Multi-state cooperative monitoring and control system for annealing and pickling line loop car

Abstract

The present application relates to the technical field of metallurgical equipment control, in particular to a multi-state cooperative monitoring and control system for an annealing and pickling line loop trolley, comprising a state sensing module, an event marking module, a multi-source correlation analysis module and a control correction module. The state sensing module is provided with vibration sensors and angle sensors at the four corners of the trolley frame and the key hinge points of the lifting mechanism, synchronously collects the structural vibration frequency spectrum and the attitude angle change sequence, and forms the mechanical state original data stream. The event marking module extracts abnormal events through time window sliding analysis, marks the high attention time and generates the correlation data snapshot. The multi-source correlation analysis module aligns and superimposes the correlation data snapshot and the operation condition parameter time sequence, identifies three types of abnormalities. The control correction module generates a closed-loop correction instruction set according to the identification result. The system realizes multi-state cooperative monitoring and accurate control of the loop trolley, and is suitable for operation monitoring and control of the annealing and pickling line loop trolley.

Description

Technical Field

[0001] This invention relates to the field of metallurgical equipment control technology, and in particular to a multi-state collaborative monitoring and control system for the looper trolley of an annealing and pickling line. Background Technology

[0002] Annealing and pickling lines are crucial for steel surface treatment in metallurgical production. The looper trolley, as the core equipment of this line, plays a vital role in regulating steel tension and ensuring production continuity; its operational status directly impacts production efficiency and product quality. Currently, the monitoring and control of the looper trolley employs a traditional single-parameter monitoring mode, deploying only a single type of sensor at a single location on the trolley frame. This only collects data on either vibration or attitude angle, failing to achieve coordinated deployment of multiple sensor types or synchronous acquisition and integration of structural vibration and attitude angle data, thus hindering the formation of comprehensive mechanical condition data.

[0003] In existing technologies, monitoring data is isolated, failing to comprehensively capture the mechanical operating status of the looper trolley and thus failing to reflect the overall operating condition of the equipment. The analysis of monitoring data relies on simple threshold judgments, without combining it with correlation analysis based on equipment operating parameters. This makes it impossible to accurately pinpoint the root cause of anomalies, resulting in the inability to promptly identify and address issues such as abnormal vibrations, structural deformations, and deviations during looper trolley operation. Furthermore, the current control method is open-loop control, unable to generate targeted corrective commands in real time based on detected anomalies, which can easily exacerbate equipment failures and affect the continuous and stable operation of the production line. Therefore, it is necessary to achieve synchronous acquisition of multi-dimensional status data of key components of the looper trolley, conduct correlation analysis combined with operating parameters, accurately identify anomalies, and perform real-time closed-loop corrections to overcome the shortcomings of existing technologies. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and propose a multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a multi-state collaborative monitoring and control system for the looper trolley of an annealing and pickling line, comprising: The state perception module deploys vibration sensors and angle sensors at the four corners of the trolley frame and the key hinge points of the lifting mechanism to synchronously collect the structural vibration spectrum and attitude angle change sequence of the trolley during operation. The structural vibration spectrum and attitude angle change sequence constitute the raw data stream of the mechanical state of the trolley. The event marking module performs time window sliding analysis on the raw data stream of the mechanical state, extracts the vibration amplitude surge event and the attitude angle discontinuous jump event, marks the time point of the vibration amplitude surge event and the attitude angle discontinuous jump event as the high attention moment, and generates the associated data snapshot of the high attention moment; The multi-source correlation analysis module uses the correlation data snapshot of the high-attention moment to reverse look up the operating condition parameters of the looper at the corresponding moment, performs time-series alignment and superposition analysis of the original mechanical state data stream and the operating condition parameters, and identifies the abnormal shaking mode caused by mechanical resonance, the structural deformation trend caused by overload impact, and the travel deviation induced by track deformation. The control correction module generates a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation.

[0006] As a further aspect of the present invention, a time-window sliding analysis is performed on the raw mechanical state data stream to extract events of sudden increase in vibration amplitude and discontinuous jumps in attitude angle, including: A fixed sampling window of several seconds is set, and the envelope of the vibration spectrum of the structure is extracted. The ratio of the peak value to the root mean square value of the envelope is calculated. When the ratio exceeds the preset mechanical health threshold, it is determined that a sudden increase in vibration amplitude has occurred within the current sampling window. Perform a first-order difference operation on the attitude angle change sequence to capture the moment when the rate of change of attitude angle exceeds a preset physical limit value per unit time, and mark the moment when the attitude angle discontinuous jump event occurs. A circular buffer is established to store the analysis results of the most recent consecutive sampling windows. When multiple consecutive sampling windows are detected to have the vibration amplitude increase event or the attitude angle discontinuous jump event, the logic for generating the high-interest moment is triggered. When the high-attention moment arrives, the original data streams of several sampling windows before and after this moment are captured, and together with the system timestamp of the event occurrence, they are packaged and encapsulated into a data snapshot associated with the high-attention moment.

[0007] As a further aspect of the present invention, by using the associated data snapshot of the high-attention moment, the operating condition parameters of the looper trolley at the corresponding moment are retrieved, including: The operating parameters include the current load curve of the traction motor, the tension feedback value of the wire rope, and the straightness detection results of the track. The system timestamp contained in the associated data snapshot of the high-concern moment is parsed and converted into a global tick clock count of the programmable logic controller of the annealing and pickling production line; Based on the global cycle clock count obtained by conversion, access the historical archive database of the production line data acquisition server, and retrieve the traction motor current timing data, wire rope tension meter reading, and track profile data obtained by the laser rangefinder corresponding to the global cycle clock count. The retrieved traction motor current timing data is filtered and denoised to remove high-frequency noise introduced by power grid fluctuations and restore the true current load curve. The retrieved track contour data is subjected to coordinate transformation processing and uniformly mapped to the local coordinate system of the looper trolley body for subsequent deformation trend analysis. The processed current load curve, wire rope tension meter reading, and track profile data are then bound to the current high-concern moment to form a complete operating condition parameter data package.

[0008] As a further aspect of the present invention, the raw mechanical state data stream and the operating condition parameters are time-aligned and superimposed for analysis to identify abnormal vibration patterns caused by mechanical resonance, including: The operating frequency of the lifting cylinder of the looper trolley is extracted from the raw data stream of the mechanical state and compared with the main peak frequency in the vibration spectrum of the structure. When the main peak frequency in the vibration spectrum of the structure is detected to be close to or equal to an integer multiple of the operating frequency of the lifting cylinder, it is determined that the conditions for mechanical resonance exist. Further analysis was conducted to determine whether the main peak frequency still existed in the vibration spectrum of the structure when the looper trolley was in a static tensioned state. If it still existed, external excitation caused by motion was ruled out, and the abnormal shaking mode was confirmed to be internal mechanical resonance. Record the load weight of the looper trolley, the hydraulic cylinder pressure, and the current temperature of the track when the internal mechanical resonance occurs, as auxiliary criteria for inducing the internal mechanical resonance.

