Method for judging deflection of fine roller in continuous annealing furnace based on Fourier transform

Abstract

The invention relates to the technical field of cold rolling continuous annealing, in particular to a method for judging deflection of a thin roller in a continuous annealing furnace based on Fourier transform, which comprises the following steps of: acquiring a torque signal of the thin roller in the production process, and performing fast Fourier transform (FFT) analysis on the torque signal of the thin roller within the acquisition time delta t; an original torque signal is converted into a frequency domain signal from a time domain signal, a frequency band closest to the rotation period of the thin roller is extracted from the torque signal by referring to the running linear speed of the thin roller, Fourier transform of a torque sequence or the frequency characteristic of a system is extracted to achieve frequency sampling, the signal amplitude of the extracted frequency band is analyzed, and the frequency of the thin roller is calculated. The unbalance loading degree is obtained, a deflection tendency chart and a scatter diagram are obtained according to Fourier transform and frequency sampling, the unbalance loading degree is compared with a dangerous threshold value, and an alarm is given if the unbalance loading degree exceeds the threshold value; according to the method, early warning is conducted on the roller with the too large deflection degree, rubbing between the roller and the strip steel is avoided, and production accidents are reduced.

Description

Technical Field

[0001] This invention relates to the field of cold rolling continuous annealing technology, and in particular to a method for judging the deflection of fine rolls in a continuous annealing furnace based on Fourier transform. Background Technology

[0002] Deflection refers to the displacement of the axis of an elastic beam in a direction perpendicular to its axis when subjected to force or experiencing non-uniform temperature changes. In a continuous annealing furnace, there is a set of inlet sealing rollers at the inlet and two sets of outlet sealing rollers at the outlet. The rapid cooling section has six sets of guide rollers, and the final cooling section has ten sets of guide rollers. These fine rollers are smaller in diameter than the strip tensioning rollers within the section, making them more prone to deflection after prolonged service under the influence of force and temperature. Furthermore, these fine rollers are close to the strip; when they deflect significantly, they can rub against the strip, causing surface quality defects and affecting the service life of the rollers.

[0003] There are many similar studies on the deflection problem of fine rollers. Patent application No. 202023008622.9 discloses a roller deflection measuring device. Using tools such as a limiting plate and a pen tip, the fixed block is moved by the handle, and the pen tip draws a curve to represent the deflection on the side plate.

[0004] Patent application number 202010419559.3 discloses a method, device and system for detecting the deflection curve of a small cantilever beam. The method uses corresponding sensors to record position data, constructs a cubic polynomial of the deflection curve, and forms the deflection curve equation.

[0005] While the methods described above help workers measure deflection, they are all measurement methods under static conditions and cannot measure or judge the deflection of fine rollers in rotating operation. Summary of the Invention

[0006] This invention provides a method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform. The method performs Fourier transform on the torque signal of the fine rollers in the continuous annealing furnace to obtain the amplitude information of each frequency component. Combined with the rotation speed in the furnace, the rotation frequency of the fine rollers is obtained. The amplitude of the torque component is used to judge whether the deflection value is too large. When the deflection exceeds the danger threshold, an alarm is triggered to avoid affecting the strip quality and the life of the fine rollers due to excessive deflection of the fine rollers, thus ensuring the smooth operation of the production process.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] A method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform includes the following steps:

[0009] S1. Collect the torque signal of the fine roller during the production process;

[0010] S2. Perform Fast Fourier Transform (FFT) analysis on the torque signal of the fine roller within the acquisition time Δt to convert the original torque signal from a time domain signal to a frequency domain signal.

[0011] S3. Based on the linear speed of the fine roller, extract the frequency band closest to the rotation period of the fine roller from the above torque signal, and extract the Fourier transform of the torque sequence or the frequency characteristics of the system to achieve frequency sampling.

[0012] S4. Analyze the signal amplitude of the extracted frequency band to obtain its off-center load degree, and obtain the scatter plot trend of deflection based on Fourier transform and frequency sampling.

[0013] S5. Compare the degree of off-center load with the danger threshold. If the threshold is exceeded, an alarm will be triggered.

[0014] Furthermore, step S2 specifically includes:

[0015] Let the torque signal during the acquisition time be x(n), which is a finite-length sequence of length M. To ensure that the periodic extension result of x(n) does not produce aliasing during subsequent frequency sampling, it is necessary to ensure that N≥M. Then, the discrete Fourier transform of N sample points in x(n) is defined as:

[0016]

[0017] in, N is the interval length of the Discrete Fourier Transform, i.e., there are N sample points in total, and N≥M; n is the sample point index, and k is the index of the N sample points.

