Welcome to HiPer it! Digest for Technology and Innovation - your go-to source for the latest news, software development, AI and digital innovation.
CEO Insights
Sometimes the best projects are the ones that almost fail. Here is how we fixed the retrofit project that went wrong (and what we learned). Most energy optimisation projects require light or deep retrofits. And while they can be designed perfectly, their actual installation can go terribly wrong. And you will not even know, because your BMS only measures a limited number of parameters. In a recent project, where a customer wanted to re-use wasted energy for hot water pre-heating, the installation team made a mistake when connecting a heat exchanger. This issue was identified during the commissioning of the project when HiPerWare Building IoT (BIoT) platform identified a change in the process pattern. Hardware installation and platform configuration takes hours instead of weeks. So as more and more retrofit projects are implemented, HiPerWare will become a unique diagnostic tool to ensure required HiPerformance operations after installation.
2 years ago we started "Energy Detectives" initiative with students in Gran Canaria to identify inefficiencies in energy consumption. Now AI anomaly detection takes care of it. We are launching a free Anomaly Detection service for your historic big data of building technical operations. Just upload your CSV file and HiPerWare AI module will take care of identifying abnormal patterns in processes. These anomalies then can be categorised for future use.
❓ Are you still manually collecting data from your energy, water or heat meters? Then you know how complex, time-consuming and error-prone this can be, especially if you have legacy analogue meters that are expensive or difficult to replace. Many building owners are lost when it comes to resource consumption. 💡 HiPer it!, the next-generation PropTech solution for real estate owners and managers offers HiPer Meter Connect that enables automatic and real-time data collection. 🚀 HiPer Meter Connect is a quick to install and affordable retrofit solution. It transforms legacy meters into smart metering devices, making your building AI ready. With built-in AI OCR technology, HiPer Meter Connect ensures accurate and reliable meter readings, significantly reducing the risk of human error. Coupled with the HiPerWare platform, the solution provides data collection, structuring and analysis. Data is transferred instantly to the HiPerWare platform, providing access to meter readings and eliminating delays in decision making and reporting. 🌟 Future-proof and optimise your assets today.
💡 As the owner of a premium property, you face a set of challenges in terms of property management, maintenance and security. Managing general repairs, air conditioning, fridges, pool maintenance and other staff can be particularly difficult, especially remotely. In addition, unoccupied properties often experience water leakages and become targets for theft or vandalism. 🙂 With HiPer Villa, you can rest assured that your property is always under digital supervision. Our AI capabilities enhance real-time monitoring, enabling the detection of operational anomalies and unauthorised use. Digital twins and advanced big data analytics streamline diagnostics, maintenance and operations, helping you avoid costly repairs and ensure optimal performance. 💰 By implementing transparent and efficient technical operations, you not only protect your property but also increase its market value. 🏡 Would you like to keep your property safe and efficient?
We're bringing the power of HiPerWare capabilities to large holiday villas! The functionality, earlier available and affordable only to large commercial buildings, is now ready for premium property care. My friends, who have large houses, told me so many "horror" stories about technical maintenance: water leakages, mold, fires, equipment failures, pool issues, unauthorised occupiers, excessive energy consumption, slow service response time. We're launching a new HiPerWare-based product - HiPer Villa - to bring peace of mind and reduce unpleasant "surprises" for owners, and provide advanced big data collection, AI analytics and digital twin capabilities to technical maintenance companies. https://lnkd.in/d7r3RQb2 If you have a holiday home/villa, how would you rate the level of technical maintenance services from 1 to 10?
🌟 Just ask HiPerBot and it quickly analyses your projects data and provides accurate and detailed information about technical processes, events and potential causes of anomalies in real-time, without the need to analyse dashboards and patterns. 🤖 HiPerBot leverages AI and machine learning capabilities to understand and interpret user questions and provide relevant answers while continuously improve its responses based on user interactions. Moreover, HiPerBot can analyse a project and make changes to the project settings. What kind of chatbot functionality are you looking for in HiPerWare?
