A Flow Rate Soft Sensor for Hydraulic Proportional Directional Valves

Flow rate is a critical control factor for hydraulic proportional directional valves. Traditional flow rate monitoring relies on flowmeters, which are inflexible and at risk of failure. To improve measurement scalability and fault tolerance, a novel soft sensor based on an embedded flow rate inferen...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE sensors journal Ročník 24; číslo 24; s. 42281 - 42288
Hlavní autoři: Ren, Wei, Su, Wenbin, Mu, Xinrong, Wei, Hongbo
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 15.12.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Accuracy
Algorithms
Artificial neural networks
Back propagation networks
Backpropagation neural network improved by wild horse optimizer (BPNN-WHO)
Circuits
embedded flow rate inference algorithm
External pressure
Fault tolerance
Flow velocity
Flowmeters
Hardware
hydraulic proportional directional valves
Hydraulic systems
Inference algorithms
Monitoring
Neural networks
Orifices
Pressure drop
Safety
Sensors
Smoothness
soft sensor
Soft sensors
Symbols
Valves
ISSN:1530-437X, 1558-1748
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Flow rate is a critical control factor for hydraulic proportional directional valves. Traditional flow rate monitoring relies on flowmeters, which are inflexible and at risk of failure. To improve measurement scalability and fault tolerance, a novel soft sensor based on an embedded flow rate inference algorithm is presented. The algorithm is taken to the STM32 and purely performs forward propagation of the backpropagation neural network improved by wild horse optimizer (BPNN-WHO). Experimental data is utilized to support the development of soft sensors. Pressure drop across the orifice and spool displacement were selected as inputs. By comparing the smoothness and distortion of signals before and after filtering, the window size of the moving average filter was determined. Hyperparameters of BPNN were optimized via WHO, and corresponding weights and biases were loaded into hardware circuits. The developed soft sensor was validated under a variety of operating conditions. Mean absolute percentage error (MAPE) of 4.79% in internal validation and 2.80% in external validation was achieved, suggesting outstanding accuracy and generalizability. The proposed soft sensor not only serves as a backup for flow rate monitoring and control but also meets a wide range of measurement requirements and offers numerous applications.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1530-437X
1558-1748
DOI:10.1109/JSEN.2024.3479216