Fig 2 - uploaded by Bahrudin Hrnjica
Content may be subject to copyright.
Source publication
This paper describes an example of an explainable AI (Artificial Intelligence) (XAI) in a form of Predictive Maintenance (PdM) scenario for manufacturing. Predictive maintenance has the potential of saving a lot of money by reducing and predicting machine breakdown. In this case study we work with generalized data to show how this scenario could lo...
Contexts in source publication
Context 1
... data represents the most important information in every PdM system. The errors are non-breaking recorded events while the machine is still operational. In the experimental data set the error date and times are rounded to the closest hour since the telemetry data is collected at an hourly rate. What we get to insight is shown in the left chart of Fig. 2. ...
Context 2
... data represents the replacements of the components due to the failure of the machines. Once the failure is happened the machine is stopped. This is a crucial difference between errors and failures. Failure distribution produced by certain component across machines is shown in the right chart of Fig. ...
Similar publications
The work aims to study the processes of production management at the multiproduct food enterprises. To improve management effectiveness, application of the created information technology to support the process of adjusting the product assortment was proposed. This is achieved by reducing the total product cost price and obtaining an additional prof...
Citations
... [ [61][62][63][64][65][66] 10. ...
... Concerns about the reliability and consistent maintenance of AI systems highlight the importance of infrastructural integrity in AI's educational utility [61][62][63][64][65][66]. Ensuring the dependability of these systems is non-negotiable, given that technological failure could lead to substantial disruptions in the learning process. ...
Despite the wave of enthusiasm for the role of Artificial Intelligence (AI) in reshaping education, critical voices urge a more tempered approach. This study investigates the less-discussed 'shadows' of AI implementation in educational settings, focusing on potential negatives that may accompany its integration. Through a multi-phased exploration consisting of content analysis and survey research, the study develops and validates a theoretical model that pinpoints several areas of concern. The initial phase, a systematic literature review, yielded 56 relevant studies from which the model was crafted. The subsequent survey with 260 participants from a Saudi Arabian university aimed to validate the model. Findings confirm concerns about human connection, data privacy and security, algorithmic bias, transparency, critical thinking, access equity, ethical issues, teacher development, reliability, and the consequences of AI-generated content. They also highlight correlations between various AI-associated concerns, suggesting intertwined consequences rather than isolated issues. For instance, enhancements in AI transparency could simultaneously support teacher professional development and foster better student outcomes. Furthermore, the study acknowledges the transformative potential of AI but cautions against its unexamined adoption in education. It advocates for comprehensive strategies to maintain human connections, ensure data privacy and security, mitigate biases, enhance system transparency, foster creativity, reduce access disparities, emphasize ethics, prepare teachers, ensure system reliability, and regulate AI-generated content. Such strategies underscore the need for holistic policymaking to leverage AI's benefits while safeguarding against its disadvantages.
... Predictive Maintenance (PM) is one way to reduce costs and downtimes by planning maintenance work based on an asset's actual condition rather than relying on fixed time-based maintenance cycles [2]. Multiple industries like aviation [3,4], manufacturing [5,6], and chemistry [7,8] vastly apply PM into their operational routine. Also, other industries like construction and facility management can benefit from improved maintenance methodologies. ...
Introduction: Due to the lack of labeled data, applying predictive maintenance algorithms for facility management is cumbersome. Most companies are unwilling to share data or do not have time for annotation. In addition, most available facility management data are text data. Thus, there is a need for an unsupervised predictive maintenance algorithm that is capable of handling textual data. Methodology: This paper proposes applying association rule mining on maintenance requests to identify upcoming needs in facility management. By coupling temporal association rule mining with the concept of semantic similarity derived from large language models, the proposed methodology can discover meaningful knowledge in the form of rules suitable for decision-making. Results: Relying on the large German language models works best for the presented case study. Introducing a temporal lift filter allows for reducing the created rules to the most important ones. Conclusions: Only a few maintenance requests are sufficient to mine association rules that show links between different infrastructural failures. Due to the unsupervised manner of the proposed algorithm, domain experts need to evaluate the relevance of the specific rules. Nevertheless, the algorithm enables companies to efficiently utilize their data stored in databases to create interpretable rules supporting decision-making.
