Research On Early Warning And Diagnosis System For Energy Efficiency Status Of Coal-fired Units

With the diversified development and fierce competition in the power generation industry,the coal-fired power unit has entered a new development trend.People have put forward higher requirements for coal-fired power plants,that is,while undertaking the task of deep peak shaving,it is also necessary to maintain low-energy,low-emission and high-efficiency operating conditions.Therefore,relying on real-time monitoring to identify the early signs of abnormal patterns,predicting the development trend of the energy efficiency status of the unit in advance,implementing preventive maintenance,and obtaining more time for the operation and maintenance personnel to do the corresponding treatment measures,so that the unit always keeps a good energy efficiency status,which is of great significance to the healthy and sustainable development of coal-fired power plants.This paper mainly studies the abnormal warning and diagnosis technology of energy efficiency status of coal-fired units.Based on the comprehensive energy efficiency index of the unit,the energy efficiency status representation quantity is proposed,and the system function realization process and method are designed.This paper mainly analyzed and teased out the energy efficiency status abnormal pattern of coal-fired units,clarified the relationship between each parameter and the abnormal pattern,and constructed the energy efficiency abnormal pattern tree;performed single classification on abnormal patterns,and an initial recognition model of the energy efficiency abnormal pattern of was constructed by one-class support vector machine algorithm.Based on the time series prediction theory,predicting status characteristic parameters,and using the predicted value as the input vector of the energy efficiency status prediction model.Using the least square support vector mechanine to construct the energy efficiency status prediction model,and using the particle swarm optimization algorithm to optimize the model parameters to achieve higher prediction accuracy.Fitting the forecast curve of energy efficiency status to determine the future energy efficiency status;Based on the abnormal pattern tree,combining the mechanism analysis with domain knowledge forms the network reasoning rules to realize the final diagnosis decision of the energy efficiency abnormal pattern.In this paper,a energy efficiency abnormal status early warning and diagnosis system is studied for a million ultra-supercritical primary intermediate reheat coal-fired units.By using the actual historical operation data of the unit to carry out the early warning and diagnosis functions,the results show that the above method can accurately predict the energy efficiency status change of the unit,find the abnormal pattern in advance,and provide relevant treatment measures.