THE AIM OF THE RESEARCH IS TO GENERATE COMPETENCE IN THE APPLICATION OF ARTIFICIAL INTELLIGENCE TO ORE PROCESSING (EXPERT SYSTEMS) AND TO DEMONSTRATE THE APPEAL OF THIS AS AN AID IN THE OPERATION OF SMALL ORE-PROCESSING MINES. THE INTRODUCTION INTO AN AUTOMATIC OPERATING SYSTEM OF HIGHLY DEVELOPED DIAGNOSTIC AND OPERATOR-ADVICE MODULES IS INTENDED TO AID THE SPREAD OF AUTOMATION WITHIN THE ORE PROCESSING INDUSTRY IN THAT MODULES OF THIS TYPE ENABLE THE PRACTICAL DIFFICULTIES ENCOUNTERED IN ON-LINE MEASUREMENT TO BE OUTWEIGHED. THE MODULAR SYSTEM ENVISAGED, BACKED-UP BY A MICRO-PROCESSOR, WOULD ALSO COST VERY LITTLE AND WOULD THEREFORE BE ABLE TO BECOME AVAILABLE TO A WIDE MARKET IN SMALL EUROPEAN MINES.
WHAT IS AT STAKE IS THEREFORE THE PARTIAL OR TOTAL AUTOMATION OF A SIGNIFICANT NUMBER OF PLANTS THROUGHOUT THE COMMUNITY THAT IS LIKELY TO YIELD PRODUCTIVITY GAINS OF THE ORDER OF SEVERAL TENS OF MILLIONS OF FF PER YEAR.
An intelligent modular driving system has been developed which allows improved operation in metallurgical installations by helping the operator to anticipate problems and find solutions. An expert system allows the simultaneous use of measurements from the installation and the results of automatic static real-time simulations in evaluations.
The objective of the project was to design, create and demonstrate an advanced control system for small size and medium size mineral processing plants, based on the use of online algorithms for plant simulation and diagnosis, and online expert systems.
A software structure (ANPROC) has been created. It handles information coming from a conventional process control system installed in a mineral processing plant and:
uses steady state simulation to foresee plant operation;
uses expertise to analyse online data and simulation results;
provides online advice (decision proposals) to the plant operators through a user friendly interface.
ANPROC is a flexible, configurable, real time software structure which includes preexisting tools for process diagnosis. It takes into account steady state simulation, empirical rules and sensors information to help operators, who have to make decisions for process improvements online.
The most innovative feature of ANPROC is the real time supervision of a steady state simulator by an expert system. To use this capability, a simulator has to be configured for a given circuit and expertise on simulation of the circuit has to be extracted and expressed as a network of rules for the expert system.
Instead of using separately and almost independently and expert module and an algorithmic module (simulation), it was found more efficient to combine these tools for dealing with the lack of sensors in the plant and providing the operators with clear and understandable advice.
The application software has been developed and simulations were done on the process giving a good representation of steady state operation.
THE RESEARCH WILL LAST FOR TWO YEARS AND WILL BE CARRIED OUT JOINTLY BY THE AUTOMATION DIVISION OF THE COPPEE LAVALIN AND BY THE BRGM.
MAIN LINES OF RESEARCH:
- OVERALL DESIGN AND PARTIAL APPLICATION OF AN INNOVATIVE "INTELLIGENT" MODULAR AUTOMATIC CONTROL SYSTEM FOR SMALL MINERAL PROCESSING PLANTS. THE SYSTEM WILL BE BACKED UP BY A PC OR COMPATIBLE MICRO-PROCESSOR.
- DESIGN AND PARTIAL INCORPORATION OF INTELLIGENT ON-LINE MODULAR INFORMATION PROCESSING MODULES:
. DIGITAL ALGORITHMS AS DIAGNOSTIC AIDS AND
. EXPERT SYSTEMS
TO ASSIST IN INDUSTRIAL OR PROCESSING OPERATIONS.
BREAKDOWN OF THE STAGES OF RESEARCH:
- PHASE 1: PRODUCTION OF THE SYSTEM STRUCTURE (COPPEE LAVALIN).
- PHASE 2: ON-LINE DIGITAL PROCESSING AND PROCESS DIAGNOSTIC MODULES (BRGM).
- PHASE 3: EXPERT SYSTEM MODULE AS AN ON-LINE DECISION-MAKING AID (COPPEE LAVALIN AND BRGM).
- PHASE 4: INDUSTRIAL TESTING AND APPLICATION.
Funding SchemeCSC - Cost-sharing contracts