[0009] As a further aspect of the present invention, the original mechanical state data stream and the operating condition parameters are time-series aligned and superimposed for analysis to identify the structural deformation trend caused by overload impact, including: Extract the current peak curves of the traction motor during the start-up and braking phases from the operating condition parameters, calculate the area enclosed by the current peak curves, and use it as the impact energy integral value of this start-up and stop process. The calculated impact energy integral value is correlated with the fatigue accumulation model of the looper trolley frame material to predict the micro-strain growth at key connection parts of the frame. Simultaneously, the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy is extracted from the attitude angle change sequence. When it is found that the micro-strain increase or the residual attitude angle offset shows a monotonically increasing trend after multiple start-stop cycles, it is confirmed that there is a structural deformation trend caused by overload impact, and further acceleration control commands are stopped.

[0010] As a further aspect of the present invention, the original mechanical state data stream and the operating condition parameters are time-series aligned and superimposed for analysis to identify the deviation induced by track deformation, including: By utilizing the characteristics of a specific frequency band in the vibration spectrum of the structure, a characteristic signal generated by poor contact between the wheel and the track is separated. The amplitude of the characteristic signal is positively correlated with the local elevation difference of the track. Based on the real-time running speed of the looper trolley in the operating condition parameters, the propagation delay of the characteristic signal in the direction of travel of the looper trolley is calculated; The deviation between the actual travel path of the looper trolley and the theoretical straight path is calculated. When the deviation value continuously exceeds the allowable tolerance and is accompanied by the enhancement of the characteristic signal, it is confirmed that there is a travel deviation induced by track deformation, and a deceleration suggestion is generated for the road section.

[0011] As a further aspect of the present invention, the step of generating a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation includes: The closed-loop correction instruction set includes the dynamic adjustment coefficient of motor torque, the delay threshold of brake action, and the extension and retraction stroke instructions of the correction hydraulic cylinder. When the identification result is the abnormal jitter mode, the dynamic adjustment coefficient of the motor torque is extracted from the closed-loop correction instruction set. The output torque fluctuation of the traction motor is reduced by decreasing the dynamic adjustment coefficient, and the delay threshold of the brake action is increased to absorb the vibration energy of the mechanical system. When the identification result is the structural deformation trend, the current closed-loop correction instruction set is frozen, and an equipment maintenance request is sent to the central scheduling system of the production line. The automatic operation mode of the looper trolley is suspended and manual intervention is awaited. When the identification result is the amount of deviation, the extension and retraction command of the correction hydraulic cylinder is activated. According to the direction and magnitude of the deviation, the correction hydraulic cylinder on one side of the looper trolley is extended and the correction hydraulic cylinder on the other side is retracted to forcibly correct the travel direction. The adjusted dynamic adjustment coefficient of the motor torque, the delay threshold of the brake action, and the extension and retraction stroke command of the correction hydraulic cylinder are packaged and sent to the field programmable gate array controller of the looper trolley for execution.

[0012] As a further aspect of the present invention, the step of generating a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation amount further includes: The preventive correction module, after identifying the abnormal jitter pattern each time, not only corrects the current motor torque, but also retrieves the frequency and duration of similar jitter events in the past period to construct a time series prediction model. Using the time series prediction model, the time window for the next severe jitter can be predicted; A few seconds before the predicted time window arrives, the dynamic adjustment coefficient of the motor torque is modified in advance to keep it at a low level, thereby achieving preventive suppression of potential resonance risks. Simultaneously, the predicted confidence level information and the current correction command are sent to the field programmable gate array controller as a basis for priority determination.

[0013] As a further aspect of the present invention, the system timestamp contained in the associated data snapshot of the high-concern moment is parsed and converted into a global clock count of the programmable logic controller of the annealing and pickling production line, including: Read the system timestamp, recorded in absolute time format, from the metadata header of the associated data snapshot at the high-concern moment; Obtain the latest heartbeat synchronization message from the programmable logic controller (PLC) of the annealing and pickling production line, and extract the reference mapping relationship between the global tick clock count and the absolute reference time maintained internally by the PLC from the message; The system timestamp is subtracted from the absolute reference time in the baseline mapping relationship to calculate the time offset of the system timestamp relative to the absolute reference time. The time offset is divided by the known refresh period of the global tick clock counter of the programmable logic controller to obtain the offset in tick count. The offset, expressed in terms of beat count, is added to the global beat clock count recorded in the reference mapping relationship to complete the conversion from the system timestamp in absolute time format to the global beat clock count used inside the programmable logic controller.

[0014] As a further aspect of the present invention, extracting the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy from the attitude angle change sequence includes: Determine the start and end times of the impact event corresponding to the impact energy integral value, and use the end time of the impact event as the starting reference point for extraction. Extract all attitude angle data within a fixed subsequent observation period starting from the initial reference point from the attitude angle change sequence to form the attitude angle subsequence to be analyzed; The attitude angle sequence to be analyzed is subjected to low-pass filtering to remove high-frequency fluctuation components caused by the normal operation of the looper trolley and sensor noise, resulting in a smoothed attitude angle sequence. In the smoothed attitude angle sequence, the last obvious turning point is identified, and the attitude angle data after the turning point remains stable within a preset steady-state determination time. Calculate the average attitude angle from the starting reference point to the turning point, and simultaneously calculate the average attitude angle during the steady-state determination time after the turning point. The difference between the mean attitude angle from the starting reference point to the turning point and the mean attitude angle within the steady-state determination time after the turning point is the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: Vibration and angle sensors are installed at the four corners of the trolley frame and key hinge points of the lifting mechanism to simultaneously collect the structural vibration spectrum and attitude angle change sequence of the trolley during operation. These two types of data are combined to form a raw mechanical state data stream. This method overcomes the limitations of single-parameter monitoring, enabling the simultaneous capture of multi-dimensional state data of key parts of the trolley. It comprehensively reflects the structural vibration and attitude changes during equipment operation, avoiding the omission of key abnormal signals caused by single-sensor monitoring. This makes the raw mechanical state data more complete and synchronized, accurately capturing subtle anomalies during equipment operation and providing comprehensive and reliable basic data for subsequent anomaly identification.

