This paper systematically explores cutting-edge fuzzy systems—such as Type-2, Pythagorean, and Neutrosophic sets—and computational frameworks like Granular Computing (GrC) and Z-numbers, aligning them with practical applications in intelligent systems. The study addresses critical gaps in translating theoretical models into real-world solutions, emphasizing methodological innovations and interdisciplinary challenges. Fuzzy systems, particularly Type-2 Fuzzy Sets (T2FS), enhance decision-making in dynamic environments by modeling second-order uncertainties, as demonstrated in adaptive control systems. Intuitionistic Fuzzy Sets (IFS) extend classical fuzzy logic by incorporating non-membership degrees, proving effective in multi-criteria decision analysis (e.g., sustainable supply chain management). Pythagorean Fuzzy Sets (PFS) further generalize IFS by allowing squared membership values, improving flexibility in high-stakes scenarios like e-commerce demand forecasting. Advanced extensions are analyzed for cybersecurity and crisis management applications, including Neutrosophic Sets (Handling indeterminacy) and soft sets (Managing incomplete data). In computational intelligence, GrC partitions data into hierarchical granules, enabling context-aware decisions in smart traffic systems. Z-numbers and D-numbers —tools for reliability-based uncertainty modeling—are evaluated for risk assessment in infrastructure projects, though their integration with Machine Learning (ML) remains underexplored. Hybrid models, such as Fuzzy-Genetic Algorithms (GAs), showcase practical benefits, reducing energy consumption in smart grids by 18. Methodological challenges include translating technical terms (e.g., "complement" in fuzzy logic) across interdisciplinary teams, requiring context-aware approaches to preserve semantic accuracy. Case studies highlight a 22% reduction in diagnostic errors using Pythagorean fuzzy systems in healthcare and optimized supplier selection in automotive supply chains via IFS. This paper underscores the transformative potential of fuzzy systems and computational intelligence while advocating for standardized frameworks, improved interpretability of hybrid models, and broader adoption of Z-numbers in industry. Future research should prioritize bridging theoretical advancements with scalable, real-world implementations to address global challenges in sustainability, healthcare, and smart infrastructure.
In an era increasingly defined by uncertainty and complexity, conventional decision-support systems often fall short in addressing imprecision, vagueness, and incomplete information. As industries evolve toward more intelligent and adaptive infrastructures, the need for robust uncertainty modeling becomes paramount. Fuzzy systems and computational intelligence have emerged as powerful frameworks to meet these demands, offering flexible and interpretable mechanisms for modeling human-like reasoning. This paper explores advanced forms of fuzzy logic—such as Type-2 Fuzzy Sets (T2FS), Pythagorean Fuzzy Sets (PFS), and Neutrosophic Sets—and their integration with computational paradigms like Granular Computing (GrC), Z-numbers, and hybrid intelligent systems. By synthesizing theoretical innovations and practical case studies across sectors such as healthcare, smart energy grids, cybersecurity, and supply chain management, this work aims to bridge the gap between abstract modeling and real-world decision-making. The following sections provide a comprehensive analysis of these frameworks, highlighting their transformative role in the next generation of decision-support systems.
Fuzzy sets and their extensions form the foundation of uncertainty modeling in computational intelligence. This section delves into their theoretical underpinnings, mathematical operations, and practical applications
Type-1 Fuzzy Sets (T1FS): The simplest form, where each element has a membership degree in the range [0,1]. For example, the statement "Temperature is high" can be modeled with a T1FS.
Limitation: Unable to handle second-order uncertainties (e.g., ambiguity in membership values themselves). T2FS: Extend T1FS by introducing a secondary membership function to model uncertainties in the primary membership. Extending T1FS makes T2FS ideal for dynamic systems like stock market prediction.
Interval T2FS: A simplified version where the secondary membership is uniform, reducing computational complexity.
Defined by membership (μ), non-membership (ν), and hesitancy (π=1−μ−ν) degrees, Intuitionistic Fuzzy Sets (IFS) capture ambiguity more effectively than T1FS
Union/intersection: Use operations like μA∪B=max(μA,μB) and νA∪B=min(νA,νB).
