1 Online detection and analysis technology Online detection and analysis technology is the basis for automation and intelligence. In the past ten years, the detection and application of single parameters such as flow, level, pressure, temperature and pH in the beneficiation process are very popular and mature, and will not be repeated here. The development and application of online analysis technology in the mineral processing process has not only stopped, but has been increasingly valued by the industry. Especially with the rapid development of applied basic research, it promotes the development and application of intelligent equipment such as high-end analysis systems. This type of equipment can detect and predict complex, comprehensive, and associated process information such as mineral characteristics, equipment status, and production indicators in production through physical measurement and mathematical modeling of one or more parameters. These equipments make it possible to intelligentize production operations, change the meaning and tasks of the beneficiation automation tradition, and shorten the economic and social benefits between mineral processing automation technology and production costs, benefits, safety, environmental protection, management, knowledge reserves, and personnel training. distance. (1) Mill condition monitoring technology Among many mineral processing equipments, the mill has the highest operating cost and the largest energy consumption. The control of the operating state and efficiency of the mill is the key to energy saving optimization of the grinding process. Three factors determine the working efficiency of the mill: the size of the mill barrel, the state of the lifting lining, the loading of the mill and the material distribution. If the equipment and state information such as the material loading amount in the mill, the material concentration granularity state in the mill, the wear level of the lining plate, etc. can be grasped in time, the operating conditions of the mill can be adjusted in time, so that the mill has the best processing capacity and the optimal operating efficiency. Maintenance and timely. Therefore, mill condition monitoring technology has always been the focus and hotspot of mining technology research. Foreign research institutes such as AMIRA, CSIRO, Outotec, COREM, and JKMRC have done a lot of research work in this area. There are many types of mill condition monitoring technologies. The International Mining Alliance AMIRA project, CSIRO's "Vibration Measurement Based Mill Load Monitoring" project, in 2006 successfully developed the inertial power supply system, accelerometer sensor group and wireless multi-channel signal acquisition system, using vibration measurement method to predict mill operation State, conducted in 2008 at the North Parkes Mine. In 2008, AMIRA used the discrete element method to simulate the dynamic characteristics between ore, steel ball and liner in the grinding process, and then established a number of prediction models for the operating parameters of the mill, including mill load, grinding size, grinding Machine liner wear condition, mill material distribution range, etc. In 2006, Outotec reported on the use of pulsation information in the power curve to predict mill loading and to develop MillSense, a mill load analyzer. They believe that the pulsation information in the power curve is periodically raised by the mill. Then it is generated by the movement process of the bottom material, so that the movement trend of the filling material can be judged by the extraction of the pulsation information and the phase change of the rotation cycle of the mill. Grinding electro-acoustic ear detection is a method that uses the acoustic method to convert the noise generated during the working process of the mill into a meter signal through the acquisition of the microphone. According to the literature, this method is used for silver Mao Nanjing Lead zinc ore concentrator Ltd., China Gold Group Mining Co., Ltd., Inner Mongolia Wu Nuge spit Hill copper-molybdenum mine concentrator such as grinding control system, played a good application effect. In China, the electric ear method is used to predict the mill load condition for a long period of time. This method can reflect the mill load condition to a certain extent. However, due to the variety of interference signals and the simple analysis and processing methods, the accuracy is affected. . In 2006, China began to research on the “grinding machine/semi- self-grinding machine load detection technology†based on the vibration signal detection and analysis of the mill wall, and successfully developed the “grinding machine load monitoring system based on vibration measurement†to realize the vibration signal acquisition and signal. Real-time processing, the equipment runs reliably for a long time in the industrial field. However, the vibration signal of the mill is seriously affected by the degree of wear of the liner, and the signal will drift for a long time as time goes by. During the 12th Five-Year Plan period, by increasing the lining wear measurement sensor, the lining wear condition is measured online, and the vibration signal