AI-based fan-shaped section driving roller anomaly detection method, system and program product

Abstract

The invention belongs to the technical field of intelligent detection, and particularly discloses an AI-based fan-shaped section driving roller anomaly detection method and system and a program product, and the method comprises the steps: carrying out the corresponding signal conversion processing and feature data extraction through collecting the operation state monitoring signal of a fan-shaped section driving roller, and obtaining a monitoring feature set; and then inputting the monitoring feature set into the trained anomaly detection model for time sequence prediction, and determining whether an abnormal condition exists according to a prediction result at the next moment, so that intelligent fan-shaped section driving roller anomaly detection can be realized, and the timeliness of fan-shaped section driving roller anomaly detection is improved. Moreover, in combination with the data analysis result of the machine learning model, an alarm is given when it is judged that the frequency of abnormal detection of the fan-shaped section driving roller within the set time period reaches the early warning frequency condition, and the accuracy of abnormal detection of the fan-shaped section driving roller can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent detection, and particularly relates to an abnormal detection method, system and program product for segment driving rolls based on AI. Background Art

[0002] In the continuous casting industry of steel, the segment driving roll is a key component of the continuous caster, and its operating state directly affects the quality of the cast slab and the production efficiency. At present, the abnormal detection of the operating state of the segment driving roll mainly relies on manual inspection and regular maintenance. This method has low detection efficiency, is difficult to detect faults in real time, and has high maintenance costs once a fault occurs. Currently, there is also a method of collecting monitoring signal values through sensors and using set comparison rules to compare the monitoring signal values with preset thresholds to detect abnormalities. However, the signal processing and analysis of this method cannot integrate more conditions for judgment, resulting in low detection accuracy. Summary of the Invention

[0003] The purpose of the present invention is to provide an abnormal detection method, system and program product for segment driving rolls based on AI to solve the above problems existing in the prior art.

[0004] To achieve the above purpose, the present invention adopts the following technical solutions: In the first aspect, an abnormal detection method for segment driving rolls based on AI is provided, including: Collecting the operating state monitoring signals of the segment driving roll based on set sampling conditions; Performing quantization conversion processing on the operating state monitoring signals to obtain a corresponding initial monitoring data set; Performing data cleaning on the initial monitoring data set to obtain a target monitoring data set, and extracting feature values from the target monitoring data set to obtain a monitoring feature set; Inputting the monitoring feature set into a trained abnormal detection model for abnormal analysis and detection to obtain a corresponding abnormal detection result; Determining whether there is an abnormal operating condition of the segment driving roll according to the abnormal detection result, and recording it as one detection abnormality when it is determined that the segment driving roll has an abnormal operating condition; When it is determined that the frequency of detection abnormalities of the segment driving roll reaches a set warning frequency condition within a set time period, an alarm message is output.

[0005] In a possible design, before inputting the monitoring feature set into a trained abnormal detection model for abnormal analysis and detection, the method further includes: Constructing a LightGBM machine learning model; Collect the historical operation status monitoring signal samples of the segment drive roll based on the set sampling conditions, and collect the true signal values at the next moment corresponding to each historical operation status monitoring signal sample; Perform quantization conversion processing on each operation status monitoring signal sample to obtain the corresponding initial monitoring data set sample; Perform data cleaning on each initial monitoring data set sample to obtain the corresponding target monitoring data set sample, extract eigenvalue from each target monitoring data set sample to obtain the corresponding monitoring feature set sample, and associate each monitoring feature set sample with the true signal value at the corresponding next moment to form a training set; Use the training set to train the LightGBM machine learning model to obtain a trained anomaly detection model.

[0006] In a possible design, when using the training set to train the LightGBM machine learning model, the method further includes: Input the monitoring feature set sample into the LightGBM machine learning model to obtain the signal prediction value at the corresponding next moment; Compare the signal prediction value at the corresponding next moment with the true signal value. When the difference between the two is greater than the set difference threshold, it is determined that an anomaly is output, and the LightGBM machine learning model is optimally adjusted according to the set model modification parameters.

[0007] In a possible design, the sampling conditions include sampling frequency, sampling time, and sampling method. After collecting the operation status monitoring signal of the segment drive roll, the method further includes: performing signal filtering processing on the operation status monitoring signal.