[0016] A time-window sliding analysis is performed on the raw mechanical status data stream to extract events such as sudden increases in vibration amplitude and discontinuous jumps in attitude angle. High-interest moments are marked and associated data snapshots are generated. The corresponding operating parameters at those moments are then retrieved using these snapshots. The raw mechanical status data stream and operating parameters are time-series aligned and overlaid to identify abnormal vibration patterns caused by mechanical resonance, structural deformation trends due to overload impact, and track deviation induced by track deformation. Based on the identification results, a closed-loop correction instruction set is generated to control the servo drive system of the looper trolley. This method achieves a deep correlation between monitoring data and operating conditions, accurately locating the root cause of anomalies and avoiding misjudgments and omissions caused by simple threshold judgments. Furthermore, the closed-loop correction instruction set can make targeted adjustments for different types of anomalies, enabling real-time correction of the looper trolley's operating status, effectively preventing anomaly escalation, ensuring equipment stability, and reducing equipment downtime. Attached Figure Description

[0017] Figure 1This is a timing diagram of the multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line described in this invention. Figure 2 A flowchart for reverse-engineering the operating parameters of the looper trolley; Figure 3 A time-series diagram for real-time vibration amplitude monitoring of the looper trolley; Figure 4 For real-time monitoring of path deviation and over-limit judgment curve of the looper trolley; Figure 5 The curve showing the time-series variation of the dynamic adjustment coefficient of the trolley motor torque. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] See Figure 1The state perception module deploys vibration and angle sensors at the four corners of the trolley's frame and key hinge points of the lifting mechanism. These sensors synchronously collect the structural vibration spectrum and attitude angle change sequence during the trolley's operation. The structural vibration spectrum and attitude angle change sequence together constitute the raw data stream characterizing the mechanical state of the trolley. The event marking module performs a time-window sliding analysis on this raw mechanical state data stream, extracting events of sudden increases in vibration amplitude and discontinuous jumps in attitude angles. The time points of these events are marked as high-interest moments, and a corresponding snapshot of the associated data is generated. The multi-source correlation analysis module uses the generated snapshots of the high-interest moments to look up the trolley's operating parameters at the corresponding times. Then, the raw mechanical state data stream and operating parameters are time-series aligned and overlaid. This analysis identifies abnormal shaking patterns caused by mechanical resonance, structural deformation trends caused by overload impact, and deviation induced by track deformation. The control correction module generates a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter patterns, structural deformation trends, and travel deviation, thereby realizing real-time adjustment and optimization of the looper trolley's operating status.

[0021] In one embodiment of the invention, a time-window sliding analysis is performed on the raw mechanical state data stream to extract events and mark moments of high interest. The length of the fixed sampling window is set to 2 seconds. Within each fixed sampling window, the system synchronously processes structural vibration spectrum data from vibration sensors. A Hilbert transform is performed on the structural vibration spectrum to extract the envelope, and the ratio of the envelope peak value to the root mean square value is calculated. When this ratio exceeds a preset mechanical health threshold, a vibration amplitude surge event is determined to have occurred within the current fixed sampling window. Attitude angle event detection is performed independently and in parallel. A first-order difference operation is performed on the attitude angle change sequence from the angle sensor to directly calculate the rate of change of angle between adjacent sampling points. When the rate of change of attitude angle exceeds a preset physical limit value for three consecutive sampling periods, the system captures this moment and marks it as the occurrence time of a discontinuous jump in attitude angle.

[0022] In some embodiments, the event marking module establishes a circular buffer with 256 storage units. This buffer stores the analysis results of the most recent 128 consecutive fixed sampling windows in chronological order. Each fixed sampling window's analysis result contains a status word that records the determination flags for sudden vibration amplitude increases and discontinuous attitude angle changes. When the circular buffer detects that the determination flag for a sudden vibration amplitude increase event is set within eight consecutive fixed sampling windows, the logic for generating a high-interest moment is triggered. Similarly, when the determination flag for a discontinuous attitude angle change event is set within three consecutive fixed sampling windows, the logic for generating a high-interest moment is also triggered.

[0023] In practice, once the logic for generating the high-interest moment is triggered, the system immediately locks the current time as the high-interest moment. Using the high-interest moment as a reference point, it backwards and forwards four fixed sampling windows of raw mechanical state data streams, simultaneously capturing four fixed sampling windows of raw mechanical state data streams. The captured raw mechanical state data streams contain the original structural vibration spectrum and attitude angle change sequence. The precise timestamp read by the system kernel when the high-interest moment occurs is appended to the data block. All data, along with the high-interest moment timestamp, is encapsulated into a data packet with specific header information; this data packet constitutes the associated data snapshot of the high-interest moment. The associated data snapshot of the high-interest moment is then pushed to a message queue for consumption by the multi-source correlation analysis module.

[0024] It is understandable that the ratio of the peak value to the root mean square value of the envelope is calculated according to the following formula:

[0025] in: This represents the ratio of the peak value to the root mean square value of the envelope. This represents the maximum amplitude value of the envelope signal extracted within a fixed sampling window. This represents the root mean square value of the envelope signal within the same fixed sampling window. The preset mechanical health threshold is an empirical constant derived from statistical analysis of the looper's historical normal operating data. In practice, the first-order difference operation is performed directly on the original sampled values ​​of the attitude angle change sequence, while the physical limit value is set based on the maximum theoretical angular velocity of the looper's mechanical structure and the system sampling period. The first-in, first-out characteristic of the circular buffer ensures that the stored data is always the latest analysis results from the continuous fixed sampling window.

[0026] In some embodiments, the encapsulation format of the high-interest moment-related data snapshot is predefined. The header of the high-interest moment-related data snapshot includes a synchronization word, a snapshot version number, and the total data length. The body of the high-interest moment-related data snapshot sequentially stores the absolute timestamp of the high-interest moment, the event type code that triggered the high-interest moment, and multiple blocks of raw sensor data from fixed sampling windows arranged in chronological order. It can be understood that the generation of the high-interest moment-related data snapshot is event-driven, executing only when the circular buffer meets the continuous event condition. This effectively filters out sporadic, isolated interference signals, ensuring that the high-interest moment-related data snapshot targeted for subsequent analysis has clear physical meaning and high analytical value.