is corrected, which greatly improves the applicability of the system. At the same time, the lining wear can be predicted and the maintenance plan can be arranged reasonably. At present, the characteristic parameter Beta of the vibration signal of the mill can be used in the grinding control system, and has been successfully applied to the grinding and feeding control loop of the concentrator such as Sanshandao Gold Mine and Jiaojia Gold Mine. Based on the research results and application practice at home and abroad, the state characteristics of the mill have multi-parameter coupling, time-varying, large lag, diversification, etc. The limitations of relying on a single detection technology or method are very large, so specific analysis of specific objects is required. According to the characteristics of the object being monitored and the specific grinding process, various technologies such as dynamics simulation, data modeling and multivariate statistical monitoring can be organically integrated to achieve better results. On the other hand, in these years, the automation system of China's mill equipment is relatively mature, including parameters such as power and current, critical operation and interlocking control of lubrication and hydraulic devices. From the perspective of production practice, these seemingly simple variables and information are also very important for the analysis of the operating state of the mill. (2) Flotation foam state analysis technology At present, flotation foam state analysis technology is mainly used to analyze the visual characteristics of flotation foam surface, which is a direct indicator of flotation conditions and process indicators. In actual mineral sorting production, flotation foam surface visual features such as color and size , flow rate, texture, etc. rely on manual observation, subjective, large error, low efficiency, can not achieve the objective evaluation and cognition of the flotation state, resulting in unstable production process, serious loss of mineral raw materials, excessive consumption of chemicals, for this reason Applying machine vision to the monitoring of the mineral flotation process can increase the recovery rate of the flotation process. In recent years, machine vision-based flotation foam surface feature monitoring technology has attracted the attention of scientific research institutions in industrialized countries and launched corresponding products. C. Aldrich et al., Ivana Jovanović et al. and South Africa's Mintek mentioned 16 foreign foam image products. However, there are few related literatures, among which Visio Froth (Metso) and FrothMaster (Outotec) have been widely used in mining enterprises abroad. The domestic Beijing Research Institute of Mining and Metallurgy, Central South University, China University of Mining and Technology, etc. also carried out research on flotation foam image processing and detection technology, and achieved certain research results. JF Reddick et al. used SmartFroth to try to predict grades by color and considered that individual color information could not be used to accurately predict concentrate grades. A. Supomo et al. used VisioFroth to measure the foam overflow rate on the rough selection tank of PT Freeport in Indonesia. By modifying the liquid level control set value and other restrictions, the yield was adjusted and the recovery rate of the rough selection process was increased. 2.4%. J. Leiva et al. used VisioFroth to measure foam transport for estimated air recovery, and I. Rojas et al. used VisioFroth to measure foam transport characteristics. E. Sanwani et al. used the JK capture method and the CSIRO conductivity method to measure and compare bubble volume fractions in the flotation cell. Nicolas Barbian et al. measured the foam thickness by means of a dedicated isolating cylinder to measure bubble stability, which is consistent with the air dispersion characteristics measured by the foam image, demonstrating that a moderate amount of aeration is the key to maintaining foam stability and recovery. Barbian. N et al. measured bubble solids loading and air recovery and introduced two parameter foam volume mineral loadings and cross-correlation poles, and studied the correlation of these two variables with flotation performance. Lin. B et al. proposed and implemented a bubble size estimation method. Nunez. F et al. developed an algorithm based on soft-measurement dynamic texture prediction for bubble velocity at sampling. MH Moys et al. improved the method of measuring foam loading. C. Marais et al. estimated the platinum grade by foam image information. Mehrshad. N et al. proposed a marker-based adaptive watershed segmentation algorithm to measure foam size distribution. A. Kramer et al. introduced the rapid surface bubble method of chemical dynamic surface tension into the flotation line to measure the surface tension of the bubble. W. Kracht et al. used a random approximation method to measure the foam size distribution. Nissinen. A used ERT measurement data to model the foam structure and combined it with foam image technology. The results show that the two have a good correlation. Jahedsaravani. A et al. used a fixed-variable method