Research On Key Technology Of The CCS Leakage Warning Based On Spatio-temporal Analysis

Carbon Capture and Storage（CCS） has been internationally recognized as a promising strategy to lower carbon dioxide emissions, but the potential leakage is the main risk faced by CCS. Developing the CCS leakage pre-warning system is very important to ensure the safety of the CCS project. The difficulties of CCS leakage pre-warning are the various leakage routes, the warning signs identification, and the leakage simulation which is affected by many factors, and so on. So this dissertation uses theoretical research, experimental simulation, numerical analysis, system development to study the key technologies of the CCS early warning system including the saptio-temporal distribution characteristics of the CO₂ leakage, the diffusion model, the warning signs identification, the source estimation and diffusion simulation based on the GIS technology, saptio-temporal analysis technology, swarm intelligence algorithm and so on. And finally we design and develop the CCS Leakage Pre-warning System（CCS-LPS） on the basis of the above research.The main achievements and conclusions are as follow:1. It developed the CO₂ diffusion model and analyzed the saptio-temporal distribution characteristics of CO₂. Firstly, several indoor experiments were done to simulate different flows of CO₂ leakage flux（different leakage rates and wind speeds）. Then the measured data were used to contrast with the existed models（Gaussian, Plate, Box） to determine available CO₂ diffusion model; and then an outdoor leakage experiment was done to exlpore the response characteristics of each sensor and saptio-temporal distribution feature of CO₂ concentration. The analysis results are to underpin for the recognition of the CO₂ leakage warning sign.2. It developed the anomaly detection model of CO₂ data streams based on fuzzy clustering. In view of the traditional anomaly detection algorithms that cannot identify the anamoly caused by CO₂ leakage, we proposed saptio-temporal anomaly detection over CO₂ data streams based on fuzzy clustering considering the saptio-temporal distribution characteristic of CO₂ diffusion. Firstly, the adaptive threshold outlier was detected by 3 σ rules; Secondly, the characteristic value of the sliding window to be detected was extracted, and then a saptio-temporal matrix between neighbor nodes in specified interval was built. It analysed the saptio-temporal correlation of adjacent nodes characteristic value based on fuzzy clustering, and then classified the results. The algorithm identifies the abnormal leakage probability according to the results of the classification. Finally, the selection of parameters of the algorithm was analyzed and the algorithm was verified by factual observation data. The results showed that the algorithm has a high detection rate and a low false rate.3. It presented an off-line CO₂ spatial outliers detection algorithm based on spatial local outlier factor（SLOF）. Considering the limitations of the traditional static threshold detection, the SLOF algorithm reflected the spatial characteristics of CO₂ monitoring data obtained in the carbon capture and storage（CCS） project. The K-Nearest Neighbour（KNN） graph was mapped by using the latitude and longitude information of each monitoring points to identify the spatial neighbourhood. Then SLOF was adopted to calculate the outlier degrees of the monitoring points and the 3σ rule was also used to identify the spatial outlier. Finally, it analysed the selection of K value and chose the suitable one. The results showed that the proposed algorithm could detect the local outliers and had a higher detection precision.4. It put forward an off-line CO₂ spatial outliers detection algorithm based on geo-statistic theory. In view of the traditional spatial outliers detection algorithm ignoring the application of spatial autocorrelation theory, it introduced some geo-statistical theories to detect spatial outlier. Firstly, the algorithm used trend analysis to identify global outliers. Secondly, it used Delaunay triangulation neighborhood relationship building the spatial neighborhood relationship, and then used the average neighborhood node’s value to replace the global outliers’ s value. Finally, it used the local Moran ’I to measure the spatial anamoly. The simulation results showed that the method had a high detection rate and a low false alarm rate, and the running time of the algorithm was not affected by the number of nodes which means it is suitable for surface domain which has vast nodes.5. It developed the CO₂ source estimation model based on the improved PSO algorithm. Taking the time and accuracy requirements of the CO₂ leakage warning system into consideration, it applied the particle swarm optimization（PSO） to calculate the leakage location and rates. Because the PSO is easy to fall into local minima and the later iteration speed is slow, it put forward the adaptive PSO algorithm based on diversity measure. Firstly, according to the feature of PSO iteration, it combined the similarity measure function and recent Rth fitness variance to construct the diversity measure, which divided the iteration course into three phases. Secondly, it proposed several improved strategies to improve the algorithm performance in each phase. Finally, the algorithm was used to estimate the leakage location and rate in the CO₂ diffusion model（plate）, and the results showed it was appropriate for the source detection of CO₂.6. It developed the CO₂ leakage pre-warning system based on GIS. On the basis of the developed key technologies of early warning, the system usesd.NET technology, ArcEngine component programming technology, and SQL Server database management technology to build the leakage warning system based on C/S architecture. The establishment of the system, which realized the monitoring platform of data, the automation of forecast warning and the visualization of diffusion simulation, is not only helpful to track the trend of all kinds of monitoring data timely, accurately, scientifically, and comprehensively, but also provides decision-making and technical support for CCS pre-warning mannagement.