... decision trees, or post-hoc techniques like text or visual explanations [5]. For instance in predictive maintenance, XAI can enhance machine downtime prediction [41]. ...
Artificial Intelligence (AI) has profound economic influence in manufacturing, but its unmindful integration can also pose societal and environmental risks. This paper provides a quantified overview of manufacturing areas that are highly advanced in AI capability research, such as maintenance. Integrating Responsible AI (RAI) in further studies of those areas is essential to mitigate risks and deliver business benefits. To enable this, manufacturing specific RAI dimensions are defined to represent accountability, explainability, fairness, human-centricity, sustainability (Green AI) and privacy & security. Further, a qualitative RAI framework consisting of responsibility areas (human involvement, decision making, business focus, system design) is proposed. Practical considerations to align the framework with manufacturing requirements are made by discussing it within an AI systems lifecycle.
... In recent years, Explainable Artificial Intelligence (XAI) has gained popularity as a manufacturing industry application. A predictive maintenance case study was carried out to show how XAI may aid in predicting maintenance activities, which used a very effective gradient boosting decision tree model to forecast tool and machine failures [14]. Similarly, STARdom's planned architecture combines XAI, active learning, and simulated reality to improve forecasts and offer guidance on making decisions for human-centered industrial systems [23]. ...
Production shifted from a product-centered perspective (mass production of one article) to a customer-centered perspective (mass customization of product variants). It also happens in energy-intensive industries, such as steel production. Mass customization companies face a challenge in accurately estimating the total costs of an individual product. Furthermore, 20% to 40% of the costs related to steel products come from energy. Increasing the product variety can cause an inevitable loss of sustainability. This paper presents machine-learning approaches to improve the sustainability of the steel production industry. It is done by finding the most accurate way to predict the power consumption and the costs of customized products. Moreover, this research also finds the most energy-efficient machine mix based on the predictions. The method is validated in a steel manufacturing Small Medium Enterprise (SME). In this research, experiments were conducted with different machine learning models, and it was found that the most accurate results were achieved using regularization-based and random forest regression models. Explainable AI (XAI) is also used to clarify how product properties influence process costs and power consumption. This paper also discusses scenarios on how the prediction of costs and power consumption can assist production planners in performing workstation selection. This research improves the production planning of customized products by providing a trustable decision support system for machine selection based on explainable machine learning models for process time and power consumption predictions.KeywordsMachine learningsustainable manufacturingmass customizationdata-driven manufacturingpower consumption predictionExplainable Artificial Intelligence (XAI)
... Despite the breadth of these challenges, which are more concerned at the enterprise level, significant research and developments are needed for application-specific sub-systems at the edge level [5]. Namely, there are current concerns in the wider academic and industrial communities, in using data-driven models that are inherently black box in nature, especially when making high-stake decisions [6]. A large number of current Industry 4.0 prototypes use black-box models such as artificial neural networks (ANNs), due to their power, scalability, and adaptability, which has resulted in great interoperability at multiple levels. ...
In the context of smart manufacturing with costly process trials (e.g., those in the aerospace sector), access to data-efficient, transparent, and diagnostic-capable inverse AI models is a key. This is because in such settings, the concurrent presence of numerous tool/material/process decision factors along with their potential uncertainties can rapidly yield the final product failure; hence, designers need to proactively predict and mitigate the underlying root causes. Towards this technological motivation, in this work, different Bayesian learning frameworks with inverse modelling capabilities have been developed and tested for a composites autoclave curing process, while assuming different training dataset size scenarios. We specifically present a new fully connected Bayesian Belief Network (BBN) architecture integrated with an expert-located proxy thermocouple within the manufacturing setup, to learn the thermal behavior of the final composite laminate rapidly. The results revealed that the integration of such expert knowledge under the BBN model can offer favorable predictive performance. It is shown that the model even with limited (e.g., 0.03% of the full set) but highly customized auxiliary data still yields useful and interpretable inverse decisions (with a maximum state likelihood of 0.9) for the process failure mitigation, thanks to the model’s apparent transparency based on the BBNs’ directed graph nature.