Modal operators: Operations like ∗, ⊙, and ⋈ refine decision-making in vague environments.
Career counseling: IFS evaluates candidates based on skills, personality, and job requirements.
Medical diagnosis: Balances conflicting symptoms and uncertain test results.
Generalizes IFS by allowing μ2+ν2≤1, providing greater flexibility in high-uncertainty scenarios
Better representation of human decision-making (e.g., "High risk" vs. "Low risk" in financial investments). Compatible with Machine Learning (ML) algorithms for predictive analytics.
Addition/multiplication: Defined using Dombi operators for complex decision-making.
Relationship with IFS: PFS is a superset of IFS, as it relaxes the constraint μ+ν≤1.
Neutrosophic sets: Incorporate indeterminacy (I) and falsehood (F) alongside truth (T), addressing inconsistent data (e.g., cybersecurity threat analysis)
Picture fuzzy sets: Extend PFS with a refusal degree, useful in voting systems or opinion polls.
Soft sets: Model uncertainty in incomplete datasets (e.g., crisis management) by parameterizing elements. Rough sets: Approximate data boundaries using lower/upper approximations, applied in fraud detection.
Operations on Fuzzy Sets:
− Union: μA∪B(x)=max(μA(x),μB(x)).
− Intersection: μA∩B(x)=min(μA(x),μB(x)).
Distance measures: Euclidean or Hamming distance metrics compare fuzzy sets for optimization tasks.
Healthcare: PFS reduces diagnostic errors by 22% in cancer screening
Energy: Hybrid Type-2 Fuzzy-genetic models cut smart grid energy use by 18% .
Supply chain: IFS optimizes supplier selection by balancing cost, quality, and delivery.
GrC is a computational paradigm that structures complex data into information granules (e.g., intervals, clusters, or fuzzy sets) to simplify decision-making in uncertain environments
This framework operates on three levels:
− Subsymbolic level: Raw data processing (e.g., sensor inputs) − Symbolic level: Abstract representations (e.g., fuzzy rules) − Knowledge level: High-level patterns (e.g., decision trees)
Applications:
− Smart traffic systems: GrC processes real-time traffic data (e.g., vehicle density, weather conditions) into granules to optimize signal timing and reduce congestion.
− Healthcare: Granular models cluster patient data (e.g., symptoms, lab results) to improve diagnostic accuracy in heterogeneous populations.
Advantages:
− Reduces computational complexity by focusing on relevant data subsets − Enhances interpretability through hierarchical abstraction
Z-numbers combine fuzzy values with reliability metrics, represented as Z=(A,B), where
− A: Fuzzy restriction (e.g., "High demand") − B: Reliability of A (e.g., "Very reliable")
Example: In energy demand forecasting, a Z-number might express:
− A: "Daily energy consumption is high."
− B: "This prediction is 80% reliable."
Challenges: Integration with ML requires robust algorithms to handle dual uncertainties (Fuzziness + reliability)
D-numbers generalize Dempster-Shafer theory to model incomplete or conflicting evidence. They assign probabilities to hypotheses without requiring an exhaustive frame of discernment
Application: Infrastructure risk assessment: D-numbers evaluate risks of bridge collapses by aggregating incomplete data (e.g., material degradation, traffic load)
Fuzzy logic enhances ML by addressing uncertainty in data and models
I. Feature selection with PFS: PFS ranks features based on membership (μ) and non-membership (ν) scores.
Example: In financial fraud detection, PFS prioritizes variables like transaction frequency (μ=0.9) over location data (μ=0.4) .
II. Uncertainty quantification in neural networks: Fuzzy Neural Networks (FNN): Combine fuzzy rules with neural architectures to model ambiguity in inputs. Example: In medical imaging, FNNs reduce false positives by assigning uncertainty scores to tumor detections .
III. Hybrid models: Fuzzy-Genetic Algorithms (GAs): Optimize parameters in dynamic systems (e.g., smart grids). A hybrid Type-2 Fuzzy-genetic model achieved 18% energy savings by balancing load distribution and renewable integration
Computational Complexity: Z-number operations increase algorithm runtime.
