Rolling mill vibration detection and correction system integrated with vibration recorder

Abstract

The invention relates to a rolling mill vibration detection and correction system integrated with a vibration recorder, and belongs to the technical field of steel rolling equipment. According to the system, a multi-dimensional signal detection unit obtains a rolling mill multi-source vibration correlation signal, the rolling mill multi-source vibration correlation signal is input into a multi-modal fusion module for time sequence correlation analysis, fusion features are extracted, and a multi-dimensional vibration detection table is output; the self-adaptive vibration correction unit combines historical data stored by the vibration recorder, pre-judges vibration evolution data through a prediction model, dynamically generates correction parameters, and adjusts rolling force, speed and roller phase; the vibration correction collaborative integration unit synchronously aligns real-time data of the recorder and corrects execution data, regularly calibrates the recorder and forms a collaborative data set; and the working condition optimization unit calls an anti-interference scheme for different working conditions, and starts grading emergency regulation and control when vibration exceeds a threshold value. Comprehensive detection, dynamic correction and multi-working-condition adaptation of vibration of the rolling mill are achieved, and the running stability of the rolling mill and the quality of rolled materials are improved.

Description

Technical Field

[0001] This invention belongs to the technical field of rolling mill equipment, specifically relating to a rolling mill vibration detection and correction system with an integrated vibration recorder. Background Technology

[0002] In steel rolling production, the rolling mill is the core equipment, and its operational stability directly determines the surface accuracy, dimensional tolerances, and mechanical properties of the rolled materials (such as strip steel and plate). Rolling mill vibration is a common and critical problem in production, manifesting as periodic vibration of the mill stand, roll system, or transmission components under rolling force. This not only leads to defects such as ripples and scratches on the surface of the rolled materials, reducing product yield, but also accelerates roll wear, bearing aging, and shortens equipment lifespan. In severe cases, it can even cause mill component breakage, resulting in production interruptions and safety hazards, significantly impacting the steel company's production capacity and economic benefits. Current technologies for detecting and correcting mill vibration still have many limitations and cannot meet the demands of high-precision, intelligent production. First, detection methods often rely on single signal acquisition, resulting in highly incomplete data. Existing technologies often only collect vibration signals from a single component of the rolling mill (such as the work roll) using vibration sensors, or only monitor single process parameters such as rolling force and speed. They cannot integrate multi-source related signals such as vibration, stress, and strip position, making it difficult to fully capture the root cause of vibration. Secondly, the vibration correction parameters are mostly fixed settings, resulting in poor adaptability. Existing correction schemes are mostly based on empirically preset adjustment values ​​for parameters such as rolling force and speed, without taking into account the differences in rolled material (such as the different mechanical properties of low-carbon steel and high-carbon steel, and their varying sensitivities to vibration) and dynamic adjustments based on real-time vibration characteristics. This leads to unstable correction effects: parameters that are effective for one type of material may exacerbate vibration or fail to completely suppress vibration when used for other materials. Third, there is a lack of a collaborative integration mechanism between vibration recording and correction execution. Although some technologies are equipped with vibration recorders, they are only used to store historical data. They do not synchronously correlate the real-time data with parameter changes during the correction execution process, nor are the recorders calibrated regularly. This results in time deviations or accuracy errors between vibration data and correction data, making it impossible to form an effective closed loop and difficult to optimize subsequent correction strategies using historical data. Fourth, the ability to adapt to different operating conditions and provide emergency control is weak. Workshop production involves various operating conditions such as normal rolling, roll changing, and strip biting. The interference signals under each condition (such as impact interference during roll changing and contact interference during biting) differ significantly. Existing technologies mostly employ uniform anti-interference schemes, failing to specifically filter interference from different operating conditions. Furthermore, emergency control is often based on a "single threshold trigger shutdown" mode, without developing control strategies according to the amplitude of the vibration. Shutting down at the slightest vibration will result in wasted production capacity, while failing to intervene in severe vibrations in a timely manner will amplify the impact of the malfunction. Summary of the Invention

