Solution Highlights
- Implemented an AI-powered solution for equipment failure prediction
- Achieved over 90% accuracy in predicting system failures ahead of time
- Helped the customer reduce maintenance costs by 5%
- Optimized the overall performance in the production line on the factory floor
About the Project
Customer:
Our customer (under NDA) is a global manufacturer of home appliances. The company is strictly committed to its quality and safety standards, so it monitors every step in the production process. By recording sensor measurements in the production line, the manufacturer aims to use this data for process optimization.
Challenges & Project Goals:
The manufacturer faced the challenge of repeated and unexpected equipment failures in the production line. These breakdowns brought high costs for the company spent on new equipment components and materials for repair or rewind, labor costs, and downtime. Inconsistent work in the production line also caused delays in shipping and affected overall plant productivity.
Solution:
To help the customer improve production line performance, we created an AI-powered solution for equipment failure prediction. The machine learning model predicted system failures ahead of time and, thus, reduced breakdowns to a minimum acceptable level by performing predictive maintenance.
Business Value Delivered:
- The ML model predicted equipment failures with over 90% of accuracy.
- By far, the AI solution helped the manufacturer reduce maintenance costs by 5% and significantly improve performance in the production line.
- The manufacturer became able to predict which parts of the equipment are most likely to fail on the factory floor and, thus, maintain or replace those parts just in time and optimize the production line performance.
Technology Solution
Together with the customer, we built an AI-powered solution to monitor the equipment state and predict equipment failures ahead of time. Here is how the project progressed:
- The Intelliarts team started with processing and analyzing the huge datasets of historical and real-time measurements collected from IoT/IIoT sensors installed in the manufacturer’s facility. We completed this by building the scalable data processing unit on top of Dask and deploying it to the AWS Fargate cluster, which allowed us to process big data efficiently.
- To increase the predictive power of ML algorithms used in the solution and improve the model performance, our team of engineers encoded categorical data using one-hot encoding and clustered samples by production flow using k-Means. We also reduced the number of numerical features that were highly linearly dependent and calculated lag features as one of the most important feature types for the solution.
- Our next step was to achieve the best results in modeling. We experimented with different algorithms and proofs of concepts and finally chose the Extreme Gradient Boosting classifier. The latter showed the best score in modeling, especially after we increased its prediction accuracy with model tuning.
- After deploying the prediction model, the Intelliarts team defined policies and helped the management of the company to teach the personnel how to use the AI-powered solution in their day-to-day work.
Business Outcomes
The customer was positively impressed with the project results. Regarding the value that the company gained:
- The ML model showed over 90% accuracy in equipment failure prediction. In the long run, this helped the manufacturer reduce maintenance costs by 5%.
- Since the company could predict which parts of the equipment were about to fail, it became able to optimize the production line performance as well as save time and money resources.
- Modeling the data would help the customer predict better quality defects in equipment parts. So, the manufacturer could follow its quality standards more efficiently and prevent poor-quality components to be used in the final product.
- As it was a full-cycle data science project, Intelliarts helped the manufacturer build a production line monitoring to provide an extra benefit to the customer. When the company sees any change in data, the manufacturer would know it immediately and be able to retrain the solution or at least look for where the solution failed.
