Search

CN-122007180-A - Rolling mill lining plate inclination detection method based on multi-axis inclination sensing

CN122007180ACN 122007180 ACN122007180 ACN 122007180ACN-122007180-A

Abstract

The invention relates to a rolling mill lining plate inclination detection method based on multi-axis inclination sensing, and belongs to the technical field of rolling mill equipment state monitoring. The method comprises the steps of obtaining initial reference inclination angle data of a lining plate detection area, generating dynamic reference data through abrasion-reference linkage self calibration, constructing a lining plate multidimensional coupling mapping model, setting a dynamic threshold system, collecting real-time inclination angle data, fusing physical quantity compensation data, preprocessing to generate effective data, comparing the effective data with the dynamic reference data, suppressing vibration interference, generating area inclination offset through area-global bidirectional iterative computation, training a trend prediction model to obtain inclination development trend, generating an abnormal judgment result through multi-area data cross check and redundancy verification, tracing induction factors, constructing a multistage early warning mechanism, triggering early warning, outputting complete abnormal information and generating an operation report. The invention improves the inclination detection precision and early warning timeliness, and provides reliable support for safe operation of the rolling mill and maintenance of the lining plate.

Inventors

  • WANG YANPENG

Assignees

  • 上海朗尚传感技术有限公司

Dates

Publication Date
20260512
Application Date
20251231

Claims (12)

