TD-SCDMA Base Station Fault Alarm Expert System Based On Neural Network

TD-SCDMA is one of the 3G international communication standards with our country independent intellectual property rights. It integrates CDMA, TDMA, FDMA and SDMA technology. In the meantime, it has many excellent features such as great capacity, strong anti-jamming capability, high spectrum efficiency, adaptive power adjustment and so on. For 3G mobile communication system is concerned, the working condition quality of base station will no doubt cause the most direct influence about the stability and reliability of the communication network.Traditional base station fault diagnosis usually adopts artificially homework way but artificial detection often is time-consuming, labor-consuming and low efficiency. Therefore how to position fault source and improve the efficiency and precision of fault detection is getting more and more people's attention.This paper puts forward a new idea applying the BP neural network algorithm of neural network and Rete pattern matching algorithm of expert system on TD-SCDMA base station fault alarm, which realize the intelligent base station fault alarm. TD-SCDMA base station fault alarm expert system based on neural network divide fault data into two kinds.One kind is facts which match with rules of expert system knowledge base. This part of facts is handed over to expert system inference machine. The other kind is facts which do not match with rules of expert system knowledge base. This part of facts is handed over to neural network inference machine. In the meanwhile, the new rules will be added to expert system knowledge base. The system makes use of neural network's associative memory and self-learning function to overcome the expert system knowledge acquisition bottleneck problems effectively, which can constantly expand base station fault alarm expert system's warning range.This paper use Java, MYSQL database, numerical simulation software MATLAB and expert system shell plugin Jess to develop TD-SCDMA base station fault alarm expert system based on neural network. The entire system is based on B/S architectureand Java language, so it has good cross-platform characteristics. Meanwhile, multifarious system upgrade and maintenance work is handed over to server to complete, which reduce the load of client and provide convenience for user's users.After the test in this machine and other LAN machines, the TD-SCDMA base station fault alarm expert system based on neural network achieved good running effect, the experiment results verify the effectiveness of the proposed algorithm.