Continual advancements in space technology are reshaping our capacity to manage Earth's affairs with precision and efficiency. Amidst the complexities of urbanization, climate change, and global connectivity, the need for sophisticated infrastructure has become increasingly apparent.&nbsp;Satellites, with advanced functionalities, have emerged as indispensable tools for data acquisition, analysis, and communication, which are crucial for informed decision-making on Earth.With their high-resolution cameras, multispectral sensors, and robust communication capabilities, we can capture and transmit data from a wide range of Earth-related activities, from urban planning and disaster management to agricultural monitoring and environmental protection.By continuously orbiting the Earth, satellites capture detailed images and other sensor data covering the whole planet’s surface. We can use this data to track changes in landscapes, monitor weather patterns, and observe natural disasters like forest fires, floods, and hurricanes in real time.This data is collected and transmitted back to Earth, where it can undergo rigorous analysis. Machine learning algorithms, data analytics, and other advanced computing techniques allow us to interpret the data with ease. For example, predictive modeling is utilized to forecast weather changes or track the migration patterns of endangered species. The ability to process real-time data enables quick decision-making, which is especially crucial during responses to natural disasters and emergencies.&nbsp;Communication is another critical functionality of satellites. They ensure communication lines are able to remain open, even in the most remote parts of the world. This facilitates global connectivity, allowing data, voice, and video communication to be transmitted across diverse geographical landscapes.<h1 dir="ltr">Maintenance in space</h1>On Earth, if something breaks down in a factory or a power plant, we don’t build a replacement. Instead, we send a repair team with spare parts, and even better yet, we monitor the components' wear and replace them before they break down. This preventive maintenance approach minimizes downtime and reduces the costs associated with sudden breakdowns.This methodology is critical in industries where reliability and uptime are essential. By continuously monitoring equipment health through sensors and predictive analytics, we can identify potential problems before they lead to operational failures. However, this approach faces unique challenges in space, where sending a repair team is not feasible due to the vast distances and extreme conditions.The transition from Earth-based maintenance to space requires innovative solutions to overcome these challenges. The harsh space environment, including vacuum conditions, microgravity, and cosmic radiation, makes traditional maintenance methods impractical. Additionally, the delay in communication between Earth and spacecraft further complicates real-time interventions, necessitating a more autonomous approach to maintenance and repair operations in space.<a href="https://www.ant61.com/" target="_blank" rel="noopener noreferrer">ANT61&nbsp;</a>develops autonomous robots that enable the maintenance of satellites without the need to put human lives at risk. ANT61 is providing solutions for both: their Beacon product allows satellite operators to understand what went wrong with their satellites and restore them back to operation.For larger and more expensive satellites, they are building robots that will dock, refuel, and, in the future, refurbish satellites, prolonging their useful life. At the core of these robots lies ANT61 BrainTM, the innovative devices that combine machine vision and decision-making technology, enabling the autonomy of these maintenance robots. Autonomy is very important as, due to the speed of light limitations, it won’t be possible to control every movement of these robots remotely from Earth.<h3 dir="ltr">BrainChip AkidaTM&nbsp;: A revolutionary neuromorphic processor for edge AI applications</h3>The first generation of the ANT61 Brain uses the <a href="https://brainchip.com/akida-neural-processor-soc/" target="_blank" rel="noopener noreferrer">BrainChip AkidaTM</a> chip for power-efficient AI and is currently on board the Optimus-1 satellite, which was deployed recently by the SpaceX Transporter-10 mission. ANT61 will test the ANT61 Brain later this year and perform training and evaluation of various neural networks that we will use for future in-orbit servicing technology demonstrations.The BrainChip Akida™ chip is a groundbreaking advancement in neuromorphic computing, designed to mimic the neural architecture of the human brain. This technology stands out for its ability to process data directly at the point of acquisition, greatly enhancing efficiency by reducing the reliance on distant cloud-based systems for data processing.Akida's unique capability lies in its event-based neural processor, which processes data only when needed, using data sparsity to significantly reduce the number of operations required. This leads to lower power consumption and heat dissipation, making it highly suitable for the demanding conditions of space. Each Akida node consists of four Neural Network Layer Engines (NPEs), which can be configured for various functions, such as convolutional layers or fully connected layers, and support incremental learning and high-speed inference. This configuration flexibility allows the chip to efficiently handle a wide range of AI tasks without needing constant connectivity or high power draw.Akida's scalable architecture enables the chip to support up to 256 nodes in a mesh network, boosting computational power while maintaining a small footprint. This scalability is crucial for space applications where efficiency and performance are vital. The entire neural network can be embedded directly within the Akida fabric, eliminating the need for frequent data transfers that consume significant power and reduce system responsiveness.Another key feature of the Akida chip is its on-chip learning capability, which allows it to adapt and improve its performance over time without external intervention. This feature is particularly beneficial in space, where conditions can change rapidly and autonomy is crucial. By processing and learning from data locally, Akida enhances the autonomy of satellite maintenance robots like those developed by ANT61, enabling them to perform complex tasks with minimal latency and enhanced reliability.The ANT61 team chose to partner with BrainChip because they believe that neuromorphic technology will bring the same exponential improvement in AI as 20 years ago. The shift from CPU to GPU opened doors to deep neural network applications, which are at the core of all AI technologies today. Neuromorphic technology is also perfect for space: the lower power consumption means less heat dissipation, and we can get up to five times more computational power for the same electrical power budget.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1715339791069-Akida_1000_chip-1.jpg" style="width: 100%px;" class="fr-fic fr-dib">Akida AKD1000 is a reference chip implemented with TSMC at 28nm<h1 dir="ltr">The ANT61 vision for the future</h1>Humanity’s expansion to the cosmos requires infrastructure that can only be built by robots. With its in-orbit servicing experience, ANT61 aims to become the main supplier of the robotic workforce for Moon and Mars installations. This will enable companies from Earth to project their ambition to space, providing valuable services and resources for future off-world factories and cities.Neuromorphic computing, exemplified by technologies like the BrainChip Akida™ chip, is set to play a pivotal role in this spacefaring era. This form of computing brings several advantages crucial for space applications, such as low power consumption, high efficiency, and the ability to perform complex computations needed for autonomy. By mimicking the human brain, neuromorphic processors can process large volumes of sensor data efficiently, making real-time decisions without the need to send data back to Earth for processing.The integration of neuromorphic computing into space robotics promises to significantly improve these machines' capabilities. Robots equipped with advanced AI processors can better handle the unpredictable nature of space environments, adapting to new challenges autonomously and performing tasks ranging from construction to maintenance and repair. This capability ensures that space habitats are built efficiently and can be sustained with minimal human oversight.ANT61's vision of utilizing these technologies to support human expansion into space reflects a forward-thinking approach that will not only facilitate the construction of off-world establishments but also ensure their long-term sustainability and success.&nbsp;

