DEVELOPMENT OF AN ARTIFICIAL INTELLIGENCE GEOINFORMATICS SYSTEM FOR SOLID MINERAL PROSPECTING

ABSTRACT

Advanced precision engineering has enabled the development of hyperspectral sensors. These sensors can be used in the collection of highly detailed spectra information from required target using non-intrusive, zero-impact and remote sensing methods. A particular usage is the use of hyperspectral sensors on satellites to obtain information about earth surfaces and minerals beneath the surface. However, the data so-obtained consist of high dimension (hundreds of bands) spectrum that needs to be processed to obtain classification of data and thus identify the present minerals. Thus, hyperspectral data dimension reduction has become an active research topic. In this work, we use a new metaheuristic, namely halfing method, to generate a characterization map of a given spectrum. The method literally ratios one half of the spectrum against the other, thus attenuating defects due to measurement conditions. Fuzzy c-means clustering was modified to obtain cluster centres from the characterization map. The obtained cluster centres were then used to classify the pixels (minerals) using unsupervised learning algorithm obtained by modifying Kohonen Self Organising Map (KSOM). Minerals of known spectral library were used to train Adapted Neuro Fuzzy Inference System (ANFIS) which was linked up with KSOM to identify the pixel (mineral) in each of classes earlier identified by KSOM. The quantity of Pixels (Mineral) in each class were also obtained. Novel minerals were also identified by the network. The instrument used for the implementation is R programming language. This research work can go a long run to boost mining sector in terms of mineral prospecting. With this software developed, it is possible to detect various mineral in a particular place in any part of the world provided their hyperspectral data is available.

CHAPTER ONE

1. INTRODUCTION

1.1              BACKGROUND OF STUDY

Artificial intelligence may be defined as the branch of computer science that is concerned with the automation of intelligent behaviour (Vandal, 2010). However, this definition suffers from the fact that intelligence itself is not very well defined or understood. Although, most of us are certain that we know intelligent behaviour when we see it, it is doubtful that anyone could come close to defining intelligence in a way that will be specific enough to help in the evaluation of a supposedly intelligent computer program, while still capturing the validity of the human mind.

Thus, the problem of defining artificial intelligence became one of defining intelligence itself. What is intelligence?

Intelligence is a capacity of a system to achieve a goal or sustain desired behaviour under conditions of uncertainty. Intelligent systems have to cope with sources of uncertainty like the occurrence of unexpected events such as unpredictable changes in the world in which the system operates and incomplete, inconsistent and unreliable information available to the system for the purpose of deciding what to do next.

Intelligent system exhibits intelligent behaviour. Intelligent behaviour if exhibited is capable of achieving specified goals or sustaining desired behavior under conditions of uncertainty even in a poorly structured environment. Such environment includes an environment where various characteristics are not measurable or where several characteristics change simultaneously and in an unexpected way and where it is not possible to decide in advance how the system should respond to every combination of events.

According to Nikola (2007), an intelligent system exhibits the following behaviour

• They should from time to time accommodate new problem solving rules.

• They should be able to analyze themselves in terms of behaviour error and success.

• Once they are to interact, they should learn and improve through interaction with the environment.
  • They should learn quickly from large amount of data.

• They should have many base exemplar storage and retrieval capability.

• They should have a parameter to present.

Agris (2006) also summarized basic features of intelligent systems as follows:

• They have the ability to generate a new knowledge from already existing one.

• They have ability to learn.

• They have ability to sense environment.

• They should have ability to act.

Many research work have been carried out in the area of intelligent system. For instance, Paul (2009) has developed an intelligent system to determine the best soil for different types of crops using artificial neural network. Also, there is an intelligent system to predict weather condition in South Korea using neuro-fuzzy method (Mesia, 2006).

Despite series of research in the area of intelligent based system, enough has not been done in terms of developing an intelligent system in geoinformatics most especially in mineral location and exploration.

Geoinformatics is a field of science that combines geodetic and spatial information processing method with computing hardware and software technology. Therefore, mineral location is one of the major areas of geoinformatics and it is the process of finding ore (commercially viable concentration of mineral) to mine. It is much more intensive and involved. The researchers have carried out different research works in the area of geoinformatics with specific regard to mineral location and explorations.

Different methods were used in such research work. Some used database/ data mining techniques or approach to determine the mineral spread in a specific location.

Some used Bayesian network classifiers (Porwal et al, 2006). Some used Logistic method (Knox-Robinson, 2000), (Luo and Dimitrakopoulas, 2003), (De Quadros et al, 2006), (Carranza et al, 2008). Artificial Neural Network was also used by Singer and Kouda (1996), Brown et al (2000, 2003), Behnia (2007), Skabar (2007), Oh and Lee (2008).

