Methods and systems of diagnosing machine components using neural networks and having bandwidth allocation
Systems and methods for data collection in an industrial environment are disclosed. A system may include a data collector to collect data from a subset of a plurality of input channels based on a selected data collection routine, and a data acquisition and analysis circuit for receiving the collected data and analyzing the collected data using an expert system analysis circuit to determine an occurrence of an anomalous condition for a machine component based on an analysis. The expert system analysis circuit may utilize a neural network. The data analysis circuit may determine an aggregate rate of data being collected and, if the aggregate rate exceeds a current bandwidth allocation rate associated with the network infrastructure, request an increase to the current bandwidth allocation rate from the network infrastructure.
1. A system for data collection in an industrial environment, the system comprising:
a data collector communicatively coupled to a plurality of input channels and to a network infrastructure, wherein the data collector collects data from a subset of the plurality of input channels based on a selected data collection routine, wherein each of the plurality of input channels communicates a respective stream of detection values for a corresponding machine component in the industrial environment; and
a data acquisition and analysis circuit for receiving the streams of detection values and structured to analyze the streams of detection values using an expert system analysis circuit, wherein the expert system analysis circuit determines a fault condition for a machine component based on an analysis of the streams of detection values, wherein the expert system analysis circuit utilizes a neural network comprising one of a probabilistic, a time delay, and a convolutional neural network, wherein the data acquisition and analysis circuit is structured to determine an aggregate rate of data being collected from the subset of the plurality of input channels, wherein if the aggregate rate exceeds a current bandwidth allocation rate associated with the network infrastructure, then the data acquisition and analysis circuit requests an increase to the current bandwidth allocation rate from the network infrastructure, and wherein a data storage has a data capacity allocation for the streams of detection values, and the data capacity allocation is increased until the current bandwidth allocation rate is increased to meet the determined aggregate rate of data.
2. The system of claim 1 , wherein the data acquisition and analysis circuit selectively eliminates collected data until the current bandwidth allocation rate is increased to meet the determined aggregate rate of data.
3. The system of claim 2 wherein the collected data are eliminated to reduce a number of monitoring points co-located on an industrial machine.
4. The system of claim 3 , wherein the monitoring points are deactivated.
5. The system of claim 2 , wherein the collected data are eliminated based on a hierarchical template that establishes a hierarchy for the collected data.
6. The system of claim 1 , wherein the data acquisition and analysis circuit modifies a data collection parameter to reduce an amount of data being collected from an individual input channel by reducing a sampling rate of the input channel or reducing a sampling resolution of the input channel.
7. The system of claim 1 , wherein data from multiple input channels are multiplexed as a fused data stream until the current bandwidth allocation rate is increased to meet the determined aggregate rate of data.
8. The system of claim 1 , wherein the neural network comprises a probabilistic neural network that determines an occurrence of the fault condition based on pattern recognition.
9. The system of claim 1 , wherein the neural network comprises a time delay neural network that determines an occurrence of the fault condition based on pattern recognition.
10. The system of claim 9 , wherein the time delay neural network is trained with machine learning.
11. The system of claim 1 , further comprising an analyzed stream of detection values that represents sound.
12. The system of claim 1 , wherein the neural network comprises a convolutional neural network that determines an occurrence of the fault condition based on pattern recognition.
13. The system of claim 12 , further comprising an analyzed stream of detection values that comprises image data.
14. The system of claim 12 , further comprising an analyzed stream of detection values that comprises video data.
15. The system of claim 1 , wherein the expert system analysis circuit analyzes a respective stream of detection values from a first and a second of a plurality of input channels to determine a relative phase at one or more times, and wherein the expert system analysis circuit further determines the fault condition in response to the relative phase determination.
16. The system of claim 1 , wherein the expert system analysis circuit further controls a plurality of data collection bands for determining collection schedules for different groupings of the plurality of input channels.
17. The system of claim 1 , wherein the data acquisition and analysis circuit determines a failure state for a machine component based on the analysis and provides the failure state to the data storage.
18. The system of claim 17 , wherein the expert system analysis circuit analyzes a first stream of detection values corresponding to a first input channel and a second stream of detection values corresponding to a second input channel for a relative phase determination and detects the failure state for the machine component using the relative phase determination.
19. The system of claim 1 , wherein the expert system analysis circuit controls data collection bands of the streams of detection values.