to change the flotation operating conditions in the laboratory, and used image technology to analyze the bubble information, and then used the neural network method to model. Ata. S et al. used a high-speed camera to capture and analyze the effect of particles on the bubbles blown out by the capillary. Morris. GDM et al. used high-speed camera flotation and verified the effect of Dippenaar's lead ore particles on bubble films in 1982 and made some different points. Beijing Mining and Metallurgical Research Institute developed the BFIPS-I flotation foam image analysis system in 2008. According to the obtained flotation foam image, the system can calculate the size, number, stability, speed, color and texture of the flotation foam. With the characteristic parameters, the system is applied in the industrial application of copper-molybdenum mixed flotation in the Dashan Concentrator of Dexing Copper Mine, and the prediction of the grade of concentrate foam by using the foam characteristic parameters is realized. In 2014, BFIPS-II flotation foam image analysis system was used in the concentrating plant of Jiaojia Gold Mine of Shandong Gold Group. By controlling the cone valve and the aeration amount, the flotation foam speed was controlled, and the flotation process was automatically controlled and achieved remarkable results. . Technical aspects of mineral flotation process monitoring, Central South University, also based on machine vision has made appropriate findings, developed a variety of mineral bauxite, copper, gold and antimony ore flotation froth image processing system bauxite float The selected foam image processing system can automatically extract foam characteristics indicating foam color, size, speed, texture, stability, fluidity, etc., and provide real-time display of foam image, characteristic curve and process parameter curve real-time display to realize the state of flotation foam. The functions of classification, identification and comprehensive evaluation and automatic generation of production reports, the system effectively improves the working environment and labor intensity of workers, provides guidance information for on-site flotation operations, and improves the automation technology level of the flotation process. In recent years, many scholars have carried out a lot of research on the extraction methods of foam texture, size, color, moving speed and other methods of flotation condition recognition. In order to accurately obtain the texture of the foam image, Gui Weihua proposed a new flotation foam image texture feature extraction method based on the gray level co-occurrence matrix method. The foam image was converted into color space, and the converted foam image was calculated. The color co-occurrence matrix, then extracts the feature statistic from the normalized color co-occurrence matrix, and finally provides the basis for the flotation optimization control based on the characteristics of the foam texture characterized by the feature statistic. Liu Jinping established the statistical distribution model of the bubble transform domain coefficient, and based on the relationship between the visual characteristics of the foam image and the flotation production conditions, the automatic identification of the health status of the drug addition process based on the statistical feature parameters of the bubble image was developed. Function and comprehensive classification and evaluation function of flotation production conditions, taking the bauxite flotation process as the object, applying the proposed method to the classification and identification of flotation process conditions, and using the collected flotation operation Knowledge, according to the working condition identification results, provides corresponding operational guidance for the flotation site operators. Based on the correlation between the foam image and the pH value of flotation pulp, Ren Huifeng proposed the extraction method of foam color, morphology and texture features, and established a soft pH measurement model based on the combination of foam image features and multi-case sub-models to realize the flotation pH of bauxite. Real-time detection of values ​​creates conditions for optimal control of the flotation process. Zhou Kaijun proposed a bubble image adaptive segmentation method based on hierarchical watershed algorithm. The concept of sample statistical distribution was introduced. The statistical characteristics of bubble average size, variance, skewness and steepness were extracted. A foam shape based on morphological signature transformation was proposed. Feature extraction method to quantify bubble morphological features. The correlation between foam morphological characteristics and mineral recovery rate was analyzed, and a mineral recovery prediction model based on least squares support vector machine was established to provide a basis for optimization operation. Lin Xia has the characteristics of large computational complexity and low precision for the texture feature extraction of existing flotation bubble images. A CCHS-based texture feature extraction method is proposed in the spatial domain of images. Aiming at the complex conditions and