[0003] To address the aforementioned problems in the prior art, this invention provides a rolling mill vibration detection and correction system integrating a vibration recorder. The objective of this invention can be achieved through the following technical solutions: Includes: a multi-dimensional signal detection unit, an adaptive vibration correction unit, a vibration correction collaborative integration unit, and a working condition optimization unit; The multi-dimensional signal detection unit acquires multi-source vibration correlation signals of the rolling mill and performs classification preprocessing according to signal type; the processed signals are input into a preset multi-modal fusion module for time-series correlation analysis, extracting fusion features and outputting a multi-dimensional vibration detection table; The adaptive vibration correction unit, based on the multi-dimensional vibration detection table and combined with the historical rolling data and vibration records stored in the vibration recorder, predicts the vibration evolution data through a prediction model; dynamically generates correction parameters based on the mapping relationship between the rolling material and vibration characteristics, and adjusts the rolling force, rolling speed and roll phase to obtain the adaptive vibration correction execution result; The vibration correction collaborative integration unit, based on the adaptive vibration correction execution result, synchronously reads the real-time data and correction execution data of the vibration recorder and performs time-series alignment; it performs periodic automatic calibration on the vibration recorder to form a vibration correction collaborative integration dataset. The operating condition optimization unit performs adaptive anti-interference processing for different operating conditions in the workshop based on the vibration correction collaborative integrated dataset. When the vibration amplitude exceeds the preset safety threshold, a graded emergency control mechanism is activated, and finally an operating condition optimization report is generated.

[0004] Specifically, the process of performing classification preprocessing according to signal type is as follows: the source flutter associated signal is identified and classified into vibration signal, stress signal, and position signal; different types of signals are processed separately, vibration signal is filtered, stress signal is smoothed, and position signal is corrected.

[0005] Specifically, the preset multimodal fusion module includes: an input layer that determines the data type of the module input, corresponding to vibration, stress, and position signals in the multi-source tremor correlation signals of the rolling mill, and assigns an independent input channel to each type of signal; a feature extraction layer that extracts features for different types of signals; a temporal correlation layer that introduces a timestamp alignment mechanism, matches feature data of different types of signals according to the acquisition time, and calculates the correlation coefficient between features to establish a temporal correlation relationship; a fusion layer that adopts a weight allocation strategy, calibrates the contribution of various signal features to tremor detection based on historical rolling data, and obtains a comprehensive fusion feature by weighting according to the contribution; and an output layer that converts the comprehensive fusion feature into a standardized data format.

[0006] Specifically, the process of performing time-series correlation analysis is as follows: First, various types of signals are organized into time-series data sequences according to the acquisition time order, and each type of time-series data sequence corresponds to an input channel of the multimodal fusion model; the multimodal fusion model extracts the feature values ​​of different types of signals at the same time node through time-series correlation, calculates the correlation coefficient between each feature value, and analyzes the trend of different signals changing over time. Specifically, the process of extracting and fusing features and outputting a multi-dimensional vibration detection table is as follows: the multi-modal fusion model integrates the correlation features of various signals and extracts comprehensive features that can reflect the vibration intensity, vibration frequency and vibration influence range of the rolling mill; the comprehensive features are organized in a preset table format, including feature name, feature value, corresponding collection location and duration information, to form the multi-dimensional vibration detection table.

[0007] Specifically, the process of predicting the evolution of vibration data through the prediction model is as follows: historical rolling data and vibration records are retrieved through the vibration recorder; the historical rolling data includes historical rolling force, rolling speed, and roll phase information; the vibration records include historical vibration amplitude and vibration duration; the historical rolling data and vibration records are input into the prediction model, and by learning the pattern of vibration variation with rolling parameters in the historical data, combined with the real-time vibration characteristics in the current multi-dimensional vibration detection table, the trend of vibration amplitude variation and the fluctuation range of vibration frequency within a preset time period are calculated. Specifically, the process of dynamically generating correction parameters is as follows: a preset rolling material attribute library is retrieved, which stores mechanical performance parameters corresponding to different rolling materials; the vibration characteristics in the multi-dimensional vibration detection table are combined with the mapping relationship library to find the associated records that match the current rolling material and vibration characteristics; based on the matched associated records, the type of rolling parameter that needs to be adjusted is determined, the adjustment range and direction of each parameter are calculated, and dynamic correction parameters are formed. Specifically, the process of adjusting the rolling force, rolling speed, and roll phase to obtain the adaptive vibration correction result is as follows: adjust the rolling force, speed, and roll phase of the rolling mill according to the correction parameters; increase or decrease the hydraulic system pressure of the rolling force; adjust the speed of the motor and the angle of the roll transmission mechanism of the roll phase; after the adjustment, feed back the actual parameter values, compare the actual parameter values ​​with the correction parameters, and record the changes in the adjusted parameters and the execution time.

