Leaching process is the first step in zinc hydrometallurgy, which involves the complex chemical reactions for dissolving zinc bearing material in dilute sulfuric acid. Ensuring the safe running of the process is a key...Leaching process is the first step in zinc hydrometallurgy, which involves the complex chemical reactions for dissolving zinc bearing material in dilute sulfuric acid. Ensuring the safe running of the process is a key point in the operation. An expert fault diagnosis system for the leaching process was proposed, which has been implemented in a nonferrous metals smeltery. The system architecture and the diagnosis procedure were presented, and the rule models with the certainty factor were constructed based on the empirical knowledge, empirical data and statistical results on past fault countermeasures, and an expert reasoning strategy was proposed which employs the rule models and Beyes presentation and combines forward chaining and backward chaining. [展开更多
在网络安全问题中,一种分布式拒绝服务(Distributed deny of services)攻击严重威胁着现有的互联网.针对DDOS攻击基于神经网络算法的防护,因为现有算法收敛性能不高,过滤DDOS攻击包的速度过慢,无法投入大规模商业使用.本文针对这个问题...在网络安全问题中,一种分布式拒绝服务(Distributed deny of services)攻击严重威胁着现有的互联网.针对DDOS攻击基于神经网络算法的防护,因为现有算法收敛性能不高,过滤DDOS攻击包的速度过慢,无法投入大规模商业使用.本文针对这个问题,提出借助SNORT入侵检测平台,利用捕捉的网络数据包进行数据规整化,利用贝叶斯模式对正常数据和异常数据进行初步分离,使得能减少冗余训练数据对神经网络的输入,之后利用改进的反向传播神经网络进行前期数据训练,使训练产生的数据对检测模型进行优化,并且自动生成防御规则.其优势在于:1)在linux系统上实现部分改进,使得现有包过滤效率增强,在攻击目标端生效之前可进行攻击拒绝;2)使用贝叶斯模型减少重复数据和不必要数据的输入,改进的神经网络算法使得训练收敛速度加快,方便规则的重新制定学习,以防新攻击.实验表明,本文方案在一定程度上提高了原有基于神经网络防护DDOS攻击的处理速度,也能够防护若干未知DDOS攻击,训练算法的收敛速度也得到进一步提升,并且该方案能在软件层面上提升自适应抗DDOS软件的性能.展开更多
文摘Leaching process is the first step in zinc hydrometallurgy, which involves the complex chemical reactions for dissolving zinc bearing material in dilute sulfuric acid. Ensuring the safe running of the process is a key point in the operation. An expert fault diagnosis system for the leaching process was proposed, which has been implemented in a nonferrous metals smeltery. The system architecture and the diagnosis procedure were presented, and the rule models with the certainty factor were constructed based on the empirical knowledge, empirical data and statistical results on past fault countermeasures, and an expert reasoning strategy was proposed which employs the rule models and Beyes presentation and combines forward chaining and backward chaining. [
文摘在网络安全问题中,一种分布式拒绝服务(Distributed deny of services)攻击严重威胁着现有的互联网.针对DDOS攻击基于神经网络算法的防护,因为现有算法收敛性能不高,过滤DDOS攻击包的速度过慢,无法投入大规模商业使用.本文针对这个问题,提出借助SNORT入侵检测平台,利用捕捉的网络数据包进行数据规整化,利用贝叶斯模式对正常数据和异常数据进行初步分离,使得能减少冗余训练数据对神经网络的输入,之后利用改进的反向传播神经网络进行前期数据训练,使训练产生的数据对检测模型进行优化,并且自动生成防御规则.其优势在于:1)在linux系统上实现部分改进,使得现有包过滤效率增强,在攻击目标端生效之前可进行攻击拒绝;2)使用贝叶斯模型减少重复数据和不必要数据的输入,改进的神经网络算法使得训练收敛速度加快,方便规则的重新制定学习,以防新攻击.实验表明,本文方案在一定程度上提高了原有基于神经网络防护DDOS攻击的处理速度,也能够防护若干未知DDOS攻击,训练算法的收敛速度也得到进一步提升,并且该方案能在软件层面上提升自适应抗DDOS软件的性能.