the robustness of the flotation foam image texture and the robustness of illumination changes, a texture feature extraction method based on the improved Gabor wavelet transform is proposed in the frequency domain of the image. Texture features are used for industrial classification. Owen Jun studied the velocity and shape feature extraction algorithm, a bubble motion velocity extraction method based on macroblock tracking, and introduced the concept of maximum potential energy to simulate the flooding process of the watershed catchment basin. The watershed segmentation method based on the maximum potential energy was used to obtain the foam. Shape features, using expert knowledge systems to provide guidance for optimal control of the flotation process. Chen Ning proposed a method for extracting texture features of flotation foam images based on color co-occurrence matrix. By analyzing the texture complexity of the foam image and the corresponding mineral position, the relationship between the two is analyzed, and the texture complexity and minerals of the foam are qualitatively pointed out. The correlation of grades gives the optimal texture complexity interval for the flotation bubble. Yan Xuemin proposed an adaptive foam image macroblock tracking method based on Fourier-Mellin transform and template matching to automatically track the deformation bubble to accurately measure the speed of the flotation bubble flow to the scraper, and based on the grayscale SIFT of the foam image. The Kalman filtering combined foam velocity feature extraction method tracks various bubble motion sub-blocks. Aiming at the problem that the characteristics of bubble stability are difficult to describe quantitatively, a method for extracting the surface deformation coefficient and breaking rate of flotation foam based on digital image processing is proposed, and the foam deformation coefficient and foam breaking rate are quantitatively described. Li Jianqi studied the definition evaluation method of flotation bubble image, the illumination uniformization method of foam image, and the multi-scale enhancement method of foam image, and further proposed a bubble image adaptive segmentation method for parameter measurement, which is floating in bauxite mine. Research has been conducted on the selection of industrial applications. Other methods based on foam image processing technology include statistical wavelet coefficients of tree wavelet transform to extract texture feature parameters of selected foam images, and ARMA (autoregressive moving average) dynamic texture model to describe the correlation between foam images. A fuzzy texture spectral texture feature extraction method based on rotation classification, a foam image classification method based on local feature Bayesian probability model, a method for extracting surface texture thickness feature of selected foam image based on improved LBP operator, and multi-resolution Wavelet transform method, SIFT-based velocity feature extraction, SVM multi-class image classification algorithm for convex-shell binary tree, wavelet transform combined with ARMA dynamic texture modeling method, adaptive fractional differential image enhancement method, color-based co-occurrence matrix (CCM) Texture feature extraction algorithm, PSO-based hybrid size bubble pattern segmentation method. (3) Thickener load monitoring technology The thickener load mainly refers to the mud storage volume, which can be reflected indirectly by measuring the truss torque, the truss current, the mud bed pressure, and the mud layer thickness. The mud bed pressure test usually needs to pre-embed the pressure sensor when installing in the cone bottom of the thickener. The thickness of the mud layer can be realized by ultrasonic level meter or immersion infrared turbidimeter. The advantage of these two methods is the reaction mud storage. More direct, the disadvantage is that a lot of calibration work is needed. It is simple and practical to judge the mud storage capacity of the thickener by using the truss torque and the truss current, but the reaction equipment state information is limited and the distortion rate is also high. In recent years, soft measurement technology has been applied to the research field of thickeners. For example, the Beijing Research Institute of Mining and Metallurgy has realized the thickness of the mud bed according to the working principle of the thickener, the input and output materials, and the current/torque of the thickener. (4) Ore size analysis The image segmentation technology is used to measure the ore particle size (blockiness) on the belt conveyor. The measurement objects include coarsely crushed, finely divided ore and crushed ore, self-grinding/semi-self-grinding machine ore, and steel balls on the belt. , ball mill feeding, etc., as a guide for grinding control, can improve the production efficiency and processing capacity of crushing and grinding. The Split-o nline Rock® Fragmentation Analysis system developed by SPLITENGEINEERING, USA, and the PlantVision system developed by KSX have been used in production practice in foreign mines. Domestic research in this area started late and is still in the stage of industrial