[0027] In one embodiment of the present invention, see [reference] Figure 2The process involves reverse-engineering the operating parameters of the looper trolley at a corresponding moment by referencing a snapshot of data associated with a high-priority moment. These operating parameters include the current load curve of the traction motor, the tension feedback value of the wire rope, and the straightness detection results of the track. The reverse-engineering operation begins by parsing the snapshot of data associated with a high-priority moment, reading the system timestamp recorded in the format "year-month-day hour:minute:second.millisecond" from the metadata header of the snapshot. The core of the time conversion is obtaining the heartbeat synchronization message issued by the programmable logic controller of the annealing and pickling production line. The heartbeat synchronization message is broadcast once per second and contains a 32-bit global clock count field and a 64-bit absolute reference timestamp field. These two fields together constitute the reference mapping relationship between the global clock count and the absolute reference time. The time offset of the absolute reference time in the system timestamp-reference mapping is calculated. This time offset is then divided by the fixed refresh period of the programmable logic controller's global tick clock counter, yielding an offset in tick counts. This offset is added to the global tick clock count recorded in the reference mapping, completing the conversion from system time to the controller's internal tick clock. The conversion relationship can be expressed as:

[0028] in: This represents the global clock count of the programmable logic controller (PLC) for the annealing and pickling production line after conversion. This represents the baseline global tick clock count obtained from the heartbeat synchronization message. This represents the system timestamp parsed from the snapshot of data associated with high-priority moments. This represents the baseline absolute reference time obtained from the heartbeat synchronization message. This indicates the fixed refresh period of the global tick clock counter of the programmable logic controller.

[0029] In some embodiments, the historical archive database of the production line data acquisition server is accessed based on the converted global beat clock count. The historical archive database uses the global beat clock count as the primary index and stores time-series data with millisecond-level precision. Traction motor current timing data, wire rope tension meter readings, and track profile data acquired by a laser rangefinder corresponding to the global beat clock count are retrieved. The traction motor current timing data comes from a Hall current sensor installed at the output of the motor driver, with a sampling frequency of 10kHz. The wire rope tension meter readings come from a tension sensor installed on the fixed pulley block bracket, with a sampling frequency of 1kHz. The track profile data is generated by scanning with a laser rangefinder array installed on the side of the looper trolley's traveling mechanism; the array acquires the contour point cloud of the track cross-section at a frequency of 100Hz.

[0030] In practice, the retrieved traction motor current timing data undergoes filtering and noise reduction. This filtering and noise reduction uses a second-order Butterworth low-pass digital filter with a cutoff frequency of 500Hz. This filter effectively eliminates high-frequency noise introduced by fluctuations in the workshop power grid, restoring the current load curve that reflects the actual load changes. The retrieved track profile data undergoes coordinate transformation. This transformation requires converting the track profile data acquired by the laser rangefinder from the laser rangefinder coordinate system to a local coordinate system of the vehicle body, with the geometric center of the looper trolley as the origin, based on the installation matrix of the laser rangefinder array on the trolley body. The processed current load curve, wire rope tension meter readings, and track profile data are then bound to the high-interest moment that triggered the query and packaged together to form a complete operating condition parameter data package.

[0031] It is understandable that the accuracy of timestamp conversion directly determines the accuracy of multi-source data timing alignment. The reference mapping relationship provided by the heartbeat synchronization message ensures that even during a short delay after a high-priority moment, a precise corresponding global tick clock count can be obtained through interpolation calculation. Optionally, when the time offset between the system timestamp and the reference absolute reference time at the high-priority moment is not an integer multiple of the fixed refresh period of the global tick clock counter, the global tick clock count obtained through conversion calculation will be... It may be a floating-point number, and the system will round down to determine a unique integer tick count for indexing the historical archive database. The historical archive database uses a circular storage structure and can store all data from the most recent 72 hours, ensuring that the operating parameters corresponding to any snapshot of data of high interest can be effectively retrieved.

[0032] In some embodiments, the wire rope tension meter readings and track profile data need to undergo unit unification and validity verification before binding. The original unit of the wire rope tension meter reading is a millivolt voltage signal, which needs to be converted into a force value in kilonewtons according to the sensor's calibration coefficient. After coordinate transformation, the track profile data undergoes outlier removal and smoothing to eliminate measurement noise. It is understood that the encapsulation format of the operating condition parameter data packet is fixed. The data packet header contains the timestamp of the source of high interest, the converted global clock count, and the length information of various data types. Optionally, a data checksum field is also retained in the data packet to ensure the integrity of the operating condition parameter data packet during subsequent transmission and processing. The entire process of reverse lookup, retrieval, processing, and encapsulation is typically completed within milliseconds to ensure that the multi-source correlation analysis module can obtain a synchronized and accurate operating condition parameter data packet in a timely manner.

[0033] In one embodiment of the present invention, the raw mechanical state data stream and operating condition parameters are time-aligned and superimposed to identify specific abnormal states. The process of identifying abnormal vibration patterns caused by mechanical resonance begins by extracting the operating frequency of the lifting cylinder of the looper trolley from the raw mechanical state data stream. The operating frequency of the lifting cylinder is obtained by performing spectral analysis on the cylinder pressure sensor signal. In a specific implementation, the three main peak frequencies with the most concentrated energy are identified from the structural vibration spectrum, and the main peak frequencies in the structural vibration spectrum are compared with the operating frequency of the lifting cylinder. For example, when the operating frequency of the lifting cylinder is 150Hz, and there is a main peak frequency of 450Hz in the structural vibration spectrum, the system determines that the main peak frequency is close to an integer multiple of the operating frequency of the lifting cylinder, indicating that there is a condition for mechanical resonance. Further analysis is conducted to see if the same main peak frequency still exists in the structural vibration spectrum obtained from the same vibration sensor when the looper trolley is in a static tensioned state. If the structural vibration spectrum still shows a significant peak at 450Hz in the static tensioned state, external excitation caused by motion is ruled out, and the abnormal vibration pattern is confirmed to be internal mechanical resonance. Record the load weight of the looper trolley, the cylinder pressure, and the current temperature of the track when internal mechanical resonance occurs.

[0034] In some embodiments, the structural deformation trend caused by overload impact is identified, and the current spike curves of the traction motor during the start-up and braking phases are extracted from the operating parameters. The current spike curves are preprocessed, and the average current value during smooth motor operation is subtracted to obtain the net impact current curve. The area enclosed by the net impact current curve over the impact duration is calculated and used as the integral value of the impact energy for this start-up and shutdown process. The calculation follows the formula:

[0035] in: This represents the integral value of the impact energy during this start-up and shutdown process. This represents the current value of the traction motor at time t. This represents the average current value of the motor during its stable operation phase. and These represent the start and end times of the impact event, respectively. The calculated impact energy integral value is correlated with the pre-stored fatigue accumulation model of the looper trolley frame material. The fatigue accumulation model takes the impact energy integral value as input and outputs the predicted value of the micro-strain growth at the key connection parts of the frame.

