Search

CN-121980298-A - Production test monitoring method and system for elevator cable

CN121980298ACN 121980298 ACN121980298 ACN 121980298ACN-121980298-A

Abstract

The invention discloses a production test monitoring method and system for an elevator cable, the method comprises the steps of receiving multi-mode monitoring data in the elevator cable production test process, generating an initial monitoring feature set based on the multi-mode monitoring data, carrying out space-time mapping on the initial monitoring feature set, constructing a monitoring feature tensor, decomposing the monitoring feature tensor into three sub-tensors of material stability, process consistency and performance reliability, carrying out real-time analysis on the sub-tensor by utilizing a time sequence anomaly detection algorithm to generate a multi-dimensional anomaly identification result, carrying out weighted fusion on the multi-dimensional anomaly identification result, calculating a comprehensive monitoring index by combining a preset quality weight system, and outputting a production test monitoring result comprising quality qualification judgment, anomaly position marking and risk grade division. By utilizing the embodiment of the invention, the real-time accurate abnormal identification and scientific judgment of the elevator cable production test can be realized, and the product qualification rate is improved.

Inventors

  • HUANG XIANMOU
  • MA BINGFENG
  • HONG BO

Assignees

  • 杭州临安森源电缆有限公司

Dates

Publication Date
20260505
Application Date
20251201

Claims (10)

