密闭舱室突发污染浓度动态预测与源项辨识

潜艇、载人航天器等密闭微环境随着人员停留时间的延长，其舱室空气污染问题已成为危害工作人员生命安全的主要因素，因此迫切需要开展快速准确的污染浓度预测以及对突发不确定污染源辨识的技术研究，并提高密闭环境主动应对突发污染的能力。对舱室污染浓度进行动态预测和污染源项强度辨识是实现舱室空气质量实时预测的关键。建立了集总污染源概念，提出了联合使用卡尔曼滤波和最小二乘算法的舱室突发污染辨识与浓度预测方法，并与建立的变结构污染浓度模型相结合，同时完成了集总污染源散发强度的动态辨识和污染浓度状态预测。另外，在突发污染源定位方面开展了前期的探讨研究工作，建立了一种新的多维浓度离散随机模型，并提出了基于多假设特征匹配的突发污染源定位方法研究。通过匹配观测数据序列与单参数（源位置）多假设获得的传感器处浓度响应序列特征来实现源项定位及散发时间估计，可初步确定源散发强度。

Prediction and Identification of Sudden Pollution Source

Such as submarines,manned spacecraft and other closed micro-environment,cabin air pollu tion has become a hazard to the safety of staff with the residence time extend.There is an urgent need for fast and accurate prediction of pollution concentration and location identification of a sudden source to im prove the closed environment active control ability for unexpected pollution.Dynamic cabin concentration prediction and pollution sources identified are a key to achieve real-time air quality forecast.A concept of lumped source and a variable structure concentration model were built to realize concentration prediction together using Kalman filtering and least-squares algorithm.In addition,a source location method was studied because it is a key link for source identification.The contaminant source location method based on multi-hypothesis source position was established and attempt to solve the source location problems.This method realizes source identification by comparing the similarity between the sensor-measured concentra tion distribution and the multiple hypothetical concentration distributions calculated at the monitoring point based on multi-hypothesis source position.The proposed method is capable of identifying a source po sition,estimating its initial emission time and approximate strength.