Search

CN-121997164-A - Park equipment abnormality monitoring method and system based on big data

CN121997164ACN 121997164 ACN121997164 ACN 121997164ACN-121997164-A

Abstract

The invention relates to the technical field of industrial Internet, and discloses a park equipment abnormality monitoring method based on big data. The method comprises the steps of obtaining multi-source equipment operation data of an industrial park, conducting time alignment and feature conversion processing to obtain time sequence feature segments, generating equipment operation reference modes according to the time sequence feature segments, storing the time sequence feature segments into a historical operation record base, comparing and judging feature vectors constructed in real time with the equipment operation reference modes to obtain potential mutation point sets, conducting historical backtracking and rule analysis on the potential mutation point sets by combining the historical operation record base to obtain abnormal confidence evaluation values, generating structured abnormal event logs according to the abnormal confidence evaluation values and triggering early warning, obtaining external feedback data, conducting integration analysis by combining the structured abnormal event logs, and outputting final monitoring reports. According to the invention, through a dynamic reference and feedback closed loop, the self-adaptive accurate monitoring of equipment abnormality is realized.

Inventors

  • WANG SHUAI

Assignees

  • 广州云物智能科技有限公司

Dates

Publication Date
20260508
Application Date
20251216

Claims (8)

  1. 1. The park equipment abnormality monitoring method based on big data is characterized by comprising the following steps of: Acquiring multi-source equipment operation data of an industrial park, and performing time alignment and feature conversion processing to obtain a time sequence feature segment; According to the time sequence characteristic segments, performing periodic pattern recognition and trend fitting processing to generate a device operation reference pattern, and storing the time sequence characteristic segments into a preset historical operation record library; acquiring equipment operation data at the current moment, constructing a feature vector, and comparing the feature vector with the equipment operation reference mode in real time and judging a dynamic threshold value to obtain a potential mutation point set; According to the potential mutation point set, combining the historical operation record library, performing historical backtracking and rule analysis, and calculating abnormal probability distribution to obtain an abnormal confidence evaluation value; According to the abnormal confidence assessment value, carrying out risk classification judgment by combining with a preset risk level standard, generating a structured abnormal event log, and triggering early warning when the high risk is judged; And acquiring external feedback data, carrying out integrated analysis and report generation processing according to the structured abnormal event log and the external feedback data, and outputting a final monitoring report.
  2. 2. The method for monitoring abnormality of industrial park equipment based on big data according to claim 1, wherein the steps of obtaining operation data of multi-source equipment of an industrial park, performing time alignment and feature conversion processing to obtain time sequence feature segments, and include: Collecting multi-source sensor signals of the industrial park, and performing time stamp alignment processing to obtain an original operation data stream; according to a preset time step, carrying out sliding window segmentation on the original operation data stream to obtain a plurality of continuous discrete data subsets; carrying out multidimensional statistical feature calculation aiming at each discrete data subset to construct a high-dimensional statistical feature vector; and performing feature projection and compression processing according to the high-dimensional statistical feature vector to obtain the time sequence feature segment.
  3. 3. The method for monitoring the abnormality of the campus equipment based on the big data according to claim 1, wherein the steps of performing periodic pattern recognition and trend fitting processing according to the time sequence feature segments, generating an equipment operation reference pattern, and storing the time sequence feature segments in a preset historical operation record library include: performing time sequence component decomposition calculation on the time sequence characteristic fragments, and separating to obtain a long-term trend component, a periodic fluctuation component and a random residual component; Performing numerical fitting and smoothing on the long-term trend component to generate a long-term trend reference curve; Performing waveform clustering and feature extraction on the periodic fluctuation component to determine a typical periodic fluctuation template; And carrying out overlapping reconstruction processing on the long-term trend reference curve and the typical periodic fluctuation template to generate the equipment operation reference mode.
  4. 4. The big data-based campus equipment anomaly monitoring method according to claim 1, wherein the steps of obtaining equipment operation data at the current moment and constructing a feature vector, comparing the feature vector with the equipment operation reference mode in real time and judging a dynamic threshold value to obtain a potential mutation point set comprise the steps of: acquiring equipment operation data at the current moment, performing feature extraction and numerical vectorization processing, and constructing a current state feature vector; according to the current state feature vector, searching and matching are carried out in the equipment operation reference mode, and a corresponding association reference feature interval is determined; Calculating the numerical difference between the current state feature vector and the associated reference feature interval to obtain a real-time deviation index; And comparing and analyzing the real-time deviation index with a preset dynamic safety threshold, and extracting sampling points exceeding the dynamic safety threshold to obtain the potential mutation point set.
  5. 5. The method for monitoring the abnormality of the campus equipment based on big data according to claim 1, wherein the step of performing historical backtracking and rule analysis according to the set of potential mutation points and the historical operation record base, calculating an abnormality probability distribution, and obtaining an abnormality confidence evaluation value comprises the following steps: According to the potential mutation point set, scene backtracking and retrieval matching are carried out in the history operation record library, and associated history scene data are extracted; Carrying out fluctuation feature statistics and reproduction analysis on the associated historical scene data to determine historical reproduction rule features; According to the historical reproduction rule characteristics, carrying out abnormal probability estimation processing to generate abnormal probability distribution; And synthesizing the abnormal probability distribution and the potential mutation point set, and carrying out multidimensional weighted quantization calculation to obtain the abnormal confidence evaluation value.
  6. 6. The big data based campus equipment anomaly monitoring method according to claim 1, wherein the performing risk classification determination according to the anomaly confidence evaluation value in combination with a preset risk classification standard, generating a structured abnormal event log, and triggering early warning when determining high risk, includes: performing interval mapping and level matching processing according to the abnormal confidence evaluation value and the preset risk level standard, and determining a risk level judgment result; carrying out data serialization and template filling processing according to the risk level judging result and a preset storage template to generate a structured abnormal event log; and carrying out threshold value verification on the risk level judging result, and if the risk level judging result falls into a preset high-risk interval, generating an instant early warning instruction comprising an abnormal type and occurrence time.
  7. 7. The big data based campus equipment anomaly monitoring method of claim 1, wherein the obtaining the external feedback data, performing an integrated analysis and report generation process according to the structured anomaly event log and the external feedback data, and outputting a final monitoring report, comprises: performing time sequence association and content matching processing on the structured abnormal event log and the external feedback data, removing false report records, and determining an abnormal event set after verification; According to the checked abnormal event set, retrieving real-time operation parameters of associated equipment, and carrying out multidimensional state fusion and fault propagation association analysis to generate state coupling abnormal details; and according to the state coupling abnormal details, carrying out data aggregation and visual rendering processing according to a preset report generation rule, and outputting the final monitoring report.
  8. 8. A big data based campus equipment anomaly monitoring system, comprising: the data preprocessing module is used for acquiring multi-source equipment operation data of the industrial park, and performing time alignment and feature conversion processing to obtain a time sequence feature segment; The reference construction module is used for carrying out periodic pattern recognition and trend fitting processing according to the time sequence feature segments, generating a device operation reference pattern, and storing the time sequence feature segments into a preset historical operation record library; The real-time monitoring module is used for acquiring equipment operation data at the current moment, constructing a feature vector, and carrying out real-time comparison and dynamic threshold judgment on the feature vector and the equipment operation reference mode to obtain a potential mutation point set; The evaluation analysis module is used for carrying out historical backtracking and rule analysis according to the potential mutation point set and the historical operation record library, calculating abnormal probability distribution and obtaining an abnormal confidence coefficient evaluation value; the early warning decision module is used for carrying out risk classification judgment according to the abnormal confidence assessment value and combining with a preset risk level standard, generating a structured abnormal event log, and triggering early warning when the structured abnormal event log is judged to be high in risk; and the report generation module is used for acquiring external feedback data, carrying out integrated analysis and report generation processing according to the structured abnormal event log and the external feedback data, and outputting a final monitoring report.

