The answer to that often comes down to data.
AI technology improves just like any technology does, so today’s limitation may be tomorrow’s breakthrough, but current AI technologies are very good at classification and translation (interpretation).
Whether you are talking about natural intelligence or artificial intelligence, nothing is intelligent that cannot learn.
Humans are born with some level of innate intelligence, and we can build on that intelligence through learning.
The only innate intelligence machines have is what we give them, and common sense is not currently on that very short list of items.
What we do provide, though, is the ability to examine examples, and create machine learning models based on the inputs and desired outputs.
For supervised learning, we provide the AI with examples.
Unsupervised learning is where you provide inputs, but not labels, and let the machine infer qualities.
This type of learning can be useful for clustering data, where data is grouped according to how similar it is to its neighbours and dissimilar to everything else.
Once the data is clustered, you can use different techniques to explore that data and look for patterns.
Reinforcement learning is where the machine makes a decision and is given a reward or retribution depending on whether the decision was a good one.
You could use reinforcement learning to teach a machine to play chess or navigate an obstacle course such as a demand-side management or power generation scheme.
Deep learning is a subset of machine learning that applies to neural networks.
Deep learning networks refer to the arrangements of the nodes, and that could definitely be the next generation of energy-related AI applicationWhen choosing sources of data for training your machine models, it is important to sample data that is representative of the data that will be encountered in production.
One challenge that can occur is when training data will not predict future input data.
This has to be addressed soon in the process.
Not necessarily that you need all the answers, but rather that you identified a process to take care of the situation.
Because garbage in equals garbage out, no AI solution can give good results from bad data.
What you can do if some of your data is bad, however, is to identify and keep your usable data, and collect (or build) new data that can be used in your solution.
You should be considering the amount, quality, and sensitivity of the data you have to work with.
The integration of Renewable Energies, easier said than doneThis is becoming more and more complex for the grid operator, but also for the smaller facilities or communities.
It now involves integrating renewable energy production assets into self-consumption (solar thermal and/or photovoltaic, production of heat from biomass or biogas, geothermal energy, …), to enable manufacturers or communities to reduce their carbon footprint, to diversify their energy mix and reduce their dependence on fossil fuels, which should allow better control of energy-related budgets over the long term.
It is, in any case, the lever to activate in priority: “the unspent energy costs nothing and does not pollute”.
On the other hand, even if there are some good successes or more specifically attractive sectors, the deployment of renewable energies and recovery technologies is still too little advanced in the industry.
There are many brakes to overcome: significant investment costs, competition current with fossil fuels and cheap electricity, the fear of operational risks associated with these innovative technologies, the lack of technical knowledge or operational capabilities, etc.
And this is where AI and data science can help greatly.
Many positive experiences can be analysed and can be considered as examples to follow in the “learning” phase.
That would aim to provide the industry with a light on the possibilities offered in the short term through recovery technologies and renewable energy, focusing on major sectors and a set of technologies.
For example, we can think of the optimization of a decentralized power generation (solar PV or wind turbine) with an intelligent energy storage system (IES).
In this scenario, if the next day’s weather forecast is communicated to the AI, the storage capacity can be prepared according to the expected state of the network.
The AI can decide to unload the storage unit (hybrid energy storage, for example) overnight so that a maximum of current can be stored there the next day.
Thanks to this control function, it would also be possible to know the status on the higher voltage grid levels.
The distribution side (Lower voltage grid) can then help maintain the voltage on the higher grid levels (and even at transmission level with a sufficient amount of capacity involved).
In surplus to that, the electricity produced is either consumed directly (priority 1) or injected into the grid or stored temporarily according to the state of the grid.
In case of local voltage problems, the locally stored electricity (battery) can be fed into the grid.
In a wider scenario, the energy provider can actively control the intelligent modules based on certain signals (weather forecast, balance group, etc.
That way, the IES positively influences the maintenance of local network voltage and, more generally, the security of supply (local and decentralized system services).
It is also an ideal platform for energy suppliers in terms of customer loyalty and the development of new products (services, assignment of contracts, etc.