[0008] As a preferred embodiment of the present invention, the specific process of the vibration correction collaborative integration unit includes: The vibration amplitude and frequency data of parts such as the mill stand and work rolls are acquired in real time from the vibration recorder; at the same time, the adjustment parameters of rolling force, rolling speed, and roll phase, as well as the actual execution results, are retrieved from the tremor correction execution module. Both types of data carry the original time stamp. The time stamps of the two types of data are uniformly converted to the workshop standard time, and the data matching at the same time point is compared. If there is a time deviation, the time stamps of the correction execution data are interpolated or fine-tuned based on the vibration recorder time to ensure that the same time point accurately corresponds to a set of vibration data and a set of correction data. A fixed calibration cycle is preset, and the process is automatically triggered when the cycle is reached: the built-in standard signal generation module calls the standard vibration signal with fixed amplitude and frequency, and inputs it into the recorder's acquisition terminal through a dedicated line; the difference in amplitude and frequency between the detection signal output by the recorder and the standard signal is compared, and the signal amplification coefficient and frequency response threshold of the recorder are automatically adjusted according to the difference until the difference meets the preset allowable range, the calibration is completed and the parameter changes are recorded; The vibration data after time alignment and the corrected execution data are summarized and superimposed with the calibration records (including calibration time, standard signal parameters, and recorder parameters before and after adjustment). All data are converted into a unified structured format, sorted by generation time, and each data entry is labeled with its type and corresponding vibration detection table number. Finally, the data is stored in the system database to form a complete vibration correction collaborative integration dataset, providing data support for subsequent operating condition optimization.

[0009] Specifically, the process of synchronously reading real-time data from the vibration recorder and corrective execution data for time alignment is as follows: real-time vibration data is obtained through the vibration recorder, and corrective execution data is obtained from the adaptive vibration correction unit. Both types of data carry time stamps. The time stamps are uniformly converted into a preset standard time. The two types of data under the same standard time are compared. If there is a time deviation, the time stamp of the corrective execution data is adjusted based on the time stamp of the vibration recorder. Specifically, the process of performing periodic automatic calibration on the vibration recorder is as follows: the system presets a fixed calibration cycle, and the calibration process is automatically triggered when the cycle is reached; a standard vibration signal is called and input to the vibration recorder's acquisition terminal through the transmission line; after receiving the standard signal, the vibration recorder outputs a detection signal, and the system compares the amplitude and frequency difference between the detection signal and the standard signal; calibration parameters are automatically generated based on the difference, and the recorder's signal amplification coefficient and frequency response parameters are adjusted until the difference meets the preset requirements.

[0010] Specifically, the process of forming the vibration correction collaborative integration dataset is as follows: summarizing the real-time vibration data after time alignment, correction execution data, and automatic calibration record data; converting data from different sources into a preset standard format and sorting them according to the time of data generation; labeling each data entry with its data type and corresponding vibration detection table number to form the vibration correction collaborative integration dataset. Specifically, the process of performing adaptive anti-interference processing for different working conditions in the workshop is as follows: determine the current working condition type in the workshop, which includes rolling condition, roll changing condition, and strip biting condition; call the corresponding anti-interference scheme for different working conditions, activate the motor base frequency interference filtering scheme for rolling condition, activate the impact interference suppression scheme for roll changing condition, and activate the contact impact interference compensation scheme for strip biting condition.

[0011] The beneficial effects of this invention are as follows: (1) By acquiring multi-source vibration, stress, and position-related signals of the rolling mill, preprocessing is performed according to the signal type (filtering is performed on vibration signals to remove environmental interference, data smoothing is performed on stress signals to eliminate instantaneous fluctuations, and deviation correction is performed on position signals to correct acquisition errors). The processed signals are then input into a preset multi-modal fusion module (the input layer allocates independent channels for each type of signal, the feature extraction layer extracts different signal features, the time-series correlation layer aligns features according to timestamps and calculates correlation coefficients to establish correlation, the fusion layer weights the contribution of historical data to obtain comprehensive features, and the output layer converts to a standard format) for time-series correlation analysis. Finally, the fusion features are extracted and a multi-dimensional vibration detection table containing feature name, value, acquisition location, and duration information is output. This can comprehensively integrate multi-source related signals of the rolling mill, break the one-sided limitation of single signal acquisition, capture the root cause of vibration, and provide accurate data support for subsequent vibration correction. (2) By combining the historical rolling data (including historical rolling force, rolling speed, and roll phase) stored in the vibration recorder with the vibration records (including historical vibration amplitude and duration), the vibration evolution data is predicted by the prediction model. Then, based on the mapping relationship between the rolling material property library (which stores the mechanical property parameters of different materials) and the vibration characteristics, the matching and associated records are searched and the adjustment amplitude and direction of each rolling parameter are calculated. After dynamically generating the correction parameters, the rolling force (increasing or decreasing the hydraulic system pressure), rolling speed (adjusting the motor speed), and roll phase (adjusting the transmission mechanism angle) are adjusted. Simultaneously, the real-time data of the vibration recorder and the correction are read. The system aligns the data according to standard time and automatically calibrates the recorder periodically (by calling the standard signal, comparing the difference between the detected signal and the standard signal, and adjusting the recorder parameters until the difference meets the requirements), forming a vibration correction collaborative integrated dataset. Based on this dataset, corresponding anti-interference schemes are invoked for different working conditions such as rolling, roll changing, and strip bite. When the vibration amplitude exceeds the safety threshold, graded emergency control is activated. This not only dynamically adapts to the material of the rolled material and the real-time vibration characteristics, avoiding the poor adaptability of fixed correction parameters, but also specifically filters working condition interference and optimizes the emergency control mode, thereby improving the stability of the rolling mill operation and the quality of rolled material production. Attached Figure Description

