A new methodology to improve interpretability in neuro-fuzzy TSK models

Velez, Miguel; Sánchez, O.; Romero, Sixto; Andújar, José Manuel

doi:10.1016/j.asoc.2009.08.027

Cited by 15 publications

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…60 So, the membership degree of the intersection point of neighboring fuzzy numbers should be less than or equal to 0.5, that is, 𝜇 Ã(𝑖𝑛𝑡𝑒𝑟𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝑝𝑜𝑖𝑛𝑡) ≤ 0.5. 61 Several researchers have proposed a value of 0.5 in modeling safety risk. 58,62 It should be noted that developing a fuzzy interval from a crisp interval may result in a linguistic label with meaningless values, or the sum of the membership degrees of neighboring fuzzy numbers for a point in the overlap may be greater than 1.0.…”

Section: Development Of Fuzzy Intervals For the Parameters Of The Req...mentioning

confidence: 99%

A risk‐based fuzzy arithmetic model to determine safety integrity levels considering individual and societal risks

Cheraghi,

Reniers,

Eslami Baladeh

et al. 2024

Quality & Reliability Eng

View full text Add to dashboard Cite

Risk‐based techniques such as risk graph and Layer of Protection Analysis (LOPA) are used to determine the Safety Integrity Level (SIL) of safety instrumented functions to ensure that risk is reduced to a tolerable level. However, these techniques have some drawbacks. For instance, they need absolute and precise numbers to evaluate SIL parameters, which are rarely available or are highly uncertain. In addition, they are incapable of considering individual and societal risks simultaneously. Moreover, risk tolerance criteria are likely to be used incorrectly in the LOPA technique, and risk graph is difficult to calibrate. In the current paper, a novel comprehensive fuzzy arithmetic model has been developed to determine the required SILs in process industries. The fuzzy required Risk Reduction Factor (RRF) is calculated for both individual and societal risks. Fuzzy numbers are developed from crisp intervals, based on the expected interval of the fuzzy numbers. Expert fuzzy‐scaled elicitation has been applied to obtain the SIL parameters. In the proposed model, the overall risk tolerance criterion and apportionment factor are defined as SIL parameters for both individual and societal risks to ensure that the applied risk criteria are compliant with the requirements of the system. In addition, an approach is introduced for determining the required SIL based on the fuzzy required RRF. The proposed methodology was demonstrated to alleviate the limitations, and thus, can be considered as a more precise alternative to the conventional methods.

show abstract

Section: Development Of Fuzzy Intervals For the Parameters Of The Req...mentioning

confidence: 99%

A risk‐based fuzzy arithmetic model to determine safety integrity levels considering individual and societal risks

Cheraghi,

Reniers,

Eslami Baladeh

et al. 2024

Quality & Reliability Eng

View full text Add to dashboard Cite

show abstract

“…The first factor, transparency, is related to interpretability of the fuzzy sets. The fuzzy sets should be interpretable as linguistic labels that are meaningful for experts in the problem under study [43]. In MGP-INTACTSKY, a specified number of fuzzy sets are assigned to each input variable that are interpretable as linguistic labels.…”

Section: Tsk Fuzzy Rule Based System: a New Structurementioning

confidence: 99%

MGP-INTACTSKY: Multitree Genetic Programming-based learning of INTerpretable and ACcurate TSK sYstems for dynamic portfolio trading

Mousavi

Esfahanipour

Zarandi

2015

Applied Soft Computing

View full text Add to dashboard Cite

“…La hibridación de estas dos técnicas se impulsó notablemente a principio de la década de los 90 con los trabajos de Jang (1991de Jang ( , 1992de Jang ( , 1993, Lin y Lee (1991), y Berenji y Khedkar (1992). Desde entonces los sistemas neuro-borrosos han permitido combinar la capacidad de adaptabilidad de las redes neuronales con la interpretabilidad y robustez intrínseca de los sistemas borroso (Pok y Xu, 1994, Vélez et al, 2010. Pero esta no es la única hibridación a la que se ha sometido a la lógica borrosa, también se ha integrado con algoritmos genéticos y bioinspirados (evolutionary computing), con técnicas probabilísticas (probabilistic computing), y con las herramientas tanto clásicas, como el consagrado PID (Ying et al, 1990, Zhao et al, 1993, Santos et al, 1996, Mudi y Pal, 1999, Tang et al, 2001), como modernas de la teoría de control (Albertos y Sala, 2004), convirtiéndose en una de las herramientas principales del denominado Control Inteligente 1 .…”

Section: Sciencedirectunclassified

Hibridación de sistemas borrosos para el modelado y control

Andújar

Barragán

2014

Revista Iberoamericana de Automática e Informática Industrial R

Self Cite

View full text Add to dashboard Cite

La lógica borrosa ha conseguido en un breve periodo de tiempo revolucionar la tecnología mediante la conjunción de los fundamentos matemáticos, la lógica y el razonamiento. Su inherente capacidad de hibridación y su robustez intrínseca han permitido a la lógica borrosa cosechar numerosos éxitos en el campo del modelado y el control de sistemas, impulsando el control inteligente. En este artículo se estudian los sistemas borrosos híbridos más usuales y su importancia en el campo del modelado y control de sistemas dinámicos. El trabajo presenta varios ejemplos que ilustran, para diferentes técnicas de hibridación, cómo éstas potencian las cualidades innatas de la lógica borrosa para el modelado y control de sistemas dinámicos. Así mismo, se incluyen más de ciento cincuenta referencias bibliográficas que permitirán al lector interesado profundizar en el campo de la lógica borrosa, y más concretamente en el de sus técnicas de hibridación con aplicación al modelado y control borroso.

show abstract

A new methodology to improve interpretability in neuro-fuzzy TSK models

Cited by 15 publications

References 15 publications

A risk‐based fuzzy arithmetic model to determine safety integrity levels considering individual and societal risks

A risk‐based fuzzy arithmetic model to determine safety integrity levels considering individual and societal risks

MGP-INTACTSKY: Multitree Genetic Programming-based learning of INTerpretable and ACcurate TSK sYstems for dynamic portfolio trading

Hibridación de sistemas borrosos para el modelado y control

Contact Info

Product

Resources

About