Systems, apparatus, and methods for adjusting parameters in response to anticipated component state
Systems, methods, and apparatus for adjusting parameters in response to a predicted anticipated state of a component are described. A system may have a data collector to collect sensor data from an industrial environment and a controller to process the data. A collection parameter for one of the input sensors may be determined, an output data pattern recognized, and a signature selected from a plurality of signatures associated with output data patterns. A neural network, trained on signatures associated with output patterns to detect a state of the industrial environment, may predict an anticipated state of the environment and a parameter may be adjusted accordingly.
1 . A monitoring system for data collection in an industrial environment, comprising:
a data collector including a controller, the data collector communicatively coupled to a plurality of input sensors operatively coupled to an industrial component in the industrial environment and configured to collect sensor data from the industrial component in the industrial environment, and
the controller including:
a data collection band circuit structured to alter at least one collection parameter for at least one of the plurality of input sensors from which to process output data based on collected sensor data and at least one of learned received output data patterns, or a state, and wherein altering the at least one collection parameter alters a physical collection of the sensor data from the at least one of the plurality of input sensors by adjusting at least one of a bandwidth for sensor data, a multiplexing configuration, a timing parameter, a frequency range, or a granularity of collection of sensor data;
a pattern recognition circuit structured to recognize at least one output data pattern based on the collected sensor data and select a signature from a plurality of signatures associated with output data patterns;
a machine learning data analysis circuit comprising a neural network structured to receive the output data patterns from the at least one of the plurality of input sensors and the selected signature, wherein:
the neural network is trained with the plurality of signatures associated with the output data patterns, including the selected signature to recognize one or more states including one or more operational states and one or more anticipated states of the industrial component, and learn the received output data patterns indicative of the one or more states; and
the trained neural network predicts an anticipated state of the industrial component based on the selected signature, which corresponds to an operational state of the industrial component within the industrial environment; and
a response circuit structured to adjust an operation of the industrial component of the industrial environment based on the predicted anticipated state of the industrial component.
2 . The system of claim 1 , wherein the operation comprises a task of the industrial component of the industrial environment.
3 . The system of claim 1 , wherein the operation is adjusted to implement at least one of: an increase in fuel efficiency, a reduction in wear, an increase of production output, an increase of an operating life of the industrial component of the industrial environment, avoidance of a fault condition, or a reduction of a load on the industrial component of the industrial environment.
4 . The system of claim 1 , wherein the operation comprises a future component design for the industrial component of the industrial environment.
5 . The system of claim 1 , wherein the one or more states of the industrial environment correspond to at least one of an outcome, or an anticipated outcome, relating to a product of the industrial environment.
6 . The system of claim 1 , wherein the machine learning data analysis circuit is further structured to learn the received output data patterns by being seeded with a model.
7 . The system of claim 1 , wherein the neural network of the machine learning data analysis circuit is further trained with the output data patterns indicative of a progress towards at least one of a goal or an alignment with a guideline.
8 . The system of claim 1 , wherein the industrial environment includes a production line.
9 . The system of claim 1 , further comprising:
the plurality of input sensors includes at least two of a sensor group having: a temperature sensor, a load sensor, an optical vibration sensor, an acoustic wave sensor, a heat flux sensor, an infrared sensor, an accelerometer, a tri-axial vibration sensor, or a tachometer.
10 . The system of claim 1 , wherein the data collection band circuit is further structured to adjust the at least one collection parameter based on one or more of the received output data patterns.
11 . The monitoring system of claim 1 , wherein, based on detected operational phase transitions, the altering the at least one collection parameter includes increasing sampling rates during detected transient conditions and decreasing sampling rates during normal operating conditions to optimize data storage while capturing critical events.