[0008] In a possible design, the performing quantization conversion processing on the operation status monitoring signal to obtain the corresponding initial monitoring data set includes: Convert the operation status monitoring signal into a digital signal, perform encoding processing on the digital signal, and convert the digital signal into an initial monitoring data set.

[0009] In a possible design, the performing data cleaning on the initial monitoring data set includes: Delete the redundant data in the initial monitoring data set during the non-casting period, use the 3sigma method to remove the abnormal data in the initial monitoring data set, and linearly interpolate to complete the missing data in the initial monitoring data set.

[0010] In a possible design, the anomaly detection model performs anomaly analysis and detection every M minutes. The warning frequency condition includes more than N detection anomalies within the set casting time period. Both M and N are integers greater than 0.

[0011] In a second aspect, an AI-based abnormal detection system for segment drive rolls is provided, including a signal acquisition unit, a conversion and processing unit, a feature extraction unit, an abnormal detection unit, an abnormal determination unit, and an abnormal alarm unit, where: The signal acquisition unit is configured to acquire the operation status monitoring signals of the segment drive rolls based on set sampling conditions; The conversion and processing unit is configured to perform quantization conversion processing on the operation status monitoring signals to obtain corresponding initial monitoring data sets; The feature extraction unit is configured to clean the data of the initial monitoring data sets to obtain target monitoring data sets, and extract feature values from the target monitoring data sets to obtain monitoring feature sets; The abnormal detection unit is configured to input the monitoring feature sets into a trained abnormal detection model for abnormal analysis and detection to obtain corresponding abnormal detection results; The abnormal determination unit is configured to determine whether there are any abnormal operation conditions of the segment drive rolls according to the abnormal detection results, and record it as one detection abnormality when it is determined that there are abnormal operation conditions of the segment drive rolls; The abnormal alarm unit is configured to output an alarm message when it is determined that the frequency of detection abnormalities of the segment drive rolls reaches the set warning frequency condition within a set time period.

[0012] In a third aspect, an AI-based abnormal detection system for segment drive rolls is provided, including: A memory for storing instructions; A processor for reading the instructions stored in the memory and executing any one of the methods in the first aspect according to the instructions.

[0013] In a fourth aspect, a computer-readable storage medium is provided, on which instructions are stored. When the instructions run on a computer, the computer is made to execute any one of the methods in the first aspect. At the same time, a computer program product is also provided. When the computer program product runs on a computer, it executes any one of the methods in the first aspect.

[0014] Beneficial effects: By collecting the operation status monitoring signals of the segment drive rolls, performing corresponding signal conversion processing and feature data extraction, a monitoring feature set is obtained. Then, the monitoring feature set is input into the trained anomaly detection model for time series prediction. Whether there is an abnormal situation is determined according to the prediction result of the next moment. The intelligent anomaly detection of the segment drive rolls can be realized, and the timeliness of the anomaly detection of the segment drive rolls is improved. Moreover, by combining the data analysis results of the machine learning model, when it is determined that the frequency of detecting anomalies of the segment drive rolls reaches the warning frequency condition within the set time period, an alarm is given, which can improve the accuracy of the anomaly detection of the segment drive rolls. Through the corresponding AI machine learning model, the time series prediction task is completed, the operation status of the segment drive rolls can be detected in real time, potential abnormal situations can be discovered and warned in time, thereby avoiding production interruption and equipment damage, and improving the stability and production efficiency of the continuous casting process. Brief Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0016] Figure 1 It is a schematic diagram of the steps of the method in Embodiment 1 of the present invention; Figure 2 It is a schematic diagram of the composition of the system in Embodiment 2 of the present invention; Figure 3 It is a schematic diagram of the composition of the system in Embodiment 3 of the present invention. Detailed Embodiments

[0017] It should be noted here that the description of these embodiment modes is used to help understand the present invention, but does not constitute a limitation to the present invention. The specific structures and functional details disclosed herein are only used to describe the exemplary embodiments of the present invention. However, the present invention can be embodied in many alternative forms and should not be construed as limited to the embodiments described herein.

[0018] It should be understood that unless otherwise clearly defined and limited, the corresponding terms should be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be an electrical connection, a direct connection, an indirect connection through an intermediate medium, or the internal connection of two components. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments can be understood according to specific situations.

