Research On Granular Computing-Based SDG Fault Diagnosis Of Steam Boiler System

Signed Directed Graph (SDG) model as an important branch of qualitative models can be used to express complicated cause-effect relationship, and has the capacity of containing large-scale potential information. When used for fault diagnosis, it not only has good completeness, strong adaptability and excellent robustness, but also can provide fault propagation paths and fault explanation. In recent years, SDG fault diagnosis gets the extensive attention of scholars, develops rapidly, and has achieved good application effect in petroleum and chemical industry. But the fault diagnodis method based on SDG model inevitably has the disadvantage of multi-meanings reasoning in qualitative method, which can result in low resolution, and there is other problems to be solved:relying on the model too much, slow diagnosis rate of large systems, bad real-time and so on.Granular Computing (GrC) which is proposed by professor T. Y. Lin in 1996, is a new way of human thinking simulating and solving complex problems in the current areas of computer intelligence, and it is an effective tool for solving complex problem, massive data mining, and fuzzy information processing. Attribute reduction algorithms in GrC can reduce redundant attributes under the premise of ensuring the ability of not changing classification, and derive the minimal attribute set. And search reasoning method can largely reduce even avoid the conflict probability of results, thus improve the reasoning resolution.In order to solve the problems exsit in SDG fault diagnosis, the attribute reduction and search reasoning of GrC is introduced in SDG fault diagnosis in this paper, and the following work is done:(1) To improve the diagnosis real-time, avoid multiple reasoning for the same fault mode and the possible combination explosion problem, a new method of SDG fault diagnosis based on decision rule is proposed, and then the deaerator system of power plant is taken as an example, we establish its SDG model by experience-based method, acquire its fault diagnosis decision table, and use the proposed method and the basic SDG fault diagnosis method respectively for diagnosis. Results show the superiority of this method.(2) A new attribute reduction algorithm based on relative granularity is proposed to remove redundant attributes in decision table, and results show that this algorithm not only has a relatively low time complexity, but also suit for both consistent and inconsistent decision tables, furthermore, it is applied in reduction of deaerator system's fault diagnosis decision table.(3) A reaserch resonning algorithm based on similarity is proposed, which obtain the most possible fault source by computing the most similarity. This method can improve resolution by effectively reducing or even avoiding conflict probability of reasoning results. And then we apply it in fault decision-making reasoning for the deaerator system.(4) A SDG fault diagnosis method based on Granular computing is proposed, and the 65t/h steam boiler system is taken as an example. Build up its SDG model, acquire its SDG fault diagnosis decision table, generate the granule base after reduction, and then search reasoning method is used to find the fault source by computing the most similarity. The simulation results illustrate this method is valid, and the results show that this method greatly improves the diagnosis rate and resolution, under this premise of reserving the advantages in SDG fault diagnosis.