Our first guest is Yuehan Wang, Global Research Director at JLL, who will share the results of their research on AI adoption in real estate. Key discussion topics are: * How AI will change the industry* Corporate enterpriseAI implementation * AI trends in the job market* AI influence on market requirements
Ongoing, in-depth big data analysis to identify anomalies and inefficiencies within building systems is the basis for reducing energy consumption. Just as there is no such thing as a perfectly healthy person, there is no such thing as a perfectly functioning building:❌ Systems are not optimally configured❌ Heating and cooling are often ON at the same time❌ Regular procedures are not followed❌ and much more. 😱 These scenarios are more common than you might think. 💡 At HiPer it! we provide building owners and operators with a clear and insightful view of the processes taking place within building systems. The data is fully integrated with the digital anatomy of a building through BIM, 360-degree panoramic photos, P&I and single line diagrams. Whether it is a property due diligence for investment, operational and energy optimisation efforts, or regular maintenance, HiPer Checkup enables AI-driven analysis of multiple patterns, identifying anomalies and inefficiencies. With HiPer Checkup you can: ✔️ Collect technical operations big data and make your building AI-Ready✔️ Immediately reduce energy consumption by an average of 12-15% ✔️ Get an audit of operational processes and recommendations for optimisation✔️ Improve discipline and reduce human error ✔️ Optimise technical operations and increase asset value.
💡 Today, industrial and commercial organisations must pay not only for the energy they consume, but also for the additional costs associated with grid usage. These costs are calculated on the basis of the maximum peak power and have a significant impact on operating budgets. 💰 At the same time, new tariffs based on dynamic energy prices are emerging. As we move away from the era of cheap night-time tariffs driven by traditional generation methods, new renewable energy sources such as wind and solar are reshaping the energy market. This shift is forcing companies to adapt their energy consumption strategies. 🤖 By using advanced technologies and dynamic energy models, companies can understand their granular power consumption, and optimise it in real time. With smart consumption, organisations can align their energy use with market prices, consuming power when demand is low and costs are low. This not only results in significant cost savings but also contributes to more sustainable energy consumption. 🌟 HiPer it! solutions allow organisations to take control of their energy consumption, reduce peak loads and take advantage of dynamic pricing.
🌟 Early detection of anomalies is key to reducing energy consumption, preventing breakdowns and extending equipment life. There are 3 key steps: 🎯 Anomaly detection at the heart of predictive maintenance. Developing an effective automatic anomaly detection algorithm involves: - Using machine learning on historical time series data to identify anomalies.- Classifying process pattern anomalies into categories. 🎯 The importance of human expertise. While AI-driven anomaly detection is powerful, human insight is essential for system refinement. A well-labelled dataset improves the accuracy of automated classifications. 🎯 Automated anomaly classification for greater efficiency. With a well-trained model, automated classification streamlines response times and maintenance efficiency through accurate anomaly categorisation and continuous learning on new datasets. 💡 The HiPerWare platform delivers fully automated AI-based anomaly detection by analysing large datasets from 4 years of operational history, ensuring quick wins after implementation.
At a recent webinar Corum Group Ltd. shared top 10 technology trends that will shape M&A in 2025. Two of these trends fully resonate with HiPer it!'s expertise and leadership: ✔️ Actionable Analytics. Big data has become one of the most valuable corporate assets. At HiPer it! we take the most of AI and Big Data based analytics to optimize building technical operations and energy consumption. Our solutions provide real-time insights to make smart decisions about energy consumption and process improvement. An important part of this is predictive maintenance, which uses big data to predict when equipment might fail. This helps avoid unexpected downtime, reduce energy and maintenance costs and extend the life of systems and assets. ✔️ Digitised environment. Businesses today understand the need to be flexible and resilient. HiPer it! solutions are an integral part of digital environment, providing solutions that simplify operations, improve teamwork and increase asset value. Using technologies such as IIoT, BIM, digital twins, 360-degree views and AI analytics, we create a connected digital environment for your assets. 🌟 HiPer it! paves the way to optimised building operations, improved productivity, energy efficiency, lower carbon emissions and cost savings in 2025 and beyond.How is your company harnessing these trends?