[0019] Specific details are provided in the following description to facilitate a complete understanding of the exemplary embodiments. However, those of ordinary skill in the art should understand that the exemplary embodiments can be implemented without these specific details. For example, a device may be shown in a block diagram to avoid obscuring the example with unnecessary details. In other embodiments, well-known processes, structures, and techniques may not be shown with unnecessary details to avoid obscuring the embodiments.

[0020] Embodiment 1: This embodiment provides an AI-based abnormal detection method for segment drive rolls, which can be applied to a corresponding abnormal detection server for segment drive rolls, such as Figure 1 As shown, the method includes the following steps: S1. Collect the operation status monitoring signals of the segment drive roll based on set sampling conditions.

[0021] Specifically, during implementation, through a corresponding data acquisition terminal, such as a data sensor, etc., based on set sampling conditions, including sampling frequency, sampling time, and sampling method, etc., the operation status monitoring signals of the segment drive roll are collected in real time, and the operation status monitoring signals of the segment drive roll are subjected to preliminary signal filtering and then transmitted to the server.

[0022] S2. Perform quantization conversion processing on the operation status monitoring signals to obtain a corresponding initial monitoring data set.

[0023] Specifically, during implementation, the server can convert the operation status monitoring signals into digital signals, perform encoding processing on the digital signals, and convert the digital signals into an initial monitoring data set suitable for storage and transmission processing. The data set may include motor torque data, casting signals, tundish car numbers, casting speeds, stopper opening data, etc.

[0024] S3. Clean the initial monitoring data set to obtain a target monitoring data set, and extract feature values from the target monitoring data set to obtain a monitoring feature set.

[0025] Specifically, since the data collected in the collection step is in milliseconds and the amount of data is large and not easy to analyze, the server needs to align the data in the initial monitoring data set to the minute level, that is, the sorted data is arranged at minute intervals, and process the data burr points and noise points in the initial monitoring data set. Then, delete the redundant data during the non-casting period in the initial monitoring data set, and use the 3sigma method to remove the abnormal data in the initial monitoring data set, and complement the missing data in the initial monitoring data set through linear interpolation to obtain the target monitoring data set. By extracting feature values from the target monitoring data set, the monitoring feature set can be obtained.

[0026] S4. Input the monitoring feature set into the trained anomaly detection model for anomaly analysis and detection to obtain the corresponding anomaly detection results.

[0027] In specific implementation, the server inputs the extracted monitoring feature set into the trained anomaly detection model for anomaly analysis and detection to obtain the corresponding anomaly detection results. Before using the anomaly detection model for data prediction, it is necessary to pre - construct a corresponding machine learning model for training. This process includes: Construct a LightGBM machine learning model. After constructing the initial model, it is necessary to obtain a training set in the manner of steps S1 - S3 to train the initial model. That is, it is necessary to collect historical operation status monitoring signal samples of the segment drive roll based on the set sampling conditions, and collect the true signal values at the next moment corresponding to each historical operation status monitoring signal sample; then perform quantization conversion processing on each operation status monitoring signal sample to obtain the corresponding initial monitoring data set sample; perform data cleaning on each initial monitoring data set sample to obtain the corresponding target monitoring data set sample, and extract feature values from each target monitoring data set sample to obtain the corresponding monitoring feature set sample, and associate each monitoring feature set sample with the corresponding true signal value at the next moment to form a training set; finally, use the training set to train the LightGBM machine learning model to obtain the trained anomaly detection model. When using the training set to train the LightGBM machine learning model, the monitoring feature set sample can be input into the LightGBM machine learning model to obtain the predicted signal value at the next moment; then compare the predicted signal value at the next moment with the true signal value. When the difference between the two is greater than the set difference threshold, it is determined that an anomaly is output, and the LightGBM machine learning model is optimized according to the set model modification parameters, so that the model can accurately predict the monitoring values in the next period of time, thereby performing effective anomaly detection.

[0028] S5. Determine whether there is an abnormal operation of the segment drive roll according to the anomaly detection results, and record it as a detection anomaly when it is determined that there is an abnormal operation of the segment drive roll.

