Explainable Artificial Intelligence for Predictive Maintenance Applications

Matzka, Stephan

doi:10.1109/ai4i49448.2020.00023

Cited by 69 publications

(40 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Our approach is evaluated using two data sets originating from PHM settings: the FordA data set obtained from (Bagnall, 2022) and the AI4I predictive maintenance dataset (Matzka, 2020) obtained from (Dua & Graff, 2017) . The FordA dataset consists of 3601 train and 1320 test instances, each consisting of 500 sensor features.…”

Section: Methodsmentioning

confidence: 99%

Certainty Groups: A Practical Approach to Distinguish Confidence Levels in Neural Networks

Lodes

Schiendorfer

2022

PHME_CONF

View full text Add to dashboard Cite

Machine Learning (ML), in particular classification with deep neural nets, can be applied to a variety of industrial tasks. It can augment established methods for controlling manufacturing processes such as statistical process control (SPC) to detect non-obvious patterns in high-dimensional input data. However, due to the widespread issue of model miscalibration in neural networks, there is a need for estimating the predictive uncertainty of these models. Many established approaches for uncertainty estimation output scores that are difficult to put into actionable insight. We therefore introduce the concept of certainty groups which distinguish the predictions of a neural network into the normal group and the certainty group. The certainty group contains only predictions with a very high accuracy that can be set up to 100%. We present an approach to compute these certainty groups and demonstrate our approach on two datasets from a PHM setting.

show abstract

Section: Methodsmentioning

confidence: 99%

Certainty Groups: A Practical Approach to Distinguish Confidence Levels in Neural Networks

Lodes

Schiendorfer

2022

PHME_CONF

View full text Add to dashboard Cite

show abstract

“…Applications of EAI include the forecast of climate-change consequences [2] and learning the mental-health impact of COVID-19 in the United States [3]. Matzka [4] presented research performed regarding explainable artificial intelligence applied for predictive maintenance applications on a specific recent dataset available on the UCI data repository [1]. The principal contribution consisted of the design of an explainable model and an explanatory interface.…”

Section: Mathematical Modeling Of Data Quality Assessmentmentioning

confidence: 99%

“…In this section, we briefly present the description of the UCI dataset that we used in the experimental data-quality-assessment evaluation. Reference [4] presents the dataset's details. Table 5 presents a snapshot of the data with included three cases (numbered with 1, 2 and 78).…”

Section: The Synthetic Dataset Used In the Evaluationmentioning

confidence: 99%

Method for Data Quality Assessment of Synthetic Industrial Data

Iantovics

Enăchescu

2022

Sensors

View full text Add to dashboard Cite

Sometimes it is difficult, or even impossible, to acquire real data from sensors and machines that must be used in research. Such examples are the modern industrial platforms that frequently are reticent to share data. In such situations, the only option is to work with synthetic data obtained by simulation. Regarding simulated data, a limitation could consist in the fact that the data are not appropriate for research, based on poor quality or limited quantity. In such cases, the design of algorithms that are tested on that data does not give credible results. For avoiding such situations, we consider that mathematically grounded data-quality assessments should be designed according to the specific type of problem that must be solved. In this paper, we approach a multivariate type of prediction whose results finally can be used for binary classification. We propose the use of a mathematically grounded data-quality assessment, which includes, among other things, the analysis of predictive power of independent variables used for prediction. We present the assumptions that should be passed by the synthetic data. Different threshold values are established by a human assessor. In the case of research data, if all the assumptions pass, then we can consider that the data are appropriate for research and can be applied by even using other methods for solving the same type of problem. The applied method finally delivers a classification table on which can be applied any indicators of performed classification quality, such as sensitivity, specificity, accuracy, F1 score, area under curve (AUC), receiver operating characteristics (ROC), true skill statistics (TSS) and Kappa coefficient. These indicators’ values offer the possibility of comparison of the results obtained by applying the considered method with results of any other method applied for solving the same type of problem. For evaluation and validation purposes, we performed an experimental case study on a novel synthetic dataset provided by the well-known UCI data repository.

show abstract

“…This validation is different from the validation in machine learning for determining the optimal model. Since authentic datasets on faulty conditions are seldom available, Matzka created a synthetic dataset based on actual predictive maintenance conditions in the industry (Matzka, 2020b). The dataset comprises 10,000 data points with six features.…”

Section: Validation Of the Hus‐ml Framework Using A Benchmark Datasetmentioning

confidence: 99%

Equipment activity recognition and early fault detection in automated construction through a hybrid machine learning framework

Harichandran

Raphael

Mukherjee

2022

Computer aided Civil Eng

View full text Add to dashboard Cite

Existing studies on automated construction equipment monitoring have focused mainly on activity recognition rather than fault detection. This paper proposes a novel equipment activity recognition and fault detection framework called hybrid unsupervised and supervised machine learning (HUS‐ML). HUS‐ML first identifies normal operations and known faulty conditions through supervised learning. Then, an anomaly detection algorithm is applied to spot any unseen faulty conditions. The framework is tested using acceleration measurements from a low‐rise automated construction system prototype. HUS‐ML outperformed the conventional machine learning approach in activity recognition and fault detection with an average F1 score of 86.6%. The conventional approach failed to detect unseen faulty operations. HUS‐ML identified known faulty operations and unseen faulty operations with F1 scores of 98.11% and 76.19%, respectively. The generalizability of the framework is demonstrated by validating it on an independent benchmark dataset with good results.

show abstract

Explainable Artificial Intelligence for Predictive Maintenance Applications

Cited by 69 publications

References 3 publications

Certainty Groups: A Practical Approach to Distinguish Confidence Levels in Neural Networks

Certainty Groups: A Practical Approach to Distinguish Confidence Levels in Neural Networks

Method for Data Quality Assessment of Synthetic Industrial Data

Equipment activity recognition and early fault detection in automated construction through a hybrid machine learning framework

Contact Info

Product

Resources

About