  1. 1. A production test monitoring method for an elevator cable, the method comprising: Receiving multi-mode monitoring data in the elevator cable production test process, and generating an initial monitoring feature set based on the multi-mode monitoring data, wherein the multi-mode monitoring data comprises raw material online detection data, production line process sensing data and finished product simulation operation test data, and each initial monitoring feature is associated with a quality monitoring dimension; performing space-time mapping on the initial monitoring feature set, constructing monitoring feature tensors, decomposing the monitoring feature tensors into three sub-tensors of material stability, process consistency and performance reliability to capture monitoring points of the whole production test flow, wherein each tensor represents space-time distribution attributes of corresponding initial monitoring features; analyzing the sub tensor in real time by using a time sequence abnormality detection algorithm to generate a multi-dimensional abnormality recognition result, wherein the time sequence abnormality detection algorithm compares the deviation of the real-time characteristic and the historical qualified baseline through a sliding window; And carrying out weighted fusion on the multidimensional abnormal recognition result, calculating a comprehensive monitoring index by combining a preset quality weight system, and outputting a production test monitoring result comprising quality qualification judgment, abnormal position marking and risk classification.
  2. 2. The method of claim 1, wherein receiving multi-modal monitoring data during an elevator cable production test generates an initial set of monitoring features based on the multi-modal monitoring data, wherein the multi-modal monitoring data includes raw material on-line detection data, process sensing data for a production line, and finished product simulation run test data, each initial monitoring feature being associated with a quality monitoring dimension, comprising: Classifying and receiving multi-mode monitoring data, collecting raw material online detection data according to detection items, sorting all working procedure sensing data of a production line according to working procedure nodes, dividing finished product simulation operation testing data according to testing types, and marking the collection time and the position of each data to form a classified monitoring data set; Preprocessing classified monitoring data, performing deviation correction on raw material data, performing time sequence smoothing on process sensing data, and repeatedly verifying finished product testing data to obtain preprocessed monitoring data; Extracting initial monitoring characteristics, and extracting key indexes from the preprocessed monitoring data, wherein raw material data are used for extracting purity fluctuation values and size deviation rates, process data are used for extracting temperature stability coefficients and speed synchronization rates, and finished product data are used for extracting performance attenuation rates and qualification times occupation ratios to form a basic characteristic list; and associating quality monitoring dimensions, matching the features in the basic feature list with preset dimensions, supplementing monitoring dimension labels corresponding to the features, and generating an initial monitoring feature set.
  3. 3. The method of claim 2, wherein the performing the space-time mapping on the initial monitoring feature set to construct a monitoring feature tensor, decomposing the monitoring feature tensor into three types of tensors of material stability, process consistency and performance reliability to capture monitoring points of the whole process of the production test, wherein each tensor represents a space-time distribution attribute of a corresponding initial monitoring feature, and the method comprises: Defining space-time mapping dimensions, wherein the time dimensions are divided into a raw material detection stage, a procedure processing stage and a finished product testing stage according to the whole production test flow, and the space dimensions are divided according to equipment/detection points to construct a space-time coordinate system; Mapping the initial monitoring features to space-time coordinates, corresponding each feature in the initial monitoring feature set to a specific point position of the space-time coordinates, and recording the numerical changes of the features at different space-time points to form a space-time feature distribution table; Constructing a monitoring feature tensor, and filling the numerical values in the space-time feature distribution table to the corresponding tensor positions by taking the time dimension, the space dimension and the feature type as three-dimensional axes to form a complete monitoring feature tensor; and decomposing and monitoring the characteristic tensor, and splitting according to the characteristic attribute, wherein the material stability tensor comprises the characteristics of a raw material detection stage and related processes, the process consistency tensor comprises the characteristics of nodes of each process, and the performance reliability tensor comprises the characteristics of a finished product testing stage, so as to obtain three types of tensors.
  4. 4. The method of claim 3, wherein the real-time analysis of the sub-tensors using a time-series anomaly detection algorithm to generate a multi-dimensional anomaly recognition result, wherein the time-series anomaly detection algorithm compares deviations of real-time features from historical acceptable baselines via a sliding window, comprising: Constructing a historical qualified baseline library, extracting normal time sequence data of three types of sub-tensors from past qualified elevator cable production batches, calculating a historical mean value and an allowable deviation range of each characteristic, and storing the historical mean value and the allowable deviation range in a classified manner according to the types of the sub-tensors to form a baseline sub-tensor library; configuring sliding window parameters, determining the size of a window according to a production test period, setting the sliding step length of the window, ensuring the full coverage of real-time data, and obtaining window configuration parameters; Comparing the real-time sub-tensor with the base line in real time, intercepting real-time sub-tensor fragments according to window configuration parameters, and performing deviation calculation on the real-time sub-tensor fragments and the sub-tensor fragments of corresponding characteristics in a base line sub-tensor library to generate a real-time deviation matrix of each characteristic; and identifying the multidimensional abnormality, marking the characteristic points exceeding the corresponding deviation threshold in the real-time deviation matrix, classifying and counting the number and the positions of the abnormality according to the sub-tensor type, and generating a multidimensional abnormality identification result.
  5. 5. The method of claim 4, wherein the performing weighted fusion on the multi-dimensional anomaly recognition result, calculating a comprehensive monitoring index by combining a preset quality weight system, and outputting a production test monitoring result including quality qualification judgment, anomaly location marking and risk classification comprises: Establishing a quality weight system, and assigning a material stability sub-tensor weight of 40%, a process consistency sub-tensor weight of 35% and a performance reliability sub-tensor weight of 25% according to the influence degree of each sub-tensor on the quality of the elevator cable to form a weight distribution table; weighting and fusing the abnormal recognition results, converting the abnormal values of all sub tensors in the multidimensional abnormal recognition results according to weights, and summing to obtain an abnormal comprehensive conversion value which is used as a weighted abnormal comprehensive value; Calculating a comprehensive monitoring index, normalizing the weighted abnormal comprehensive value to 0-100, comparing the qualified condition of the quality judgment by a qualified threshold value, and outputting the comprehensive monitoring index and a qualified judgment result; marking the abnormal position and classifying the risk level, positioning the abnormal space-time position according to the multi-dimensional abnormal recognition result, classifying the risk level according to the comprehensive monitoring index, integrating the qualification judgment, the abnormal position mark and the risk level, and outputting the production test monitoring result.
  6. 6. A production test monitoring system for an elevator cable, the system comprising: The system comprises a receiving module, a processing module and a processing module, wherein the receiving module is used for receiving multi-mode monitoring data in the elevator cable production test process and generating an initial monitoring feature set based on the multi-mode monitoring data, the multi-mode monitoring data comprises raw material online detection data, production line process sensing data and finished product simulation operation test data, and each initial monitoring feature is associated with a quality monitoring dimension; the mapping module is used for carrying out space-time mapping on the initial monitoring feature set, constructing a monitoring feature tensor, decomposing the monitoring feature tensor into three sub-tensors of material stability, process consistency and performance reliability so as to capture the monitoring key points of the whole production test flow, wherein each tensor represents the space-time distribution attribute corresponding to the initial monitoring feature; The analysis module is used for carrying out real-time analysis on the sub-tensor by utilizing a time sequence abnormality detection algorithm to generate a multi-dimensional abnormality recognition result, wherein the time sequence abnormality detection algorithm compares the deviation of the real-time characteristic and the historical qualified baseline through a sliding window; And the output module is used for carrying out weighted fusion on the multi-dimensional abnormal recognition result, calculating the comprehensive monitoring index by combining a preset quality weight system, and outputting a production test monitoring result comprising quality qualification judgment, abnormal position marking and risk classification.
  7. 7. The system according to claim 6, wherein the receiving module is specifically configured to: Classifying and receiving multi-mode monitoring data, collecting raw material online detection data according to detection items, sorting all working procedure sensing data of a production line according to working procedure nodes, dividing finished product simulation operation testing data according to testing types, and marking the collection time and the position of each data to form a classified monitoring data set; Preprocessing classified monitoring data, performing deviation correction on raw material data, performing time sequence smoothing on process sensing data, and repeatedly verifying finished product testing data to obtain preprocessed monitoring data; Extracting initial monitoring characteristics, and extracting key indexes from the preprocessed monitoring data, wherein raw material data are used for extracting purity fluctuation values and size deviation rates, process data are used for extracting temperature stability coefficients and speed synchronization rates, and finished product data are used for extracting performance attenuation rates and qualification times occupation ratios to form a basic characteristic list; and associating quality monitoring dimensions, matching the features in the basic feature list with preset dimensions, supplementing monitoring dimension labels corresponding to the features, and generating an initial monitoring feature set.
  8. 8. The system according to claim 7, wherein the mapping module is specifically configured to: Defining space-time mapping dimensions, wherein the time dimensions are divided into a raw material detection stage, a procedure processing stage and a finished product testing stage according to the whole production test flow, and the space dimensions are divided according to equipment/detection points to construct a space-time coordinate system; Mapping the initial monitoring features to space-time coordinates, corresponding each feature in the initial monitoring feature set to a specific point position of the space-time coordinates, and recording the numerical changes of the features at different space-time points to form a space-time feature distribution table; Constructing a monitoring feature tensor, and filling the numerical values in the space-time feature distribution table to the corresponding tensor positions by taking the time dimension, the space dimension and the feature type as three-dimensional axes to form a complete monitoring feature tensor; and decomposing and monitoring the characteristic tensor, and splitting according to the characteristic attribute, wherein the material stability tensor comprises the characteristics of a raw material detection stage and related processes, the process consistency tensor comprises the characteristics of nodes of each process, and the performance reliability tensor comprises the characteristics of a finished product testing stage, so as to obtain three types of tensors.
  9. 9. A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method of any of claims 1-5 when run.
  10. 10. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the method of any of claims 1-5.