[0029] In specific implementation, the server can determine whether there is an abnormal operation of the segment drive roll according to the anomaly detection results (i.e., the predicted value), and record it as a detection anomaly when it is determined that there is an abnormal operation of the segment drive roll. The anomaly detection model performs anomaly analysis and detection every several minutes, such as every 5 minutes.

[0030] S6. When the frequency of detection anomalies of the segment drive roll reaches the set warning frequency condition within the set time period, output an alarm message.

[0031] In specific implementation, when the server determines that the frequency of detection abnormalities of the sector segment drive roller within the set time period reaches the set warning frequency condition, it outputs an alarm message. For example, if it is determined that if detection abnormalities occur more than 6 times within 1 hour during the pouring time period, an alarm will be issued once, and combined with the pouring start signal and the intermediate ladle car number for analysis, when it is in the non-pouring time period or the intermediate ladle replacement time period, no alarm will be issued even if the detection abnormality frequency reaches the set warning frequency condition. When the abnormality is handled on site and the fault is confirmed and removed on the system, the alarm will be lifted. If the detection abnormality still occurs more than 6 times within the next hour, the alarm will continue to be issued.

[0032] Embodiment 2: This embodiment provides an AI-based sector drive roller abnormality detection system, such as Figure 2 As shown, it includes a signal acquisition unit, a conversion processing unit, a feature extraction unit, an abnormality detection unit, an abnormality determination unit and an abnormality alarm unit, wherein: A signal acquisition unit, used for acquiring a running status monitoring signal of a sector drive roller based on a set sampling condition; A conversion processing unit, used to perform quantitative conversion processing on the operating status monitoring signal to obtain a corresponding initial monitoring data set; A feature extraction unit is used to perform data cleaning on the initial monitoring data set to obtain a target monitoring data set, and to extract feature values ​​from the target monitoring data set to obtain a monitoring feature set; An anomaly detection unit, used to input the monitoring feature set into the trained anomaly detection model to perform anomaly analysis and detection, and obtain the corresponding anomaly detection result; an abnormality determination unit, for determining whether the segment drive roller has an abnormal operation according to the abnormality detection result, and when it is determined that the segment drive roller has an abnormal operation, recording it as a detection abnormality; The abnormality alarm unit is used to output an alarm message when it is determined that the frequency of detection abnormalities of the sector segment drive roller within a set time period reaches a set early warning frequency condition.

[0033] Embodiment 3: This embodiment provides an AI-based sector drive roller abnormality detection system, such as Figure 3 As shown, at the hardware level, it includes: Data interface, used to establish data connection between the processor and the data acquisition terminal; A memory for storing instructions; The processor is used to read the instructions stored in the memory and execute the AI-based sector drive roller abnormality detection method in Example 1 according to the instructions.

[0034] Optionally, the system further includes an internal bus through which the processor can be interconnected with the memory and the data interface. The internal bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0035] The memory can include, but is not limited to, a random access memory (RAM), a read only memory (ROM), a flash memory, a first input first output (FIFO) memory, and / or a first in last out (FILO) memory, etc. The processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0036] Embodiment 4: This embodiment provides a computer-readable storage medium, on which instructions are stored. When the instructions run on a computer, the computer is caused to execute the AI-based abnormal detection method for segment driving rolls in Embodiment 1. Among them, the computer-readable storage medium refers to a carrier for storing data, and can include, but is not limited to, a floppy disk, an optical disc, a hard disk, a flash memory, a USB flash drive, and / or a memory stick, etc. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0037] This embodiment also provides a computer program product. When the computer program product runs on a computer, it executes the AI-based abnormal detection method for segment drive rolls in Embodiment 1. Among them, the computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0038] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The AI-based sector drive roller anomaly detection method is characterized by: include: Collecting the operating status monitoring signal of the sector drive roller based on the set sampling conditions; Perform quantization conversion processing on the operation status monitoring signal to obtain the corresponding initial monitoring data set; Perform data cleaning on the initial monitoring data set to obtain the target monitoring data set, and perform feature value extraction on the target monitoring data set to obtain a monitoring feature set; Input the monitoring feature set into the trained anomaly detection model to perform anomaly analysis and detection, and obtain the corresponding anomaly detection results; Determine whether the segment drive roller has an abnormal operation according to the abnormal detection result, and record it as a detection abnormality when it is determined that the segment drive roller has an abnormal operation. When it is determined that the frequency of detection anomalies of the sector segment drive roller within a set period of time reaches the set warning frequency condition, an alarm message is output.