trial research. (5) Slurry particle size measurement The particle size of the grinding process is an important indicator directly related to the concentrate grade and metal recovery rate of the ore dressing. The on-line detection of the particle size is of great significance for the optimal control of the grinding process and the improvement of the concentrate grade and metal recovery rate. Relevant companies and scientific research institutions at home and abroad have developed stable detection schemes and products using various measurement principles such as ultrasonic attenuation, direct diameter measurement, and laser diffraction. Ultrasonic particle size analyzer The ultrasonic particle size analyzer is mainly composed of a sampling device, an air eliminator, a sensor (ultrasonic probe), an electronic processing device, and a display instrument. The slurry from the process flow enters the air eliminator through the sampling device, removes the air bubbles mixed into the slurry, and flows into the sensor for detection. In order to overcome the influence of the slurry concentration, the sensor also needs to detect the ultrasonic attenuation caused by the concentration to correct the measurement result. The ultrasonic attenuation principle includes On-line Particle size analysis by Ultrasonic Spectroscopy (OPUS), PSM-400 Particle Size Analyzer from American Thermoelectric Company, and domestic equipment including DF developed by Oriental Measurement and Control Company. -PSM online ultrasonic particle size analyzer, etc. These products have application cases in domestic mining enterprises. 2. Direct caliper particle size analyzer The direct caliper particle size analyzer (also called mechanical particle size analyzer) generally consists of a sampling device, a flow stabilizer, a calibration sampler, a measuring head, an electronic control display unit and the like. The measuring part of the core detecting component is composed of a motor, a speed reducing mechanism, a cam, a measuring plunger, a differential transformer and a measuring slot. The rotary motion of the motor is converted into the vertical movement of the plunger in the measuring slot by the motor, the speed reducing mechanism, the cam and the plunger, and the ceramic measuring head is driven to complete the measuring action. The direct caliper type particle size analyzer does not need a degassing device, and is not affected by the magnetic effect of the slurry and the impurities in the slurry. The influence of the concentration change is also not sensitive. It can be seen from the related reports that the number of popularized applications is much larger than that of other measurement principles. . 3. Laser diffraction particle size analyzer The laser diffraction particle size analyzer is based on the light scattering of mineral particles. The scattering model is based on the Mie theory, and the model width depends on the particle size. One of the advantages of the laser diffraction analysis principle is that it can give a full-grain distribution structure without calibration, and its repeatability and precision are good over a wide range of particle sizes, but its disadvantage is the sample being analyzed. The quantity is very small, and the effect of its application in the industry is mixed. The PSI500 particle size analyzer based on the light diffraction principle of Finnish Outotec has been reported in the application of the Yongping Copper Mine Concentrator in China, but the product is not seen. Other companies promote the application. 4. Soft measurement technology Based on the characteristics of the grinding circuit, Northeastern University uses a multi-input layer neural network and genetic algorithm to propose a hybrid algorithm that uses real-coded genetic algorithm to train multi-input layer neural networks, and establishes a neural network soft-measurement of grinding granularity. The model and the effectiveness of the method were verified by field data verification and practical application in a large concentrator. (6) Pulp grade measurement Online and real-time analysis of pulp grade plays a key role in guiding production, saving chemicals, controlling product quality and improving recovery. The online and current-carrying X-ray fluorescence analyzer is a large-scale instrument for continuous process industrial process parameter analysis integrating electronic, nuclear, automatic control and precision instrument processing. This type of instrument is continuous and automatic in the production process. Multi-element composition analysis is widely used in the process analysis of metallurgy, mineral processing, chemical industry, building materials and other industries. The slurry grade analyzer has two methods: wavelength dispersive X-ray fluorescence analysis (WDXRF) and energy dispersive X-ray fluorescence analysis (EDXRF); and from the method of obtaining X-ray fluorescence, there are radioisotopes and X-rays. X-ray fluorescence analysis of two different excitation sources. The former has a radioisotope online X-ray fluorescence analyzer represented by Amdel Corporation of Australia; the latter is led by the Outokumpu Company of Finland, which uses the X-ray tube as