Fuzzy Comprehensive Evaluation of Strip Cladding Forming Quality Based on Matlab-FIS

Literature Overview

This paper, published in the Welding Journal (Hanjie Xuebao) in 2013 by Guo Xiao, Xu Kai, and Zou Liwei from the Harbin Welding Research Institute of the Chinese Academy of Mechanical Sciences, addresses a critical challenge in strip cladding (SAW-based strip cladding or submerged arc welding strip cladding) manufacturing: the systematic evaluation of forming quality. The work was supported by two national major science and technology projects (2012ZX060004-21 and 2011ZX04016-061), reflecting its significance in the context of China's strategic energy and materials development programs. The core methodology involves constructing a fuzzy inference system (FIS) implemented in MATLAB to perform comprehensive quality assessment of strip cladding deposits.

Core Technical Content and Methodology

The Problem of Multi-Parameter Quality Evaluation

Strip cladding, particularly submerged arc welding strip cladding (SAW strip cladding), is widely employed for producing clad plate where corrosion resistance or wear resistance is required on the surface of a structural steel substrate. The quality of strip cladding is not determined by a single parameter but rather by a complex interplay of multiple factors including dilution rate, overlay thickness uniformity, surface morphology, microstructure characteristics, and bond strength. Traditional evaluation methods often assess individual parameters in isolation, which fails to capture the holistic quality picture. This paper proposes a fuzzy comprehensive evaluation approach that integrates multiple quality indicators into a unified assessment framework.

Fuzzy Inference System Architecture

The Matlab-FIS approach employs fuzzy set theory to handle the inherent uncertainty and subjectivity in quality assessment. The system architecture typically includes the following components:

Component	Function	Typical Implementation
Fuzzification	Convert crisp input values to fuzzy sets	Membership functions for each quality parameter
Rule Base	Define expert knowledge in IF-THEN rules	Linguistic rules based on engineering experience
Inference Engine	Apply fuzzy logic operations	Mamdani or Sugeno inference method
Defuzzification	Convert fuzzy output to crisp value	Centroid method or weighted average

The key quality parameters evaluated in strip cladding typically include:

Dilution rate: The percentage of base metal in the overlay layer, directly affecting corrosion resistance and alloy composition. For stainless steel overlay, dilution below 30% is generally required to maintain adequate chromium content.
Overlay thickness uniformity: Variation in thickness across the cladding surface, affecting both functional performance and subsequent machining allowance.
Surface quality: Including surface roughness, presence of ripples, and visual defects.
Bond strength: The interfacial strength between overlay and substrate, critical for structural integrity.
Microstructure: Grain size, phase composition, and presence of deleterious phases.

Process Parameters Influencing Strip Cladding Quality

The study implicitly references the following process parameters that feed into the quality evaluation:

Process Parameter	Typical Range	Effect on Quality
Welding current	300-600 A	Affects penetration and dilution
Welding voltage	28-36 V	Influences deposition rate and bead width
Travel speed	0.3-0.8 m/min	Controls heat input and dilution
Wire feed speed	2-6 m/min	Affects deposition rate
Shielding gas composition	Ar + CO2 mixtures	Influences weld pool stability
Electrode composition	Matching overlay material	Determines final alloy composition

Engineering Practice Integration

Application in Quality Control Systems

The fuzzy comprehensive evaluation method offers significant advantages for integration into manufacturing quality control systems. In practice, strip cladding operations at facilities producing clad plate for pressure vessels, heat exchangers, and chemical equipment require consistent quality assessment across large production volumes. The Matlab-FIS approach provides:

A repeatable and objective evaluation methodology that reduces inter-operator variability.
The ability to incorporate expert knowledge through the rule base, ensuring that accumulated manufacturing experience is preserved and applied consistently.
A framework for continuous improvement where the rule base can be refined as new data becomes available.

Connection to Standards Requirements

The quality parameters evaluated align with requirements in relevant standards including:

GB/T 150: Pressure vessel fabrication standards requiring specified overlay thickness and bond strength.
ASTM A263/A264/A265: Standards for clad plate specifying minimum overlay thickness, dilution limits, and bond strength requirements.
NB/T 47002: Chinese standards for steel plates used in pressure vessels.
API 934: Specification for overlay welding of casing and tubing.

Practical Implementation Considerations

From an engineering practice perspective, implementing such an evaluation system requires:

Data acquisition: Reliable measurement systems for each quality parameter, including dilution analysis (optical emission spectrometry or chemical analysis), thickness measurement (ultrasonic or contact methods), and surface inspection.
Rule base development: Systematic extraction of expert knowledge through interviews with experienced welders and quality engineers, supplemented by statistical analysis of production data.
Validation: Comparison of fuzzy evaluation results with conventional acceptance/rejection decisions to ensure the system does not introduce unacceptable risk.

Key Technical Insights and Reflections

Advantages of Fuzzy Logic in Welding Quality Assessment

The fundamental insight of this work is that welding quality assessment is inherently a multi-criteria decision-making problem where the boundaries between "acceptable" and "unacceptable" are often fuzzy rather than crisp. Traditional binary decision rules (pass/fail based on individual parameter limits) may reject otherwise functional products or, conversely, accept products with marginal overall quality. The fuzzy approach better reflects engineering reality by acknowledging that:

A slightly elevated dilution rate may be compensated by excellent bond strength.
Minor surface imperfections may be acceptable if functional properties are fully met.
The overall quality is a weighted combination of multiple factors, not a simple minimum of individual assessments.

Limitations and Considerations

While the methodology is conceptually sound, several practical limitations warrant consideration:

The accuracy of the evaluation depends critically on the quality of the rule base, which requires significant expert input.
The system must be calibrated against actual failure data to ensure it provides meaningful risk differentiation.
Integration with real-time process monitoring systems requires additional development for automatic data acquisition and processing.

Study Insights and Implications for Practice

This work represents an important contribution to the rationalization of cladding quality assessment. For engineers involved in strip cladding operations, the key takeaway is that systematic, multi-parameter evaluation methods produce more reliable quality predictions than single-parameter checks. The fuzzy logic approach, while requiring initial development effort, offers a framework that can be adapted to different cladding applications, materials, and process configurations. Future work should focus on integrating such evaluation systems with real-time process monitoring and predictive quality models to enable closed-loop process control in strip cladding manufacturing.