Approachable AI Applied: Quality Assurance
Machine data through sensors and connected devices are generating a wealth of information that can be converted into value-added knowledge for your organization. Traditional approaches to data can allow organizations to monitor for defined conditions but these approaches are naturally backward looking. The introduction of machine learning to your data can enable predictive analytics and forward-looking analysis to further unlock insights.
Traditionally, it has been difficult to implement and scale the usage of machine learning primarily due to technical complexities and a shortage of AI talent across industries.
This series aims to feature use cases showing how Elipsa enables Approachable AI to allow an organization's existing talent to become AI talent and apply advanced analytics without needing to write code.
As referenced in our blog post on Defect Detection, the American Society for Quality (ASQ) suggests that the Cost of Quality is usually around 15–20% of sales, often as high as 40% in some organizations. The cost of poor quality includes both internal and external failure costs. In other words, organizations are spending a considerable amount of money correcting defects found both on the factory floor and in the customer’s hands. This not only leads to a direct monetary loss but also indirectly in the form of reputational risk.
Multiple industries from manufacturing, to mining, to 3-D printing, suffer from quality control issues. Often times processes consist of multiple time-consuming steps. Companies look to optimize their processes but often times the final result is still of insufficient quality, leading to increased waste, increased cost, and decreased revenue.
With predictive analytics, an organization could utilize the information at various stages of the production process such as material attributes and machine settings to predict the final outcome quality. This prediction would enable companies to dial in their machines to help ensure a quality result every time.
Currently, organizations try to minimize costs by following specific processes and set points. However, these set points are not always optimal. Machine learning enables organizations to find the optimal set points for each run.
Data scientists can build predictive models to fit your needs, but those resources are expensive and hard to come by.
Certain vendors provide the ability to monitor quality with AI but they require an infrastructure overhaul.
Elipsa’s goal with Approachable AI is to allow users to build custom quality prediction models with their existing software/data and without the need for a data scientist.
Problem: Quality Assurance
In our example, we explore a data set from a mining process with the main goal being to predict how much impurity is in the final ore concentrate. In this particular example, we would look to predict the level of impurities every hour based on the current % of contaminants in the material and the current machine settings in order to give engineers advanced notice as to whether they need to make adjustments to help minimize the amount of contaminants in the end result.
The resulting lower level of contaminants would help the environment while also increasing the value of the final product as the price of the product is directly dependent on the contaminant levels.
The dataset consists of 501,132 data points, or point-in-time snapshots, containing the percentage of contaminants of the input material, the current value of 19 sensors and machine settings, and the amount of contaminants in the output to use for prediction.
Elipsa’s Approachable AI Applied
From this information, you could utilize Elipsa to predict the value of contaminants in the output. However, in our example use case, we define a threshold of the level of contamination that is acceptable. Instead of predicting the % of contaminants, we are predicting whether the final contaminant amount will be over our defined threshold.
Because we are looking to build a model to detect whether a value will be over a threshold, we are going to choose to predict an event (over the threshold or not) vs predicting the value itself.
In our case, we have defined the acceptable threshold of silica in the final product as less than 3%. Our dataset contains a column, named ‘Poor’, indicating whether the silica % is over 3 (a value of 1 in the column) or not (a value of 0).
Once the dataset is upload, we simply need to select which column we are looking to predict. In our case, it is the column indicating whether the final result is poor quality or not.
From there, select which columns you would like to as potential predictors and build the model. The Elipsa AI engine runs through a series of models and data science experiments, finding the model that best matches your data with the most accurate predictive results.
Model creation is performed in the background and the user is alerted upon completion. The result is a series of statistics detailing how well your model performed on new data. This information will allow users to assess whether the results are good enough to deploy this model to production.
In our example, the model was able to correctly predict whether the final result would be of poor quality 95.23% of the time. Engineers could use predictions to optimize the settings on their machines, ensuring quality results and effectively eliminating the production of poor quality materials.
With high accuracy received, Elipsa users can easily deploy this model to the cloud or to their own edge devices with a push of a button. Once deployed, user’s can connect the cloud or edge to their existing applications via Elipsa APIs in order to make predictions on future steaming data.
We were able to build a highly accurate model to predict whether an end product will be below an acceptable quality threshold. In addition, we were able to deploy this model to production without changes to existing infrastructure.
The use of AI and predictive analytics for quality assurance would allow an organization to reduce the waste and costs associated with poor quality. In our example, with price tied to the level of contaminants, it also enables the organization to charge more for the end product and increase revenue.
With Approachable AI, Elipsa seeks to turn an organization’s existing talent into AI talent enabling them to extract critical insights at scale.
For more information book a demo @ www.elipsa.ai