[0012] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0013] Figure 1 This is a system architecture diagram of a mill vibration detection and correction system integrating a vibration recorder according to the present invention; Figure 2 This is a data flow diagram of a mill vibration detection and correction system with an integrated vibration recorder according to the present invention. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0015] Please see Figure 1-2 A mill vibration detection and correction system integrating a vibration recorder; Includes: a multi-dimensional signal detection unit, an adaptive vibration correction unit, a vibration correction collaborative integration unit, and a working condition optimization unit; The multi-dimensional signal detection unit acquires multi-source vibration correlation signals of the rolling mill and performs classification preprocessing according to signal type; the processed signals are input into a preset multi-modal fusion module for time-series correlation analysis, extracting fusion features and outputting a multi-dimensional vibration detection table; The adaptive vibration correction unit, based on the multi-dimensional vibration detection table and combined with the historical rolling data and vibration records stored in the vibration recorder, predicts the vibration evolution data through a prediction model; dynamically generates correction parameters based on the mapping relationship between the rolling material and vibration characteristics, and adjusts the rolling force, rolling speed and roll phase to obtain the adaptive vibration correction execution result; The vibration correction collaborative integration unit, based on the adaptive vibration correction execution result, synchronously reads the real-time data and correction execution data of the vibration recorder and performs time-series alignment; it performs periodic automatic calibration on the vibration recorder to form a vibration correction collaborative integration dataset. The operating condition optimization unit performs adaptive anti-interference processing for different operating conditions in the workshop based on the vibration correction collaborative integrated dataset. When the vibration amplitude exceeds the preset safety threshold, a graded emergency control mechanism is activated, and finally an operating condition optimization report is generated.

[0016] Specifically, the process of performing classification preprocessing according to signal type is as follows: the source flutter associated signal is identified and classified into vibration signal, stress signal, and position signal; different types of signals are processed separately, vibration signal is filtered, stress signal is smoothed, and position signal is corrected.

[0017] In this embodiment, the multi-source vibration correlation signal of the rolling mill refers to various multi-dimensional acquisition signals directly related to the rolling mill vibration phenomenon, specifically including three core signals: vibration signals: vibration state data captured by vibration acquisition components at parts such as the rolling mill stand, work rolls, and roller conveyor, reflecting the vibration intensity and frequency of the components; stress signals: stress data of the rolling mill's stressed components (such as rolls and stands) obtained by stress acquisition components, reflecting the transmission and distribution of force during the rolling process; and position signals: strip position information data captured by position acquisition components along the strip transport path, reflecting the position offset and attitude change of the strip during the transport process.

[0018] In this embodiment, the historical rolling data and vibration records refer to the collection of past rolling process-related data stored in the vibration recorder; the historical rolling data refers to the records of the core process parameters of the rolling mill over a period of time, including historical rolling force (hydraulic system pressure value), historical rolling speed (motor speed), historical roll phase (transmission mechanism angle), and other parameters that directly affect the rolling state; the vibration records refer to the records of key characteristics of the mill vibration phenomenon over a period of time, including historical vibration amplitude, historical vibration duration, historical vibration frequency and corresponding occurrence time and acquisition location information.

[0019] In this embodiment, the real-time data of the vibration recorder refers to the rolling mill operating status data collected by the vibration recorder in real time. The core of the data is the real-time vibration amplitude and frequency of key parts such as the mill stand and work rolls. Each data point carries a collection time stamp. The correction execution data refers to the parameter adjustment related data generated during the adaptive vibration correction process. This includes dynamically generated correction parameters such as the rolling force adjustment amplitude, rolling speed adjustment direction, and roll phase adjustment angle, as well as the actual parameter values ​​and adjustment execution time records fed back by each control module after adjustment. Each data point also carries a time stamp.