🤝 The combined strengths and expertise in asset and property management will change the way asset management is approached. This collaboration aims to empower asset managers by providing powerful tools for data-driven analysis and measurable insights that optimise technical processes and ensure energy efficiency. 💡 🏢 Celsios enables users to build portfolios and manage assets seamlessly from the start. The platform offers valuable suggestions and facilitates careful comparison of different providers, easing the burden traditionally associated with property management. 🚀 This partnership marks a significant step forward in the field of asset management and amplifies the opportunities for unparalleled efficiency in asset management, allowing property managers to focus on what really matters: decarbonisation, energy and operational efficiency – future-proofing and maximising the potential of their assets.
☀️ ❄️ 🏢 Energy should only be used when there is a real need for it. This is not just a trend—it's a necessity for sustainable and cost-effective development. Imagine a scenario where lighting, heating or cooling of spaces, and ventilation automatically adjust to real-time conditions: the number of people in a room, shifts, weather conditions, levels of natural light, and season. Such an “on-demand” approach will significantly reduce unnecessary costs and reduce carbon footprint. HiPer it! provides energy experts with powerful tools for analysing real time energy consumption and building operating conditions. We offer clear and visual dynamic and retrospective digital trail of processes such as: ⚡ Energy consumption 🌡️ Temperature 🌧️ Humidity, CO2 and air quality levels 💡 Lighting in workspaces and common areas With collected technical big data HiPerWare platform enables identification of inefficient modes of operation and energy consumption. Let's take a step together towards smarter and more efficient energy use!
HiPer it! - an expert in AI and big data DigitalTwins for optimising energy and technical operations, is proud to announce the launch of a partnership with PMTech, aimed at implementing Connected BIM solutions to improve the efficiency of building operations and reduce energy and operational costs. PMTech, a leading engineering firm specialising in an integrated BIM approach for the AEC industry and the lifecycle of real estate projects, sees this partnership as a unique opportunity to enrich its developed BIM models with the HiPerWare platform. "The collaboration with HiPer it! provides an excellent opportunity to optimise operations and improve energy efficiency in our projects," said Evgeniy Bakhovchuk, CEO of PMtech Engineering. The HiPerWare platform, developed by HiPer it!, enables real-time big data and AI-based analytics tools that significantly improve building technical operations. It is integrated with building representation: BIM, 360° panoramic photos, P&I and single-line diagrams and provides the full “digital trail” to determine the cause-and-effect relationships of any ongoing building processes. "We are pleased to build our partnership with PMTech, which will allow us to increase the number of joint projects and demonstrate the value of BIM models combined with live data during building operations," commented Arseny Tarasov Global CEO of HiPer it! The collaboration between PMTech and HiPer it! opens new horizons for customers of both companies, enabling them to use advanced technologies for sustainable development and cost reduction. By joining forces in the field of ConnectedBIM, the companies aim to create more powerful and environmentally friendly solutions for efficient building operations.
In today's world, energy optimisation is a top priority for manufacturing companies, commercial building owners, and public and institutional building operators. In many buildings that were built over 10-15 years ago, Energy waste is a common effect of design decisions made when energy efficiency was not the top priority. One example is heat from the operation of cooking or production equipment. Energy waste is worthless, harmful or dangerous excess energy that must be disposed of, incurring additional costs and posing significant environmental risks. Energy waste is usually released into the environment, increasing greenhouse gas emissions and causing thermal pollution. For example, data centres and manufacturing facilities consume huge amounts of energy, much of which is converted into heat that is ultimately discarded. This process not only results in energy waste, but also requires additional energy to manage and dispose of excess heat. The good news is that you can turn energy waste into a valuable energy asset - just like recycling traditional waste. HiPer it! solutions help companies identify their "energy waste", assess its potential for reuse and recycling, and implement energy and operational optimisation initiatives. With HiPer NetZero offering you can measure and visualise how energy is generated, distributed, used, stored and lost with a dynamic Business Energy Model (BEM). Compared to traditional energy audits, HiPer NetZero provides additional value and savings through Dynamic Energy Audit. The HiPerWare platform provides big data and AI-based analytics to reduce energy costs and carbon footprint. In addition, HiPer it! energy experts provide comprehensive retrofit roadmap development and full project lifecycle management. Here is an example of a major energy optimisation project at a discrete manufacturing company. The client wanted to achieve Grean Deal targets and reduce energy costs. The Business Energy Model created with HiPerWare identified inefficiencies in the design and technical operations. The proposed retrofit solution reduced energy costs by 66% and CO2 emissions by 100%. Let's turn energy waste into an asset!