[0036] In practice, the residual attitude angle offset of the looper trolley after being subjected to the integrated value of impact energy is extracted from the attitude angle change sequence. The start and end times of the impact event corresponding to the integrated value of impact energy are determined, and the end time of the impact event is used as the starting reference point for extraction. All attitude angle data within a fixed subsequent observation time period starting from the starting reference point are extracted from the attitude angle change sequence to form the attitude angle subsequence to be analyzed. The subsequent observation time period is usually set to 5 seconds. The attitude angle subsequence to be analyzed is subjected to low-pass filtering using a Butterworth filter with a cutoff frequency of 2Hz to filter out high-frequency fluctuation components caused by the normal operation of the looper trolley and sensor noise, resulting in a smoothed attitude angle sequence. In the smoothed attitude angle sequence, the last obvious inflection point is identified. The attitude angle data after the inflection point remains stable within a preset steady-state determination time, which is set to 2 seconds. The mean attitude angle from the starting reference point to the inflection point is calculated, and the mean attitude angle within the steady-state determination time after the inflection point is also calculated. The difference between the mean attitude angle from the starting reference point to the turning point and the mean attitude angle within the steady-state determination time after the turning point is the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy.

[0037] It is understandable that determining the structural deformation trend is a process based on historical comparison. The system maintains a circular queue to store the predicted values ​​of micro-strain growth and residual attitude angle offsets calculated in the most recent consecutive start-stop cycles. When the system finds that the predicted value of micro-strain growth shows a monotonically increasing trend in ten consecutive start-stop cycles, or that the residual attitude angle offset shows a monotonically increasing trend in five consecutive start-stop cycles, it confirms the existence of a structural deformation trend caused by overload impact. Optionally, the criterion for determining a monotonically increasing trend does not require that each value be strictly greater than the previous one. Instead, a linear regression method is used to calculate the slope of the sequence of values ​​over time. When the slope exceeds a set positive threshold, a monotonically increasing trend is determined to exist. In some embodiments, the fatigue accumulation model used to correlate the predicted micro-strain growth is a database lookup table established offline based on the SN curve of the frame material and Mainner's linear cumulative damage theory. The impact energy integral value is used as an index to directly retrieve the corresponding predicted value of micro-strain growth. It is understandable that the residual attitude angle offset extraction algorithm can effectively separate the plastic deformation component caused by impact from the elastic vibration component during operation. The setting of the steady-state determination time window ensures that the extracted offset is a stable measurement value after the system returns to a quasi-static state.

[0038] See Figure 3In the real-time vibration amplitude time-series analysis of the looper trolley with multi-parameter collaborative monitoring, the system visualizes the time-domain signals collected by the frame vibration sensors, intuitively capturing the abnormal fluctuation characteristics of the mechanical state. The curve in the figure shows the dynamic change of the vibration amplitude (unit: g) during the operation of the looper trolley from 0 to 12 seconds. The overall amplitude remains within the normal fluctuation range of ±0.5g, with only one sharp increase in amplitude exceeding 2.2g at about 7 seconds, which meets the criteria for a sudden increase in vibration amplitude. The extraction of this event relies on the time window sliding analysis technique: the system processes the envelope of the vibration signal with a fixed-length sampling window, calculates the ratio of the envelope peak value to the root mean square value, and when this ratio exceeds the preset mechanical health threshold, it is marked as a high-interest moment and a related data snapshot is generated. By performing time-series alignment and overlay analysis of this sudden vibration event with operating parameters such as traction motor current and wire rope tension, the cause of the anomaly can be further identified: if the moment corresponds to a current spike during the motor's starting / braking phase, it points to structural vibration caused by overload impact; if the vibration peak frequency is an integer multiple of the lifting cylinder's operating frequency, further verification of the spectral characteristics under static tension is needed to confirm whether it is internal mechanical resonance. This visualization chart provides an intuitive time-series feature input for the multi-source correlation analysis module and is the core data carrier for identifying abnormal vibration patterns, structural deformation trends, and travel deviation.

[0039] In one embodiment of the present invention, the amount of wheel deviation induced by track deformation is identified, and the characteristic signal generated by poor contact between the wheel and the track is separated using the characteristics of the 50Hz to 150Hz frequency band in the structural vibration spectrum. The structural vibration spectrum is acquired by vibration sensors installed at the four corners of the frame. By performing bandpass filtering and envelope analysis on the original spectrum, the energy amplitude of a specific frequency band is extracted as an intensity index of the characteristic signal. The amplitude of the characteristic signal is positively correlated with the local elevation difference of the track traversed by the wheel; that is, the more significant the local depression or convexity of the track, the higher the amplitude of the characteristic signal. An example of the correspondence between the characteristic signal amplitude and typical local elevation differences of the track is shown in Table 1.

[0040] Table 1: Correspondence between characteristic signal amplitude and local elevation difference of the track

[0041] In some embodiments, the propagation delay of the characteristic signal in the direction of travel of the looper trolley is calculated, and the real-time operating speed of the looper trolley is considered in the operating condition parameters. The operating speed is calculated from the encoder feedback pulse frequency of the servo driver. The calculation of the propagation delay is based on the spatial wavelength of the characteristic signal on the track and the operating speed of the looper trolley, and is expressed by the following formula:

[0042] in: This represents the propagation delay of the characteristic signal in the direction of travel of the looper trolley. This represents the spatial wavelength of the characteristic signal exhibited in the structural vibration spectrum by wheel excitation caused by periodic track deformation. This represents the real-time operating speed of the looper trolley, obtained from the operating condition parameters. The characteristic signal spatial wavelength... This data was obtained by analyzing stable vibration modes that repeatedly occurred in historical data and were caused by known track welds or fixed defects. The values ​​reflect the typical intervals of track deformation. For example, when the spatial wavelength of the characteristic signal... The real-time operating speed of the looper trolley is 2 meters. The calculated propagation delay at a speed of 1 m / s For 2 seconds. Optional, the spatial wavelength of the characteristic signal. The value can be determined by performing spectral analysis on multiple sets of historical data and taking the average spatial period of the main frequency components.