  1. 1. A rolling mill lining plate inclination detection method based on multi-axis inclination sensing is characterized by comprising the following steps: S1, acquiring initial reference inclination angle data of a lining plate detection area of a rolling mill, introducing a wear-reference linkage self-calibration mechanism, correcting the initial reference inclination angle data to generate dynamic reference data, constructing a lining plate multidimensional coupling mapping model, and setting a dynamic threshold system based on historical fault and normal working condition data; S2, acquiring real-time inclination angle data of each detection area in the running process of the rolling mill, synchronously fusing physical quantity compensation data and carrying out a preprocessing flow; S3, based on the comparison result of the effective data and the dynamic reference data, suppressing vibration interference signals, and generating regional inclination offset through comparison operation; combining the stress distribution characteristics of the lining plate in the lining plate multidimensional coupling model, adopting region-global bidirectional iterative computation logic to optimize region weight; And S4, generating an inclination abnormality judgment result through multi-region data cross check and redundancy verification, tracing the induction factors of the inclination abnormality, setting a multi-stage early warning mechanism according to the severity degree and the development trend of the abnormality based on the dynamic threshold system, triggering early warning, generating a multi-stage early warning signal, outputting the tracing conclusion of the abnormal region, the inclination parameters and the induction factors of the abnormality, and generating an operation detection report.
  2. 2. The method of claim 1, wherein in the step S1, the calibration process of the abrasion-reference linkage self-calibration mechanism specifically comprises continuously obtaining the accumulated operation time length of the lining plate and the actual abrasion amount data recorded in each maintenance operation, establishing a database by combining the abrasion resistance characteristics of the lining plate materials, excavating the corresponding relation between the accumulated operation time length and the actual abrasion amount through a correlation analysis algorithm, performing coupling analysis on inclination angle change data to generate inclination angle correction coefficients of different abrasion stages, calling the corresponding correction coefficients in stages based on the abrasion process of the operation of the lining plate, performing successive correction on the initial reference inclination angle data, comparing the current inclination angle data with the corrected dynamic reference data, and verifying the correction effect.
  3. 3. The method of the multi-dimensional coupling mapping model of the lining plate is characterized in that in S1, the construction process of the multi-dimensional coupling mapping model of the lining plate specifically comprises the steps of obtaining lining plate material characteristic parameters, actual gap data during installation, load data of different operation stages of a rolling mill, rotation speed fluctuation data and environment temperature and humidity continuous monitoring data in a classifying mode, denoising and normalizing various data, eliminating invalid interference data, calculating correlation coefficients with inclination angle change of the lining plate to generate impact weights on the inclination angle change, dividing analysis units based on different working condition combinations, establishing quantitative mapping relations between each impact factor and the inclination angle change in each analysis unit, integrating the quantitative mapping relations, and generating the multi-dimensional coupling mapping model of the lining plate.
  4. 4. The method of claim 1, wherein in the step S1, the updating process of the dynamic threshold system comprises the steps of periodically obtaining the data of the current operation abrasion stage of the lining plate and the real-time change data of the environmental temperature and humidity, calling the working condition data similar to the current abrasion stage and the environmental condition in a preset historical database, extracting the corresponding threshold parameters, combining the current operation data and the historical similar working condition data, adjusting the upper limit range and the lower limit range of the threshold, substituting the new threshold parameters into the actual detection process, and monitoring the early warning accuracy and the false alarm rate.
  5. 5. The method according to claim 1, wherein in S2, the step of fusing the physical quantity compensation data specifically includes acquiring temperature data, vibration intensity data and load fluctuation data in a stress area of the lining plate, distributing corresponding fusion weights for the temperature data, the vibration intensity data and the load fluctuation data based on the influence weights of the physical quantity in the lining plate multidimensional coupling mapping model, and performing weighted fusion on the physical quantity compensation data and the real-time inclination angle data through calculation.
  6. 6. The method according to claim 1, wherein in S2, the specific implementation process of the preprocessing flow includes the steps of conducting noise reduction processing on the converged inclination angle data, identifying and filtering interference signals through analyzing frequency characteristics of the data, dynamically adjusting filtering cut-off frequency, selecting a sliding window with fixed length based on a time sequence data smoothing algorithm, calculating the mean value or the median of the data in the window to replace original data in the center of the window, checking time sequence continuity of the data after processing is completed, and filling break points by adopting a linear interpolation method when data break points exist.
  7. 7. The method according to claim 1, wherein in S3, the specific process of suppressing the vibration interference signal includes obtaining vibration intensity data in real time during the running process of the rolling mill, identifying the type, the frequency range and the amplitude change rule of the vibration signal through spectrum analysis, adjusting parameters of the working condition self-adaptive anti-interference noise reduction technology, and monitoring the amplitude change of the inclination effective signal in real time.
  8. 8. The method according to claim 1, wherein in the step S3, the specific implementation process of the region-global bidirectional iterative computation logic includes assigning an initial weight to each local detection region based on the stress distribution characteristics of each detection region in the liner plate multidimensional coupling mapping model, performing weighted summation on the tilt offset of each local region and the corresponding initial weight to obtain an initial global synthesized tilt state, computing the deviation value of the tilt offset of each local region and the initial global synthesized tilt state, adjusting the weight of the corresponding region, substituting the adjusted region weights into the weighted summation operation again to obtain a new global synthesized tilt state, repeating the deviation computation, the weight adjustment and the global synthesis steps until the difference value of the global synthesized tilt states obtained in two adjacent times is smaller than a preset judgment standard, stopping iteration, and outputting a final global synthesized tilt state.
  9. 9. The method of claim 1, wherein in S3, the training process of the trend prediction model specifically comprises the steps of generating a time sequence data set based on historical inclination data accumulated by long-term operation of a lining plate in a time sequence sorting mode, synchronously obtaining working condition related data and environment influence data of each section of the historical inclination data corresponding to a period, preprocessing to remove abnormal values, fill up missing values, carrying out data normalization processing, mapping the data to a unified range, dividing a training set and a verification set, constructing a model frame by adopting a time sequence analysis algorithm, inputting the training set data, iteratively adjusting core parameters of the model to minimize prediction errors of the verification set into a target optimization model, generating a development trend prediction model, and pre-judging the change amplitude and the development direction of inclination of the lining plate under different working conditions.
  10. 10. The method according to claim 1, wherein in S4, the specific implementation process of the multi-region data cross check and redundancy check comprises the steps of extracting the inclination offset, time sequence change curve and local abnormal characteristic parameters of each detection region, dividing check groups according to classification modes of adjacent regions, symmetrical regions and stress-related regions, carrying out bidirectional comparison on inclination data of each region in the same group, calculating synchronous deviation of inclination change among the regions, simultaneously calling redundant backup data of each detection region, carrying out point-by-point comparison on real-time acquisition data and the redundant backup data, removing abnormal region data, and incorporating the data passing through the cross check and the redundant backup comparison in the group into abnormality judgment analysis.
  11. 11. The method of claim 1, wherein in S4, the specific process of tracing the inclination anomaly induction factors comprises the steps of extracting working condition data, abrasion related data and environment influence data of an inclination anomaly occurrence period from the lining plate multidimensional coupling mapping model, comparing the working condition data, the abrasion related data and the environment influence data item by item with corresponding data under normal working conditions, calculating an offset value, checking one by one based on the influence weight priority of each factor in the lining plate multidimensional coupling mapping model, analyzing the causal relationship between the checked anomaly factors and the inclination anomaly to obtain core induction factors and related induction factors, and generating an anomaly induction factor tracing chain.
  12. 12. The method of claim 1, wherein in S4, the implementation process of the multi-stage early warning mechanism comprises dividing early warning grades by combining inclination deviation standards, inclination development trend prediction results and influence ranges of inclination abnormality on rolling mill operation of different sections in the dynamic threshold system, and integrating abnormal region positions, specific inclination parameters, abnormality induction factor tracing conclusions and inclination development trend prediction results when any level early warning is triggered to generate an early warning information packet.