Interpretability: Hybrid models (e.g., FNNs) often act as "Black boxes," complicating trust in critical applications like healthcare.
By expanding these frameworks, computational intelligence bridges the gap between theoretical models and real-world decision-support systems, particularly in dynamic or data-scarce environments
Hybrid models combine fuzzy systems with other computational intelligence techniques to address complex decision-making challenges. These integrations enhance adaptability, accuracy, and scalability in dynamic environments
Concept: Merge fuzzy logic with Artificial Neural Networks (ANNs) to handle uncertainty in training data
Application:
− Healthcare: FNN reduces diagnostic errors in medical imaging by quantifying uncertainty in tumor detection. − Finance: Predict stock market trends by integrating fuzzy rules with neural network pattern recognition.
Concept: Use GAs to optimize fuzzy system parameters (e.g., membership functions)
Application:
− Energy systems: A Type-2 Fuzzy-genetic model reduced energy consumption in smart grids by 18% by balancing load distribution and renewable integration.
− Supply Chain: Optimize inventory management by evolving fuzzy rules for demand forecasting.
Concept: Particle Swarm Optimization (PSO) algorithms refine fuzzy controllers in real-time dynamic systems.
Application:
− Robotics: Adaptive path planning for drones in uncertain environments using Type-2 Fuzzy-PSO
Translating theoretical fuzzy models into practical applications faces interdisciplinary and linguistic barriers
Issue: Technical terms like "Complement" (Complementary) or "Membership function" (Membership function) may have different interpretations across disciplines
Example: In mathematics, the complement of a fuzzy set A is 1−μA(x), but engineers might interpret it as a binary "Absence".
Solution: Develop context-aware glossaries to standardize terms for interdisciplinary teams .
Issue: Translating fuzzy concepts (e.g., "High risk" vs. "Low risk") into multilingual environments can lead to ambiguity.
Example: The Persian term "Probability" (Probability) is often conflated with "Uncertainty" (Uncertainty) in technical documents
Solution: Use visual aids (e.g., fuzzy membership graphs) to clarify abstract concepts during cross-cultural collaborations.
Challenge: Implementing fuzzy systems in large-scale applications (e.g., smart cities) requires balancing computational efficiency with accuracy.
Case study: A Type-2 fuzzy system for traffic management in Tehran reduced latency by 30% using edge computing.
Data sensitivity: Fuzzy systems in healthcare or surveillance must anonymize data while preserving decision accuracy.
Example: Neutrosophic sets anonymize patient data by masking indeterminate values in diagnostic systems. Bias mitigation: Address biases in training data (e.g., socioeconomic factors in credit scoring) using Pythagorean Fuzzy fairness metrics.
This section provides detailed real-world applications of fuzzy systems and computational intelligence, demonstrating their transformative impact across industries. Each case study aligns with theoretical frameworks discussed earlier and references practical implementations from recent research
Background: Diagnostic errors in medical imaging (e.g., tumor detection) often stem from ambiguous data and subjective interpretations.
Methodology:
− PFS were integrated with neural networks to analyze MRI scans.
− Membership (μ) and non-membership (ν) values quantified uncertainty in tumor boundaries, while hesitation margins (π) accounted for indeterminate regions.
Outcome:
22% reduction in diagnostic errors compared to traditional methods.
Improved accuracy in distinguishing malignant vs. Benign tumors, particularly in early-stage cases .
Challenge: Balancing cost, quality, and delivery reliability in supplier selection for automotive manufacturing. Solution: IFS modeled criteria such as:
− μ: Supplier cost efficiency − ν: Risk of delays − π: Ambiguity in quality certifications
Result:
15% improvement in supply chain efficiency by prioritizing suppliers with optimal μ−ν scores.
Reduced procurement costs while maintaining quality standards.
Problem: Optimizing energy distribution in smart grids with fluctuating renewable energy inputs.
Approach: Type-2 Fuzzy-GA combined:
− Type-2 Fuzzy logic: Handled uncertainties in solar/wind forecasts.
