On the road to intelligence with smart PV

Find out how AI-powered photovoltaics have become the main future trend for energy.

By Xu Yingtong, President, Ascend Computing Business, Huawei & Chen Guoguang, President, Smart PV Business, Huawei

AI is playing a significant role in helping us deal with new challenges, pursue better lives, explore science, and coexist harmoniously with nature. During the COVID-19 outbreak, AI has formed a vital part of prevention and control measures, including analyzing the virus, developing vaccines, enabling intelligent temperature measurements, and providing auxiliary diagnosis.

enterprise private lines

Alongside 5G and big data, AI is an important part of China's New Infrastructure initiative to drive new technologies, new industries, and new economies. We believe that AI will spawn new technologies, products, industries, business, and business models, which in turn will transform production, distribution, exchange, and consumption.

AI's enabling power is improving value creation methods across all industries in completely new ways, and our collective mission today is leveraging this innovation to transform the world.

Since unveiling its AI strategy in October 2018, Huawei has been steadily and methodically implementing the strategy and promoting AI product development and commercialization. In 2019, with the launch of the Ascend AI processor, the all-scenario AI computing framework MindSpore, a full line up of Atlas products, and Ascend-based cloud services, Huawei completed its full-stack, full-scenario AI solution – a solution that today has been widely deployed in all industries, making AI ubiquitous.

In recent years, problems like global resource shortages and environmental degradation have intensified and we now stand at a fork in the road for our survival. The only way to deal with the global resource crisis and climate change is to develop new energy and fundamentally change the current energy supply model. Led by photovoltaics (PV), renewables are set to be the main future trend in energy. But the main challenge lies in increasing the penetration of PV and transforming it from grid parity to a high-quality mainstream power source.

AI + PV: An industry first

Digital transformation is the foundation of intelligence. And the higher the degree of digitalization, the easier it is to realize intelligence. With the development of digital IT, Huawei's Smart PV has remained at the forefront of three eras of PV development: one, the digital + PV era; two, the Internet + PV era, and three, today’s AI + PV era.

In 2014, Huawei pioneered intelligence in PV with the launch of the Smart PV solution. At the core of the solution was the string inverter. Smart PV transformed these string inverters into sub-array sensors, supporting precise information collection for each string, essentially enabling intelligent perception of PV power plants, in turn facilitating E2E digital transformation.

Between 2015 and 2018, Huawei integrated more digital technologies into PV, including wireless private network technologies, M-BUS technologies, Smart I-V Curve Diagnosis, and the intelligent O&M cloud center, and integrated PV into the agriculture and fishery industries.

Covering device, edge & cloud

In 2019, Huawei unveiled the first-ever Smart PV solution with AI. In 2020, Huawei further integrated Smart PV and its full-stack, all-scenario AI solution by creating core architecture for device-edge-cloud collaboration that will maximize the value of each PV plant and accelerate the intelligent evolution of the industry.

On the device side, Huawei has upgraded PV inverters to serve as smart PV controllers. This enables high-precision, real-time data collection, the real-time control of string-level energy yield optimization, real-time DC arc detection, and real-time response to grid-tied control. It also gives the system real-time inference, execution, and self-closed-loop control capabilities.

At the edge, Huawei has embedded an AI inference module into the sub-array controller to upgrade it with intelligence. The modules collect real-time device data and an AI model infers optimal power generation in real time, enabling optimal power generation and grid-tied control of sub-arrays.

In the cloud, Huawei has deployed an AI training and inference platform on the management system, supporting continuous AI training and optimization of the AI algorithm model without needing to modify existing devices. This enables system energy yields and potential fault diagnosis accuracy to be constantly improved and ensures device and edge inference models are promptly updated in batches for efficient collaboration.

Huawei's full-stack, all-scenario AI solutions have been used across a wide array of industries, including electric power, manufacturing, and healthcare, considerably enhancing the level of intelligence in these sectors. For instance, China Southern Power Grid has utilized the Atlas 200-based intelligent O&M and inspection system. As part of the system, AI cameras installed on power transmission towers can check power transmission lines in real time and detect foreign objects and damage. The unmanned inspection of transmission lines reduces risk to engineers from climbing up pylons and proximity to live lines, significantly improving both O&M efficiency and safety.

enterprise private lines

50-percent higher O&M efficiency 

Over the next 5 to 10 years, the world's major nations and regions will adopt renewable energy as the main power source for their electric grids. PV is set to become the mainstay of renewable energy. Ensuring the safety, reliability, and cost-efficiency of power generation assets will thus become a major challenge.