Photo by Stephen Dawson on UnsplashState of play: a smart dashboard of innovative technologies, integrated and tailored to the needsFirst, alongside conventional solutions there are numerous relatively mature innovative technologies to produce and self-consume energy, be it heat (at temperature levels compatible with the most uses), cooling or electricity.
And, “vis-à-vis” each unitary need, there are even several alternative solutions.
In the first approach, these technologies can be classified into three complementary categories:Technologies providing so-called low-temperature heat, such as geothermal energy (very low energy), recovery on drying steam or solar thermal, which are adapted to uses such as domestic hot water production, space heating, or low-temperature industrial processes, such as pasteurization in the food industry.
As a reminder, Concentrated solar power, geothermal heat with very high heat (shaded part), and that low and medium energy geothermal energy are presented in the cartography but are not this study.
Solar Energy Generating Systems (SEGS) solar complex in northern San Bernardino County, California.
The first commercial parabolic trough power plants with a total of 354 megawatts went online in California.
Gov — BLM — BUREAU OF LAND MANAGEMENT — http://www.
html, Public Domain https://commons.
php?curid=159578902) Technologies providing so-called high-temperature heat, such as biomass, biogas or smoke recovery from certain furnaces, which make it possible to respond to needs, particularly those that can be found in metallurgy, glass or chemistry.
3) Technologies to produce electricity, useful for all processes studied.
These can be covered for example with solar photovoltaic, wind energy or biomass or biogas cogeneration unit.
Source: Adapted from The National Energy Education Project (public domain)Issues related to the integration of technologies: carbon footprint, competitiveness, conditions integration and exploitationA survey carried out in 2018 among many manufacturers in Europe (ref: www.
fr/mediatheque), showed that these technologies are already deployed on the ground, often successfully, despite the difficulties encountered.
The interviewed industrials evoke a first family of stakes: the reduction of CO2 emissions, the development of a responsible corporate image, which can confer a marketing advantage, while coherence with their environmental and societal commitments.
The authors added:« Le niveau de déploiement de ces technologies est assez inégal, notamment du fait du niveau historique de leur compétitivité respective face aux énergies conventionnelles, et cela, malgré la visibilité sur les coûts de production qu’apporte une solution EnR&R (indépendants de la fluctuation du prix des énergies fossiles).
» or, in the Shakespeare language: The level of deployment of these technologies is quite uneven, in particular because of the historical level of their competitiveness compared with conventional energy sources, despite the visibility on the costs of production that brings a Renewable Energy or Energy Recovery solution (independent of the fluctuation of the price of fossil fuels).
However, even if the industrials interviewed mention environmental and branding issues, they agree that competitiveness issues are of the first order: in other words, the deployment of Renewables, or Energy Recovery technologies are only done if it contributes to their competitiveness.
Photo by paolo candelo on UnsplashOn the road to competitivenessFor renewable energies (RE), it’s always a challenge to be considered an add-on to competitiveness.
Especially when there are no subsidies available, under current market conditions, with a very low gas price, and apart from a few special cases, they are globally less competitive than traditional reference solutions (electricity and natural gas).
It can be noted that these conventional energies are often accessible at even lower cost for large consumers, benefiting from better supply contracts.
In this context, public policies, and the support mechanisms associated with them, but also maximization of the outcomes with AI and energy storage (IES) play a major role.
Under the impetus for subsidies and technology integration, many solutions can bring competitiveness to industrial.
But RE coupled with IES should be seen as a team member into the bigger competitive race.
In addition, investing in an energy asset is often a mid-term or a long-term choice, with longer depreciation period.
A choice today will have consequences for the next 5, 10, 15 years or more.
Choosing to invest in one or more Renewable or Recovery assets today may be a long-term paying choice because it allows to diversify the energy mix and decrease the dependence on traditional energies, whose prices may be particularly erratic.
Moreover, such a strategy allows also, at least in part, to overcome the increase in the price of fuel, electricity or even to a lesser extent the CO2 credit market.
The integration of technologies must be part of a global reflection on the needs and possibilities of a site.
Integrate on an industrial site a solar installation, a wind turbine, a biomass boiler, or a methanization plant, demand for space and adapted infrastructure and even more a good evaluation (if not a good prediction) of the future outcomes.