an excitation source to produce the Curry series. Flow X-ray fluorescence analyzer. Because wavelength dispersive X-ray fluorescence analysis is superior to energy dispersive X-ray fluorescence analysis in terms of counting rate, resolution, and measurement speed. Over the years, Outokumpu has been committed to the development, development and production of the Courier family, becoming the world's leading manufacturer of current-carrying wavelength dispersive X-ray fluorescence analyzers, Courier series of grade analyzers in Fankou lead-zinc mine Application in automatic detection technology for mineral processing. Of course, Australia's Amdel, the United States, Denver, and the Swedish company Boliden have developed their own current-carrying X-ray fluorescence grade analysis systems, each with its own characteristics and widely used in countries around the world. China has supported the project of “Development of Current-Carrying X-Fluorescence Grade Analysis System†in the “Eleventh Five-Year†and “863†research projects. The project was undertaken by the Beijing Research Institute of Mining and Metallurgy and successfully developed the BOXA-type current-carrying X-ray fluorescence grade analyzer. The analyzer system includes a single sampler, multiplexer, analyzer control unit, analyzer probe and analyzer management station. 5 parts. The instrument can measure 24 slurry flow channels, 5 metal elements, measurement accuracy: high grade pulp 2% ~ 4%, low grade pulp 4% ~ 6%. A grade analysis system can be configured with up to 24 primary samplers, belonging to 4 multiplexers, and the primary sampler and multiplexer are controlled by the analyzer control unit to complete sampling and flushing according to the measurement needs. The analyzer control unit includes a human-computer interaction interface and a modular controller, and the modules are uniformly scheduled and work together. The analyzer probe consists of a high-precision high-voltage source, an X-ray tube, a spectroscopic crystal and an X-ray detector, a refrigeration system, and a temperature control system. Through the analyzer's management station, parameters such as parameter design, regression model analysis, historical data statistics, and reporting can be performed. At present, the BOXA analyzer has been promoted and applied in more than a dozen concentrating plants at home and abroad, and has achieved good application results. At the same time, the domestic Sinosteel Maanshan Mine Research Institute Co., Ltd. Dandong Oriental Measurement and Control Technology Co., Ltd. has also successfully developed an energy dispersive analyzer product using a nuclear radiation source as an excitation source. (7) Level measurement Outotec's new product, LevelSense, released in 2013, is used to detect the thickness of the flotation foam layer. It uses electrical impedance imaging technology and is probe-type. It claims that the product is more accurate and reliable for liquid level detection. Deep research. Pursiainen was commissioned by Outotec to develop a spray cleaning unit for LevelSense. Heiskanen Kari instructs students to use LevelSense to study the relationship between conductivity and process parameters. A. Nissinena proposes two simplified model algorithms for the probe ERT probe to improve the speed of the operation. Antti Nissinen et al. published an in-depth development on the use of LevelSense to measure foam size and foam loading. J Kourunen studied the gas-liquid two-phase flow imaging using the test conditions provided by Outotec et al., and at the end of the article, Numcore and Outotec were successfully applied in the three-phase flow in subsequent studies. Suzanna Ridzuan Aw et al. reviewed the application of electrical impedance imaging technology, mostly focusing on gas holdup and fluid measurement, as well as bubble rise rate and bubble size measurement. (8) Analysis of ore grade on the belt Belt ore grade analysis of the current US-Thermo ECA's most mature products, the product uses the prompt gamma neutron activation analysis (PGNAA), mainly used in coal and cement raw materials testing, Cottle, I in 2007 It introduced the series products of iron ore eighth meeting. Jia Wenbao et al. reported on the mainland China Power MJA coal online detection device, applied to a number of power plants. Gao Xiang et al. studied the application of PGNAA technology to coal analysis. Jia Wenbao et al. conducted in-depth research on the impact and correction of coal weight on PGNAA analysis. Song Qingfeng and others used PNGAA technology to detect copper- nickel ore and considered it to meet the needs of on-site inspection and control. CANGZHOU XINFENG PLASTIC CO.,LTD , https://www.xinfengplastics.com
Mineral processing automation online detection and analysis technology
In the past ten years, the development speed and popularization rate of beneficiation automation have never been seen before. The new concentrating plant has synchronized the construction of automation systems without exception. The old concentrating plant has also continuously upgraded, upgraded and upgraded through the construction of automation systems. However, the application and implementation effects are different.