After extensive consultations with global industry experts, Huawei believes that the rapid development of emerging ICT like AI, cloud, big data, and 5G, as well as power electronics technology, will enable smart PV to create value in four major ways: lower LCOE, increase power-grid "friendliness", achieve intelligent convergence, and boost security and trustworthiness.

In the PV industry, labor costs are the main focus of cost control. PV plants have traditionally required human labor to complete a large amount of O&M tasks, which often involves high levels of risk, repetition, and precision. The COVID-19 pandemic this year has led to a shortage of workers, reigniting discussion in the industry around PV-station O&M. Having previously conducted numerous unmanned O&M trials in response to the pandemic, Huawei and its many PV customers successfully implemented unmanned O&M. Huawei's AI BOOST Smart I-V Curve Diagnosis supports one-click remote scanning of all PV strings and can automatically output a diagnostic report for a 100 MW power plant in 15 minutes, including maintenance requirements, fault diagnosis, and precise fault location for plant engineers. Unmanned inspections improve PV power station O&M efficiency by more than 50 percent.

As labor costs increase, the traditional maintenance method involving site visits by experts will be gradually phased out. AI technologies that integrate extensive expert experience will conduct diagnosis and make decisions in place of O&M specialists, while heavy, repetitive O&M workloads will be achieved by drone inspection and robot-based automatic O&M instead of human workers. As simple, repetitive, and hazardous tasks are increasingly completed by AI systems, this will reduce labor costs and improve efficiency, while enhancing speed and accuracy. For instance, drones equipped with HD or infrared cameras negate the need for manual inspections and can complete real-time failure analysis and diagnosis with much higher efficiency and accuracy. Unmanned is certain to be a future development trend in PV plants.

AI can be integrated into all aspects of the power grid. Our Smart DC System (SDS) integrates previously independent components, brackets, and inverters to form a closed-loop system of bifacial PV modules, trackers, and multi-channel MPPT smart PV controllers, which optimizes the entire DC generation system. The SDS also improves on the traditional astronomical algorithm. Its smart PV controller acts like a smart brain that can self-learn a tracking optimization algorithm and continually evolve. AI training and modeling use a neural network to adjust the trackers to the optimal angle to maximize the potential of every string in a PV station.

In Suixi, Anhui, tests showed a 1.31 percent increase in actual energy yield over 183 days. In tests on PV solar farms owned by China General Nuclear Power Group and Huanghe Hydropower, energy yield was increased by 0.5 percent to 1 percent.

Huawei's smart solar inverters integrate world-leading algorithm architecture and greater control capabilities, bringing impedance reshaping AI technology and leading grid-connected algorithms, such as dynamic damping adaptation, into the PV industry. The self-learning AI can identify the electrical features of a PV plant and automatically match the grid-connected algorithm to the power grid. Huawei's industry-leading solar inverters also support high-voltage, direct current (HVDC) scenarios, a minimum power grid short circuit ratio (SCR) of 1.5, high-penetration power without derating, a better connection to weak power grids, and fault ride-through (FTR) capability.

In the future, the grid-tied control capability of solar inverters needs to be continuously improved to support connection to weaker power grids, ensuring that PV plants can run stably without disconnecting from the power grid even in the case of low SCR. Huawei's Smart PV has helped PV stations evolve from grid adapters to grid supporters. Now, it can help them reconstruct the grid and form independent networks so they can transform PV from grid parity to a high-quality mainstream power source.

Ensuring safety is also very important. In an industry-first, Huawei has integrated AI algorithms into the arc-fault circuit interrupter (AFCI), providing more accurate arc detection and faster fault protection, to fully secure the safety of distributed PV systems.

AI BOOST's AFCI has three unique features. First, powered by the AI model, it can continuously and efficiently learn to recognize more arc features, fueled by a database of over 1 million arc features. Second, the strong computing power enables AFCI to actively identify, and analyze more than 92 arc characteristics comparison points, and identify arcs with high accuracy. This means no false positives or false negatives, delivering all-around upgraded protection. Third, it can shut down and cut off the power supply within 0.5 s, which is much quicker than the industrial standard of 2.5 s, helping to prevent fires and ensure rooftop PV plants are safe. We believe that our new AFCI standard is set to become the industry standard both in China and internationally.

AI + PV = a green smart world

In the future, AI will be as pervasive as water and electricity. In the PV industry, it will become the new main engine of intelligent transformation. 2020 marks the first year when cloud, AI, and 5G are truly integrating. We believe that we’re poised to become an intelligent society in the next 30 years. Digital technology and new energy are reshaping the world. When PV meets AI, the PV industry will break through its limits and be restructured and redefined. When AI meets PV, a greener world where all things are sensing, connected, and intelligent will come into being.