This website uses cookies. By continuing to browse the site, you agree to our use of cookies. Read our privacy policy

Artificial intelligence, big data and the Internet of Things can eliminate this urban flood that harms fish, drinking water and aesthetics.

Many cities south of the Yangtze River have extensive water networks dotted with scenic lakes and lakeside restaurants and entertainment venues for tourists. The increase in urban income is accompanied by the proliferation of blue-green algae in spring, summer and autumn, destroying the aquatic ecology and discouraging tourists with unpleasant appearance and smell.

In the past, algae management relied on manual observation, manually operated unmanned aerial vehicle (UAV) inspections, and waterweed harvesters. However, these methods are not enough because the algae is blowing in the wind.

The digital foundation of Huawei's smart city approach can be applied to local ecology through drones enhanced by artificial intelligence. Calling and coordinating UAV fleet management and route planning functions make regular, mission-driven algae inspections possible.

Examining the difference between the optical properties of blue-green algae blooms and unaffected water shows that the refraction of algae is significantly increased in the infrared and near-infrared bands, but the remote sensing reflectance (Rrs) of the green band is higher because the algae contain chlorophyll- a. The most common way to distinguish between algae and normal water is to reconstruct simple features such as band variance based on data from different bands and multispectral imaging. Classical imaging segmentation methods or classification methods (such as support vector machines (SVM)) used for machine learning can be used to identify differences.

Cameras installed on drones usually use visible light images, which requires machine vision. Unlike the multispectral image, the visible light image does not cover the infrared band, and the green band is basically indistinguishable, so it is difficult to accurately identify blue-green algae. Since the drone is deployed outdoors, its detection function is also affected by factors such as shooting angle, distance, and lighting conditions. Due to strong light reflection, the color of some algal blooms may be almost the same as the surrounding water, so the sample image cannot accurately show the color change. City managers must frequently change the camera in the drone or install multiple cameras on a single drone, which will increase the recognition efficiency, but will reduce the efficiency, and is limited by the battery life of the drone.

In order to solve this problem, we must determine the best technical path for visual recognition based on visible light. Huawei Intelligent Operations Center (IOC) has big data and artificial intelligence capabilities. It uses deep neural network algorithms and multi-scale fusion networks to extract algae image data sets from various terrains from simple water surfaces and coastlines to more complex waters. Intermediate zone or coastline communities. The complex model coupled with continuous learning and training allows Huawei's IOC to detect algae more accurately and efficiently.

Figure 1: Algae floating on the lake (simple identification) 

Figure 2: Algae floating on the surface of a narrow waterway (more complex identification) 

Figure 4: Algae on the nearby river identified by a drone driven by artificial intelligence

Figure 5: Changes in the bio-optical properties (Rrs) of muddy water, algae and vanity in the water

IOC enables municipalities to deploy large-scale smart services. Its AI platform can centrally develop algorithm models and provide full life cycle management services. IOC integrates machine vision recognition capabilities into various APIs, which can be orchestrated and coordinated to realize the reuse, combination, and large-scale development of user applications. The solution minimizes manual intervention and operation and maintenance costs, and supports integrated operations. 

Video data collection: Video Cloud supports H.264 and H.265 video encoding formats, receives and stores real-time video streams from drones via the Internet, and exposes video collection and retrieval services to the AI ​​platform through APIs.

5G connection: Relying on the low latency and high bandwidth of 5G, the drone supports 4K ultra-high-definition image transmission and artificial intelligence real-time identification and warning.

AI algorithm scheduling: The AI ​​platform can call the video stream data sent by the drone, and calculate the longitude, latitude, severity and distribution of the algal bloom based on the video timestamp and flight data. It calculates the cumulative algae area discovered by drones, captures algae images and video clips, and sends different levels of alerts to the work order preprocessing and distribution system.

Application integration and message integration: The API gateway serves as a service platform to receive the algae identification data sent by the AI ​​platform. It encapsulates other service logic data, such as real-time drone location, and uses messaging middleware services to put the data in the message queue. Then it sends the message queue to the visual user interface or other service system that subscribes to the message service.

Big data analysis and processing: The deep learning module supports algae recognition algorithms. Based on data governance, the seaweed data is packaged into different themes, the data dimensional structure is defined according to different service requirements, and customized content is provided.

UAV fleet management capabilities: UAV hangars and management software provided by Huawei's partners enable fully autonomous drone scheduling and driving. The hangar supports automatic take-off, landing and charging. Flight data packets can be sent back to the IOC in real time to get through the dispatching capabilities.

Optimized flight: IOC controls the hangar, and formulates the flight route and mission plan of the fleet according to factors such as key areas, time and tasks. The drone takes off automatically, and the artificial intelligence algorithm automatically calculates the area and location of the algae and algae protein concentration. Automated incident reports and work orders can achieve timely, accurate and efficient harvesting. The International Olympic Committee regulates the process of sending information such as algae alerts and ticket closure statistics to different users in the form of text messages or printed reports.

In 2020, an IOC was deployed in a city of Jiangsu, and two drones covered 19 square kilometers of water in the city, with an average of 12 sorties per day. Without increasing personnel or ships, the system has harvested 36,000 tons of blue algae and decomposed 15.1 tons of algae mud since the system was deployed, and the efficiency has increased by more than 45%. 

Seasonal changes and local weather patterns can affect the growth of blue-green algae, but mainly industrial wastewater and domestic sewage have caused the major surge. For example, wastewater can cause too much nutrients such as nitrogen and phosphorus, which in turn encourages river basins where businesses and restaurants are concentrated to bloom throughout the year. The harvest team can clean these areas every day, but the flowers will recover the next day. Therefore, the solution is to control the source of pollution. 

In order to take full advantage of technology, water needs to be able to purify itself through the restored aquatic food chain. The discharge of industrial wastewater and domestic sewage must be controlled, and the wastewater must be treated to remove the nutrients that cause the surge of cyanobacteria. The IOC's big data platform conducts vertical and horizontal data mining, and generates blue-green algae alerts and cleanup tasks based on sewage discharge standards and sewage monitoring and corporate violation data provided by various departments. IOC can analyze companies located along the river in order to treat wastewater from the source.

For data quantification and modeling, a water pollution model can be created based on data obtained by IoT sensors on unmanned surface vessels (USV) during river navigation, such as algae protein density, pH, electrolyte, chlorophyll, and turbidity. These data can train and optimize the artificial intelligence algae visual recognition model to identify the enterprises that discharge wastewater, the components in the wastewater, the discharge location, and the raw materials and processes they use. These data can be used to study the relationship between these factors and the distribution, area and concentration of seasonal algae outbreaks, and the number of harvesting operations, and can help identify illegal discharges through hidden pipelines or leaks. This process requires traversing various combinations of possibilities, simulation-based reasoning and decision-making.

The USV can send close-range monitoring data back to the IOC through the 5G network. The drone can automatically respond to the request of the unmanned boat and follow their trajectory in the air. The deep detection and wide-area monitoring of UAV and USV can make water resources management more time-saving and intelligent.

The thermal mapping service provided by high-altitude remote sensing satellites can be used to regularly analyze the distribution of algae on the surface of large water bodies. Together with weather data, this enables target trend analysis, mission planning and route design for cruise flight and harvesting operations, thereby minimizing the time and labor required for these operations. This kind of world-earth synergy is not only suitable for cyanobacteria identification, but also for urban water resources management, ecological restoration, environmental protection, disaster prevention and other fields.

Read more about Huawei's smart operation center solutions.