Description

Production test monitoring method and system for elevator cable Technical Field The invention belongs to the technical field of monitoring, and particularly relates to a production test monitoring method and system for an elevator cable. Background The elevator cable is used as a key transmission component for elevator operation, the quality of the elevator cable directly influences the safety and the service life of the elevator, and the production test and the monitoring are required to cover the whole flow of raw materials, production process and finished product performance. The existing monitoring method is mainly aimed at a single link (such as detecting only the finished product resistance or the raw material purity), lacks multi-mode data fusion capability, cannot correlate potential influences of raw material quality, process parameters and finished product performance, and is easy to have risks of 'partially qualified but wholly invalid'. Meanwhile, the traditional monitoring relies on manual sampling detection, has hysteresis, is difficult to capture the time sequence abnormality of a production line (such as insulating layer defect caused by temperature fluctuation of a certain process), lacks a unified weight system in abnormality judgment, has insufficient differentiation degree on key indexes (such as cable fatigue resistance) and secondary indexes, and is easy to misjudge or miss-judge. These problems lead to unstable production yield of elevator cables, and increase later operation and maintenance cost and potential safety hazards. Disclosure of Invention The invention aims to provide a production test monitoring method and system for an elevator cable, which are used for solving the defects in the prior art, realizing real-time accurate abnormal identification and scientific judgment of the elevator cable production test and improving the product qualification rate. One embodiment of the present application provides a production test monitoring method for an elevator cable, the method comprising: Receiving multi-mode monitoring data in the elevator cable production test process, and generating an initial monitoring feature set based on the multi-mode monitoring data, wherein the multi-mode monitoring data comprises raw material online detection data, production line process sensing data and finished product simulation operation test data, and each initial monitoring feature is associated with a quality monitoring dimension; performing space-time mapping on the initial monitoring feature set, constructing monitoring feature tensors, decomposing the monitoring feature tensors into three sub-tensors of material stability, process consistency and performance reliability to capture monitoring points of the whole production test flow, wherein each tensor represents space-time distribution attributes of corresponding initial monitoring features; analyzing the sub tensor in real time by using a time sequence abnormality detection algorithm to generate a multi-dimensional abnormality recognition result, wherein the time sequence abnormality detection algorithm compares the deviation of the real-time characteristic and the historical qualified baseline through a sliding window; And carrying out weighted fusion on the multidimensional abnormal recognition result, calculating a comprehensive monitoring index by combining a preset quality weight system, and outputting a production test monitoring result comprising quality qualification judgment, abnormal position marking and risk classification. Optionally, the receiving the multi-mode monitoring data in the elevator cable production test process generates an initial monitoring feature set based on the multi-mode monitoring data, where the multi-mode monitoring data includes raw material online detection data, sensing data of each process of a production line and finished product simulation operation test data, and each initial monitoring feature is associated with a quality monitoring dimension, and includes: Classifying and receiving multi-mode monitoring data, collecting raw material online detection data according to detection items, sorting all working procedure sensing data of a production line according to working procedure nodes, dividing finished product simulation operation testing data according to testing types, and marking the collection time and the position of each data to form a classified monitoring data set; Preprocessing classified monitoring data, performing deviation correction on raw material data, performing time sequence smoothing on process sensing data, and repeatedly verifying finished product testing data to obtain preprocessed monitoring data; Extracting initial monitoring characteristics, and extracting key indexes from the preprocessed monitoring data, wherein raw material data are used for extracting purity fluctuation values and size deviation rates, process data are used for extracting temperature stability coefficient