[0020] In this embodiment, the process of the multimodal fusion module performing time-series correlation analysis and extracting fusion features is as follows: First, the vibration, stress, and position signals after classification and preprocessing are organized into time-series data sequences according to the acquisition time order, and an independent input channel of the multimodal fusion module is assigned to each type of sequence; the module extracts the feature values ​​of signals from different channels at the same time node through the time-series correlation layer, calculates the correlation coefficient between various feature values, analyzes the synchronous trend of signal changes over time, and completes the time-series correlation analysis; then, the fusion layer integrates these correlation features, focuses on extracting comprehensive features that can reflect the intensity, frequency, and range of influence of mill vibration, organizes the features according to a preset format, and labels the feature name, value, acquisition location, and duration information.

[0021] Specifically, the preset multimodal fusion module includes: an input layer that determines the data type of the module input, corresponding to vibration, stress, and position signals in the multi-source tremor correlation signals of the rolling mill, and assigns an independent input channel to each type of signal; a feature extraction layer that extracts features for different types of signals; a temporal correlation layer that introduces a timestamp alignment mechanism, matches feature data of different types of signals according to the acquisition time, and calculates the correlation coefficient between features to establish a temporal correlation relationship; a fusion layer that adopts a weight allocation strategy, calibrates the contribution of various signal features to tremor detection based on historical rolling data, and obtains a comprehensive fusion feature by weighting according to the contribution; and an output layer that converts the comprehensive fusion feature into a standardized data format.

[0022] Specifically, the process of performing time-series correlation analysis is as follows: First, various types of signals are organized into time-series data sequences according to the acquisition time order, and each type of time-series data sequence corresponds to an input channel of the multimodal fusion model; the multimodal fusion model extracts the feature values ​​of different types of signals at the same time node through time-series correlation, calculates the correlation coefficient between each feature value, and analyzes the trend of different signals changing over time. Specifically, the process of extracting and fusing features and outputting a multi-dimensional vibration detection table is as follows: the multi-modal fusion model integrates the correlation features of various signals and extracts comprehensive features that can reflect the vibration intensity, vibration frequency and vibration influence range of the rolling mill; the comprehensive features are organized in a preset table format, including feature name, feature value, corresponding collection location and duration information, to form the multi-dimensional vibration detection table.

[0023] Specifically, the process of predicting the evolution of vibration data through the prediction model is as follows: historical rolling data and vibration records are retrieved through the vibration recorder; the historical rolling data includes historical rolling force, rolling speed, and roll phase information; the vibration records include historical vibration amplitude and vibration duration; the historical rolling data and vibration records are input into the prediction model, and by learning the pattern of vibration variation with rolling parameters in the historical data, combined with the real-time vibration characteristics in the current multi-dimensional vibration detection table, the trend of vibration amplitude variation and the fluctuation range of vibration frequency within a preset time period are calculated. Specifically, the process of dynamically generating correction parameters is as follows: a preset rolling material attribute library is retrieved, which stores mechanical performance parameters corresponding to different rolling materials; the vibration characteristics in the multi-dimensional vibration detection table are combined with the mapping relationship library to find the associated records that match the current rolling material and vibration characteristics; based on the matched associated records, the type of rolling parameter that needs to be adjusted is determined, the adjustment range and direction of each parameter are calculated, and dynamic correction parameters are formed. Specifically, the process of adjusting the rolling force, rolling speed, and roll phase to obtain the adaptive vibration correction result is as follows: adjust the rolling force, speed, and roll phase of the rolling mill according to the correction parameters; increase or decrease the hydraulic system pressure of the rolling force; adjust the speed of the motor and the angle of the roll transmission mechanism of the roll phase; after the adjustment, feed back the actual parameter values, compare the actual parameter values ​​with the correction parameters, and record the changes in the adjusted parameters and the execution time.

[0024] As a preferred embodiment of the present invention, the specific process of the vibration correction collaborative integration unit includes: The vibration amplitude and frequency data of parts such as the mill stand and work rolls are acquired in real time from the vibration recorder; at the same time, the adjustment parameters of rolling force, rolling speed, and roll phase, as well as the actual execution results, are retrieved from the tremor correction execution module. Both types of data carry the original time stamp. The time stamps of the two types of data are uniformly converted to the workshop standard time, and the data matching at the same time point is compared. If there is a time deviation, the time stamps of the correction execution data are interpolated or fine-tuned based on the vibration recorder time to ensure that the same time point accurately corresponds to a set of vibration data and a set of correction data. A fixed calibration cycle is preset, and the process is automatically triggered when the cycle is reached: the built-in standard signal generation module calls the standard vibration signal with fixed amplitude and frequency, and inputs it into the recorder's acquisition terminal through a dedicated line; the difference in amplitude and frequency between the detection signal output by the recorder and the standard signal is compared, and the signal amplification coefficient and frequency response threshold of the recorder are automatically adjusted according to the difference until the difference meets the preset allowable range, the calibration is completed and the parameter changes are recorded; The vibration data after time alignment and the corrected execution data are summarized and superimposed with the calibration records (including calibration time, standard signal parameters, and recorder parameters before and after adjustment). All data are converted into a unified structured format, sorted by generation time, and each data entry is labeled with its type and corresponding vibration detection table number. Finally, the data is stored in the system database to form a complete vibration correction collaborative integration dataset, providing data support for subsequent operating condition optimization.