The integration of the cleantech assets may also require a significant overhaul of distribution associated with it.
Depending on the topology of the site concerned, integration constraints, location of needs, changes may not be marginal.
In a number of cases, it can even be necessary to deploy new grid or to install storage capacities (because of the time lag between when heat recovery can be achieved and when energy recovered can be used, or because of the variability of some RE).
Artificial “augmented” Intelligence is required to efficiently perform all this.
And this is emphasized by the fact that, in general, due to their relative lack of flexibility, Renewables require more specific operational know-how than their traditional competitors (electricity grid, conventional power generation, heat produced from natural gas).
Credit: Optimizing power generation with Hybrid Energy Storage and AI (CC BY-NC-SA 2.
0), © Smart Phases Inc.
(DBA Novacab)The strengths of one counterbalance the weaknesses of the othersIndeed, conventional technologies have the advantage of being (almost) always available, responsive and flexible.
They easily adapt to even rapid fluctuations in load and activities and are therefore able to provide load curves with a lot of responsiveness.
This is not the case for the majority of Renewable technologies.
Solar and wind technologies, on the other hand, are typically variable, producing only electricity and/or heat in the presence of sun and/or wind.
It is the same with the global process of production and combustion of biogas or biomass, which may be subject to the availability of resources local (liquid effluents or various wastes).
However, it is possible to overcome these difficulties by intelligently combining energies technologies with heat recovery technologies and renewable energies, to provide the different uses of the site.
As an example, we can consider the association of district heat with energy storage solutions to cope with the variability of production of certain renewable energy assets or asynchronous heat recovery solutions.
To facilitate the integration of renewable energy, industrialists will have to rely on the contribution of all the ecosystem: technology providers, service providers, financing actors through innovative business models, in order to share the risks and overcome the difficulties mentioned.
Such projects might seem quite complex to implement for us mortal human, but with the help of AI, they can bring high-performance levels, both from an economic and an environmental point of view.
In order to give the most value, it is important to focus on improving training times for our AI and squeezing the most insight from smaller quantities of data.
The end results are solutions that require fewer data to build, are faster to train and deploy, and that protect your intellectual property.
In short, a mix of energy storage (electric, thermal, hybrid, mechanical, etc.
) and AI are jointly able to solve this downside of RE.
The combination of conventional technologies and Renewables with Intelligent Energy storage technologies facilitates the integration and exploitation of these.
Indeed, while all these technologies might appear implicitly in a competitive situation to each other, these technologies must above all be seen as complementary to the each other, just as they can be complementary to conventional sources (electricity and gas in particular).
The whole challenge lies in the construction of an energy system in which the assets complement each other so that the strengths of one counterbalance the weaknesses of the others (each technology brings its share of advantages and disadvantages, which need to be assessed and taken into case by case, at each industrial site), to enable the deployment of efficient energy solutions of an operational and economic point of view.
And at each industrial site corresponds a specific solution.
Multiple data sources are key here to optimize the solution and AI is necessary to deal with this mountain of information.
As Edmund Hillary once said: When you go to the mountains, you see them and you admire them.
In a sense, they give you a challenge, and you try to express that challenge by climbing them.
The Energy Transition is the mountain range ahead.
It challenges us, we need to better join our strengths and “augmented intelligence” to climb it the best way possible!Photo by Hu Chen on Unsplash___________________________________________________________________This article is part of a series on Artificial Intelligence and Energy Storage by Stephane Bilodeau, ing.
Eng, PhD, FEC.
Chief Technology Officer, Smart Phases (Novacab), Fellow of Engineers Canada and expert contributor to Energy Central.
You can find other articles here:Artificial intelligence in a “no choice but to get it smart” energy industry!The Renewable Energy boom is a great opportunity for the power sector & the fight against climate change In a smart…towardsdatascience.
comThe Combo effect against the Butterfly effectArtificial Intelligence to boost Energy Storage that will give resilience and efficiency to the Renewable Energy…towardsdatascience.
comNo Fast (Enough) Energy Transition without Intelligent Energy Storage!AI and Energy Storage fast require strong coupling for an efficient transition with Renewables against the furious…towardsdatascience.
com.. More details