news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

source: people's daily overseas edition

left: atmospheric remote sensing satellite data receiving station. map: pan xutao right: beijing's atmospheric "three monitoring" linkage precise pollution control dispatch platform. image provided by beijing ecological environment monitoring center

from "looking forward to blue sky" every day to "taking photos of blue sky" frequently, beijing citizens' sense of blue sky has been greatly improved. today, beijing's air pollution prevention and control has entered a stage of further improvement at a relatively low concentration level, and the space for further improvement is narrowing and the difficulty of improvement is increasing. under the new situation, how can the public enjoy more blue sky and white clouds?

beijing's ecological and environmental department pioneered the "supervision-monitoring-inspection" linkage model (referred to as the "three-monitoring" model) in the country, using big data, artificial intelligence and other technologies, with a new monitoring network, intelligent analysis technology, and an efficient dispatching system to support the continuous improvement of beijing's air quality.

data shows that the "three monitoring" model has intelligently mined and pushed more than 10,000 problem clues, and has played a practical role in pollution process response and normalized application.

“smart tentacles” to obtain massive amounts of data

on the streets of beijing, citizens can see a special-shaped van with various sampling tubes on the top of the vehicle and the words "atmospheric cruise monitoring" sprayed on the body. after the air is sucked into the car, it is analyzed by the mass spectrometer and data can be generated in as fast as 5 seconds to reflect the real-time concentration of vocs (volatile organic compounds) around the car. the data monitored by the cruise vehicle will be uploaded to the "three monitoring" large model system in real time to provide "food" for large model training.

data is the cornerstone and foundation of the big model, and data quality is crucial to the big model. based on the existing three-dimensional atmospheric "sky-ground" monitoring network, the "three-monitor" big model innovatively builds a multi-data perception system of "observing from the sky, patrolling on the ground, connecting data, and verifying electricity", integrating more than 200,000 intelligent sensing devices to achieve the aggregation of hundreds of millions of data every day.

satellite remote sensing can "see from the sky" and intelligently identify more than 10 types of targets such as bare land, with an identification accuracy of 90%. the uav "patrols on the ground" and measures vocs while walking, achieving "second-level response - intelligent tracing - closed-loop supervision".

"many technical means such as navigation and remote sensing are like the 'intelligent tentacles' of the 'three monitoring' large model, 'feeding' real-time data to the large model." wang xiaoju, director of the comprehensive (big data) office of beijing ecological environment monitoring center, said that the large model processes and analyzes data like a sophisticated brain to discover clues to problems in a timely manner.

"the 'three monitoring' big model is also connected to more than 50 types of multi-source databases, such as air quality monitoring and various pollution source data, and to an environmental protection professional knowledge base of more than 1 million documents such as standards, regulations, research reports, and papers." shen xiu'e, director of the beijing ecological environment monitoring center, said that these data support the "three monitoring" big model to perform intelligent analysis, prediction and forecasting, precise tracing, clue mining and other functions.

self-developed algorithms help with accurate analysis

excellent algorithms are the guarantee for efficient training of large models. focusing on the needs of refined urban management at low concentration levels, the beijing ecological environment monitoring center has built an intelligent recognition algorithm library and independently developed 26 types of problem clue mining algorithms, such as excessive vehicle emissions and asynchronous production and treatment of enterprises, to dynamically track high-value bubbles and excessive emissions.

on june 28, the "three monitoring" large model system found that a heavy-duty diesel vehicle had abnormal nitrogen oxide emissions. the system not only recorded the company and license plate number of the vehicle, but also clearly displayed the vehicle's driving trajectory through a picture. shen xiu'e introduced that the "three monitoring" large model system uses obd remote online monitoring technology to grasp the driving routes and emission conditions of 180,000 heavy-duty diesel vehicles in use in beijing, and processes hundreds of millions of data per second.

the "three monitoring" model can also monitor the electricity consumption of enterprise facilities at the minute level. through intelligent matching and analysis of the model, clues to the problem of the asynchrony between production and treatment of enterprises can be found. "using the facility electricity consumption monitoring technology, we have included more than 1,000 key enterprises that emit air pollutants and more than 6,000 monitoring points into the monitoring. once the electricity meter of the production facility of an enterprise is running but the electricity meter of the pollution control facility is not moving, it will be captured and recorded by the system." shen xiu'e said.

on june 20, the power consumption monitoring of the facilities found that the printing press and other production facilities of a company were running day and night, but the pollution control facilities were suspected to be turned on only during the day and not at night. after accurate analysis by the big model, the clues were pushed to the district-level law enforcement department for handling.

"facility electricity consumption monitoring and the big model are mutually supportive. facility electricity consumption monitoring provides professional, high-quality data input for the big model, and the professional algorithms embedded in the big model analyze the electricity consumption monitoring data, extract clues and analyze conclusions." shen xiu'e said, the "three monitoring" big model provides data support for precise and rapid response and effective scheduling of pollution control, and promotes the gradual shift of ecological and environmental governance from "flood irrigation" to "precision drip irrigation."

information integration enables efficient scheduling

based on comprehensive monitoring and perception, intelligent analysis and judgment, the "three monitoring" model has achieved efficient and precise scheduling of atmospheric environment management.

the beijing atmospheric "three monitoring" linkage precise pollution control dispatch platform clearly displays various information data. on the left side of the screen are real-time air quality and future forecast results. the middle of the screen gathers meteorological data, pollution source data, dispatch information, etc. the right side of the screen displays clues to four major categories of problems, including mobile sources, industrial sources, dust sources, and high environmental values, as well as the results of precise pollution control dispatch.

the "three monitoring" model played an important role in beijing's 2024 summer volatile organic compound control campaign. the "three monitoring" model accurately locates high-value areas of vocs, intelligently analyzes the source of pollution, and automatically transfers clues to law enforcement departments for disposal.

li xiang, captain of the first brigade of beijing ecological and environmental protection comprehensive law enforcement corps, described the process of discovering and handling a problem: "the system prompted that a high-value point of vocs appeared next to an auto repair company. this high value was likely caused by this auto repair company, so we conducted an on-site inspection."

entering the auto repair company, li xiang and other law enforcement officers came to the spray paint room for inspection and found that the activated carbon of some equipment had exceeded the replacement cycle, and there were gaps in the carbon disc, causing untreated exhaust gas to be discharged from the gaps. the law enforcement officers required the company to make immediate rectifications and handed it over to the relevant local departments for corresponding penalties in accordance with the law.

according to reports, the precise pollution control and dispatching platform realizes the visual integration of "one network for monitoring, one map for results, and one list for problems", and has achieved rapid evaluation of air quality, comprehensive analysis of the pollution situation, intelligent tracing of pollution problems, coordinated dispatch of urban areas and towns, and dynamic tracking of activity levels. it can effectively support the comprehensive analysis and supervision of major pollution sources.

after the handling, the law enforcement officers fed back the on-site verification information to the "three monitoring" large model database. the monitoring personnel conducted an on-site review and found that the vocs concentration had reached the normal level.

the "three monitoring" large model and other scientific and technological means have effectively improved the efficiency of pollution control and protected beijing's ecological environment. according to data from the beijing municipal ecological environment bureau, from january to july this year, the average concentration of fine particulate matter (pm2.5) in beijing was 33 micrograms per cubic meter, and the number of good days accumulated to 150, an increase of 7 days year-on-year. (pan xutao and hua yiran)

people's daily overseas edition (page 08, august 30, 2024)