[0018] Furthermore, step S3 specifically includes extracting the torque signal x(n) within the acquisition time as an absolutely summable signal, whose z-transform is expressed as:

[0019]

[0020] Where z is a complex frequency domain variable;

[0021] A finite-length sequence must satisfy the absolute summability condition, meaning its region of convergence includes the unit circle. In the z-plane, the z-transform is calculated around N equidistant points on the unit circle, let...

[0022]

[0023] Using symbols The above formula can be rewritten as:

[0024]

[0025] For the z-transform of sequence x(n), extract The discrete Fourier transform can be obtained from the sample values ​​of these specific points. Therefore, on the unit circle of the z-plane, take the kth point at equal intervals with the argument ω that is closest to the rotation frequency of the fine roller, calculate its z-transform, and obtain the kth sample value X(k) of DFT[x(n)].

[0026] For a finite-length sequence x(n), when the frequency domain sampling theorem is satisfied, the N-point frequency domain sample X(k) represents the characteristics of the selected sequence without distortion. Therefore, the N sampled values ​​X(k) can completely express the entire X(z) function and frequency response X(e^(-z / k)). -jw ).

[0027] Furthermore, the scatter plot of the deflection is the percentage by which the deflection amplitude exceeds the natural frequency amplitude.

[0028] Furthermore, the danger thresholds include historical record thresholds and screen alarm thresholds;

[0029] The historical record threshold is used to record historical data when the absolute deflection value of the fine roller is greater than the historical record threshold;

[0030] If the on-screen alarm threshold is higher than the historical record threshold, a real-time on-site alarm will be triggered.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] Fourier transform is performed on the torque signal of the fine roller in the continuous annealing furnace to obtain the amplitude information of each frequency component. Combined with the rotation speed in the furnace, the rotation frequency of the fine roller is obtained. The amplitude of the torque component is used to determine whether the deflection value is too large. When the deflection exceeds the danger threshold, an alarm is triggered to avoid affecting the strip quality and the life of the fine roller due to excessive deflection of the fine roller, and to ensure the smooth operation of the production process.

[0033] When applied to actual production, the degree of deflection of the fine rollers can be seen intuitively, and the condition of each fine roller can be observed laterally. Early warnings can be given for fine rollers with excessive deflection to prevent them from rubbing against the strip steel and reduce the occurrence of production accidents. Attached Figure Description

[0034] Figure 1 This is a flowchart of the method of the present invention.

[0035] Figure 2 This is the flexural scatter plot trend diagram described in this invention. Detailed Implementation

[0036] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0037] See Figure 1This is a flowchart of the method of the present invention. The present invention provides a method for determining the deflection of fine rollers in a continuous annealing furnace based on Fourier transform, comprising the following steps:

[0038] S1. Collect the torque signal of the fine roller during the production process, and obtain the signal corresponding to the torque of the fine roller and time through the fine roller motor.

[0039] S2. Perform Fast Fourier Transform (FFT) analysis on the torque signal of the fine roller within the acquisition time Δt to convert the original torque signal from a time-domain signal to a frequency-domain signal. The Fast Fourier Transform is a fast computational form of the Discrete Fourier Transform. Let the torque signal within the acquisition time be x(n), a finite-length sequence of length M. To ensure that the periodic extension results of x(n) do not produce aliasing during subsequent frequency sampling, N ≥ M must be guaranteed. Therefore, the Discrete Fourier Transform of N sample points in x(n) is defined as:

[0040]

[0041] in, N is the interval length of the Discrete Fourier Transform, i.e., there are N sample points in total, and N≥M; n is the sample point index, and k is the index of the N sample points.

[0042] S3. Based on the linear velocity of the reference fine roller, extract the frequency band closest to the rotation period of the fine roller from the above torque signal; frequency sampling refers to extracting the Fourier transform of the torque sequence or the frequency characteristics of the system. The extracted torque signal within the acquisition time is x(n) absolutely summable, and its z-transform is expressed as:

[0043]

[0044] Where z is a complex frequency domain variable;

[0045] A finite-length sequence must satisfy the absolute summability condition, meaning its region of convergence includes the unit circle. In the z-plane, the z-transform is calculated around N equidistant points on the unit circle, let...

