Dr Ntokozo Mthembu, Advisor to the ODI Board, writes:
“In today’s rapidly evolving industrial landscape, manufacturers looking to maintain their competitiveness must prioritise achieving operational excellence. Key performance indicators (KPIs) like Overall Equipment Effectiveness (OEE) are crucial in assessing and improving manufacturing efficiency. However, with the integration of artificial intelligence (AI) technologies, traditional methods of calculating OEE are evolving to provide more accurate insights and promote higher levels of optimisation. This article delves into the calculation of AI-inspired Overall Equipment Effectiveness and its significance for modern factories.
Understanding Traditional OEE Calculation
Before exploring AI-enhanced approaches, it is essential to grasp the fundamentals of traditional OEE calculation[1].
OEE is a metric that measures the effectiveness of manufacturing processes by three critical factors[2].
Availability: The percentage of time that equipment is available for production. It takes into consideration both scheduled and unforeseen downtime including maintenance, changeovers, and breakdowns.
Performance: The ratio of actual production speed to the ideal or maximum achievable speed. Performance losses may result from equipment slowdowns, minor stoppages, or suboptimal operation.
Quality: The proportion of good-quality products produced relative to the total output. Quality losses encompass defects, rework, and scrap.
The formula for calculating OEE is:
ππΈπΈ=π΄π£πππππππππ‘π¦ΓπππππππππππΓππ’ππππ‘π¦OEE=AvailabilityΓPerformanceΓQuality
Traditionally, this calculation provides valuable insights into overall manufacturing efficiency, helping identify areas for improvement and optimization.
Advancements with AI in OEE Calculation
The advent of AI has revolutionised how OEE is calculated and utilised within manufacturing environments. AI-driven approaches offer several enhancements over traditional methods, including:
Predictive Maintenance: AI algorithms can analyse equipment sensor data in real-time to predict potential failures before they occur[3]. By proactively addressing maintenance needs, factories can minimise downtime and maximise equipment availability[4].
Dynamic Performance Analysis: Unlike static performance assessments in traditional OEE calculations, AI enables dynamic analysis of production speeds and identifies optimal operating conditions in real-time[5]. This capability allows for immediate adjustments to maximise performance efficiency.
Quality Assurance and Defect Detection: AI-powered vision systems can inspect products with unparalleled accuracy, detecting defects and anomalies in real-time[6]. By integrating quality assessment directly into the manufacturing process, factories can reduce waste and improve overall product quality.
Data Integration and Analytics: AI facilitates the integration of diverse data sources, including equipment telemetry, production logs, and external factors like weather or market demand. Advanced analytics algorithms can then uncover hidden patterns and insights, enabling data-driven decision-making for optimizing OEE[7].
Calculating AI-Inspired OEE
To calculate AI-inspired OEE, factories integrate AI technologies into each component of the traditional OEE formula:
AI-Enabled Availability: Utilising predictive maintenance algorithms, factories can accurately predict equipment downtime and schedule maintenance activities during optimal production windows[8]. This proactive approach minimises unplanned downtime, maximising equipment availability.
Dynamic Performance Optimisation: AI algorithms continuously monitor equipment performance metrics and production conditions, dynamically adjusting operating parameters to optimise performance in real-time[9]. By leveraging machine learning techniques, factories can identify patterns and trends to enhance production efficiency continuously.
AI-powered Quality Assurance: Integrating AI-based vision systems into the production line enables real-time inspection and quality assessment of products[10]. Any deviations or flaws are found right away, enabling quick remedial action to preserve high standards and reduce quality losses. By incorporating AI advancements into each component of the OEE calculation, factories can achieve higher levels of operational excellence, driving greater efficiency, productivity, and competitiveness in today’s dynamic manufacturing landscape.
AI algorithms and techniques that are frequently applied in OEE calculations and related areas
There isn’t a single “best practice” algorithm for calculating Overall Equipment Effectiveness (OEE) using artificial intelligence (AI). Instead, the choice of algorithm depends on various factors such as
- the type of manufacturing process,
- the availability of data,
- computational resources,
- and the specific goals of the organisation.
However, I can outline some common AI algorithms and techniques that are frequently applied in OEE calculations and related areas.
