| The fault wave recording application in Power Grid reforms the former protection information system to isolate record wave part and is established as an independent record wave system. The data collecting equipment is high-speed, multi-channel, continuous and multi-terminal equipment whose data annually has reached TB-level data. It is challeng work for the new system will provide massive record data's storage, efficient data's retrieval and distributed computing for upper layer's applications, such as the relay protection. Meanwhile the faulty recorder integration application is a typical cyber-physical system in power grid which integrates collecting transient data, massive data storage and retrieval. And data processing spreads different security zones because physical system of power grid in China is close relation with cyber system. A lot of security accidents showed that the network technology and modern information technology bring many negative effects to power grid because deeply coupling between physical system and cyber system.This article is based on faulty recorder integration application.1 study deeply common transient data format (IEEE COMTRADE) and provide some methods to solve problems such as the massive high-speed distributed storage, efficient retrieval on NoSQL, distributed processing based on Hadoop and network security assessment. My works are as follows:(1)For the sake of the requirement of effectively I/O and fault-tolerance with massive COMTRADE data, we present a massive data collection and storage model based on Lustre file system. It connects collection equipments and combines a huge number of small size COMTRADE files to big size COMTRADE files and design bonding NIC which can improve I/O efficiency obviously. Fault tolerance algorithm provides consistency and reduces errors.(2)We provide an index framework for non-structural massive COMTRADE data encapsulated by compressed WAH bitmap which solves the traditional indexing defect, that is latency and delayed response. Massive COMTRADE data indexing makes sense for improving performance and using compression-word aligned bitmap indexes for requirement of multi-dimensional data retrieval. Two Phase Commit Protocol (2PC) provide consistency among index data, metadata and raw data.(3)We build storage systems and distributed fault collection system based on Hadoop distributed computing framework. Distributed fault data's collection works assign to different nodes for computing massive COMTRADE data. We rebuild a new algorithm to decompose and distribute different computing nodes to achieve fast computation. To combine small COMTRADEs to a big file, this method can speed up storage efficiency. We build Web platform which include interfaces, such as storage, index, download, distributed computing data and waving picture's showing.(4)We study assessment method of power grid's network topology and information vulnerabilities. When information security vulnerability is exploited, vulnerability's accumulated risk has close relationship with the dependency relation between nodes. Triple dependency rates between the nodes will be taken into account. The calculation of the transformation matrix which will be computed in different level requirements should be completed.(5)We provide a trust model to assess security in power grid information networks that computates trust of node when vulnerabilities are exploited under intrusion tolerance environment. This trust model based on evidences of data link layer is most commonly trust model in IP distributed information network of power grid. It is true that the evidences have continuous and stable characteristics and evidences can't be obtained from IP layer or wouldn't be trusted because of malicious attacks. The method evaluates trust relationship of nodes in the distributed IP network by defining function of link state quality evaluation and function of trust attitude with the continuous link layer evidences. |
PDF Full Text Request