12 . A monitoring apparatus, comprising:
a data collector including a controller, the data collector communicatively coupled to a plurality of input sensors operatively coupled to an industrial environment, and
the controller including:
a data collection band circuit structured to implement an adaptative sampling protocol that adjusts at least one collection parameter for at least one of the plurality of input sensors based on detected operational phase transitions from which to process output data, wherein the adaptive sampling protocol increases sampling rates during detected transient conditions and decreases sampling rates during normal operating conditions to optimize data storage while capturing critical events;
a pattern recognition circuit structured to recognize at least one output data pattern and select a signature from a plurality of signatures associated with output data patterns;
a machine learning data analysis circuit structured to receive the output data patterns from the at least one of the plurality of input sensors and the selected signature, wherein:
the machine learning data analysis circuit is trained with a portion of the plurality of signatures associated with the output data patterns, including the selected signature to recognize one or more states, including one or more operational states and one or more anticipated states of the industrial environment; and
the trained machine learning data analysis circuit continuously predicts multiple time-period anticipated states of the industrial environment based on the selected signature, which corresponds to an operational state of a component within the industrial environment, wherein the multiple time-period predictions enable both immediate and long-term operational planning; and
a response circuit structured to adjust an operation of the component of the industrial environment based on the multiple time-period predicted anticipated states.
13 . The apparatus of claim 12 , wherein the operation comprises a task of the component of the industrial environment.
14 . The apparatus of claim 12 , wherein the operation is adjusted to implement at least one of: an increase in fuel efficiency, a reduction in wear, an increase of production output, an increase of an operating life of the component of the industrial environment, an avoidance of a fault condition, or a reduction of a load on the component of the industrial environment.
15 . The apparatus of claim 12 , wherein the one or more states correspond to at least one of an outcome, or an anticipated outcome, relating to a product of the industrial environment.
16 . The apparatus of claim 12 , wherein the machine learning data analysis circuit is further structured to learn the received output data patterns by being seeded with a model.
17 . The apparatus of claim 12 , wherein the machine learning data analysis circuit is further trained with the output data patterns indicative of a progress towards at least one of a goal or an alignment with a guideline.
18 . The apparatus of claim 12 , wherein the industrial environment includes a production line.
19 . The apparatus of claim 12 , further comprising:
the plurality of input sensors includes at least two sensors selected from a group of sensors including: a temperature sensor, a load sensor, an optical vibration sensor, an acoustic wave sensor, a heat flux sensor, an infrared sensor, an accelerometer, a tri-axial vibration sensor, or a tachometer.
20 . A method for data collection in an industrial environment, comprising:
collecting sensor data from a plurality of input sensors operatively coupled to the industrial environment, the plurality of input sensors communicatively coupled to a data collector;
implementing an adaptive sampling protocol that adjusts at least one collection parameter for at least one of the plurality of input sensors from which to process output data based on the collected sensor data, wherein the at least one collection parameter alters a physical collecting of sensor data by adjusting, for at least one sensor, at least one of: a bandwidth for sensor data, a multiplexing configuration, a timing parameter, a frequency range, or a granularity of collection of sensor data;
receiving the collected sensor data from the at least one of the plurality of input sensors;
performing a pattern recognition operation on the collected sensor data to recognize at least one output data pattern and select a signature from a plurality of signatures associated with output data patterns;
performing a machine learning operation to train a neural network with the signatures associated with the output data patterns, including the selected signature to recognize one or more states of a component of the industrial environment;
continuously predicting multiple time-period anticipated states of the industrial environment using the trained neural network based on the selected signature which corresponds to an operational state of the component within the industrial environment; and
adjusting an operation of the component of the industrial environment in response to the predicted anticipated state of the component.
21 . The method of claim 20 , wherein the adjusting the operation comprises implementing at least one of: increasing a fuel efficiency, reducing wear of the component, increasing a production output of the industrial environment, increasing an operating life of the component, avoiding a fault condition, or reducing a load on the component.
22 . The method of claim 20 , wherein the adjusting the operation comprises at least one of: scheduling a maintenance for the component, providing a future component design for the industrial environment, or ordering at least one of a new component or a replacement component.
23 . The method of claim 20 , wherein the performing the machine learning operation to train the neural network further comprises training with the at least one output data pattern indicative of a progress toward at least one of: a goal, or an alignment with a guideline.
24 . The method of claim 20 , wherein the industrial environment includes a production line.
25 . The method of claim 20 , wherein the adaptive sampling protocol increases sampling rates during detected transient conditions and decreases sampling rates during normal operating conditions to optimize data storage while capturing critical events.