Predictive diagnostics
It can have a significant economic effect on an industrial enterprise.
This approach allows you to predict possible equipment and machine failures before they occur, which can reduce equipment downtime and increase productivity.
Here are some examples of the economic benefits of predictive diagnostics::
All this can lead to a significant economic effect on an industrial enterprise. This is confirmed by studies and reports that show that using predictive diagnostics can reduce equipment maintenance costs by 10% -40%, and equipment downtime by 50%. It also allows you to increase equipment productivity by 3% -5%, reduce repair costs by 12% -15%, and reduce the cost of spare parts and materials by 3% -5%.
Here are some examples of the economic benefits of predictive diagnostics::
- Reduction of equipment downtime.
- Reduce the cost of repairs.
- Product quality improvement.
- Optimization of production processes.
- Reduced maintenance costs;
All this can lead to a significant economic effect on an industrial enterprise. This is confirmed by studies and reports that show that using predictive diagnostics can reduce equipment maintenance costs by 10% -40%, and equipment downtime by 50%. It also allows you to increase equipment productivity by 3% -5%, reduce repair costs by 12% -15%, and reduce the cost of spare parts and materials by 3% -5%.
Masyutin Alexey Aleksandrovich
Head of the Center for Artificial Intelligence,
National Research University "Higher School of Economics"
National Research University "Higher School of Economics"
Certainly, the future development of heavy industry lies with analytical systems that use machine learning.
AI technologies show high efficiency in the tasks of diagnostics and predicting breakdowns and, at the same time, are more flexible in relation to the initial conditions than the "classical" methods of analyzing diagnostic signals.*
I believe that the use of Russian intelligent data processing solutions in production is very promising and corresponds to the priorities of the national economy development.
AI technologies show high efficiency in the tasks of diagnostics and predicting breakdowns and, at the same time, are more flexible in relation to the initial conditions than the "classical" methods of analyzing diagnostic signals.*
I believe that the use of Russian intelligent data processing solutions in production is very promising and corresponds to the priorities of the national economy development.
Vibromonitoring is a technology for multiple (continuous or periodic) monitoring of object vibration parameters with analysis of changes in their values over time. A mandatory condition for vibration monitoring is the accumulation of measurement results in a database with their binding to a specific unit, point and direction of vibration monitoring.
Assessment of the technical condition is performed by the maximum value of the vibration velocity recorded at the control points. Acceptable vibration levels are determined from GOST ISO 10816
Assessment of the technical condition is performed by the maximum value of the vibration velocity recorded at the control points. Acceptable vibration levels are determined from GOST ISO 10816
Vibration diagnostics – a method of diagnostics of dynamic equipment, based on the study of vibration parameters created either directly by operating equipment, or superimposed secondary vibration. Vibration diagnostics of equipment is the main method for determining the technical condition of dynamic units, which gives a high probability of detecting equipment defects.
The transition from the time domain of the representation to the frequency domain (spectra) is performed using the Fast Fourier Transform FFT. Spectral analysis is a signal processing technique that allows you to identify the frequency composition of a signal.
The transition from the time domain of the representation to the frequency domain (spectra) is performed using the Fast Fourier Transform FFT. Spectral analysis is a signal processing technique that allows you to identify the frequency composition of a signal.
Predictive analytics is a class of data analysis methods used to predict the future behavior of objects and subjects in order to make optimal decisions. Predictive analytics uses statistical methods, data mining, game theory, machine learning, and neural networks, and evaluates current and historical facts to make predictions about future events.
There are four main stages of implementing predictive analysis methods: data collection, data processing, data analysis,and forecasting.
Global practice shows that the construction of predictive models describing the behavior of most types of dynamic equipment requires a joint analysis of the following information, grouped by sources:
Automated process control system: information about the technological process and current operating modes – instantaneous discrete values of vibration, speed, temperature, current, pressure, etc.
Mobile MRO: visual inspection results, discrete values of vibration and temperature obtained during daily inspections by technological or operational personnel.
Portable diagnostic devices: results of periodic measurements of various dynamic data (vibration signals and spectra, thermograms, UZK maps, etc.).) by diagnostic services or shop staff.
Stationary diagnostic systems: results of continuous measurements of various dynamic data (vibration signals and spectra, etc.).) the most critical and critical equipment in real time.
MRO modules (general factory and shop) : information about ongoing maintenance and repairs, input monitoring, load balancing, and alignment.
There are four main stages of implementing predictive analysis methods: data collection, data processing, data analysis,and forecasting.
Global practice shows that the construction of predictive models describing the behavior of most types of dynamic equipment requires a joint analysis of the following information, grouped by sources:
Automated process control system: information about the technological process and current operating modes – instantaneous discrete values of vibration, speed, temperature, current, pressure, etc.
Mobile MRO: visual inspection results, discrete values of vibration and temperature obtained during daily inspections by technological or operational personnel.
Portable diagnostic devices: results of periodic measurements of various dynamic data (vibration signals and spectra, thermograms, UZK maps, etc.).) by diagnostic services or shop staff.
Stationary diagnostic systems: results of continuous measurements of various dynamic data (vibration signals and spectra, etc.).) the most critical and critical equipment in real time.
MRO modules (general factory and shop) : information about ongoing maintenance and repairs, input monitoring, load balancing, and alignment.
Estimation of the remaining resource based on the forecast of parameters
The remaining resource is the total operating time of an object from the moment of monitoring its technical condition to the transition to the limit state.
The estimation of the remaining resource is based on the forecast of growth of the measured and calculated diagnostic parameters of the system. Based on historical measurement data, a line of the predicted state of the parameter (degradation model)\is plotted on the parameter trend graphs.
Predicted failure time – the point on the graph where the predicted state line crosses the control limit for this parameter.
One of two laws can be chosen for the degradation model:
Linear degradation: The forecast is a straight line whose slope angle is determined by historical data. This is usually applied if the system does not accumulate damage (degradation).
Exponential degradation: The forecast is an exponent of. This is usually applied if the system can accumulate damage.
The estimation of the remaining resource is based on the forecast of growth of the measured and calculated diagnostic parameters of the system. Based on historical measurement data, a line of the predicted state of the parameter (degradation model)\is plotted on the parameter trend graphs.
Predicted failure time – the point on the graph where the predicted state line crosses the control limit for this parameter.
One of two laws can be chosen for the degradation model:
Linear degradation: The forecast is a straight line whose slope angle is determined by historical data. This is usually applied if the system does not accumulate damage (degradation).
Exponential degradation: The forecast is an exponent of. This is usually applied if the system can accumulate damage.
Maintenance strategies
Positive:
- Full use of the machine's resource
- No monitoring costs
- Secondary damage is possible
- Repairs take time, unplanned downtime
- Need a spare parts warehouse
- Possible failure of the car for a long time
Positive:
- Less chance of secondary damage
- Downtime and repairs on schedule
- No monitoring costs
- Incomplete use of the machine resource
Positive:
- Full use of the machine's resource
- Spare parts and repairs according to the condition of the car
- No secondary damage
- Scheduled time to stop
- Monitoring costs
Positive:
- Full use of the machine's resource
- Spare parts and repairs according to the condition of the car
- No secondary damage
- Scheduled time to stop
- Extended operating time of the equipment
- Monitoring costs
THEY TRUST US