[0046]

[0047] Using symbols The above formula can be rewritten as:

[0048]

[0049] It can be seen that for the z-transform of sequence x(n), the extraction The discrete Fourier transform can be obtained from the sample values ​​of these specific points. Therefore, on the unit circle of the z-plane, take the kth point at equal intervals with the argument ω that is closest to the rotation frequency of the fine roller, calculate its z-transform, and obtain the kth sample value X(k) of DFT[x(n)].

[0050] For a finite-length sequence x(n), when the frequency domain sampling theorem is satisfied, the N-point frequency domain sample X(k) represents the characteristics of the selected sequence without distortion. Therefore, the N sampled values ​​X(k) can completely express the entire X(z) function and frequency response X(e^(-z / k)). -jw ).

[0051] S4. Analyze the signal amplitude of the extracted frequency band to obtain its off-center load degree, see... Figure 2 Based on the Fourier transform and frequency sampling described above, a scatter plot of deflection was obtained, which represents the percentage of deflection amplitude exceeding the natural frequency amplitude, 26%.

[0052] S5. Compare the degree of off-center load with the danger threshold. If the threshold is exceeded, an alarm is triggered. There are three threshold rules: protection threshold, historical record threshold, and screen alarm threshold. According to the principle of signal amplitude-frequency decomposition, the zero-frequency component is the mean of the torque signal, that is, the basic component when the signal vibration is not considered. Since the load conditions of different fine rollers are different and the motor output is different, the magnitude of their torque zero-frequency component is also different. Therefore, the selection of the screen alarm threshold for deflection requires statistical analysis of historical data and threshold calculation based on the percentage of the zero-frequency component.

[0053] To prevent false alarms caused by abnormal data, when the zero-frequency component is very small, the deflection is very small, which will generate many false alarms. Therefore, the system sets a protection threshold. Only when the zero-frequency component is greater than the protection threshold will subsequent judgment be performed.

[0054] A historical record threshold is set to collect data for trend analysis of deflection. The historical record only records the absolute deflection value of the fine roll, and is mainly used to analyze the maximum historical peak value of deflection. The historical record threshold is used to record historical data when the absolute deflection value of the fine roll is greater than the historical record threshold. This is beneficial for accident tracing and regular inspection of the fine roll status, and provides a basis for roll replacement or maintenance.

[0055] The real-time alarm threshold for the screen is the real-time alarm on site. Therefore, to avoid frequent alarms, if the screen alarm threshold is higher than the historical record threshold, it will be considered a real-time alarm on site.

[0056] The following embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments. Unless otherwise specified, the methods used in the following embodiments are conventional methods.

[0057]

Example

[0058] This invention provides a method for determining the deflection of fine rollers in a continuous annealing furnace based on Fourier transform, comprising the following steps:

[0059] S1. Collect the torque signal of the fine roller during the production process, and obtain the signal corresponding to the torque of the fine roller and time through the fine roller motor.

[0060] S2. Perform Fast Fourier Transform (FFT) analysis on the torque signal of the fine roller within a 15-second sampling period. The sampling period is 65ms, with a total of 230 samples. Convert the original torque signal from the time domain to the frequency domain. Let the torque signal within the sampling time be x(n), defined as a finite-length sequence of length M. To ensure that the periodic extension result of x(n) does not produce aliasing during subsequent frequency sampling, it is necessary to ensure that N≥M. Therefore, the discrete Fourier transform of N sample points in x(n) is defined as:

[0061]

[0062] in, N is the interval length of the Discrete Fourier Transform, i.e., there are N sample points in total, and N≥M; n is the sample point index, and k is the index of the N sample points.

[0063] S3. Based on the linear velocity of the reference fine roller, extract the frequency band closest to the rotation period of the fine roller from the above torque signal; frequency sampling refers to extracting the Fourier transform of the torque sequence or the frequency characteristics of the system. The extracted torque signal within the acquisition time is x(n) absolutely summable, and its z-transform is expressed as:

[0064]

[0065] Where z is a complex frequency domain variable;

[0066] A finite-length sequence must satisfy the absolute summability condition, meaning its region of convergence includes the unit circle. In the z-plane, the z-transform is calculated around N equidistant points on the unit circle, let...

[0067]

[0068] symbol The above formula can be rewritten as:

[0069]

[0070] It can be seen that for the z-transform of sequence x(n), the extraction The discrete Fourier transform can be obtained from the sample values ​​of these specific points. Therefore, on the unit circle of the z-plane, take the kth point at equal intervals with the argument ω that is closest to the rotation frequency of the fine roller, calculate its z-transform, and obtain the kth sample value X(k) of DFT[x(n)].

