PLAiN is an end-to-end solution for advanced production scheduling (APS) in manufacturing. It is designed to optimize production efficiency, increase manufacturing bandwidth, and reduce costs. It is deployed in five phases, starting with data clean up and ending with the deployment of a fully functioning APS module connected live to the real manufacturing plant. In phase one, the production data typically stored in an ERP backend database is cleaned up to ensure high-quality data that can be used for the next phases of the implementation. This includes data acquisition, exploratory data analysis, resolution of data quality issues, and more.
Once the data is cleaned up, phase two starts. A digital shadow (or a twin) of the production process is created. The digital shadow is validated against real-life production data to ensure that it is an accurate representation of the production process. Once the digital twin is validated, it serves for simulations and training of the AI agent. The agent is trained using reinforcement learning, a technique similar to the one used for chess-playing AI. The agent performs hundreds of thousands of simulations to learn the principles of the plant functioning and to become capable of creating efficient production schedules that increase the manufacturing bandwidth and reduce the costs of production. The agent can leverage different optimization functions (energy saving, warehouse optimization, maximum production bandwidth, etc.) as its targets in the learning process. Once the agent is fully trained, it gets connected to the ERP backend DB and it starts generating production schedules almost in real-time.
In the final phase, the front-end web application is customized for the customer’s needs. Alternatively, the ERP system front-end can serve to interact with the APS module.
PLAiN offers a complete set of standard APS features, including forward and backward planning, addressed and non-addressed planning, flexible production batches, grouping of production orders to minimize changeover times, MRP planning, and capable-to-promise planning. Our solution also indicates risks in the production schedule, is faster than standard APS solutions, and allows for cooperation between human production planners and the AI agent. It can be integrated via standard interfaces such as TCP/IP, gRPC, or HTTP2.
The implementation typically takes less than a year, including all phases of the implementation. This can vary depending on the size and complexity of the manufacturing plant, but we have experience in delivering the solution in a timely and efficient manner. Overall, PLAiN is a next-generation production scheduling solution that uses the state-of-the-art deep learning AI algorithms to create efficient production schedules that increase manufacturing bandwidth and reduce costs. Its end-to-end implementation process ensures high-quality data and accurate representation of the production process in the form of a digital twin, resulting in a fully functioning APS solution connected live to the real manufacturing processes.
IPR / Licence
The solution is deployed as a Cloud service or it can be deployed within the customer’s server infrastructure. It requires an environment for running Docker images. It is connected to the back end ERP database or its clone. It comes with our own simple front-end application OR the ERP system GUI can be altered/used to interact with the solution via a defined API. SW architecture documentation and HW infrastructure requirements are available upon request.
The implementation cost is 60-85k EUR. Yearly license and M&S (including rights to updates) cost 8.000-12.500 EUR. Once the solution is installed, the customer pays a yearly license fee.