RESEARCH ON DATA COMPRESSION ALOGRITHM AND COMMUNICATION NETWORK SIMULATION IN POWER SYSTEM

Power data compression is a new research topic for power system. It becomes more and more important with the extension of power grid, the development of power information system, and the application of wide area information. Nowadays many new automatic supervisory control and relay protection devices are used in power system. On one hand, they improve the standard of automation and information of power system operation and management. On the other hand, they generate large amount of data and have increasing demand for remote communication, which brings heavy burden to the data communication and storage of power system. Data compression is an efficient method to deal with the problem. The research and application of data compression technology is of great importance for reducing the data storage burden, improving the real-time index of power communication, accelerating the power information development, as well as improving the standard of power system operation and management. Based on the achievement of former researchers, the paper manages to investigate power data compression from the view of power system, and put much attention to the characteristics of power waveform data acquired from digital sampling. After studying in detail the theory and application of data compression, the paper proposes novel methods for both lossless and lossy power data compression. Finally, with the help of advanced communication network simulation software, the paper simulates the power communication network, and analyzes the effect of data compression. The paper first analyzes the application of Huffman lossless data compression algorithm in power system, and points out that the compression ratio of Huffman algorithm is low. By studying the magnitude and probability characteristics of power waveform data, as well as the information entropy of information theory, the paper shows the application of different character codes could influence the compression ratio. Then the paper performs theoretical deduction, and validates the advantage of hexadecimal characters over decimal characters for improving compression ratio of Huffman algorithm for power waveform data. In order to greatly improve the compression ratio of lossless data compression, the paper takes the advantage of delta modulation method that is usually used for the lossy compression of voice signal, and proposes a high-order delta modulation algorithm suitable for power waveform data, and it is improved to become a lossless compression algorithm with high compression ratio. The proposed algorithm applies high-order differential and reducing operation that requires little computation. It effectively reduces the magnitude of power data, so it could reduce the number of characters and bits required for compression coding. The proposed algorithm is particularly suitable for dealing with the digital sampling data that is provided by modern power devices with high sampling frequency, and it can output high-order differential data series suitable for the combined application of Huffman compression algorithm. In the field of power data lossy compression, the paper studies the wavelet packet compression method and points out its shortcomings. The study of various entropy functions for searching the best wavelet packet tree is emphasized, since it will help make the best use of the features and advantages of wavelet packet method over wavelet method. The paper finds that the best wavelet packet tree found by the commonly used Shannon entropy has poor compression effect, and the threshold entropy is better. Based on the findings and the thought of TCM (Trellis Code Modulation), the paper tries to achieve global optimization instead of local optimization for each step of wavelet packet compression algorithm, and proposes a wavelet packet unified threshold compression method. The proposed method uses the same threshold for both steps of searching the best wavelet packet tree and reserving the necessary wavelet packet coefficients, so it ensures the best wavelet packet tree found is also the tree with least amount of wavelet packet coefficients. Thus it achieves high compression ratio with little error. Simulation shows the proposed method achieves the highest compression ratio and the error is small, comparing with the results derived from the wavelet compression algorithm with the five entropy functions provided by MATLAB. In view of the present power communication network and the requirement of wide area information communication, the paper makes use of the advanced communication network simulation software OPNET, and simulates power data communication of both local network inside substations and wide area network among substations. The communication of power quality data and relay protection data are taken as examples to analyze the communication performance of different types of communication networks, to calculate the time delay indexes, and performs quantitative analysis to the effect of data compression technology for improving the real-time property of power system communication.

Alternate abstract:

电力系统数据压缩是目前新兴的研究课题，它随着电网规模的扩大、电力信息化的发展、以及广域信息的应用而变得越来越重要。近年来，大批新型自动监控和保护装置被应用于电力系统，一方面它们提高了电力系统运行管理的自动化和信息化水平，另一方面它们产生的大量数据以及对远程通信的需求，给电力系统的数据通信和存储造成了很大负担。数据压缩是解决这一问题的有效措施。研究和应用数据压缩技术对减少数据存储的负担，提高电力通信的实时性，加快电力信息化的发展，提高电力系统运行管理水平都具有重要意义。 本文在认真总结前人研究成果的基础上，努力从电力专业角度出发来研究数据压缩问题，充分考虑数字采样后得到的电力波形数据自身特点，对数据压缩方法进行广泛深入的理论分析和应用研究，对电力数据的无损压缩和有损压缩都提出了有效的新方法。最后借助先进的通信网络仿真软件对电力通信网进行仿真，验证了采用数据压缩技术后所取得的良好效果。 本文首先分析了Huffman无损数据压缩算法在电力数据压缩中的应用现状，指出Huffman算法存在压缩比小的缺点。通过充分考虑电力波形数据的幅值和出现概率特点，结合信息论中信息熵，指出对电力波形数据采用不同进制的字符编码能够影响压缩比，推导并验证了在Huffman算法中对常见的电力波形数据采用十六进制字符编码比十进制字符编码具有更好的压缩效果。 为了进一步提高无损压缩的压缩比，本文借鉴通信理论中常用于语音信号有损压缩的增量调制法，提出了适合电力波形数据的高阶增量调制方法，并将其发展为一种压缩比很高的无损压缩算法。该算法采用高阶差分和还原运算，计算量小，能有效减小电力数据的幅值从而减少压缩编码所需字符数和比特数；特别适用于现代电力设备高采样率时采集的数字信号，而且输出的高阶差分序列有利于结合使用Huffman压缩算法。 在电力数据有损压缩领域，本文针对发展前景良好的小波包压缩方法，分析了目前研究中存在的不足，提出应当注意研究各种熵函数以搜索最优小波包树，以充分发挥小波包方法不同于小波方法的特色和优势。本文指出目前常用Shannon熵搜索到的最优小波包树的压缩效果并不好，而threshold熵的效果更好。本文进一步借鉴通信理论中格状编码调制的研究思路，力图将传统小波包压缩算法中各阶段的分别优化统一为整体优化，提出了基于threshold熵的小波包统一阈值压缩方法。该方法在搜索最优小波包树阶段和取舍小波包系数阶段采用统一的阈值，从而保证搜索到的最优小波包树也就是保留小波包系数数量最少的树，因此压缩比很高，而且误差较小。仿真实验表明：与目前利用MATLAB提供的5种熵函数实现的小波包压缩算法相比，本文方法的压缩比最高，且压缩比相同时的重构误差较小。 考虑到电力数据通信网络现状以及广域信息传输的需求，本文利用先进的通信仿真软件OPNET仿真了电力数据在变电站内的局域网通信以及变电站间的广域网通信，并以电能质量和继电保护数据的通信为例，分析不同类型通信网络的通信性能，计算数据通信的延时指标，对数据压缩技术提高电力通信实时性的效果进行了量化分析。