Description

Rolling mill lining plate inclination detection method based on multi-axis inclination sensing Technical Field The invention belongs to the technical field of rolling mill equipment state monitoring, and particularly relates to a rolling mill lining plate inclination detection method based on multi-axis inclination sensing. Background The rolling mill is used as core production equipment in the metallurgical industry, the running stability of the rolling mill directly influences the quality and production safety of products, and the lining plate is used as a key stress part of the rolling mill and plays an important role in protecting a roller, transmitting rolling force and guaranteeing rolling precision. Under the working conditions of long-term heavy load, high-speed operation and material impact, the lining plate is easy to incline due to factors such as uneven wear, change of installation clearance, fluctuation of load and the like, if the lining plate cannot be detected and interfered in time, unbalance of stress of a roller and increase of product size deviation are caused, and safety accidents such as falling of the lining plate, shutdown of a rolling mill and the like are caused when the lining plate is serious, so that great economic loss is caused. At present, the inclination detection of a lining plate of a rolling mill mainly depends on manual inspection, single threshold alarming or simple sensing detection modes, and has the technical defects that firstly, the traditional detection method mainly adopts fixed reference data for comparison, the reference drift problem caused by long-term abrasion of the lining plate is not considered, as the operation time is accumulated, the detection precision is continuously reduced, misjudgment or missed judgment is easy to occur, secondly, the traditional early warning mechanism mostly sets a threshold value based on single inclination offset, the comprehensive consideration of inclination development trend and abnormal influence range is lacked, only simple alarming can be realized, a grading response scheme cannot be provided according to the abnormal severity, the operation personnel is difficult to accurately grasp the treatment priority, thirdly, early warning information only comprises basic abnormal prompting, the critical information such as abnormal region positioning, induction factor tracing and the like is not integrated, the operation personnel needs to additionally spend a large amount of time for checking, the treatment opportunity is delayed, fourthly, the detection mechanism lacks closed loop optimization capability, the detection and early warning process is mutually independent, the treatment effect cannot be fed back to a threshold value adjustment link, the early warning adaptability is poor, the rolling mill is difficult to adapt to complex and variable operation working conditions, and fifth, the partial detection method is not suitable for the interference caused by the operation of the rolling mill, the operation of the full running, the temperature change, the disturbance factor is difficult to accurately influence, and the vibration influence is further caused. Therefore, aiming at the problems of fixed reference, single early warning, incomplete information, lack of closed-loop optimization, weak anti-interference capability and the like in the existing rolling mill lining plate inclination detection technology, a detection method capable of dynamically adapting to the state of the lining plate, grading accurate early warning, integrating complete abnormal information and realizing closed-loop optimization needs to be developed so as to improve the accuracy of rolling mill lining plate inclination detection and the timeliness and effectiveness of early warning and ensure safe and stable operation of the rolling mill. Disclosure of Invention In order to solve the problems in the prior art, the invention provides a rolling mill lining plate inclination detection method based on multi-axis inclination sensing, The aim of the invention can be achieved by the following technical scheme: a rolling mill lining plate inclination detection method based on multi-axis inclination sensing comprises the following steps: S1, acquiring initial reference inclination angle data of a lining plate detection area of a rolling mill, introducing a wear-reference linkage self-calibration mechanism, correcting the initial reference inclination angle data to generate dynamic reference data, constructing a lining plate multidimensional coupling mapping model, and setting a dynamic threshold system based on historical fault and normal working condition data; S2, acquiring real-time inclination angle data of each detection area in the running process of the rolling mill, synchronously fusing physical quantity compensation data and carrying out a preprocessing flow; S3, based on the comparison result of the effective data and the dynamic reference data, sup