We know that Optimization tools typically prescribe best-action recommendations, based on certain goals, conditions and potential trade offs. I am an Operations Research enthusiast so optimization methods are close to my heart. However, I believe that optimization methods, when combined with predictive tools, have the potential to deliver greater value to organizations by helping them achieve realistic and feasible optimized business decisions.
What can be the structure of such a synergy ?
Creating synergy between these two types of tools can add great value to good decision making process. A predictive analytics tool can offer insights into future scenarios and an optimization tool can then prescribe optimal recommendations for responding to these scenarios.
Ideally, the tools should talk each other on a continuous basis. Technology infrastructure capabilities these days can allow such interfaces. The frequency of data exchange needs to be determined on the level and requirement of synergy between the tools.
Example of applications in Manufacturing
According to a recent report from IDP and SAP, 60% of manufacturers will be using analytics data tracked using connected devices by 2020. Now, there are mundane examples of synergies between predictive and optimization tools, like predicting the demand and then optimizing your production lines using the predicted demand.
However, an interesting application can be to minimize manufacturing disruptions by incorporating data from IoT sensors dynamically into optimization and/or scheduling tools to create solutions that takes into account real world manufacturing challenges.
Innovative organizations around the world are already leveraging predictive analytics for preventive maintenance. Leveraging IoT based sensors, predictive algorithms identify possible machine breakdowns. Condition based preventive maintenance can then be done to avoid actual breakdowns. If such an algorithm is combined/integrated with a manufacturing process scheduling optimization tool, it will lead to creation of more realistic schedules and will deliver significant efficiencies.
Examples of applications in Logistics Optimization
Incorporating AI into logistics route optimization algorithms can help incorporate factors and/or inputs that are generally not considered by typical VRP algorithms. Examples are Weather, typical traffic patterns on certain routes, driver behavior and vehicle performance data pertaining to each equipment. Significant reduction in fuel cost and drive time can be realized. Also, by using IoT sensors that monitor vehicle performance and driver behavior, and combining this data with logistics asset optimization and driver scheduling tools, can help direct behavioral aspects of drivers and also help optimize fuel consumption and reducing maintenance costs.
Example of application in Yard optimization
Conventional yard optimization tools leverage data from RFID tags to track assets in the yard. However, these tools still present many challenges. Visibility issues like what material is in the trailers that have been dropped in the yard still prevail. In case of reefer trailers, there may be malfunctioning of trailers leading to loss of products being stocked in the trailers.
IoT sensors on assets as well as on the pallets/packaging inside the trailers can provide very specific data that can help Yard optimization tools to prescribe optimal decisions, thereby reducing assets, increasing asset utilization and increasing yard efficiency.
Embrace the integration strategy
The list of applications can be really long. The gist however is that while developing your organization’s AI talent pool, don’t ignore the old school optimization techniques.
Instead, try to leverage the best of both by integrating these capabilities.