Optimal Operational Control Software System For Hematite Grinding Process

As one of the key mineral processing procedures, the grinding process is used to provide raw material for the subsequent beneficiation operation. During this process, run-of-mine from crushing procedure continues to be comminuted to finer particle size such that the valuable mineral and gangue can be dissociated by each other. The product particle size and production efficiency of the grinding process has a great influence on product quality (such as the recovery rate of valuable minerals) and production capacity of whole concentration plant. China is rich in iron ore resources, but most of them are hematite ore, which are characteristics of low-grade, fine-grained, non-homogenous distribution, composition complexity and instability. In order to obtain the qualified grinding particle size (GPS), a closed-loop circuit with a ball mill and a spiral classifier is usually applied. The technological objectives of such process are to control the GPS within its desired range and as close as its target value, while avoiding mill load abnormal condition so as to accomplish the safe, stable and successive operation under the situation of frequent fluctuations of ore particle size and grindability.The hematite grinding process (HGP) is a multiple production equipment connected industrial process, which has a comprehensive complexity, mainly in:i) the operation model with mill ore feed rate, mill water flow rate and classifier overflow slurry density as input and GPS as output has features of strong nonlinearities, multivariable couplings, unclear grinding mechanism, and unable modeling, and also its dynamic characteristics are affected by ore particle size and grindability; ii) the magnetic agglomerate usually appears in the strong magnetic particles, which makes the online particle size analyzer difficult measure actual particle size accurately. In actual production process, the feedback information of GPS only relies on the laboratory assay. But the assay period is about 2 hours that is far longer than the closed-loop optimization control period that is about 15 minutes. Thus, the assay value of GPS cannot be used in optimization control; iii) when ore particle size and grindability have changed, the improper setpoints of the mill ore feed rate, mill water flow rate and classifier overflow slurry density may lead to "under load" or "over load" abnormilities of the mill, and even result in mill belly or empty tamping accidents, which seriously affect the safety and stability of grinding process operation.The above complex characteristic makes the existing grinding operational optimization and control approaches as well as the software product difficult to be applied in HGP. Therefore, the optimal operational control of HGP challenges the existing control technology. From the algorithm point of view, to achieve the control objectives of HGP, diverse complex algorithms, actual data and knowledge must be incorporated to develop the optimal operational control method integrating the operation index prediction, optimization control, fault diagnosis and self-healing control. From the software point of view, although the existing commercial software for the operational optimization and control provides abundant human-computer interaction interfaces, they cannot support the extensions and modifications of control algorithms, also the usage of third party software for algorithm developing and solving. Moreover, it is difficult to use the general algorithm configuration software platforms of DCS/PLC control system to realize the optimal operational control since those software platforms are unable to meet the requirements for complex algorithm developing and running. Consequently, it is necessary to develop optimal operational control method and software system in accordance with domestic hematite grinding process to serve the mineral processing industry.This dissertation is supported by the National Key Basic Research Program of China (973 Program) "Research on the key technology and hardware-in-the-loop simulation experiment of integrated control system with safety, cooperativity and usability (2009CB320604)". It focuses on a comprehensive study of optimal operational control software system for HGP. The major contributions of the work are summarized as follows:1. A soft-sensing algorithm which is composed of a main model and an error compensation model for GPS is proposed. In the main model, the dynamic process of GPS is modeled based on the mass balance principle, and the model parameters are corrected using prey-predator optimization method. In the error compensation model, an online robust random vector functional link network (i.e., OR-RVFLN) using nonparametric kernel density estimation and weighted least squares algorithms is employed.2. A data-driven optimal operational control method of HGP, which includes loop setpoint optimization, load abnormal condition diagnosis and self-healing control, is proposed. To maintain the GPS inside its desired range and as close as its target value, loop setpoint optimization incorporates a quadratic performance index with respect to the deviation between the real value and target value of GPS and employs cascade neural network to optimize the control loop setpoints online under the condition of known desired GPS. To detect and deal with load abnormal conditions, the load abnormal condition diagnosis and self-healing control adopts rule based reasoning technology to identify the abnormal conditions in real time, and correct the current setpoints based on the identification results using case based reasoning technology. The outputs of the control loop track the corrected setpoints, thereby forcing the process to recover from the abnormal conditions.3. A configuration software platform for the research of optimal operational control of HGP is developed to realize the GPS soft-sensing, loop setpoint optimization, load abnormal condition diagnosis and self-healing control algorithms, etc. The developed software platform is composed of several functional modules, such as graphical configuration module for optimal operational control algorithm, algorithm library module, algorithm solver module, control strategy check and automatic execution module, data visualization and analysis module. The proposed methods are developed using this software platform and tested in a hardware-in-the-loop simulation experiment system. The results show the usability and effectiveness of the proposed method and developed software platform.4. An industrial application-oriented optimal operational control software system of HGP is developed. The software system includes the following functional modules:data entry module, optimization condition judgment module, GPS soft-sensing module, loop setpoint adjustment module, operation guidance module, operation log management module, etc. The software system has been successfully applied into a practical grinding process of a hematite processing plant in China. The on-line GPS estimation performance of the system has been validated using the actual industrial data first, and then the system has been used to assist the on-site operator to adjust the control loop setpoint on-line. The application has brought the enhancements in grinding product qualification rate and productivity, which shows the usability and effectiveness of the developed software system.