Based on the content of data analysis, PHM can be divided into six themes: sensor data processing, condition monitoring, health management, fault diagnosis, fault prediction, and operation and maintenance optimization. These themes are based on the monitoring data of the equipment and environment and the full-dimensional historical data of the equipment (here called the equipment profile model.) The PHM analysis framework is shown in Figure 4-4. This framework differs from the system of OSA-CBM [8]: ① Separate fault diagnosis from fault prediction, and carry out fault diagnosis after the fact and fault prediction beforehand, and its analysis algorithms are obviously different; ② In terms of data collection, separate monitoring data from equipment management data, which is usually independent in the IT system, and the data modeling techniques are different.
Condition monitoring detects condition abnormalities in a timely manner based on recent condition monitoring information. Compared with the classic SCADA and DCS threshold alarms, PHM has to deal with more complex alarm rules such as anomaly pattern detection based on multiple time series, anomaly trend identification, eliminating false alarms based on multi-sensor fusion and trend analysis, and eliminating alarm storms (Alarm Shower) by merging alarm information based on the propagation mechanism of faults. Sometimes statistical analysis methods can also adaptively provide more appropriate threshold estimates for threshold alarm rules for SCADA and DCS.
Fault prediction predicts the remaining life, time to failure, and risk of failure of equipment based on indications of signs of failure and modeling of the degradation process of the equipment. When a system, subsystem, or component may have minor defects and early failures or gradually degrade to a state where it cannot fulfill its specified function with optimal performance, these minor defects, early failures, or performance degradation can be detected through the selection of relevant detection methods and the design of a prediction system that enables equipment maintenance personnel to predict the time of failure, and thus take a series of preventive maintenance measures without having to wait until the failure occurs before responding reactively. Reactive response. Signs of failure are anomalies that can be observed before a failure mode occurs or in the early stages of a failure mode's evolution.
Health management refers to monitoring the operating parameters of a system in various ways while the systems are in operation or in working condition, and determining whether the system can work properly under the current conditions (mission capability). Health assessment and diagnosis opens a new way to improve the reliability, maintainability and effectiveness of equipment. In order to avoid the failure of certain operational processes and cause the paralysis of the entire equipment system, it is necessary to quickly deal with the failure after it occurs, to maintain the basic functions normal, to improve the utilization efficiency of the equipment and the use of security, to ensure safe and reliable operation.
The premise of fault diagnosis is to understand the failure mechanism of the equipment. For example, turbine, compressor and other fluid rotating machinery, abnormal vibration and noise signals in the time domain and frequency domain for fault diagnosis provides important information, which is only part of the fault information; fluid machinery load and media temperature, pressure, flow rate changes, the operating state of the equipment has an important impact on the equipment is often an important factor leading to abnormalities in the equipment and operation of the destabilizing factors. Therefore, the fault diagnosis of rotating equipment, only in the acquisition of the machine's steady state data, transient data, process parameters and operating conditions and other information on the basis of the fault signs through the calculation, fault-sensitive parameter extraction and comprehensive analysis of judgment, in order to determine the cause of the failure, to reach a diagnosis in line with the reality of the conclusions, and put forward measures to manage.
On the basis of the above analysis, according to the characteristics and complexity of the system, suitable methods can be used to achieve operation and maintenance optimization. Integrating multiple objectives such as cost, time, effectiveness, and equipment lifetime, and considering constraints such as resource constraints, time window requirements, and compliance requirements, maintenance staff allocation, maintenance schedules, and other logistical support activities can be optimized. Broad optimization also includes providing knowledge of the O&M process, etc., to improve maintenance efficiency.
In many scenarios, enterprises want data analysts to sort out the planning of intelligent O&M and form intelligent O&M big data analysis questions from the perspective of enterprise ecology and operations.There are three main application modes for PHM, as shown in Table 4-9.
Recommended reading: "Industrial Big Data Analytics Practices" for an all-round understanding of industrial data analytics related knowledge.
Edited just now