What is Predictive Maintenance and what are the technical systems?

Predictive Maintenance (PredictiveMaintenance, referred to as PdM) is based on the state (ConditionBased) maintenance, in the operation of the machine, its main (or need) parts of the regular (or continuous) condition monitoring and fault diagnosis, to determine the state of the equipment, predict the state of the equipment future development trends, based on the state of the equipment development trends and possible failure modes, pre-developed predictive maintenance plan to determine when and how the machine should be repaired and the necessary technical and material support. Based on the state development trend and possible failure mode of the equipment, the predictive maintenance program is formulated in advance to determine the time, content, mode and necessary technical and material support that the machine should be repaired. Predictive maintenance set equipment condition monitoring, fault diagnosis, fault (state) prediction, maintenance decision support and maintenance activities in one, is an emerging maintenance mode.

Predictive maintenance is not only in the name of the name is different, in the concept of the connotation and extension of the difference, so there is a narrow and broad concept of predictive maintenance.

The narrow sense of predictive maintenance based on "condition monitoring", emphasizing "fault diagnosis", refers to irregular or continuous condition monitoring of equipment, according to the results, to identify the state of the equipment with or without abnormalities or fault trends, and then timely Arranging for maintenance. Predictive maintenance in the narrow sense of the maintenance cycle is not fixed, only through the monitoring and diagnosis of the results to the timely scheduling of maintenance programs, which emphasizes the monitoring, diagnosis and maintenance of the trinity of the process, this idea is widely used in process industries and mass production methods.

Generalized predictive maintenance will be condition monitoring, fault diagnosis, condition prediction and maintenance decision-making multi-bit integration, condition monitoring and fault diagnosis is the foundation, condition prediction is the focus of the maintenance decision-making to come to the final maintenance activities required. Predictive maintenance in the broad sense is a systematic process, it will be the maintenance management into the scope of predictive maintenance, the overall consideration of the entire maintenance process, until the maintenance activities related to the content.

Corrective Maintenance (CorrectiveMaintenance), also known as after-the-fact maintenance (Break-downMaintenance), is "fault-based maintenance (FailureBased)" approach, which is based on whether the equipment is intact or whether the maintenance, only to the extent that the equipment can be used. It is based on whether the equipment is intact or whether the maintenance can be used as a basis, only in part or all of the equipment failure and then restore its original state, that is, after the use of bad repair, belongs to the unplanned maintenance.

Preventive maintenance (PreventiveMaintenance), also known as timed maintenance, is based on time (TimeBased) maintenance, it is based on the production plan and experience, according to the specified time intervals for stopping to check, disassemble, and replace parts and components, in order to prevent damage, secondary destruction and production losses. This maintenance method is also commonly used by the current planned maintenance or regular maintenance, such as large, medium and small repairs.

Predictive maintenance technology system:

1, condition monitoring technology

Condition monitoring technology has developed to the present, in various engineering fields have formed their own monitoring methods, condition monitoring methods based on the different means of condition detection and is divided into a number of species, commonly used, including: vibration monitoring method, noise monitoring method, temperature monitoring method, pressure monitoring method, fluid analysis monitoring method, acoustic emission monitoring method. method, acoustic emission monitoring method.

2, fault diagnosis technology

Speaking only of "fault diagnosis", it is a newly developed science, and more and more attention, especially in the continuous production system, fault diagnosis has a very important meaning. According to the principle of diagnosis, fault diagnosis can be divided into: time-frequency diagnosis, statistical diagnosis, information theory analysis and other artificial intelligence methods (expert system diagnosis, artificial neural network diagnosis, etc.), fuzzy diagnosis, gray system theory diagnosis and integrated diagnosis (such as fuzzy expert system diagnosis, neural network expert system diagnosis, fuzzy neural network diagnosis, etc.).

3, state prediction technology

State prediction is based on the operation of the equipment information, assessment of the current state of the components and expected future state. The commonly used methods are time series model prediction method, gray model prediction method and neural network prediction method. And there are generally three basic ways for the development of prediction methods: physical models, knowledge systems and statistical models. In practical application, the three paths can be integrated together to form a hybrid fault prediction technology that combines the traditional physical model and intelligent analysis methods, and can deal with digital and symbolic information, which is more effective for realizing predictive maintenance.

4, maintenance decision support and maintenance activities

Maintenance decision-making is from the personnel, resources, time, cost, efficiency and other aspects, from multiple perspectives, according to the results of condition monitoring, fault diagnosis and condition prediction of the maintenance feasibility analysis, to set out the maintenance plan to determine the maintenance of resources to ensure that the maintenance activities of the time, location, personnel and content. Maintenance decision-making methods generally have fault tree reasoning method, mathematical model analysis method, Bayesian (Bayes) network method (applicable to the expression and analysis of uncertainty and probabilistic things) and intelligent maintenance decision-making method.