Evidence theory model and Geographical Information System (GIS) were also used. GIS which works on the principle of database has also been used for processing and combining data for mineral allocation (Partington, 2010). Apart from GIS approach, some other approaches have also been used to solve this problem: Weight of evidence model (Agterberg and Bonham, 2005), (Jianping et al, 2005), (Nykanen and Raines, 2006), (Porwal et al, 2006), (Roy et al, 2006) (Nykanen and Ojala, 2007), (Raines et al, 2007), (Oh and Lee, 2008), (Harris et al, 2008), (Benomar et al, 2010).

Apart from the different types of methods used, mineral exploration is a multidisciplinary task requiring the simultaneous consideration of numerous geophysical, geological and geochemical dataset (Knox et al, 2003). The size and complexity of regional exploration data available to

geologist are increasing rapidly from a variety of sources such as remote sensing, airborne geophysics, large commercially available geological and geochemical data (Brown et al, 2010). This demands for more effective integration and analysis of regional and geospatial data with different formats and attributes. This needs spatial modelling techniques regarding the association of mineral occurrence with various geological features in qualitative manner. Moreover, reliable geoinformation in form of geological and mineral potential map is very important for exploration and the development of mineral resources.

In a country like Nigeria, mineral resources development has played an important role for sustainable economic development. For the past years, petroleum has been the only mineral which the country relies upon economically. There is high tendency for this mineral resource to be exhausted or for the world market price for this product to fall. If it happens, the economy of the country will be grounded. Therefore ,there is a need for Nigeria to continue searching for more minerals.

Several geological, geochemical and geophysical data have been collected to carry out further research on how to locate and explore other minerals apart from petroleum. Today, several of such data are acquired from various conventional exploration and mining companies. Most of the information is dispersed and kept in analog format. The truth of the matter is that there is no serious effort to convert these data and integrate information for the purpose of locating other minerals, determining the volume or quantity of such mineral, generating a mineral map to aid mining and exploration process. Again, the generated or existing geological information including geological maps do not contain enough relevant and sufficient information to aid location and exploration.

Recently, the National Space and Research Centre was established in Abuja in the Northern part of Nigeria. This centre covers most part of the country with microwave remote sensing data set. But these data sets are neither being processed nor integrated with the available geological data to produce sufficient information for mineral location and exploration.

In this research work, an intelligent geoinformatics system is developed to train hyperspectral remote sensing data set using; Halving algorithm, modified Fuzzy-C means algorithm, Kohonen Self Organizing Map and Adapted Neuro-Fuzzy Inference System (ANFIS). The hyperspectral remote sensing data set will be collected from Nevada, USA satellite. The data is to be collected from Nevada because hyperspectral data is not yet available in Nigeria and since this type of research is not expected to be a local research, data from any part of the world can be used.

With series of research to develop intelligent system, there has not been an attempt to develop such a system in the area of mineral prospecting using the type of automated method we have used. This research work is to develop an intelligent system which combines different algorithms to solve the complex problem of geoinformatics with specific regard to mineral prospecting. The use of the stated methods will go a long way to generate a better result or output.

Unlike previous works that mostly generate numerical solutions, this research work will generate better solutions and comprehensive mineral prediction or location map. This will be so since the intelligent system does not depend on statistical distribution and analyses of data like the previous ones. The system will not depend on the statistical distribution of data neither will it make use of database approach but rather it will be made to be intelligent. The intelligence system developed can be a useful tool in solving series of geoinformatics problems that have to do with mineral prospecting in Nigeria and beyond.

1.2              STATEMENT OF THE PROBLEM

Hyperspectral data are characterized by large size and complexity. This poses a serious problem since data of very large size and complexity will be very difficult to process to obtain relevant information. Despite the fact that the available data are of high dimension and of great complexity, they still contain relevant information in term of mineral prospecting. How do we reduce the size of such data without losing some of the vital information contained? Even after dimensional reduction, how do we process such data to obtain the classes, types and volume of mineral in each of the classes that are present in such data?

Unfortunately, various methods used by different researchers are not sophisticated enough to handle this important task. The major challenge with this research is how to use AI principles to develop an intelligent software that can be used to process hyperspectral data to obtain relevant and vital information that can be used in the area of geoinformatics especially in the area of mineral prospecting.

1.3              AIM AND OBJECTIVES OF THE STUDY

The aim of this research work is to develop intelligent geoinformatics system that can be used for solid mineral prospecting

The objectives of this research work are to develop a system that can:

1. Offer an intelligent based geoinformatics data/information.
2. Generate from hyperspectral data, numerical solutions that determine the location of existing minerals and the quantity of minerals that are present in a particular location.
3. Generate mineral location map.
4. Detect unknown or noble minerals as a group of minerals that is detected for the first time and therefore the name could not be found among the existing minerals used to train the network of the developed system.

