Research And Development Of Real-time Tracking And Identification Technology And System Of Boiler Incoming Coal Quality

The power sector is a major consumer of coal,accounting for more than half of total annual coal consumption.The supply chain for coal has become increasingly tight and has led to price increases due to the tough economic environment and the outbreak of the epidemic.Coal consumption accounts for more than 60%of the cost of power generation in coal-fired power plants.Due to policy,economic and geographical factors,the quality of coal fed into the furnace is diversified.Resulting in a variety of coal quality,poor quality of coal combustion,long-term deviation from the design coal.This makes it difficult to adjust the combustion parameters according to the design scheme when the boiler is in operation.Therefore,the quality of incoming coal is an important parameter that directly affects the safe and economic operation of coal-fired units.The majority of coal-fired power plants in China still use manual methods of coal quality testing,which have a lag in guiding boiler operation adjustments.Therefore,an intelligent real-time coal quality tracking and identification technolo gy is needed.This will enable rapid and accurate detection of changes in coal quality when they occur.The operating parameters of the pulverising or air distribution system can then be changed in time for different coal quality characteristics,so that the combustion performance and economy of the boiler can be improved.This research work has a high practical application value.This paper builds on the innovative idea of applying electrostatic signals from pulverised coal to identify coal quality changes proposed by previous authors.It breaks away from the traditional method of incoming coal quality detection and identifies the various coal quality component parameters as a whole.The work is carried out in five areas:studying the mechanism of electrostatic method coal quality recognition,designing experiments for the determination of recognition mechanism variables,designing of a real time tracking and identification scheme for incoming coal quality,example applications and analysis and software development for incoming coal quality recognition.Firstly,the theoretical study found that there is a correspondence between the coal quality and the influence of its electrostatic signal.By analysing the factors influenced by the strength of the electrostatic signal,a mechanism for coal quality identification based on the electrostatic method is proposed.Secondly,an experimental platform based on electrostatic method for gas-solid two-phase flow was designed and built.By means of controlled variables,group experiments were conducted for the effects of four variables,namely,coal quality,flow rate,concentration and fineness,on the strength of electrostatic signals respectively.The results verified the rationality and validity of the identification mechanism.Then,a 600MW coal-fired#2 unit was used as the research object.A full-section non-contact pulverised coal measurement system was selected according to the identification mechanism and installed in the primary air duct for real-time measurement of pulverised coal.The data was cleaned and analysed for correlation,delay and other characteristics.The rationality of the identification mechanism was further verified.The training samples of different coal qualities were then fed into five machine learning classifiers for comparison experiments.The effectiveness of the electrostatic method combined with the machine learning algorithm for incoming coal quality recognition was demonstrated,with the LSTM classification recognition significantly outperforming the other models.Finally,the design and development of the electrostatic coal quality identification software was completed based on PyQt5.Its functionality allows real-time measurement data from the hardware system to be directly operated by the software for coal quality identification.It provides convenience to the user side of the operation.Finally,based on machine learning classification algorithm,an electrostatic method coal quality recognition software was developed and the interface design of the software was completed using PyQt5.The software can visualise the incoming coal quality recognition by extracting and training modelling the coal powder measurement data.