[0043] In practical implementation, the deviation between the actual travel path of the looper trolley and the theoretical straight path is calculated. The actual travel path is obtained by integrating the continuous distance measurements taken by the laser displacement sensor mounted on the trolley body from the lateral reference plane of the track. The theoretical straight path is determined by the production line installation coordinate system. The deviation calculation is performed in real time, and the vertical distance of the current position of the looper trolley relative to the theoretical straight path is calculated as the instantaneous path deviation. When the absolute value of the instantaneous path deviation continuously exceeds the allowable tolerance of 5 mm within a continuous 3-meter travel distance, and the amplitude of the characteristic signal extracted from the structural vibration spectrum within this interval is enhanced by more than 50% compared to the baseline amplitude of the smooth track section, the system confirms the existence of a travel deviation induced by track deformation. It can be understood that the baseline value of the characteristic signal amplitude is the average value statistically determined when the looper trolley runs on a known straight track section. In some embodiments, the system generates a deceleration suggestion for the identified road section with travel deviation. The target speed value of the deceleration suggestion is calculated based on the proportion of the path deviation exceeding the limit and the proportion of the characteristic signal enhancement. Optionally, the formula for calculating the target speed value is:

[0044] in: This is the suggested target speed. This is the current real-time running speed. It is the normalized path deviation exceeding the limit ratio. It is the normalized feature signal enhancement ratio. and These are weighting coefficients. It's understandable that using path deviation and vibration characteristic signals simultaneously as criteria improves the accuracy and reliability of identifying track deformation-induced deviation.

[0045] See Figure 4In the process of identifying deviation in the multi-state collaborative monitoring and control system for the looper trolley, real-time path deviation monitoring is one of the core criteria. The curve, with the travel distance as the horizontal axis and the path deviation as the vertical axis, intuitively presents the degree of deviation between the looper trolley's trajectory and the theoretical straight path within its entire travel range of 0-20 meters. It also overlays a ±5 mm allowable deviation threshold and an 8-11 meter over-limit section marker, providing a visual basis for determining deviation induced by track deformation. From the curve characteristics: in the normal section (0-8 meters, 11-20 meters): the path deviation fluctuates within the ±3 mm range, not reaching the ±5 mm allowable tolerance. This corresponds to the looper trolley running on a straight track section, with stable wheel-track contact and no significant deviation trend. Exceeding Limit Section (8-11 meters): The path deviation continuously exceeds the +5 mm upper limit, with a peak deviation approaching 9.5 mm, and remains in an exceeding limit state within a continuous 3-meter travel distance. Combined with the vibration characteristic signal enhancement criterion described in the background technology, it can be confirmed that this section exhibits a deviation induced by local track deformation. Threshold and Exceeding Limit Logic: The ±5 mm dashed line represents the system's preset path deviation tolerance. When the absolute value of the deviation continuously exceeds this threshold within a continuous 3-meter travel distance, and the amplitude of the synchronously extracted 50-150 Hz frequency band vibration characteristic signal is enhanced by more than 50% compared to the baseline, the system will trigger the deviation judgment logic and generate a deceleration correction command for this exceeding limit section. By adjusting the servo drive system and the correction hydraulic cylinder's action, the deviation expansion is suppressed, ensuring operational safety. The curve provides intuitive visualization of the path deviation data and precise marking of the exceeding limit section, providing time-aligned deviation characteristic data for the multi-source correlation analysis module. It is a key visualization carrier for the monitoring of the looper trolley's travel status and closed-loop control correction.

[0046] In one embodiment of the present invention, a closed-loop correction instruction set for controlling the servo drive system of the looper trolley is generated based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation. The closed-loop correction instruction set includes a dynamic adjustment coefficient for motor torque, a delay threshold for brake action, and extension / retraction commands for the correction hydraulic cylinder. When the identification result is an abnormal jitter pattern, a preset base value for the dynamic adjustment coefficient of motor torque is extracted from the closed-loop correction instruction set, for example, a base value of 1.0. This dynamic adjustment coefficient is multiplied by a suppression factor between 0.1 and 0.8 to reduce the final output value of the dynamic adjustment coefficient, thereby reducing the output torque fluctuation of the traction motor. At the same time, the delay threshold for brake action is increased from the standard value of 50 milliseconds to the range of 100-200 milliseconds to absorb the vibration energy of the mechanical system. When the identification result is a structural deformation trend, the system immediately freezes the closed-loop correction instruction set to be issued and sends an equipment maintenance request containing the equipment number, the abnormality type "structural deformation trend", and a timestamp to the central scheduling system of the production line. Subsequently, the automatic operation mode of the looper trolley is paused and switched to a safety holding state. When the identification result is the amount of deviation, the system activates the extension and retraction command of the correction hydraulic cylinder and calculates the correction amount according to the direction and magnitude of the deviation. For example, if the control logic detects that the looper trolley has deviated 5 mm to the left of the track, it will generate a command to retract the correction hydraulic cylinder on the left side of the looper trolley by 8 mm and extend the correction hydraulic cylinder on the right side of the looper trolley by 8 mm to generate a torque that forcibly corrects the direction of travel.

[0047] In some embodiments, the adjusted motor torque dynamic adjustment coefficient, brake action delay threshold, and correction hydraulic cylinder extension / retraction stroke command are packaged into a standard format data frame. The data frame includes a command header, the looper trolley device ID, various command parameters, and a cyclic redundancy check code. The packaged data frame is sent to the looper trolley's field-programmable gate array (FPGA) for parsing and execution via a real-time industrial Ethernet connection. Upon receiving the data frame, the FPGA distinguishes the command type based on the command header and writes the motor torque dynamic adjustment coefficient, brake action delay threshold, and correction hydraulic cylinder extension / retraction stroke command into the corresponding control registers, driving the corresponding power units to operate. The system also includes a preventative correction module. After each abnormal jitter pattern is identified, the preventative correction module not only calculates and issues correction commands for the current abnormal jitter pattern but also retrieves the occurrence time and duration of all similar jitter events within the past 24 hours to construct a time-series-based prediction model. The time-series prediction model uses an autoregressive method, taking the time interval sequence of preceding abnormal jitter events as input to predict the time window for the next severe jitter event. The prediction formula is expressed as:

[0048] in: Indicates the predicted start time of the next severe jitter event. This indicates the time when the most recently identified abnormal jitter event occurred. This represents the time when the i-th abnormal jitter event occurs in the historical sequence. This represents the total number of historical events used to build the model. This is a configurable sensitivity coefficient. This represents the standard deviation of a series of historical event time intervals. Within the prediction time window... A few seconds before the arrival of the resonance, the system intervenes in advance, modifying the dynamic adjustment coefficient of the motor torque to a lower fixed value, keeping it at a low level, thereby achieving preventive suppression of potential resonance risks.