2. The AI-based sector drive roller anomaly detection method according to claim 1, characterized in that: Before inputting the monitoring feature set into the trained anomaly detection model for anomaly analysis and detection, the method further includes: Build LightGBM machine learning model; Collect historical operation status monitoring signal samples of the sector drive roller based on the set sampling conditions, and collect the real value of the signal corresponding to the next moment of each historical operation status monitoring signal sample; Perform quantization conversion processing on each running status monitoring signal sample to obtain the corresponding initial monitoring data set sample; Perform data cleaning on each initial monitoring data set sample to obtain the corresponding target monitoring data set sample, extract feature values ​​from each target monitoring data set sample to obtain the corresponding monitoring feature set sample, and associate each monitoring feature set sample with the corresponding true value of the signal at the next moment to form a training set; The LightGBM machine learning model is trained using the training set to obtain a trained anomaly detection model.

3. The AI-based sector drive roller anomaly detection method according to claim 2 is characterized in that: When training the LightGBM machine learning model using the training set, the method further includes: Input the monitoring feature set samples into the LightGBM machine learning model to obtain the signal prediction value corresponding to the next moment; The predicted signal value corresponding to the next moment is compared with the actual signal value. When the difference between the two is greater than the set difference threshold, the output is judged to be abnormal, and the LightGBM machine learning model is optimized according to the set model modification parameters.

4. The AI-based sector drive roller anomaly detection method according to claim 1, characterized in that: The sampling conditions include sampling frequency, sampling time and sampling mode. After collecting the running status monitoring signal of the sector drive roller, the method further includes: performing signal filtering processing on the running status monitoring signal.

5. The AI-based sector drive roller anomaly detection method according to claim 1, characterized in that: The step of performing quantization conversion processing on the operation status monitoring signal to obtain a corresponding initial monitoring data set includes: The operation status monitoring signal is converted into a digital signal, and the digital signal is encoded and processed to convert the digital signal into an initial monitoring data set.

6. The AI-based sector drive roller anomaly detection method according to claim 1, characterized in that: The data cleaning of the initial monitoring data set includes: The redundant data in the non-casting period in the initial monitoring data set were deleted, and the 3sigma method was used to remove the abnormal data in the initial monitoring data set, and the missing data in the initial monitoring data set were supplemented by linear interpolation.

7. The AI-based sector drive roller anomaly detection method according to claim 1, characterized in that: The anomaly detection model performs an anomaly analysis and detection every M minutes, and the warning frequency condition includes more than N detection anomalies occurring within a set pouring time period, where M and N are both integers greater than 0.

8. The AI-based sector drive roller anomaly detection system is characterized by: It includes a signal acquisition unit, a conversion processing unit, a feature extraction unit, an abnormality detection unit, an abnormality determination unit and an abnormality alarm unit, wherein: A signal acquisition unit, used for acquiring a running status monitoring signal of a sector drive roller based on a set sampling condition; A conversion processing unit, used to perform quantitative conversion processing on the operating status monitoring signal to obtain a corresponding initial monitoring data set; A feature extraction unit is used to perform data cleaning on the initial monitoring data set to obtain a target monitoring data set, and to extract feature values ​​from the target monitoring data set to obtain a monitoring feature set; An anomaly detection unit, used to input the monitoring feature set into the trained anomaly detection model to perform anomaly analysis and detection, and obtain the corresponding anomaly detection result; an abnormality determination unit, for determining whether the segment drive roller has an abnormal operation according to the abnormality detection result, and when it is determined that the segment drive roller has an abnormal operation, recording it as a detection abnormality; The abnormality alarm unit is used to output an alarm message when it is determined that the frequency of detection abnormalities of the sector segment drive roller within a set time period reaches a set early warning frequency condition.

9. The AI-based sector drive roller anomaly detection system is characterized by: include: A memory for storing instructions; A processor is used to read the instructions stored in the memory and execute the AI-based sector drive roller abnormality detection method according to any one of claims 1 to 7 according to the instructions.

10. A computer program product, characterized in that When the computer program product runs on a computer, the AI-based sector drive roller abnormality detection method described in any one of claims 1 to 7 is executed.

Images