CAJ | 학술논문

为准确辨识原油管道调合过程的压力、液位等关键运行状态，文中提出了一种基于多传感器信息的状态推理辨识新方法。首先，根据传感器事先获得的物理量值与对应的实际运行状态，建立相应的隶属度曲线；在此基础上，利用传感器实时采集当前物理量，获取各传感器对状态子集的基本概率分配；然后，根据各传感器信息的相关性对其进行修正、排序及推理辨识；推理辨识过程中，对于焦元中的一致性部分采用乘性算子，对冲突部分在引起冲突的焦元间进行加性局部分配。实验结果表明，新的状态推理辨识方法可以充分利用原油管道调合过程多传感器的一致性或冲突性信息，提高状态辨识的收敛速度，减小误判率，降低决策风险，增强系统可靠性。
위준학변식원유관도조합과정적압력、액위등관건운행상태，문중제출료일충기우다전감기신식적상태추리변식신방법。수선，근거전감기사선획득적물리량치여대응적실제운행상태，건립상응적대속도곡선；재차기출상，이용전감기실시채집당전물리량，획취각전감기대상태자집적기본개솔분배；연후，근거각전감기신식적상관성대기진행수정、배서급추리변식；추리변식과정중，대우초원중적일치성부분채용승성산자，대충돌부분재인기충돌적초원간진행가성국부분배。실험결과표명，신적상태추리변식방법가이충분이용원유관도조합과정다전감기적일치성혹충돌성신식，제고상태변식적수렴속도，감소오판솔，강저결책풍험，증강계통가고성。
In order to accurately identify such key operation parameters as pressure and liquid level in the blending process of crude oil pipelines,a new method of status reasoning and identifying based on multi-sensor information is presented. In this method,first,according to the sensor values obtained previously and the actual corresponding operation status,a membership degree curve is established,on the basis of which the current real-time physical va-lues are sampled by sensors,and the basic probability assignment of status subsets is obtained. Then,the modify-ing,ordering,reasoning and identification of evidence sources are carried out according to the correlation of sensor information. Moreover,multiplicative operator is utilized for the consistency of focus elements,and plus partial al-location is conducted for the conflict part among focus elements in the reasoning process. Experimental results indi-cate that the proposed method fully takes advantage of the multi-sensor consistency or conflict information in the blending process of crude oil pipelines. As a result,it helps increase the convergence speed,decrease the misjudg-ment rate,reduce the decision-making risk and enhance the system reliability of status identification.