OpenTelemetry is an open standard for collecting, processing and exporting telemetry (logs, metrics and traces) from software. This standard helps organisations gain a complete view of system performance, which is especially important for distributed applications and microservice architectures. 🌟 With OpenTelemetry there are significant improvements in - detection and resolution of problems - performance of our services - overall platform stability and speed With this new functionality, we will be able to respond even faster to any disruptions, making your journey to energy efficiency smoother than ever!
I have started making selection and buying decision, using AI tools, like Grok and Perplexity. And most of your customers will start doing the same soon, so you have to update your marketing materials ASAP. Within our internal HiPer it! AI Transformation initiative we have identified certain gaps in the way how we look in the eyes of LLMs. Grok, for example, focuses on content in X and websites, while Perplexity reads from well-established platforms like Medium, LinkedIn, and your company's website. ChatGPT does not have real-time access to social media platforms. During our research we realised, that things that are obvious to us, are not 100% clear to LLMs during their research. This lead to re-design of our website and a different approach to posting on social media. Just like on the illustration below you can confuse AI tools with inaccurate information representation. For us this exercise was a real eye-opener. Try to ask LLM what it thinks about capabilities of your product. How happy are you with the answers?
In the previous post we discussed how monitoring tools can identify ML model degradation. Now, in the 7th post of the AI and ML for PropTech series, we'll explore the strategies that can be used to update and refine models over time to maintain model accuracy. According to HiPer it! AI expert Pavel Filonov, choosing the right strategy or combination of strategies is both an art and a science. Here are the suggestions: ✔ Adapt to evolving data Continuously and incrementally update your model with new data to capture changing patterns and trends. This should be done when your factory starts producing a new product line, or when your office building's HVAC controls receive a software update, or in any other case that can cause concept drift. ✔ Address new requirements Regularly optimise hyper-parameters to improve model performance. ✔ Implement feedback loops Implement real-time monitoring of ML model metrics and feedback mechanisms to quickly identify and address problems. ✔ Implement product lifecycle management Like any other piece of software, the ML model needs to perform a standard set of PLM actions such as third-party software updates, version control and management, documentation updates, ongoing maintenance by your MLOps team. By integrating these and other strategies, you can ensure that your ML models remain effective, accurate and aligned with evolving business needs.
Thermal insulation of building envelope performance (building thermal envelope). The effectiveness of a building's thermal envelope is critical to energy retention/savings. Any variation in energy use, such as heat loss through walls, windows, doors, roofs and ventilation systems, can have a significant impact on overall performance. Every modern building has an energy passport (BEP) that shows its annual energy demand per sq m and the energy efficiency class assigned to it. However, BEP doesn’t provide information about current real building performance. It contains static data and represents the average energy performance of a building as designed, just as a car's manufacturer metrics are not equal to its current fuel consumption. Real-time analysis of energy consumption data is critical for energy efficient buildings, helping to define real energy performance in different conditions. It incorporates live data from building systems, reflecting current energy use, updating it and providing a more accurate picture of energy losses. At HiPer it! we provide real-time big data analytics that reflect the current state of energy use. We know how a building should perform in different weather conditions based on the Building Energy Passport. We measure energy consumption and compare it with the target consumption in the current climate and weather conditions. The target energy consumption value is determined based on statistical climate data for the location and energy passport data.Is your building energy performance better or worse than designed?
The latest release of the HiPerWare platform introduces updates to the 360panorama functionality, giving you even more ways to efficiently interact with your assets! Here's what's new: 🌐 iFrame Content: Embed virtually any web page directly into your panorama. This is a game changer for seamlessly integrating external services and data. 📼 Video streaming: We've added support for mjpeg streams, and soon you'll be able to use RTSP with WebRTC conversion. View and analyse events in real time right in your panorama. 🔗 Links: Classic links to documents, files or equipment websites can now be accessed directly in the panorama - important information is just a click away. ⚡ Explore the new features today, contact us for a free demo!
