The Research And Application Of Incremental Updating Data Mining In Power Plant Operation Optimization

Power plant operation optimization aimed at reducing coal consumption by adjusting unit to optimal state. And with the development of database and computer technique, DCS is widely used in power system and large amount of data are accumulated by power enterprise. These data contained a great deal of information and power system can get rule and knowledge from it to optimize operation. Data mining can finish this task. Data mining is a powerful tool to get valuable information to help improving enterprise decision-making, finding exceptional circs and forecasting future mode, etc.According to the application of data mining in power system, this paper analyzed the following questions in detail:1. The paper induced incremental updating data mining technique into power plant operation optimization and proposed overall structure of optimal target value based on incremental updating data mining. The paper indicated that the optimal targets values were not invariable and they varied with different unit state. Re-mining all these data again was time wasting, so this paper proposed incremental updating data mining technique to make full use of primary mining results to avoiding re-mining all the data. This method can improve mining efficiency greatly.2. The paper proposed advanced incremental updating quantity association rule mining algorithm to maintain of optimal target value when the database changed. Subjection function was introduced into the mining process and it was indicated that the selection of subjection function should according to actual instance. Experiments showed that this algorithm could improve mining efficiency greatly.3. The paper analyzed the relativity between power plant data and veracity of the data. Data is the basis of data mining. So the data must be tested before mining. Power plant data is large and miscellaneous. Through data checking, data quality can be improved. Relativity exists between power plant data widely, and correlations analysis can help to wipe off outlying and redundant attribute before mining. All these can help to improve the efficiency and quality of mining results.4. Partial correlation analysis was proposed to find the relations between different parameters. The relation between power plant parameters is very complicated, and partial correlation analysis is more powerful in finding real relationship compared with simple correlation analysis.5. The paper analyzed the effect of oxygen content on unit efficiency and determined optimal oxygen contens at different unit states by increment updating data mining algorithm. This method had higher efficiency and got optimal oxygen content on the basis of actual equipment. So mining results is reachable in practice.