[0049] It is understandable that the preventative correction module sends the calculated prediction confidence information along with the current correction command to the field-programmable gate array controller. The prediction confidence information is based on the standard deviation of the historical sequence. and the number of data points used for prediction The calculations serve as the basis for instruction priority determination within the field-programmable gate array (FPGA) controller. Optionally, when the prediction confidence level is lower than a preset threshold, the FPGA controller can choose to ignore the preventative correction instruction and only execute real-time correction instructions for the already occurring abnormal jitter patterns. In some embodiments, the time series prediction model is continuously updated, recording the occurrence time whenever a new abnormal jitter pattern is identified. It is then added to the historical sequence, becoming the oldest data point. If removed, the model recalculates its predictions based on the latest data within the sliding window. This design allows the preventative correction module to adapt to changes in the loopback vehicle's operating state, continuously optimizing prediction accuracy. Optionally, when the field-programmable gate array (FPGA) controller receives both real-time and preventative correction commands simultaneously, it processes them according to its embedded priority logic. Typically, real-time correction commands have the highest execution priority, followed by preventative correction commands, but their control timing can be superimposed or merged.

[0050] See Figure 5In the dynamic torque adjustment control strategy of the looper trolley motor, the suppression effect on abnormal jitter modes can be intuitively reflected by the time-series curve of the torque adjustment coefficient. Specifically, the base torque coefficient is kept constant at 1.0, serving as the baseline output under normal operating conditions. When an abnormal jitter mode caused by mechanical resonance is detected, the system multiplies the dynamic adjustment coefficient by a suppression factor in the range of 0.2 to 0.8, forming the fluctuating adjusted coefficient curve shown in the figure. From a time-series perspective, the adjusted coefficient shows peak values ​​(approximately 0.8) at multiple time points, such as 0.02 hours, 0.12 hours, 0.15 hours, and 0.20 hours, corresponding to the triggering times of real-time suppression commands. It remains at or below 0.4 in the ranges of 0.03 to 0.09 hours and 0.23 to 0.25 hours, dropping to 0.2 in some periods, demonstrating a strong suppression effect on the risk of continuous resonance. It maintains a relatively high level of 0.8 in the range of 0.26 to 0.28 hours, reflecting the control logic of early intervention by the preventative correction module to maintain the coefficient within a low fluctuation range. The curve fully illustrates the control strategy of coordinated real-time and preventative correction: real-time correction rapidly lowers the coefficient to suppress torque fluctuations after a vibration event has occurred, while preventative correction predicts the risk window based on historical vibration sequences and maintains the coefficient within a stable range in advance, ultimately minimizing the output torque fluctuations of the traction motor and absorbing the vibration energy of the mechanical system. In terms of parameter configuration, the suppression factor range (0.2~0.8), the base coefficient (1.0), and the preventative intervention lead time jointly determine the fluctuation amplitude and steady-state range of the curve, and are the core design variables for balancing vibration suppression effectiveness and system response efficiency.

[0051] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications 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 protection scope of the present invention.

Claims

1. A multi-state collaborative monitoring and control system for the looper trolley of an annealing and pickling line, characterized in that, The system includes: The state perception module deploys vibration sensors and angle sensors at the four corners of the trolley frame and the key hinge points of the lifting mechanism to synchronously collect the structural vibration spectrum and attitude angle change sequence of the trolley during operation. The structural vibration spectrum and attitude angle change sequence constitute the raw data stream of the mechanical state of the trolley. The event marking module performs time window sliding analysis on the raw data stream of the mechanical state, extracts the vibration amplitude surge event and the attitude angle discontinuous jump event, marks the time point of the vibration amplitude surge event and the attitude angle discontinuous jump event as the high attention moment, and generates the associated data snapshot of the high attention moment; The multi-source correlation analysis module uses the correlation data snapshot of the high-attention moment to reverse look up the operating condition parameters of the looper at the corresponding moment, performs time-series alignment and superposition analysis of the original mechanical state data stream and the operating condition parameters, and identifies the abnormal shaking mode caused by mechanical resonance, the structural deformation trend caused by overload impact, and the travel deviation induced by track deformation. The control correction module generates a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation.

2. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 1, characterized in that, A time-window sliding analysis was performed on the raw mechanical state data stream to extract events of sudden increases in vibration amplitude and discontinuous jumps in attitude angle, including: A fixed sampling window of several seconds is set, and the envelope of the vibration spectrum of the structure is extracted. The ratio of the peak value to the root mean square value of the envelope is calculated. When the ratio exceeds the preset mechanical health threshold, it is determined that a sudden increase in vibration amplitude has occurred within the current sampling window. Perform a first-order difference operation on the attitude angle change sequence to capture the moment when the rate of change of attitude angle exceeds a preset physical limit value per unit time, and mark the moment when the attitude angle discontinuous jump event occurs. A circular buffer is established to store the analysis results of the most recent consecutive sampling windows. When multiple consecutive sampling windows are detected to have the vibration amplitude increase event or the attitude angle discontinuous jump event, the logic for generating the high-interest moment is triggered. When the high-attention moment arrives, the original data streams of several sampling windows before and after this moment are captured, and together with the system timestamp of the event occurrence, they are packaged and encapsulated into a data snapshot associated with the high-attention moment.

3. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 2, characterized in that, By using the associated data snapshots of the high-interest moments, the operating parameters of the looper trolley at the corresponding moments can be retrieved, including: The operating parameters include the current load curve of the traction motor, the tension feedback value of the wire rope, and the straightness detection results of the track. The system timestamp contained in the associated data snapshot of the high-concern moment is parsed and converted into a global tick clock count of the programmable logic controller of the annealing and pickling production line; Based on the global cycle clock count obtained by conversion, access the historical archive database of the production line data acquisition server, and retrieve the traction motor current timing data, wire rope tension meter reading, and track profile data obtained by the laser rangefinder corresponding to the global cycle clock count. The retrieved traction motor current timing data is filtered and denoised to remove high-frequency noise introduced by power grid fluctuations and restore the true current load curve. The retrieved track contour data is subjected to coordinate transformation processing and uniformly mapped to the local coordinate system of the looper trolley body for subsequent deformation trend analysis. The processed current load curve, wire rope tension meter reading, and track profile data are then bound to the current high-concern moment to form a complete operating condition parameter data package.

4. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 3, characterized in that, The raw mechanical state data stream and the operating condition parameters are time-aligned and overlaid for analysis to identify abnormal vibration patterns caused by mechanical resonance, including: The operating frequency of the lifting cylinder of the looper trolley is extracted from the raw data stream of the mechanical state and compared with the main peak frequency in the vibration spectrum of the structure. When the main peak frequency in the vibration spectrum of the structure is detected to be close to or equal to an integer multiple of the operating frequency of the lifting cylinder, it is determined that the conditions for mechanical resonance exist. Further analysis was conducted to determine whether the main peak frequency still existed in the vibration spectrum of the structure when the looper trolley was in a static tensioned state. If it still existed, external excitation caused by motion was ruled out, and the abnormal shaking mode was confirmed to be internal mechanical resonance. Record the load weight of the looper trolley, the hydraulic cylinder pressure, and the current temperature of the track when the internal mechanical resonance occurs, as auxiliary criteria for inducing the internal mechanical resonance.

5. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 4, characterized in that, The raw mechanical state data stream is time-aligned and overlaid with the operating condition parameters to identify the structural deformation trend caused by overload impact, including: Extract the current peak curves of the traction motor during the start-up and braking phases from the operating condition parameters, calculate the area enclosed by the current peak curves, and use it as the impact energy integral value of this start-up and stop process. The calculated impact energy integral value is correlated with the fatigue accumulation model of the looper trolley frame material to predict the micro-strain growth at key connection parts of the frame. Simultaneously, the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy is extracted from the attitude angle change sequence. When it is found that the micro-strain increase or the residual attitude angle offset shows a monotonically increasing trend after multiple start-stop cycles, it is confirmed that there is a structural deformation trend caused by overload impact, and further acceleration control commands are stopped.

6. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 5, characterized in that, The raw mechanical state data stream and the operating condition parameters are time-aligned and overlaid for analysis to identify the deviation caused by track deformation, including: By utilizing the characteristics of a specific frequency band in the vibration spectrum of the structure, a characteristic signal generated by poor contact between the wheel and the track is separated. The amplitude of the characteristic signal is positively correlated with the local elevation difference of the track. Based on the real-time running speed of the looper trolley in the operating condition parameters, the propagation delay of the characteristic signal in the direction of travel of the looper trolley is calculated; The deviation between the actual travel path of the looper trolley and the theoretical straight path is calculated. When the deviation value continuously exceeds the allowable tolerance and is accompanied by the enhancement of the characteristic signal, it is confirmed that there is a travel deviation induced by track deformation, and a deceleration suggestion is generated for the road section.

7. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 6, characterized in that, Based on the identified abnormal jitter patterns, structural deformation trends, and deviation, a closed-loop correction instruction set is generated for controlling the servo drive system of the looper trolley, including: The closed-loop correction instruction set includes the dynamic adjustment coefficient of motor torque, the delay threshold of brake action, and the extension and retraction stroke instructions of the correction hydraulic cylinder. When the identification result is the abnormal jitter mode, the dynamic adjustment coefficient of the motor torque is extracted from the closed-loop correction instruction set. The output torque fluctuation of the traction motor is reduced by decreasing the dynamic adjustment coefficient, and the delay threshold of the brake action is increased to absorb the vibration energy of the mechanical system. When the identification result is the structural deformation trend, the current closed-loop correction instruction set is frozen, and an equipment maintenance request is sent to the central scheduling system of the production line. The automatic operation mode of the looper trolley is suspended and manual intervention is awaited. When the identification result is the amount of deviation, the extension and retraction command of the correction hydraulic cylinder is activated. According to the direction and magnitude of the deviation, the correction hydraulic cylinder on one side of the looper trolley is extended and the correction hydraulic cylinder on the other side is retracted to forcibly correct the travel direction. The adjusted dynamic adjustment coefficient of the motor torque, the delay threshold of the brake action, and the extension and retraction stroke command of the correction hydraulic cylinder are packaged and sent to the field programmable gate array controller of the looper trolley for execution.

8. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 7, characterized in that, The step of generating a closed-loop correction instruction set for controlling the servo drive system of the looper trolley based on the identified abnormal jitter pattern, structural deformation trend, and travel deviation amount also includes: The preventive correction module, after identifying the abnormal jitter pattern each time, not only corrects the current motor torque, but also retrieves the frequency and duration of similar jitter events in the past period to construct a time series prediction model. Using the time series prediction model, the time window for the next severe jitter can be predicted; A few seconds before the predicted time window arrives, the dynamic adjustment coefficient of the motor torque is modified in advance to keep it at a low level, thereby achieving preventive suppression of potential resonance risks. Simultaneously, the predicted confidence level information and the current correction command are sent to the field programmable gate array controller as a basis for priority determination.

9. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 8, characterized in that, Parse the system timestamps contained in the associated data snapshots of the high-priority moments and convert them into global tick clock counts for the programmable logic controller of the annealing and pickling production line, including: Read the system timestamp, recorded in absolute time format, from the metadata header of the associated data snapshot at the high-concern moment; Obtain the latest heartbeat synchronization message from the programmable logic controller (PLC) of the annealing and pickling production line, and extract the reference mapping relationship between the global tick clock count and the absolute reference time maintained internally by the PLC from the message; The system timestamp is subtracted from the absolute reference time in the baseline mapping relationship to calculate the time offset of the system timestamp relative to the absolute reference time. The time offset is divided by the known refresh period of the global tick clock counter of the programmable logic controller to obtain the offset in tick count. The offset, expressed in terms of beat count, is added to the global beat clock count recorded in the reference mapping relationship to complete the conversion from the system timestamp in absolute time format to the global beat clock count used inside the programmable logic controller.

10. The multi-state collaborative monitoring and control system for the looper trolley of the annealing and pickling line according to claim 9, characterized in that, From the attitude angle change sequence, the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy is extracted, including: Determine the start and end times of the impact event corresponding to the impact energy integral value, and use the end time of the impact event as the starting reference point for extraction. Extract all attitude angle data within a fixed subsequent observation period starting from the initial reference point from the attitude angle change sequence to form the attitude angle subsequence to be analyzed; The attitude angle sequence to be analyzed is subjected to low-pass filtering to remove high-frequency fluctuation components caused by the normal operation of the looper trolley and sensor noise, resulting in a smoothed attitude angle sequence. In the smoothed attitude angle sequence, the last obvious turning point is identified, and the attitude angle data after the turning point remains stable within a preset steady-state determination time. Calculate the average attitude angle from the starting reference point to the turning point, and simultaneously calculate the average attitude angle during the steady-state determination time after the turning point. The difference between the mean attitude angle from the starting reference point to the turning point and the mean attitude angle within the steady-state determination time after the turning point is the residual attitude angle offset of the looper trolley after being subjected to the integral value of the impact energy.

Images