[0025] Specifically, the process of synchronously reading real-time data from the vibration recorder and corrective execution data for time alignment is as follows: real-time vibration data is obtained through the vibration recorder, and corrective execution data is obtained from the adaptive vibration correction unit. Both types of data carry time stamps. The time stamps are uniformly converted into a preset standard time. The two types of data under the same standard time are compared. If there is a time deviation, the time stamp of the corrective execution data is adjusted based on the time stamp of the vibration recorder. Specifically, the process of performing periodic automatic calibration on the vibration recorder is as follows: the system presets a fixed calibration cycle, and the calibration process is automatically triggered when the cycle is reached; a standard vibration signal is called and input to the vibration recorder's acquisition terminal through the transmission line; after receiving the standard signal, the vibration recorder outputs a detection signal, and the system compares the amplitude and frequency difference between the detection signal and the standard signal; calibration parameters are automatically generated based on the difference, and the recorder's signal amplification coefficient and frequency response parameters are adjusted until the difference meets the preset requirements.

[0026] Specifically, the process of forming the vibration correction collaborative integration dataset is as follows: summarizing the real-time vibration data after time alignment, correction execution data, and automatic calibration record data; converting data from different sources into a preset standard format and sorting them according to the time of data generation; labeling each data entry with its data type and corresponding vibration detection table number to form the vibration correction collaborative integration dataset. Specifically, the process of performing adaptive anti-interference processing for different working conditions in the workshop is as follows: determine the current working condition type in the workshop, which includes rolling condition, roll changing condition, and strip biting condition; call the corresponding anti-interference scheme for different working conditions, activate the motor base frequency interference filtering scheme for rolling condition, activate the impact interference suppression scheme for roll changing condition, and activate the contact impact interference compensation scheme for strip biting condition.

[0027] In this embodiment, different workshop operating conditions refer to different operating scenarios in the workshop during the steel rolling production process, specifically including three core operating conditions: the rolling condition refers to the normal operating scenario of the rolling mill continuously rolling strip steel. At this time, the various components of the rolling mill operate stably, mainly facing fixed frequency interference such as the motor base frequency; the roll changing condition refers to the maintenance scenario of the rolling mill replacing rolls. At this time, the equipment involves disassembly, installation and other operations, mainly facing instantaneous impact interference; the strip biting condition refers to the starting scenario where the strip steel just enters the roll biting area. At this time, the strip steel and the rolls make initial contact, mainly facing contact impact interference.

[0028] In this embodiment, based on the current rolling rhythm, equipment operating status, and process instructions of the mill, the operating condition is determined to be either rolling, roll changing, or strip biting. If it is rolling, the preset motor base frequency interference filtering parameters are immediately retrieved, and the collected signal is filtered to remove fixed interference components that are consistent with the motor operating frequency. If it is roll changing, the impact interference suppression scheme is activated, and the preset buffer threshold is called to remove the instantaneous high-amplitude interference signal generated during the roll changing operation, so as to avoid interference affecting the accuracy of the data. If it is strip biting, the contact impact interference compensation scheme is activated, and the signal fluctuation generated when the strip first contacts the roll is compensated and corrected based on the preset compensation value of the historical biting condition data, so as to ensure that the collected signal can truly reflect the actual operating status of the mill.

[0029] In this embodiment, the tiered emergency control mechanism refers to a preset mechanism that executes corresponding control measures in different levels according to the degree to which the mill tremor amplitude exceeds the safety threshold. The specific tiers and corresponding measures are as follows: Mild exceedance: The vibration amplitude slightly exceeds the safety threshold. Parameter fine-tuning is initiated, adjusting only core parameters such as rolling force and rolling speed by a small margin. Production does not need to be interrupted. Moderate exceedance: The vibration amplitude significantly exceeds the safety threshold. Speed ​​reduction check and control are initiated, reducing the rolling speed to a safe range while simultaneously monitoring the vibration trend. If the exceedance continues, proceed to the next step. Severe exceedance: The vibration amplitude severely exceeds the safety threshold. Emergency shutdown control is initiated, immediately stopping the mill operation and triggering an alarm to prevent equipment damage or batch product defects.

