China is forging ahead to rekindle the flame of passion and dream. The world's eagerly awaited Beijing Winter Olympic Games open on February 4, 2022. Looking at this winter Olympic Games, many new technologies applied in athletes’ participation process, audiences’ watching experience and epidemic prevention means, have made amazing appearances and are deeply integrated with the events, adding unique digital intelligence charm to the Winter Olympic Games. From now on, we will jointly explore and record the story of intelligent technology workers helping the Beijing Winter Olympics through the column of CAAI & Winter Olympics.

November 18, 2021, Chongli, Hebei. A burst of cheers broke out on the temporary stand of Yunding ski resort, the core area of the Winter Olympic Games. In the next two days, the group of about 40 people applauded, jumped and cheered in the cold wind of more than ten degrees below zero from time to time, making a huge "noise". This is the research group of the "science and technology Winter Olympics" special project "key technology of intelligent IOT monitoring and early warning of temporary facilities under severe cold conditions" to test the safety of temporary buildings on site.

China will offer a simple, safe and wonderful Olympic Games to the world. The safety of the Winter Olympic Games and the Winter Paralympic Games cannot be guaranteed without intelligent technology. Technologies such as intelligent building, intelligent transportation, intelligent broadcasting and intelligent equipment have appeared in the Beijing Winter Olympic Games. Even the temporary buildings of the Winter Olympic are adopted visual intelligent IOT operation and maintenance platform to provide security.

"These temporary buildings and facilities are currently using intelligent IOT monitoring technology." Guo Maozu, member of the Standing Committee of CAAI Machine Learning Special Committee and Dean of the school of Electrical and Information Engineering of Beijing Jianzhu University, said.

Compared with the Winter Olympic Games in Turin, Pingchang and Sochi, the Beijing 2022 Winter Olympic Games and winter Paralympic Games have complex terrain, more spectators, and larger quantities and scale of temporary facilities. Around the venues, stands, toilets and broadcasting towers are mostly temporary buildings. In severe cold and strong wind conditions, ensuring the safety of temporary buildings is naturally a top priority.

Previously, China has special sensors to monitor the inclination and vibration of temporary facilities under severe cold conditions. However, there are still urgent problems to be solved in the aspects of structural safety of temporary facilities under the severe cold conditions of the Winter Olympic Games and the influence of large passenger flow, spectator induced stand vibration, abnormal aggregation, escape and event source reasoning based on artificial intelligence method.

In order to solve the above problems, from the beginning of 2020, Guo Maozu led the team to cooperate with Zhongbei University, Chongqing University, Dalian University of technology and Shenyang University of architecture to "Customize" the temporary building security system for the Winter Olympic Games.

They went to Yanqing competition area and Zhangjiakou competition area for repeated investigation, installed sensors such as strain, displacement, inclination and acceleration for temporary photography platforms and temporary stands of ski resorts, and monitored and evaluated the safety of temporary facilities in real time.

The research group focuses on the realization of the functions of intelligent monitoring, safety operation and maintenance and early warning of temporary facilities under severe cold conditions, and develops the status information monitoring sensor of temporary facilities; study the rapid identification technology of temporary facility load based on data fusion and the identification and location technology of structural hazard sources; study the risk prediction technology of temporary facilities failure caused by abnormal crowd activities; build an intelligent IOT operation and maintenance platform for temporary facilities based on big data.

"Severe cold conditions directly affect the monitoring accuracy and standby time of conventional sensors." Guo Maozu said, "If we don't develop special sensors during the competition between Yanqing and Chongli, we can't guarantee the monitoring accuracy and support the continuous monitoring during the competition."

During large-scale sports events, factors such as large passenger flow and audience induction will also bring potential safety hazards to the structure of temporary facilities. Under the condition of large passenger flow, the load distribution of temporary facilities is uneven, which is easy to form local overload. The audience's collective action, abnormal gathering, abnormal escape and other activities are also easy to form local load reinforcement. If the abnormal load and structural damage of temporary facilities can be identified in time, it will have a decisive impact on avoiding major accidents.

Based on these special needs, the team developed special sensors suitable for the severe cold conditions of the Winter Olympic Games, and used grid technology to realize the load identification and structural damage identification of temporary facilities; the emergency identification system caused by the failure of temporary facilities is developed, and the crowd spatiotemporal activity detection method based on the combination of deep learning and optical flow method is used to identify and locate the abnormal aggregation, confusion and abnormal interaction with temporary facilities. Using the self-supervised spatiotemporal trajectory representation learning method, the abnormal activity trajectory and abnormal pattern are identified, and the deep neural network combining spatiotemporal and external characteristics is used to predict the activity trajectory of the population.

From November 18 to 20, 2021, the team installed two groups of four special sensors at the key positions of the two stands of Yunding ski resort in Chongli competition area to collect stand status data. At the same time, laser scanning equipment is used to map and model the temporary facilities such as two stands, Olympic Village restaurant, two-story shed house of cross-country project and stand of cross-country project, and analyze the status data of these temporary facilities to ensure the safety performance of these temporary facilities.

"The results have achieved good results in the relevant snow fields, snow trails and race end areas." Tian Le, one of the members of the team and the school of electrical and information engineering of Beijing Jianzhu University, said, "the test effect has been highly recognized by the construction party."

In Guo Maozu's view, as one of the many scientific and technological applications to escort the safety of the Winter Olympic Games, the rapid identification of temporary facility loads and the identification and positioning of structural hazard sources based on data fusion, and the prediction technology of facility failure risk caused by abnormal crowd activities based on machine learning not only solve the urgent needs of the operation and maintenance management of temporary facilities in the competition area of the Winter Olympic Games and the winter Paralympic Games, it can also be extended to the intelligent operation and maintenance of similar stadiums and temporary facilities for large-scale activities, which has good social and economic benefits.