Addressing the sustainability and energy efficiency of residential, commercial, public and industrial buildings, it is not enough to simply follow the prescriptions of standards and periodic expert recommendations. It is absolutely essential to approach this complex and dynamic task in a systematic and comprehensive manner, identifying strategic and tactical goals, ways to achieve them, and metrics that confirm whether a goal is being achieved throughout the life cycle of the facility. HiPer it! is the first to develop such a systematic and comprehensive method - Dynamic BEM (Business Energy Model) audit. The method identifies, analyses and prevents energy losses in these 6 key areas: 1️⃣Thermal insulation of building envelope performance (building thermal envelope) 2️⃣ Dynamic energy consumption (based on real demand) 3️⃣ Elimination of energy waste 4️⃣ Efficient dynamic generation of heat and cold 5️⃣ Peak shaving and smart energy consumption 6️⃣ Technical operations optimisation in building systems Which of these metrics do you consider to be the most important for you?
The AI algorithm, trained on big data set of operating patterns, analyses pattern of operations. The model identified abnormal behaviour- although the temperature is still within the threshold, the behaviour of the system is abnormal. However, this anomaly is difficult to see with the naked eye. These variations indicate suboptimal equipment performance or even early signs of potential failure, helping to avoid system failure and costly maintenance.
Our expectations were exceeded by the professionalism of the organisers of this leading European event in the PropTech industry. In a vibrant business atmosphere, we had dozens of valuable meetings with press, partners, clients and investors. It was a good opportunity to explore innovative solutions from colleagues in the industry and to define new vectors for HiPer it!'s development strategy. The revamped format of PropTech Connect offered attendees a wide range of opportunities for networking, knowledge sharing and meaningful discussion. We saw the great interest in the Innovation Stage format in general and the brilliant presentation from Kate, Co-founder of HiPer it!, in particular. ⚡ Analysing the interest in our solutions, we noticed that the demonstration of the AI Engine built into the platform made the biggest impression. It provides live notifications of anomalies in time-series data from processes, as well as insights into the environmental and financial metrics of the managed assets. It's no surprise that this capability reinforces HiPer it!'s position as a leader in decarbonisation, energy and operational efficiency solutions. 🤝 It was great to connect with so many knowledgeable professionals and we look forward to the new exciting projects and the future collaboration.
Excited to share a case study from a major manufacturing company that set out to meet Green Deal targets at their state-of-the-art facility. By leveraging HiPerWare, they were able to collect and structure critical energy and operational data, leading to the development of a comprehensive Business Energy Model that analyzed all energy flows. 💡 One of the major manufacturing companies in Germany had a goal of achieving Green Deal targets at its modern facility. HiPerWare was used to collect and structure energy and operational data, develop a comprehensive Business Energy Model to analyse all energy flows. ⚡ The findings showed that excess heat from the production processes was being wasted into an underground river instead of being used for heating in the workshop. The first picture shows the initial situation. 🍀 The proposed energy retrofit solution allowed for a 66% reduction in energy costs, complete elimination of CO2 emissions and a payback period of less than 3 years. Please refer to pictures 2 and 3 to explore the energy reuse solution. HiPerWare ensures efficient launch of the new energy solution and sustaining HiPerformance operations after the implementation.