− GAs: Evolved optimal load distribution strategies.
Outcome:
18% reduction in energy waste by dynamically balancing grid loads.
Enhanced integration of renewable energy sources (e.g., solar farms).
Context: Managing congestion in Tehran’s traffic systems using real-time data.
Implementation: GrC processed sensor data (Vehicle density, weather) into hierarchical granules.
3-level framework:
− Subsymbolic: Raw traffic flow data − Symbolic: Fuzzy rules for signal timing (e.g., "IF traffic density is high, THEN extend green light") − Knowledge: Predictive models for peak-hour congestion
Impact:
30% reduction in average commute time during peak hours.
Lowered fuel consumption and emissions through adaptive traffic lights .
Challenge: Identifying sophisticated cyberattacks in networks with incomplete or conflicting data.
Method: Neutrosophic sets modeled:
− T: Probability of a threat − I: Indeterminacy (e.g., ambiguous network logs) − F: False alarm likelihood
Result:
25% higher detection rate for zero-day attacks compared to traditional systems.
Reduced false positives by incorporating indeterminacy metrics.
Application: Optimizing irrigation and pesticide use in drought-prone regions.
Process:
− Pythagorean Fuzzy logic analyzed soil moisture (μ) and crop stress (ν) data.
− Automated drones adjusted water/fertilizer distribution based on fuzzy rules.
Outcome:
35% reduction in water usage while maintaining crop yields.
Minimized environmental impact through targeted pesticide application.
The integration of fuzzy systems and computational intelligence has demonstrated transformative potential in addressing uncertainty and complexity across diverse domains. This paper systematically explored advanced fuzzy frameworks—such as T2FS, PFS, and Neutrosophic sets —alongside computational tools like GrC, Z-numbers, and D-numbers, highlighting their theoretical and practical significance. Key findings and implications are summarized below:
Theoretical Advancements: T2FS and PFS extend classical fuzzy logic by modeling higher-order uncertainties, enabling robust decision-making in dynamic environments (e.g., stock market prediction, e-commerce demand forecasting). Neutrosophic sets further address indeterminacy, proving critical in cybersecurity and crisis management.
Computational frameworks: GrC simplifies complex data hierarchies for context-aware systems (e.g., smart traffic control), while Z-numbers and D-numbers enhance reliability-based modeling in risk assessment.
Healthcare: Pythagorean fuzzy systems reduced diagnostic errors by 22% in cancer screening by quantifying uncertainty in medical imaging.
Energy: Hybrid Type-2 Fuzzy-GAs achieved 18% energy savings in smart grids by optimizing load distribution and renewable integration.
Supply chain: IFS improved supplier selection in the automotive industry by balancing cost, quality, and delivery reliability.
Interdisciplinary translation: Technical terms (e.g., "Complement" in fuzzy logic) require context-aware standardization to avoid misinterpretation in cross-cultural collaborations .
Integration gaps: Z-numbers and ML remain underexplored, limiting their adoption in real-time decision systems.
Ethical concerns: Balancing innovation with privacy (e.g., FPV drones in surveillance) and bias mitigation in AI-driven fuzzy models demand rigorous frameworks.
Standardization: Develop unified protocols for Z-numbers and hybrid models to enhance industrial adoption. Interpretability: Improve transparency in AI-fuzzy systems (e.g., FNN) to build trust in critical applications like healthcare.
Scalability: Expand applications to emerging fields such as precision agriculture and climate modeling, leveraging Pythagorean fuzzy systems for high-stakes decision-making.
Fuzzy systems and computational intelligence are pivotal in bridging the gap between theoretical models and real-world challenges. By addressing translational barriers, enhancing interdisciplinary collaboration, and prioritizing ethical considerations, these technologies can drive sustainable progress in smart infrastructure, healthcare, and global supply chains. Future research must focus on scalable implementations to tackle pressing global issues, ensuring that uncertainty modeling evolves from a niche field to a cornerstone of next-generation decision-support systems.
Mojtaba Nasehi is the sole author. The author read and approved the final manuscript.
The author declare no conflicts of interest.