Description

Park equipment abnormality monitoring method and system based on big data Technical Field The invention relates to the technical field of industrial Internet and big data analysis, in particular to a method and a system for monitoring abnormality of park equipment based on big data. Background At present, in an intelligent management system of a modern industrial park, real-time monitoring and fault early warning of equipment operation states are core links for guaranteeing production continuity and safety. Along with the promotion of industry 4.0, the scale of equipment in a park grows exponentially, the generated operation data is huge in quantity and complex in source, and how to accurately identify abnormal signals from the massive data is directly related to the operation and maintenance of enterprises. In one prior art, device monitoring schemes typically rely on a Microcontroller (MCU) deployed at the edge or a conventional PLC system to perform basic data acquisition and decision. Most of these systems employ monitoring logic based on fixed rules or preset static thresholds, i.e. when the acquired parameters (e.g. temperature, vibration frequency) exceed preset upper and lower limits, an alarm is triggered. The static strategy ignores performance drift generated by aging and abrasion of equipment in the long-term operation process and dynamic influence of different working conditions (such as load change and environmental temperature and humidity fluctuation) on normal operation parameters of the equipment. Due to the lack of deep mining capability on the periodic mode and the long-term trend behind the massive time sequence data, the system cannot accurately distinguish normal parameter fluctuation and potential fault symptoms, so that when facing complex and changeable running environments, key abnormal signals are often missed due to the fact that information cannot be effectively integrated, or false alarms are frequently generated due to the fact that threshold setting is stiff. Therefore, the technical problems of high abnormality monitoring false alarm rate and difficult accurate identification of potential hidden danger exist in the prior art. Disclosure of Invention The invention provides a method and a system for monitoring abnormality of park equipment based on big data, which are used for solving the technical problems that in the prior art, the abnormality monitoring false alarm rate is high and potential hidden danger is difficult to accurately identify. In order to solve the technical problems, the invention provides a method for monitoring abnormality of park equipment based on big data, comprising the following steps: Acquiring multi-source equipment operation data of an industrial park, and performing time alignment and feature conversion processing to obtain a time sequence feature segment; According to the time sequence characteristic segments, performing periodic pattern recognition and trend fitting processing to generate a device operation reference pattern, and storing the time sequence characteristic segments into a preset historical operation record library; acquiring equipment operation data at the current moment, constructing a feature vector, and comparing the feature vector with the equipment operation reference mode in real time and judging a dynamic threshold value to obtain a potential mutation point set; According to the potential mutation point set, combining the historical operation record library, performing historical backtracking and rule analysis, and calculating abnormal probability distribution to obtain an abnormal confidence evaluation value; According to the abnormal confidence assessment value, carrying out risk classification judgment by combining with a preset risk level standard, generating a structured abnormal event log, and triggering early warning when the high risk is judged; And acquiring external feedback data, carrying out integrated analysis and report generation processing according to the structured abnormal event log and the external feedback data, and outputting a final monitoring report. In a second aspect, the present invention provides a big data based campus equipment anomaly monitoring system, comprising: the data preprocessing module is used for acquiring multi-source equipment operation data of the industrial park, and performing time alignment and feature conversion processing to obtain a time sequence feature segment; The reference construction module is used for carrying out periodic pattern recognition and trend fitting processing according to the time sequence feature segments, generating a device operation reference pattern, and storing the time sequence feature segments into a preset historical operation record library; The real-time monitoring module is used for acquiring equipment operation data at the current moment, constructing a feature vector, and carrying out real-time comparison and dynamic threshold judgment on the