AI/ML models are more like cars than traditional software. A new car loses 10% of its value as soon as it leaves the dealership and continues to lose it over time. Similarly, a new AI/ML model begins to degrade as soon as it's deployed in the real world. In the 6th post of AI ML for PropTech series let’s look into why it happened. The reasons for the model degradation are: 🌟 Gradual or sudden concept drift. Consider an industrial site: manufacturing a new item started, the operating mode of machines and equipment changed, what was relevant last month for the energy consumption or engineering systems operations, might not be today. It is really impacting the quality of the model. 🌟 Data distribution drift. Imagine an office building: energy consumption AI/ML model based on various features such as temperature, occupancy, time of day, and equipment usage patterns might suffer from seasonal changes, occupancy patterns, equipment upgrades and so on. 🌟 Data quality issues This encompasses a range of issues related to the accuracy, completeness, and reliability of input data. Sensors providing data for the AI/ML model may lose accuracy or be accidentally displaced during maintenance work. Data streams can also be interrupted due to power outages or connectivity issues. AI/ML model monitoring tools should mitigate the risks above. Here are the main goals of the monitoring tools: 🌟 Issue Detection It is “hygienic” task, that should be fulfilled to identifying problems with production models. It can alert a wide range of issues, from decrease in model accuracy to increase in missing data volume. 🌟 Root Cause Identification It should help to pinpoint problem sources, such as low-performing segments of the model or corrupted features. 🌟 Model Behavior Analysis This is a more advanced feature that provides insights into user interactions and shifts in the model's environment, aiding in adapting and improving performance. 🌟 Action Suggestions Tools can suggest specific actions, like switching to a fallback system or retraining if performance drops below a threshold. 🌟 Performance Insight Tracking of the ongoing performance for future analysis or audits. There are plenty of monitoring tools that can be used, including open-source options like Red Hat OpenShift, Evidently Observability, and others. These tools help to identify instances of model degradation and provide insights into why it happens. To use a car analogy: it’s like the red lights on the car’s dashboard that alert you to check oil, brakes, and tires when the car experiences wear and tear. With a car, you need to change the oil, swap the tires, or replace the brakes to ensure safety and efficiency. But what should be the corrective actions for an AI/ML model? Let’s explore this in the next post of the series—Continuous Learning and Improvement.
Deployment and integration of AI / ML models into the PropTech platforms “Most of the failures in ML development come not from developing poor models but from poor productization practices” wrote McKinsey partners Eric Lamarre, Kate Smaje and Rodney W. Zemmel in MLOps so AI can scale. So, in the 5th post of the AI ML for PropTech series, let's take a look at a real-world successful case of AI/ML integration for business use within the PropTech platform HiPerWare. Our experience shows that writing reproducible, maintainable, and modular data science code is challenging. Software frameworks simplify this by applying software engineering principles - modularity, separation of concerns, and versioning - to ML code. Here is the list of actions that should be performed during the deployment of the AI/ML model: - Integration with real-time APIs and data sources - On-the-fly data processing and enrichment - Post-processing - Initiating actions or responses In addition, some IT-specific actions must also be performed: - Autoscaling - Model containerization - Adding an automation framework Moving from small-scale data science experiments and model development to production may require refactoring code, changing frameworks, etc., resulting in significant engineering and programming work. This can lead to significant delays or even failure of the entire solution. To eliminate these risks, the HiPer it! team uses a containerized solution to provide the same software environment for staging and production. Currently, AI/ML capabilities and models are successfully integrated into the HiPerWare platform. It enables in-depth analysis of data for engineering system operations and energy consumption by leveraging its ability to learn patterns, detect anomalies and predict future behavior through: Pattern Recognition: ML algorithms analyze historical data to identify normal operating and energy consumption patterns. By learning these patterns, the ML model establishes a baseline of what typical behavior looks like. Anomaly Detection: Once normal behavior is understood, ML models can detect deviations from this baseline that may indicate equipment malfunction or energy inefficiency. Predictive maintenance: ML models can predict potential failures by analyzing trends and patterns in data over time and identifying "early" hidden signs. Root cause analysis: When anomalies are detected, ML can help determine the root cause. By analyzing correlations and dependencies within the data, ML models can identify which factors are likely contributing to the observed anomalies. It may seem that once AI/ML capabilities are integrated into production software, the job is done. But this is not the case. The AI/ML software cannot be left unattended, as it is likely to degrade and its performance will deteriorate. In our next post, we will discuss why this happens and what needs to be done to prevent it.
