Research On Fault Location Model Of Telecom Service Operation Support System Based On Integrated Computing

Telecom business operation support system(BOSS) trend as loosely structured under the premise of centralized data. Since the data centralized and dimension increases, loosely structured cause user’s data is prone to the fault which is hard to locate, and witch is a huge challenge to enhance of the quality of the customer service of telecom companies. From the perspective of BOSS troubleshooting, the key of improve the quality of customer service is to positioning the number of different type of failures in System quickly and accurately.In order to meet the mission requirements of to locate a variety of faults quickly, this study presents a cascading neural network integrated structure. By dimension separation, the single neural network of single fault location single neural network is integrated to the cascading structure to achieve a single fault detection one by one in order to the fault location. Fault location of China Mobile Business Operation Support System is the object of the study, and 6000 CRBT data of China Mobile Co. is used as a sample, and the proposed model was tested and validatedFirstly, this article describes the current framework telecom business operation support system, pointing out that the current user data due to mistakes caused by loose and billing errors and other issues, explained the significance of the study. At present, according to the research system and the reliability evaluation methods and user carrying capacity and other aspects of doing induction and summarized, existing fault location and the presence of passive latency and other shortcomings. Then, we propose a neural network integrated fault localization waterfall model, discusses the model system for inter-module communication delay between modules feasibility own fault and fault localization module interface, and the model for a detailed presentation, including high-dimensional data structures needed to integrate the dimension of separation, training and integrated computing unit cell neural network. Then, using MATLAB as an experimental tool to locate the fault model was implemented for the experiment was designed and carried out the data collection business training and testing the neural network from Jiangmen Mobile BOSS database(negative samples collected from the currently known failure data). It was found that in contrast to a single neural network and its integrated structure, waterfall-type neural networks for fault location has a more substantial upgrade. Finally, the model proposed by the system implementation.