... It needs to be mentioned that due to the current inability of machines to explain their decisions and interaction with humans, the effectiveness of the mentioned systems is restricted. Applying explainable AI and ML help physicians and medical staff to understand, appropriately trust, and effectively interact with these technologies [10]. ...
... On the other hand, in the medical field as well applying explainable AI and ML are important to help physicians and medical staff to understand, appropriately trust, and effectively interact with such methods [10]. We will describe the XAI tools and techniques in Section 3.1.1 to 3.1.11 ...
This paper investigates the applications of explainable AI (XAI) in healthcare, which aims to provide transparency, fairness, accuracy, generality, and comprehensibility to the results obtained from AI and ML algorithms in decision-making systems. The black box nature of AI and ML systems has remained a challenge in healthcare, and interpretable AI and ML techniques can potentially address this issue. Here we critically review previous studies related to the interpretability of ML and AI methods in medical systems. Descriptions of various types of XAI methods such as layer-wise relevance propagation (LRP), Uniform Manifold Approximation and Projection (UMAP), Local Interpretable Model-agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), ANCHOR, contextual importance and utility (CIU), Training calibration-based explainers (TraCE), Gradient-weighted Class Activation Mapping (Grad-CAM), t-distributed Stochastic Neighbor Embedding (t-SNE), NeuroXAI, Explainable Cumulative Fuzzy Class Membership Criterion (X-CFCMC) along with the diseases which can be explained through these methods are provided throughout the paper. The paper also discusses how AI and ML technologies can transform healthcare services. The usability and reliability of the presented methods are summarized, including studies on the usability and reliability of XGBoost for mediastinal cysts and tumors, a 3D brain tumor segmentation network, and the TraCE method for medical image analysis. Overall, this paper aims to contribute to the growing field of XAI in healthcare and provide insights for researchers, practitioners, and decision-makers in the healthcare industry. Finally, we discuss the performance of XAI methods applied in medical health care systems. It is also needed to mention that a brief implemented method is provided in the methodology section.
... There are several approaches to maintenance and asset management, including running equipment until failure (reactive maintenance), scheduled or preventive maintenance, and predictive maintenance [12,13,14]. Predictive maintenance anticipates equipment failure or maintenance needs ahead of time in order to avoid unplanned downtime [15]. Predictive maintenance solutions typically use historical and current data gathered from a given asset to predict and estimate future maintenance needs [1]. ...
div class="section abstract"> With the developments of Industry 4.0, data analytics solutions and their applications have become more prevalent in the manufacturing industry. Currently, the typical software architecture supporting these solutions is modular, using separate software for data collection, storage, analytics, and visualization. The integration and maintenance of such a solution requires the expertise of an information technology team, making implementation more challenging for small manufacturing enterprises. To allow small manufacturing enterprises to feasibly obtain the benefits of Industry 4.0 data analytics, a full-stack data analytics framework is presented, and its performance evaluated as applied in the common industrial analytics scenario of predictive maintenance. The predictive maintenance approach was achieved by using a full-stack data analytics framework comprised of the PTC Inc. Thingworx software suite. When deployed on a lab-scale factory, there was a significant increase in factory uptime in comparison with both preventive and reactive maintenance approaches. The predictive maintenance approach simultaneously eliminated unexpected breakdowns and extended the uptime periods of the lab-scale factory. This research concluded that similar or better results may be obtained in actual factory settings, since the only source of error on predictions in the testing scenario would not be present in real world scenarios. An analysis of the effect of downtime period durations and discussion on the cost of reactive maintenance and associated breakdowns is also presented.