1.4              SIGNIFICANCE OF THE STUDY

Data Reduction: The choice of data in terms of its representation and selection is one of the important issues to be considered during hyperspectral data processing. This can actually determine whether the problem is solvable or not. Performing such data reduction in term of its

dimensionality by the developed system can be of numerous advantages during the computational process. Hence, the research provided a new direction on how to reduce hyperspectral data for further research work.

Reduction in computational process time: Hyperspectral data by virtue of their nature are extremely very large. Large data can take a longer time for processing. This leads to a drastic reduction in the efficiency of the system. Therefore, it is always better to reduce the dimension of the data during preprocessing stage so as to reduce the processing time. The developed system is able to carry out data reduction to a reasonable size thereby reducing the computational processing time.

Improved result/output: The developed system produced an hybridized algorithm which is an improvement on the existing ones. It is an hybridization of different Artificial Intelligence algorithm. By using appropriate data, the algorithm will be able to learn faster and better. The developed system because of the way it was designed and implemented is able to bring an improved result or output compared to the existing system.

Simplified model: It is always better to construct a simple model. The simpler the model, the better the output or result generated. The developed system made use of highly simple model. This assists the system in generating a good result.

The developed system is able to use both supervised and unsupervised leaning algorithm to train hyperspectral data in other to classify and recognize different types of mineral that are present. The developed system is made to be efficient enough to detect strange data set i.e. data set that do not belong to a particular class in a classified hyperspectral data set.

Though, the intelligent system we have developed is applied in the area of mineral prospecting yet, with some modifications, it can be applied in some other areas of geoinformatics and mining.

To ensure a continued supply of mining products, it is necessary to discover new mineral deposit in addition to those currently being mined. Successful exploration therefore ensures the future of the individual and the world economic well being.

Considering the application domain, the new system has other numerous advantages. For instance, mining industry is very important in the development of any nation. The industry obviously employs reasonable number of the nation workforce. It is perhaps not surprising that its importance to everyday life is still poorly understood and appreciated by people.

Mining is not confined to large scale ore operations, nickel and gold production from goldfields, or bauxite mining to produce alumina. It also includes the mining of silica for the glass industry to produce drinking glasses, car windscreens and window panes. Also, the aggregate used to build roads, clay for house bricks, roof tiles and crockery, copper for electrical wire, and the

exotic element like tantalum and yttrium necessary for production of capacitors and other products essential for modern semiconductor technology are products of solid minerals. It also includes coal, petroleum and natural gas that provide power and warmth for the community and a host of associated by-product such as plastics and synthetic fibres.

In order to maintain our living standard, we must continue mining and this requires continued exploration for new deposits of all types. Mineral location or exploration is like looking for a needle in a haystack. So, it is important to keep searching. Moreover, everyone in the modern world, depends heavily on the product of mining. The development of commercially viable mineral deposit is also a key factor in achieving a sound economy. To ensure a continued supply of mining products, it is necessary to discover new mineral deposit to replace those currently being mined. Successful exploration therefore ensures the future of the individual and the world economic well being.

Mineral location or exploration is a scientific investigation of the earth crust to determine if there are mineral deposits present that may be commercially developed. To be able to find new deposit, explorers must have access to the land. This will only be permitted if exploration can be carried out with negligible impact on the natural environment. Modern location methods like the one used in this work is capable of discovering deeply concealed deposits which have eluded earlier explorers.

Almost everything that we eat, drink, live in, fly in depends on the products of the mineral industry for either its components, its production or its source of energy. The exploration, mining and mineral processing industry exist because the consumers demand these product. Mineral occur in earth crust in rare concentration known as mineral deposits. Mining is the process of removing these deposits from the ground. Every deposit, no matter how large, has a finite life and will one day be exhausted. It will not be wise to fold our arms until the minerals are exhausted. There is a need to ensure a continued supply of mineral to meet the need of a growing population. Different types of methods have previously been applied to solve the problem of mineral location/ exploration. For instance, previous authors have used. e.g. Evidence Weight Method, Bayesian Theory, Tree diagrams, Neural Network, GIS e.t.c. Obviously, most of these methods are statistically based method which may bring a lot of inaccuracies and bias in the result obtained. The GIS method also depends purely on database. In other words, the results generated are not producing enough information for the mining industry to locate minerals. This calls for further research work.

Therefore, an attempt is being made to use Artificial intelligent method to solve the problem of mineral location. The system is expected to produce a better result and therefore better information for mineral industry. If the mining industries are boosted, it provides the basic needs for the people; there will be good source of income for the government, individuals, parastaters

e.t.c. This will also provide better job opportunities for the citizen. In fact, it will go a long run to boost the economy of the country at large.