In the AI and ML for PropTech series, the 4th post explores the evaluation and validation of models. Ensuring the model accurate and effective in managing property energy and operations is crucial. As the PropTech expert Ivan Grigoryev from HiPer it! team emphasizes, a non-optimal model can result in two significant consequences. Firstly, it may lead to increased energy consumption and carbon footprint, resulting in excessive utility costs and non-compliance with environmental regulations. Secondly, it can lead to missed anomalies, resulting in financial losses. Here are some key points that HiPer it! AI expert Pavel Filonov suggest considering when choosing the AI model in PropTech:⭐ Focus on the model's ability to identify energy consumption and technical operation anomalies and detect early potential faults early, especially for high-stakes scenarios like preventing equipment failures.⭐ Implement cross-validation – to test the models multiple times on different subsets of data to see how it performs across various scenarios. This helps ensure that the chosen model is reliable and not just tailored to a specific set of data.⭐ To estimate the room for the performance improved by fine-tuning model settings. For the relatively small building or asset almost every model works well, but for the large one with thousands or even tens of thousands of parameters involved, performance may become in issue over time.⭐ Conduct testing in real-world conditions after selecting the best model. This involves deploying the model in a live environment and monitoring its performance. It means ensuring that the model accurately predicts energy usage and system faults over time, adapting to changes in building usage patterns. In summary, choosing the right AI model in PropTech involves a comprehensive approach that considers performance, reliability and real-world validation. Now when the most appropriate AI model is chosen, let's discuss how it should be integrated in the PropTech energy and operation optimization software, like the HiPerWare. Stay tuned for the next post in this series, where we’ll dig deeper into it.
The recent joint apc|m Europe Conference + Smart Systems Integration Conference (SSI) was a great opportunity to share expertise in AI technologies, discuss technical challenges that arise in various business processes and learn how to overcome them with the help of AI. The topics ranged from specific problems with physical and chemical processes in chip printing, to building sensors with low energy consumption, and specific examples of practical applications. Leonid Lopatin and Pavel Filonov presented a study on the application of modern time-series changepoint detection algorithms to sensor data from building management systems. The thorough and scientific nature of the study was based on an extensive data set collected over several years on the real-world examples. ⭐ The HiPer it! study attracted a lot of interest from the conference attendees! It demonstrated a comprehensive approach to tackling the modern challenge of reducing energy consumption and sustaining high-performance operations with the help of AI-based technologies. 🎯 Get your copy of the study by contacting us in the comments.
Without properly completing this step, even the latest and greatest scientific AI algorithms become entirely ineffective ("garbage in – garbage out," as they say). But jokes aside, why is this step so crucial? Firstly, because approximately 80% of the time and effort in AI/ML projects is devoted to data preparation. Secondly, improperly prepared, low-quality data will result in inaccurate models, making their results unusable. Therefore, before training the model, it's essential to prepare the raw data: clean it, format it correctly, structure it, label it. Importantly, this should be a collaborative effort between Property Industry experts and AI experts. And yes, HiPer it! has both teams in place. So, what exactly needs to be done? 1. A massive amount of data, commonly referred to as BigData, will be collected. Data from facility management software like EMS, BMS, and SCADA is insufficient for AI training as it only captures a tiny fraction of the overall context. Data will be collected at high frequencies from carefully chosen key equipment points through the installation of sensors, detectors, and probes to compile a comprehensive Big Data set. 2. To ensure data readiness for AI model training, the initial focus lies in cleanup: rectifying errors, filling gaps, and maintaining consistency. Both software tools and human expertise to be employed to filter out incorrect data, correct errors, and align timelines. Furthermore, validation of data's relevance to the problem and its accuracy in representing real-world phenomena is required. The avoidance of biased or incomplete data is paramount, as they can significantly skew AI model outcomes. 3. Once data is well-prepared, the next step involves structuring. Data lake, created at the previous stage, to be transformed into a structured Data Base format. Data is labeled with appropriate tags or categories to facilitate supervised learning. 4. Annotation and enrichment represent a pivotal stage in AI model training. Here, metadata or annotations are integrated, offering vital context and depth to the dataset, thereby enhancing comprehension and performance. In practice, Property Industry experts meticulously mark and label data segments classifying specific scenarios, such as excessive energy consumption, equipment malfunction, faults, or conversely, normal operation and energy consumption patterns. Now armed with the appropriate dataset, the AI model training is ready to be kickstarted.. Stay tuned for the next post in this series, where the intricacies of the training process will be delved into.
