Sampsa Laine

Visualization of particle size and shape distributions using self-organizing maps

In pharmaceutical process technology, characterization of the sizes and shapes of different particles is essential. However, comparisons and analysis of different size and shape characteristics of particles are very difficult. In this investigation, we used the self-organizing map (SOM) to visualize the size and shape distributions obtained with image analysis (IA) of a series of model particles and particles created by fluidized bed granulation. Thereafter, the SOM visualization was compared to principal component analysis (PCA) results of the same data. This study shows that the self-organizing map is a useful and interpretive method for analysis of large data sets of particle size and shape distributions. The results indicate that the self-organizing map was capable of creating an intuitive presentation of the differences in the studied particle populations. The choice of data analysis tools should always be made with great consideration.

Ore Type based Expert Systems in Mineral Processing Plants

Artificial intelligence (AI) includes excellent tools for the control and supervision of industrial processes. Several thousand industrial applications have been reported worldwide. Recently, the designers of the AI systems have begun to hybridize the intelligent techniques, expert systems, fuzzy logic and neural networks, to enhance the capability of the AI systems. Expert systems have proved to be ideal candidates especially for the control of mineral processes. An expert system based on on-line classification of the ore type has been developed. Self-organizing maps (SOM) are used for pattern recognition of the type of feed. The system has been tested in two concentrators, the Outokumpu Finnmines Oy, Hitura Mine and Outokumpu Chrome Oy, Kemi Mine. The methodology for the development of the ore type based expert system is presented and the preliminary results obtained in the above plants are described.

Visualization of fluid-bed granulation with self-organizing maps

The degree of the instrumentation of pharmaceutical unit operations has increased. This instrumentation provides information of the state of the process and can be used for both process control and research. However, on-line process data is usually multidimensional, and is difficult to study with traditional trends and scatter plots. The Self-Organizing Map (SOM) is a recognized tool for dimension reduction and process state monitoring. The basics of the SOM and the application to on-line data collected from a fluid-bed granulation process are presented. As a batch process, granulation traversed through a number of process states, which was visualized with SOM as a two-dimensional map. In addition, it is demonstrated how the differences between granulation batches can be studied. The results suggest that SOM together with new in-line process analytical solutions support the in-process control of the pharmaceutical unit operations. Further, a novel research tool for understanding the phenomena during processing is achieved.

On-line determination of ore type using cluster analysis and neural networks

Abstract Expert systems have proved to be excellent tools for the control of mineral processes. An expert system based on on-line classification of the ore type has been designed and is described in this paper. The neural approach to computation has emerged in recent years for dealing with the sort of problems for which more conventional solutions have proven ineffective. In the study a comparison between the on-line cluster algorithm and the Kohonen feature map for ore type classification is presented. The study was carried out with measurement data from the Outokumpu Hitura mine.

On-line determination of the concentrator feed type at Outokumpu Hitura mine

Abstract Variability in process feed poses a process control problem. Processes using natural material as their feed are particularly prone to feed type variability. Usually, the process operators are responsible for dealing with these problems. Operators would benefit from an expert system showing the current feed type and its proper treatment. Unfortunately, on-line detection of the feed type is difficult because explicit on-line measurements of the feed type seldom exist. This paper presents a method for creating such information based on the data from standard on-line measurements. Finding the correct combination of measurements is crucial. The method described in this paper shows how off-line information, such as on-site laboratory analysis results, can be used in this task. The paper is based on experiences acquired from the concentrator of the Outokumpu Hitura mine.

Recent Developments in Designing ORE Type Based Expert System at the Hitura and Kemi Concentrators

Abstract Expert systems have proved to be excellent tools for the control of mineral processes. An expert system based on on-line classification of the ore type has been developed. The neural approach to computation has emerged in recent years for dealing with the sort of problems for which more conventional solutions have proven ineffective. This study has used a neural network for classification. The system has been tested in two concentrators, The Outokumpu Hitura mine and Outokumpu Chrome Kemi mine. The preliminary results are described in the paper.

Combining off-line and on-line information in process study using the self-organizing map (SOM)

This paper presents how process control problems can be studied and solved by combining offline and online information. Offline information is accurate and versatile, it is used to define the problem. Online information describes the state of the process in real time. The online variables containing information of the offline defined problem are selected using a variable selection algorithm. These variables are used to create an online observer of the problem. This observer can be used to solve process control problems. The main algorithms used in this paper are the variable selection technique and the self-organizing map (SOM). The methodology is illustrated using the case of the concentrator of the Outokumpu Hitura mine.

Using SOM-based data binning to support supervised variable selection

We propose a robust and understandable algorithm for supervised variable selection. The user defines a problem by manually selecting the variables Y that are used to train a Self-Organizing Map (SOM), which best describes the problem of interest. This is an illustrative problem definition even in multivariate case. The user also defines another set X, which contains variables that may be related to the problem. Our algorithm browses subsets of X and returns the one, which contains most information of the user’s problem. We measure information by mapping small areas of the studied subset to the SOM lattice. We return the variable set providing, on average, the most compact mapping. By analysis of public domain data sets and by comparison against other variable selection methods, we illustrate the main benefit of our method: understandability to the common user.

Selecting the variables that train a self-organizing map (SOM) which best separates predefined clusters

The paper presents how to find the variables that best illustrate a problem of interest when visualizing with the self-organizing map (SOM). The user defines what is interesting by labeling data points, e.g. with alphabets. These labels assign the data points into clusters. An optimization algorithm looks for the set of variables that best separates the clusters. These variables reflect the knowledge the user applied when labeling the data points. The paper measures the separability, not in the variable space, but on a SOM trained into this space. The found variables contain interesting information, and are well suited for the SOM. The trained SOM can comprehensively visualize the problem of interest, which supports discussion and learning from data. The approach is illustrated using the case of the Hitura mine; and compared with a standard statistical visualization algorithm, the Fisher discriminant analysis.

Visual approach to supervised variable selection by self-organizing map.

Practical data analysis often encounters data sets with both relevant and useless variables. Supervised variable selection is the task of selecting the relevant variables based on some predefined criterion. We propose a robust method for this task. The user manually selects a set of target variables and trains a Self-Organizing Map with these data. This sets a criterion to variable selection and is an illustrative description of the users problem, even for multivariate target data. The user also defines another set of variables that are potentially related to the problem. Our method returns a subset of these variables, which best corresponds to the description provided by the Self-Organizing Map and, thus, agrees with the users understanding about the problem. The method is conceptually simple and, based on experiments, allows an accessible approach to supervised variable selection.