摘要
当前信息安全态势评估方法存在评估结果与实际值拟合度低、容错性能差、能耗高的问题,提出基于神经网络的变电站告警信息安全态势智能评估方法。利用消除模糊区域,扩大神经网络吸引范围,实现神经网络容错性分析。采用布谷鸟算法求取构建神经网络的最优权值和阈值,通过浮点数编码方式对布谷鸟进行编码,以神经网络的总误差值越小越好为依据,设计适应度函数,基于适应度函数的确定来设置布谷鸟种群规模和发现概率。通过编码和参数设定,初始化布谷鸟算法,根据迭代寻优完成最优权值和阈值求解。将历史告警信息安全态势值序列为当作输入数据输入至利用最佳权值和阈值构建的且已经训练好的神经网络,并将神经网络容错性分析结果引入信息安全态势评估中,实现变电站告警信息安全态势智能评估。实验结果表明,该方法评估准确性和容错性好、评估能耗低。
The current information security situation assessment method has the problems of low fitness,low fault tolerance and high energy consumption. A neural network-based intelligent assessment method for substation alarm information security situation is proposed. By eliminating the fuzzy region,the neural network attracting range is expanded,and the neural network fault tolerance analysis is realized. The cuckoo algorithm is used to obtain the optimal weight and threshold of the neural network. The cuckoo is coded by floating-point number coding. The fitness function is designed based on the fitness function based on the smaller the total error value of the neural network. The function is determined to set the cuckoo population size and probability of discovery. Through the coding and parameter setting,the cuckoo algorithm is initialized,and the optimal weight and threshold solution are completed according to the iterative optimization. The historical alarm information security situation value sequence is input as input data to the trained neural network constructed with the optimal weight and threshold,and the neural network fault tolerance analysis result is introduced into the information security situation assessment to realize the substation alarm. Intelligent assessment of information security posture. The experimental results show that the method has good accuracy and fault tolerance,and low energy consumption.
作者
肖艳炜
XIAO Yanwei(State Network Zhejiang Electric Power Co.Ltd,Hangzhou 31007,China)
出处
《自动化与仪器仪表》
2019年第1期148-151,共4页
Automation & Instrumentation
关键词
神经网络
变电站
告警信息
安全态势
智能评估
neural networks
substation
alarm information
security situation
intelligent evaluation