[0071] For a finite-length sequence x(n), when the frequency domain sampling theorem is satisfied, the N-point frequency domain sample X(k) represents the characteristics of the selected sequence without distortion. Therefore, the N sampled values ​​X(k) can completely express the entire X(z) function and frequency response X(e^(-z / k)). -jw ).

[0072] S4. Analyze the signal amplitude of the extracted frequency band to obtain its off-center load degree. Based on the Fourier transform and frequency sampling mentioned above, obtain the scatter plot trend of deflection, that is, the percentage of deflection amplitude exceeding the natural frequency amplitude, 26%.

[0073] S5. Compare the degree of off-center load with the danger threshold. If the threshold is exceeded, an alarm is triggered. In this method, three threshold rules are set: protection threshold, historical record threshold, and screen alarm threshold. According to the principle of signal amplitude-frequency decomposition, the zero-frequency component is the mean of the torque signal, that is, the basic component when the signal vibration is not considered. Since the load conditions of different fine rollers are different and the motor output is different, the magnitude of their torque zero-frequency component is also different. Therefore, the selection of the screen alarm threshold for deflection requires statistical analysis of historical data and threshold calculation based on the percentage of the zero-frequency component to complete the setting.

[0074] The fine roll deflection analysis program ran and detected a severe problem of uneven loading on the fine roll. The corresponding indicator light on the fine roll turned red in the alarm screen. In this example, the deflection alarm occurred on the fine roll in the slow cooling section. Verification revealed scratches on the strip surface.

[0075] The above embodiments are merely illustrative of the principles and effects of the present invention, as well as some examples of its application, and are not intended to limit the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform, characterized in that, Includes the following steps: S1. Collect the torque signal of the fine roller during the production process; S2. Perform Fast Fourier Transform (FFT) analysis on the torque signal of the fine roller within the acquisition time Δt to convert the original torque signal from a time domain signal to a frequency domain signal. S3. Based on the linear speed of the fine roller, extract the frequency band closest to the rotation period of the fine roller from the above torque signal, and extract the Fourier transform of the torque sequence or the frequency characteristics of the system to achieve frequency sampling. S4. Analyze the signal amplitude of the extracted frequency band to obtain its off-center load degree, and obtain the scatter plot trend of deflection based on Fourier transform and frequency sampling. S5. Compare the degree of off-center load with the danger threshold. If the threshold is exceeded, an alarm will be triggered.

2. The method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform according to claim 1, characterized in that, Step S2 specifically includes: Let the torque signal during the acquisition time be x(n), which is a finite-length sequence of length M. To ensure that the periodic extension result of x(n) does not produce aliasing during subsequent frequency sampling, it is necessary to ensure that N≥M. Then, the discrete Fourier transform of N sample points in x(n) is defined as: in, N is the interval length of the Discrete Fourier Transform, i.e., there are N sample points in total, and N≥M; n is the sample point index, and k is the index of the N sample points.

3. The method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform according to claim 2, characterized in that, Step S3 specifically includes extracting the torque signal x(n) within the acquisition time, which is absolutely summable, and its z-transform is expressed as: Where z is a complex frequency domain variable; A finite-length sequence must satisfy the absolute summability condition, meaning its region of convergence includes the unit circle. In the z-plane, the z-transform is calculated around N equidistant points on the unit circle, let... Using symbols The above formula can be rewritten as: For the z-transform of sequence x(n), extract The discrete Fourier transform can be obtained from the sample values ​​of these specific points. Therefore, on the unit circle of the z-plane, take the kth point at equal intervals with the argument ω that is closest to the rotation frequency of the fine roller, calculate its z-transform, and obtain the kth sample value X(k) of DFT[x(n)]. For a finite-length sequence x(n), when the frequency domain sampling theorem is satisfied, the N-point frequency domain sample X(k) represents the characteristics of the selected sequence without distortion. Therefore, the N sampled values ​​X(k) can completely express the entire X(z) function and frequency response X(e^(-z / k)). -jw ).

4. The method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform according to claim 1, characterized in that, The scatter plot of the deflection shows the percentage by which the deflection amplitude exceeds the natural frequency amplitude.

5. The method for judging the deflection of fine rollers in a continuous annealing furnace based on Fourier transform according to claim 1, characterized in that, The danger thresholds include historical record thresholds and video alarm thresholds; The historical record threshold is used to record historical data when the absolute deflection value of the fine roller is greater than the historical record threshold; If the on-screen alarm threshold is higher than the historical record threshold, a real-time on-site alarm will be triggered.

Images