</div
... In light of these central challenges, the field of explainable artificial intelligence (XAI) has recently emerged as a research area exploring various approaches. Explainable and trustworthy data-driven methodology can improve decision-making for fault diagnosis as well as the broader field of prognostics and health management (PHM), enabling cost-saving intelligent predictive maintenance and effective resource allocation strategies (Hrnjica and Softic 2020). ...
... XAI methods vary by aspects such as explanation scope, model specificity, and the location of the explanations (Ahmed et al. 2022). In recent years, XAI methods have been developed for several applications including manufacturing cost estimation (Yoo and Kang 2021), predictive maintenance (Hrnjica and Softic 2020;Serradilla et al. 2021), privacy-preserving industrial applications (Ogrezeanu et al. 2022), semiconductor defect classification , and quality management in semiconductor manufacturing (Senoner et al. 2021). Most of these applications attempt to quantify feature attributions for predictions and global feature importance, which historically have been model-specific and derived from nonparametric, decision tree-based modeling (Ahmed et al. 2022). ...
Data-driven artificial intelligence models require explainability in intelligent manufacturing to streamline adoption and trust in modern industry. However, recently developed explainable artificial intelligence (XAI) techniques that estimate feature contributions on a model-agnostic level such as SHapley Additive exPlanations (SHAP) have not yet been evaluated for semi-supervised fault diagnosis and prognosis problems characterized by class imbalance and weakly labeled datasets. This paper explores the potential of utilizing Shapley values for a new clustering framework compatible with semi-supervised learning problems, loosening the strict supervision requirement of current XAI techniques. This broad methodology is validated on two case studies: a heatmap image dataset obtained from a semiconductor manufacturing process featuring class imbalance, and a benchmark dataset utilized in the 2021 Prognostics and Health Management (PHM) Data Challenge. Semi-supervised clustering based on Shapley values significantly improves upon clustering quality compared to the fully unsupervised case, deriving information-dense and meaningful clusters that relate to underlying fault diagnosis model predictions. These clusters can also be characterized by high-precision decision rules in terms of original feature values, as demonstrated in the second case study. The rules, limited to 1-2 terms utilizing original feature scales, describe 12 out of the 16 derived equipment failure clusters with precision exceeding 0.85, showcasing the promising utility of the explainable clustering framework for intelligent manufacturing applications.
... Furthermore, the 5G and XAI-enabled manufacturing systems can perform focused analytics using a reliable AI-integrated prediction model for maintenance tasks. It can develop a flexible production environment for future product trends and customer needs [99]. b. ...
The emergence of Explainable Artificial Intelligence (XAI) has enhanced the lives of humans and envisioned the concept of smart cities using informed actions, enhanced user interpretations and explanations, and firm decision-making processes. The XAI systems can unbox the potential of black-box AI models and describe them explicitly. The study comprehensively surveys the current and future developments in XAI technologies for smart cities. It also highlights the societal, industrial, and technological trends that initiate the drive towards XAI for smart cities. It presents the key to enabling XAI technologies for smart cities in detail. The paper also discusses the concept of XAI for smart cities, various XAI technology use cases, challenges, applications, possible alternative solutions, and current and future research enhancements. Research projects and activities, including standardization efforts toward developing XAI for smart cities, are outlined in detail. The lessons learned from state-of-the-art research are summarized, and various technical challenges are discussed to shed new light on future research possibilities. The presented study on XAI for smart cities is a first-of-its-kind, rigorous, and detailed study to assist future researchers in implementing XAI-driven systems, architectures, and applications for smart cities.
... It is obvious that XAI has a beneficial impact on ML security. Examples show that it is successfully adopted as predictive maintenance to predict machine errors or tool failures in manufacturing [88], and as fault-detection in real chiller systems [89]. Because the reasons behind how ML model acts are able to be explained by XAI, it also benefits the governance and compliance aspect in SecMLOps. ...