[0030] In this embodiment, taking the scenario of rolling medium-thick plates of material A in a hot-rolled medium-thick plate workshop as an example, the specific implementation includes: Multi-dimensional signal acquisition and preprocessing: Vibration (vibration state of components), stress (force value of rolls), and position (transmission offset of medium and heavy plates) signals are acquired through acquisition components in the mill stand, work rolls, and medium and heavy plate transmission path; and processed according to type—vibration signals are filtered, stress signals are smoothed, and position signals are corrected for deviation, resulting in clean processed signals; Multimodal fusion and detection table output: The processed signal is input into the multimodal fusion module. First, it is organized into a time series data sequence according to the acquisition time interval t. The features of different types of signals are aligned by timestamps, and the correlation coefficient between features is calculated to complete the time series correlation analysis. Then, the fusion features such as jitter intensity and jitter frequency are extracted by weighting according to the contribution of each type of signal to jitter detection (based on historical data calibration). The fusion features are organized into a multi-dimensional jitter detection table according to the preset format. Vibration prediction and correction parameter generation: Combining historical data stored in the vibration recorder (historical rolling data of material A such as rolling force, rolling speed, and roll phase, as well as corresponding vibration records with vibration amplitude and duration), the prediction model is input to learn the law of vibration change with parameters, and predict future vibration evolution data (including vibration amplitude change trend and vibration frequency fluctuation range); the mechanical property parameters of material A are retrieved, and the adjustment range of rolling force, rolling speed, and roll phase is calculated according to the mapping relationship of "material mechanical parameters - vibration characteristics - parameter adjustment direction", generating dynamic correction parameters; Data Collaboration and Recorder Calibration: Synchronously read the real-time vibration data (including real-time amplitude and frequency) and correction execution data (adjusted actual parameter values ​​and execution time) from the vibration recorder, convert the time stamps of the two types of data to the workshop standard time and align them; automatically calibrate the recorder according to a preset cycle—call the built-in standard signal S into the recorder, compare the difference between the detection signal and the standard signal S, adjust the recorder parameters until the difference meets the requirements, and summarize the data to form a vibration correction collaborative integrated dataset; Operating condition optimization and emergency control: Based on the collaborative integrated dataset, if the current condition is rolling, the motor base frequency interference filtering scheme is activated; if the condition is changing rolls, the impact interference suppression scheme is activated; if the condition is strip biting in, the contact impact interference compensation scheme is activated. If the vibration amplitude exceeds the safety threshold, the parameters are fine-tuned if it is slightly exceeded, the speed is reduced for monitoring if it is moderately exceeded, and the machine is shut down urgently if it is severely exceeded. Finally, an operating condition optimization report is generated.

[0031] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A mill vibration detection and correction system integrating a vibration recorder, characterized in that, include: Multi-dimensional signal detection unit, adaptive vibration correction unit, vibration correction collaborative integration unit, and operating condition optimization unit; The multi-dimensional signal detection unit acquires multi-source vibration correlation signals of the rolling mill and performs classification preprocessing according to signal type; the processed signals are input into a preset multi-modal fusion module for time-series correlation analysis, extracting fusion features and outputting a multi-dimensional vibration detection table; The adaptive vibration correction unit, based on the multi-dimensional vibration detection table and combined with the historical rolling data and vibration records stored in the vibration recorder, predicts the vibration evolution data through a prediction model; dynamically generates correction parameters based on the mapping relationship between the rolling material and vibration characteristics, and adjusts the rolling force, rolling speed and roll phase to obtain the adaptive vibration correction execution result; The vibration correction collaborative integration unit, based on the adaptive vibration correction execution result, synchronously reads the real-time data from the vibration recorder and the correction execution data, and performs time alignment. Perform regular automatic calibration on the vibration recorder to generate a vibration correction collaborative integrated dataset; The working condition optimization unit performs adaptive anti-interference processing for different working conditions in the workshop based on the vibration correction collaborative integrated dataset. When the vibration amplitude exceeds the preset safety threshold, the tiered emergency control mechanism is activated, and an operational condition optimization report is generated.

2. The system according to claim 1, characterized in that, The specific process of performing classification preprocessing according to signal type is as follows: the source flutter associated signal is identified and classified into vibration signal, stress signal, and position signal; different types of signals are processed separately, vibration signal is filtered, stress signal is smoothed, and position signal is corrected.

3. The system according to claim 1, characterized in that, The preset multimodal fusion module specifically includes: an input layer, which determines the data type of the input signal, corresponding to vibration, stress, and position signals in the multi-source tremor correlation signals of the rolling mill, and assigns an independent input channel to each type of signal; a feature extraction layer, which extracts features for different types of signals respectively; a temporal correlation layer, which introduces a timestamp alignment mechanism, matches the feature data of different types of signals according to the acquisition time, and calculates the correlation coefficient between features to establish a temporal correlation relationship; a fusion layer, which adopts a weight allocation strategy, calibrates the contribution of various signal features to tremor detection based on historical rolling data, and obtains a comprehensive fusion feature by weighting according to the contribution; and an output layer, which converts the comprehensive fusion feature into a standardized data format.

