| In China,the data science and technology,represented by big data,cloud platform and knowledge discovery,has been widely concerned by people,under the new situation of promoting the ten year programme of "Made in China 2025",enhancing the innovation and productivity of the real economy.The data science and technology involve multiple subject areas such as information,geography,biology,energy,medicine,sociology and so on.The informatization of coal preparation plant has also made great progress in china.The network environment of office,control and monitoring,have established successively.The automation control has realized basically,and established a large number of information management system.With the development of information technology,the following problems are also exposed.Such as numerous information systems exist independently,the intelligence level of the applications is very low,and the production process has also not been realized intelligent.Aiming at the information island in the development of informationization in coal preparation plant and different demand side decentralized data acquisition,in this paper,the information of coal preparation is classified according to the working function of each department,and gives a brief description of the business and basic data for each management department.With the example of coal preparation equipment,this paper introduces the classification standard and coding rules.The specific methods of basic data acquisition and exchange in coal preparation plant are studied,and then the process variables in the PLC controller and the data information in the different relational databases are collected.The real-time storage rules and the historical storage rules are established for the collected information,and a unified data collection method is realized.A unified data access interface is provided for different basic information users by using Web Service technology,to provide support of data activation and knowledge discovery for coal mining researchers and production managers.The core of the information fusion and activation is to classify,sort,gather and regression the data of the data warehouse through the method of data mining,mining the potential knowledge,and feedback valuable information.In this paper,the knowledge activation system of coal preparation plant is divided into three main areas: intelligent monitoring,intelligent management(production analysis and decision)and intelligent control(process simulation and optimization),and the detection of high risk area for production line based on image recognition,intelligent control of dense medium and flotation in coal preparation process,optimization of product structure and production cost in management and other contents are involved.The information fusion and activation of mining group and coal industry are mainly based on higher level production,planning and management decision support.In the application of knowledge activation system,the data should be fully understood to achieve unified planning,design and implementation,pay attention to the construction of the standard,and reserve related personnel technology.The second part of the paper introduces the specific process of data fusion and activation combined with the production process practical matter of dense medium coal separation.The production system of Tang Kou coal preparation plant was taken as the research object to study the situation of the circulatory medium density given hysteresis.The paper chooses the raw coal and the product for a period of time as the experimental object to analyze the composition of raw coal ash and raw coal density.The results show that the change of ash content in raw coal has a significant linear correlation with the yield and ash content of +1.8kg/L density,and the linear relationship between the cumulative yield of-1.3—-1.8kg/L density is better than the accumulated ash content at different densities.It shows that,to a certain extent,the float-sink compose of raw coal can be characterized by the raw coal ash,that is,there is a significant correlation between the raw coal ash,clean coal ash and circulating medium density.Combined with the results of experimental analysis,the related data of the gravity separation in Tang Kou coal preparation plant is collected and stored,involving belt scales,ash content and density meter and other measuring instrument.The actual production data were analyzedaccording to the actual situation,and the missing points in the acquisition process are completed by means of the average value of the adjacent points.The abnormal value of belt weighing for raw coal,clean coal and middings coal is handled to zero.At the same time,considering the problems of the ash meter in the case of no production still has data,the raw coal ash content is cleaned.In view of the large error of the ash gauge in the Tangkou coal preparation plant,,the operational principle of ash meter was analyzed in this paper.The result of quick ash of coal quality assay was compared with the on-line ash data,according to the results of the correlation analysis of the data,it is shown that the error of ash detection in coal preparation plant is highly correlated with the coal flow rate on the belt scale.The different curve models were established by the on-line ash content,belt scale and the assay ash to perform a fitting regression.It shows that the cubic curve model has a good fitting effect where the RMSE is 0.3335.The LSSVM algorithm is used to re-train and predict the online data,and the predicted RMSE is 0.1747,which is better than the gray value predicted by the curve regression model.Therefore,it is recommended that the LSSVM algorithm can be used to adjust the ash content in real time in the actual production process,which is combined with the scale of the belt and the assay ash.After the completion of the data preparation,the belt weighing,and scale of ash meter and density of circulating medium are selected as input matrix,the Microsoft neural network algorithm,Microsoft logic regression algorithm and Microsoft decision tree algorithm are used to construct the data mining model.From the results of mining and mining accuracy,Microsoft neural network algorithm is more suitable for the prediction of circulating medium density.When the different input is missing,the prediction accuracy of algorithm changes not very large and the root mean square error is stable at 0.02.Further research shows that the data matrix for training is not the more the better,the more unstable production data,or the more data that can not reflect the current production situation,have great influence on the accuracy of the results,it requires the application of the process should pay more attention to the timeliness of data,and ensure the data rolling update timely.Moreover,the production data of coal preparation plant has certain time characteristics.The online data related to the product has a fluctuating delay compared with the data of raw coal quantity and raw coal ash content.It can improve the prediction accuracy of data mining results which the input training data matrix after phase space reconstruction.After the data mining model is completed,the mining results need to be embedded in different application systems of the coal preparation plant to realize industrial applications,closed-loop analysis The related applications of information fusion and activation are demonstrated at the end of this paper,involved the application of data acquisition,the web platform for data acquisition and exchange,data acquisition terminal data acquisition B/S architecture exchange Web platform,the control application based on C/S architecture and the mobile application for coal preparation plant based on Android. |
PDF Full Text Request