Predictive Maintenance Algorithms: These algorithms utilise machine learning techniques, such as regression analysis, time series forecasting, and anomaly detection, to predict equipment failures before they occur[11]. By analysing historical data on equipment performance, maintenance logs, and sensor readings, predictive maintenance models can estimate the remaining useful life of machinery and schedule maintenance activities to prevent downtime.
Dynamic Optimization Algorithms: AI-driven optimisation algorithms, including genetic algorithms, simulated annealing, and reinforcement learning, can dynamically adjust production parameters to maximise OEE in real-time[12]. These algorithms continuously analyse production data, such as throughput rates, cycle times, and quality metrics, to identify optimal operating conditions and adapt production schedules accordingly.
Machine Learning for Quality Control: Machine learning algorithms, such as convolutional neural networks (CNNs) and support vector machines (SVMs), are widely used for automated quality inspection and defect detection in manufacturing processes[13]. By analysing images or sensor data from production lines, these algorithms can identify defects, anomalies, and quality issues in real-time, allowing for immediate corrective actions to maintain high-quality standards and minimise quality losses.
Anomaly Detection Algorithms: Anomaly detection techniques, such as Isolation Forest, One-Class SVM, and Autoencoders, can identify unusual patterns or deviations in production data that may indicate equipment malfunctions, process inefficiencies, or quality issues[14]. By flagging anomalous events in real-time, these algorithms enable proactive interventions to prevent downtime and quality defects, thereby improving OEE.
Time Series Analysis: Time series analysis methods, including autoregressive integrated moving average (ARIMA), exponential smoothing, and Fourier analysis, are used to model and forecast production metrics over time[15]. By analysing historical OEE data and identifying underlying trends, seasonal patterns, and cyclic variations, time series models can provide insights into long-term performance trends and inform strategic decision-making for optimising OEE.
Big Data Analytics: Big data analytics platforms, such as Apache Hadoop and Spark, enable the processing and analysis of large volumes of manufacturing data from diverse sources, including equipment sensors, production logs, and enterprise systems[16]. By leveraging distributed computing and parallel processing capabilities, these platforms can uncover hidden insights, correlations, and causal relationships in manufacturing data, thus facilitating data-driven decision-making for improving OEE.
It’s important to note that the effectiveness of AI algorithms in OEE calculations depends on the quality and availability of data, the complexity of the manufacturing process, and the specific objectives of the organisation. Therefore, organisations should carefully evaluate and customise AI algorithms based on their unique requirements and constraints to achieve optimal results in optimising Overall Equipment Effectiveness.
Conclusion
As factories embrace digital transformation and Industry 4.0 initiatives, the calculation of Overall Equipment Effectiveness is evolving with the integration of AI technologies. AI-inspired OEE offers a more nuanced and dynamic approach to assessing manufacturing efficiency, enabling proactive maintenance, dynamic performance optimiSation, and real-time quality assurance. By harnessing the power of AI, factories can unlock new levels of operational excellence, driving continuous improvement and sustainable growth in the modern industrial era.”
To read Ntokozo’s article about Measuring Machines & Equipment Effectiveness, click here.
RESOURCES:
[1] https://www.oee.com/
[2] https://www.oee.com/calculating-oee/
[3] https://www.ppmashow.co.uk/press-release/ai-in-manufacturing
[4] https://tractian.com/en
[5] https://www.aveva.com/en/solutions/digital-transformation/artificial-intelligence/
[6] https://neurosys.com/blog/ai-defect-detection-in-manufacturing#
[7] https://www.linkedin.com/pulse/revolutionizing-manufacturing-unleashing-power-advanced- analytics-yevqc/
[8] https://www.myaifrontdesk.com/blog/24-7-availability
[9] https://www.plantengineering.com/articles/ai-driven-solutions-for-enhanced-plant-automation-productivity/
[10] https://www.linkedin.com/pulse/industrial-automation-redefined-leveraging-ai-vision-systems/
[11] https://www.linkedin.com/pulse/machine-learning-algorithms-predictive-maintenance-hope-edet-2tuje/
[12] https://arxiv.org/pdf/1901.02256
[13] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6512995/
[14] https://medium.com/@venujkvenk/anomaly-detection-techniques
[15] https://www.datamation.com/big-data/what-is-time-series-analysis/
[16] https://www.intellectyx.com/apache-hadoop-big-data-analytics-for-manufacturing/