4. The system according to claim 1, characterized in that, The specific process of performing time-series correlation analysis is as follows: First, various types of signals are organized into time-series data sequences according to the acquisition time order. Each type of time-series data sequence corresponds to an input channel of the multimodal fusion model. The multimodal fusion model extracts the feature values ​​of different types of signals at the same time node through time-series correlation, calculates the correlation coefficient between each feature value, and analyzes the trend of different signals changing over time.

5. The system according to claim 1, characterized in that, The specific process of extracting and fusing features and outputting a multi-dimensional vibration detection table is as follows: The multi-modal fusion model integrates the correlation features of various signals and extracts comprehensive features that can reflect the vibration intensity, vibration frequency and vibration influence range of the rolling mill; the comprehensive features are organized in a preset table format, including feature name, feature value, corresponding collection location and duration information, to form the multi-dimensional vibration detection table.

6. The system according to claim 1, characterized in that, The specific process of predicting the evolution of vibration data through the prediction model is as follows: historical rolling data and vibration records are retrieved through the vibration recorder; the historical rolling data includes historical rolling force, rolling speed and roll phase information; the vibration records include historical vibration amplitude and vibration duration; The historical rolling data and vibration records are input into the prediction model. By learning the pattern of vibration variation with rolling parameters in the historical data and combining it with the real-time vibration characteristics in the current multi-dimensional vibration detection table, the trend of vibration amplitude and the fluctuation range of vibration frequency within a preset time period are calculated.

7. The system according to claim 1, characterized in that, The specific process of dynamically generating correction parameters is as follows: A preset rolling material attribute library is retrieved, which stores the mechanical performance parameters corresponding to different rolling materials; combined with the vibration characteristics in the multi-dimensional vibration detection table, a matching record matching the current rolling material and vibration characteristics is searched in the mapping relationship library; based on the matched matching record, the type of rolling parameter to be adjusted is determined, and the adjustment range and direction of each parameter are calculated to form dynamic correction parameters.

8. The system according to claim 1, characterized in that, The specific process of adjusting the rolling force, rolling speed, and roll phase to obtain the adaptive vibration correction execution result is as follows: adjust the rolling force, speed, and roll phase of the rolling mill according to the correction parameters; increase or decrease the hydraulic system pressure of the rolling force; adjust the speed of the motor and the roll phase by adjusting the angle of the roll transmission mechanism; after the adjustment is performed, the actual parameter values ​​are fed back, the actual parameter values ​​are compared with the correction parameters, and the changes in the adjusted parameters and the execution time are recorded.

9. The system according to claim 1, characterized in that, The specific process of synchronously reading real-time data from the vibration recorder and corrective execution data for time alignment is as follows: real-time vibration data is obtained through the vibration recorder, and corrective execution data is obtained from the adaptive vibration correction unit. Both types of data carry time stamps. The time stamps are uniformly converted into a preset standard time. The two types of data under the same standard time are compared. If there is a time deviation, the time stamp of the corrective execution data is adjusted based on the time stamp of the vibration recorder.

10. The system according to claim 1, characterized in that, The specific process of performing periodic automatic calibration on the vibration recorder is as follows: the system presets a fixed calibration cycle, and the calibration process is automatically triggered when the cycle is reached; a standard vibration signal is called and input to the vibration recorder's acquisition terminal through the transmission line; after receiving the standard signal, the vibration recorder outputs a detection signal, and the system compares the amplitude and frequency difference between the detection signal and the standard signal; calibration parameters are automatically generated based on the difference, and the recorder's signal amplification coefficient and frequency response parameters are adjusted until the difference meets the preset requirements.

11. The system according to claim 1, characterized in that, The specific process for forming the vibration correction collaborative integrated dataset is as follows: summarizing the time-aligned real-time vibration data, correction execution data, and automatic calibration record data; converting data from different sources into a preset standard format and sorting them according to the data generation time order; Each data entry is labeled with its data type and corresponding vibration detection table number to form the vibration correction collaborative integration dataset.

12. The system according to claim 1, characterized in that, The specific process of implementing adaptive anti-interference processing for different working conditions in the workshop is as follows: determine the current working condition type in the workshop, which includes rolling condition, roll changing condition, and strip biting condition; call the corresponding anti-interference scheme for different working conditions, activate the motor base frequency interference filtering scheme for rolling condition, activate the impact interference suppression scheme for roll changing condition, and activate the contact impact interference compensation scheme for strip biting condition.

Images