Empirical modeling for processing parameters' effects on coating properties in plasma spraying process. (August 2015)

• Record Type: Journal Article
• Authors: Wu, Zhenhua
• Is Part Of: Journal of manufacturing processes. Volume 19(2015)
• Page Start: 1
• Page End: 13
• Publication Date: 2015-08
• Subjects: Empirical modeling -- Plasma spraying -- Processing parameters -- Coating properties -- Coating mechanism; Production management -- Data processing -- Periodicals; Manufacturing processes -- Periodicals; Procestechnologie; Productietechniek; Production -- Gestion -- Informatique -- Périodiques; Fabrication -- Périodiques; Manufacturing processes; Production management -- Data processing; Periodicals; 670.5
• DOI: 10.1016/j.jmapro.2015.03.007
• Languages: English
• ISSNs: 1526-6125

Abstract

Highlights: Solid implementation of ANFIS and NN models to identify the pattern between processing parameters and coating properties. Systematical evaluation on model complexity and generalization with bootstrap re-sampling and cross-validation approaches. Detailed interpretation on physical mechanism between processing parameters and coating properties. Abstract: In order to model the effects of processing parameters (primary gas flow rate, stand-off distance, powder flow rate, and arc current) on the plasma spraying coating properties (thickness, porosity and micro-hardness), adaptive neural fuzzy inference system (ANFIS) and neural network (NN) based empirical models were proposed to estimate process parameters and understand the spraying process. To overcome the difficulty of the small size of sample data, and to balance the trade-off between model complexity and prediction accuracy, the bootstrap method was applied for the resampling technique, and cross validation was applied for the performance evaluation. The ANFIS and NN models were compared on the performance metrics of (1) mean square error (MSE), and (2) determination coefficient ( R 2 ). With the limited size of experiment data, both models illustrated high accuracy. In the training stage: on the R 2, ANFIS has the value of 1, and NN has the minimum value of 0.84; on the MSE, ANFIS has the minimum value of 1.3e−5, and NN has the minimum value of 0.32. In the validation stage: on the R 2, ANFIS has the minimum mean value of 0.42, and NN has the minimum mean value of 0.512; on the MSE, ANFIS has the minimum mean value of 23.67, and NN has the minimum mean value of 89.50. The comparisons illustrated that ANFIS model showed significant superiority over the NN model. This may be due to the fact that ANFIS combines the strength of NN's learning capability and fuzzy logic's knowledge interpretation ability. With the obtained ANFIS model, the physical mechanisms – including (1) melting states of particles, (2) loading effect, and (3) oxidation – were interpreted as processing parameters' effects on the coating properties. The empirical models and that physical mechanism are viable to be effectively integrated with feedback control strategy to regulate the coating quality in plasma spraying process.