs DFT transformations on the even-numbered and odd-numbered subsequences of the decomposition, respectively, to obtain a DFT result of length N / 2. Let x... even (k) and x odd (k) are the DFT results for even-numbered and odd-numbered terms, respectively; S13, Merge: Merge the DFT results of even-numbered and odd-numbered terms to obtain the complete DFT result.

7. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 6, characterized in that, In step S2, a notch filter is used to denoise the opening signal of the stopper rod in real time. The transfer function of the notch filter is: In equation (10), ξ1 and ξ2 are related to the parameters of the notch filter, including the center frequency w of the notch filter. n The depth of the trap and the bandwidth of the notch filter. b ; Wherein, the center frequency w of the notch filter n The trap depth and the bandwidth w of the notch filter are obtained using the FFT algorithm in step S1. b Selection based on experience.

8. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 7, characterized in that, The center frequency w of the notch filter n The trap depth and the bandwidth w of the notch filter b The relationship with ξ1 and ξ2 is as follows:

9. The method for identifying the state of the stopper rod in a continuous casting mold according to claim 7, characterized in that, In step S3, the crystallizer is considered as a typical integral element, as shown in equation (12) below: In equation (12), Q in Q is the inflow rate of the crystallizer. out A is the outflow rate of the crystallizer. m The cross-sectional area of ​​the crystallizer; If the liquid level in the crystallizer is stable at this time, then: The flow balance equation of the crystallizer is then approximately: Q in ≈Q out (14) in: Combining equations (14) and (15), the identification formula for the flow coefficient λ of the stopper rod is: In equation (16), th is the thickness of the crystallizer, w is the width of the crystallizer, v is the drawing speed, and h is the opening of the stopper rod.

Images