1.5              SCOPE OF THE WORK

In this research work, an intelligent system that can process hyperspectral data has been developed. The system was applied in the area of mineral prospecting using hyperspectral data for Cuprite from Nevada, USA as the case study. Hyperspectral Data from other countries or locations can still be used. Though we have applied the intelligent system in one area of geoinformatics, with some modifications it can still be applied in some other areas e.g. it could identify different types of soil, vegetation, land topology in a given hyperspectral data.

1.6              DEFINITION OF TERMS

Expert System: This is a computer program designed to simulate the problem solving behaviour of human who is an expert in a narrow domain or discipline.

Artificial Intelligence: This could be defined as the ability of computer software and hardware to do those things that we as human being recognize as intelligent behavior.

Geo-informatics: This is a field of science that combines geodetic and spatial information processing method with computing hardware and software.

Intelligence: This is the capacity of a system to achieve a goal or sustain desired behavior under condition of uncertainty.

Intelligent System: This is the system that exhibits intelligent behavior.

Mineral Location: This is the act of detecting geographical areas where minerals could be sited and explored.

Mineral: This is a naturally occurring inorganic solid, with definite chemical composition and an ordered atomic arrangement e.g. oil, granite, gold, charcoal e.t.c.

Geology: This is the study of earth crust, its rocks and its history.

Geographical Information System: This is the type of system that deals with spatial and semantic data and provide means to analyze them, using computer hardware and software tools.

Geologic modelling or Geomodelling: This is the applied science of creating computerized representations of portions of the Earth’s crust based on geophysical and geological observations made on and below the Earth surface.

Spectra Signature: They are specific combinations of reflected and absorbed electromagnetic radiation at varying wavelengths which can uniquely identify an object.

Hyperspectral Data: It consists of large numbers of narrow spectra channel from optical wavelength range.

Data Clustering: It is the process of dividing data element into groups or classes such that items in the same class are as similar as possible and items in different classes are as dissimilar as possible.

Unmixing: It is a method used for estimating and measure abundant fraction in a mixed mineral.

Cluster: Clusters can be defined as objects belonging most likely to the same distribution/group.

Clustering: This is otherwise known as Cluster Analysis. It is the task of grouping a set of objects in such a way that objects in the same group (called cluster) are more similar (in some sense or another) to each other than to those in other groups (clusters).

Cluster Center: Mean squared distance from each data point to its nearest center.

Mineral: A mineral, or pel, (picture element) is a physical point in a raster image, or the smallest addressable element in a display device. It is the smallest controllable element of a picture represented on the screen.

Spectrum: An array of entities, as light waves or particles, ordered in accordance with the magnitudes of a common physical property, as wavelength or mass.

Mineral Map: Maps describing the location or the conditions of formation of mineral deposits. These maps are prepared on the basis of records of mineral deposits and data obtained from geological surveying, prospecting, and exploration.

Membership Function: The membership function of a fuzzy set is a generalization of the indicator function for a fuzzy set which assigns a truth value (0 or 1) to each element in a classical set.

Rule Surfaces: A ruled surface can always be described (at least locally) as the set of points swept by a moving straight line. For example, a cone is formed by keeping one point of a line fixed whilst moving another point along a circle. In algebraic geometry, ruled surfaces were originally defined as projective surfaces in projective space containing a straight line through any given point.

Inference Engine: An inference engine is a computer program that tries to derive answers from a knowledge base. It is the “brain” that expert systems use to reason about the information in the knowledge base for the ultimate purpose of formulating new conclusions. Inference engines are considered to be a special case of reasoning engines, which can use more general methods of reasoning.

Module: Each of a set of standardized parts or independent units that can be used to construct a more complex structure.

Supervised Learning: Supervised learning is the machine learning task of inferring a function from labeled training data. The training data consist of a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal).

Supervised Learning Algorithm: Analyzes the training data and produces an inferred function, which can be used for mapping new examples.

Unsupervised Learning: Unsupervised learning refers to the problem of trying to find hidden structure in unlabeled data. Since the examples given to the learner are unlabeled, there is no error or reward signal to evaluate a potential solution. This distinguishes unsupervised learning from supervised learning and reinforcement learning.

Back Propagation: An abbreviation for “backward propagation of errors”, is a common method of training artificial neural networks. From a desired output, the network learns from many inputs, similar to the way a child learns to identify a dog from examples of dogs.

Novel Mineral: This is a new mineral that has not been earlier detected by the network simply because it is not part of the trained data.

₦5,000.00 – DOWNLOAD FULL PROJECT DOWNLOAD FULL PROJECT Added to cart

---

---

---

---

Related Project Topics :