If you get large amounts of waste every hour in industrial processing, you would certainly appreciate an improved process. Mathematical models of process monitoring may, however, be very complicated, or they operate only in certain conditions. This is why they give very little information or none at all, although it would be extremely useful.

The cleaning process of waste water involves various factors that affect the end product i.e. the cleaned water. Furthermore, biological processes tend to be nonlinear. Processes that have a large number of inter-dependent quantities require specialists to monitor them. But usually the situations change rapidly, and the process controller is seldom completely familiar with the whole process.

What is Neural-Mill©

Neural-Mill© is our solution to problems like this. Neural Mill© is a neural network system that has been specially developed for process control. However, you can use the system for learning any kind of tasks. The input file including measurement values of the process can actually include anything from currency exchange rates to information needed when producing cement! This means it can include any kind of numerical information, although Neural Mill© has been developed especially for industrial processes.

How can I use the program

Neural-Mill© has been developed especially for learning, analyzing and monitoring your process. It is an off-line process, which requires a user for final process monitoring.

There are two main technologies for neural networks, (1) Multilayer Perceptron (MLP) and (2) self-organizing map (SOM). The latter is more visual and is better adapted to process monitoring. Neural Mill© is a self-organizing map.

Learning is automatic

The most important aspect of a self-organizing map like this is automatic learning. You do not have to program functions to the PC, because it learns itself. Another important fact is that neural networks are very useful in real processing, because learning is based on empirical material. This makes neural networks much more reliable than e.g. estimated calculations in processing models. Thirdly, it learns also nonlinear processes. This is important, because most of the inter-relationships between process quantities are nonlinear and complicated.

Neural networks are intelligent, just like us humans. This does not mean that technology will replace humans. Our extremely perceptive senses work well in everyday life, but things like industrial processes can be done more effectively by neural networks.

Neural Mill© offers you an effective and visual answer to complicated and nonlinear industrial processes. It finds new options for process control and gives you a better way to monitor your process.