Remote satellite maintenance is crucial to critical data acquisition and global communication


Advancing Earth Management Through Neuromorphic Computing Powered Satellites

The globalization of supply chains is being formed on production lines that are able to supply only the required quantity of products at the time required. This means that an obstacle arising in one place in the production line may strike a major blow to the entire supply chain.The causes of production delays and stoppages include conflicts, disasters, raw material shortages, and equipment and device failures. In particular, production stoppages and the outflow of defective products due to equipment and device failures may lead to major problems for manufacturers. These problems include the expense required to recover those defective products and opportunity losses due to product supply stoppages. Therefore, equipment maintenance, which bears responsibility for the stable operation of the production line, is an important issue that holds the key to the life of manufacturers.The importance of equipment maintenance has been strongly expressed up to now. The techniques of equipment maintenance have also been advancing from breakdown maintenance to preventive maintenance and then onto predictive maintenance as well in recent years.We introduce in an easy-to-understand manner equipment maintenance that is continuing to evolve in this way—from basic knowledge such as the types of equipment maintenance and their advantages and disadvantages to the technologies to realize the predictive maintenance that is considered the ideal form of equipment maintenance.<h2>What Is Equipment Maintenance?</h2>Equipment maintenance refers to the activities to keep various production equipment in factories functioning normally. It is also called “machine maintenance.” The purposes of equipment maintenance are cited as being, for example, to prolong the life of machine components used in equipment, to shorten machine and device downtime, and to prevent unexpected failures in addition to repairing equipment that has failed.Moreover, equipment maintenance up to now has mainly focused on machines and devices. Now, the focus is also on the software that controls the Programmable Logic Controllers (PLCs) and machines and devices with the conversion of regular factories into smart factories using industrial robots, machine vision, and other technologies.There are other terms similar to maintenance such as “upkeep” and “care,” but these are the same as equipment maintenance. The reason is that they all involve inspecting equipment and then servicing or repairing it as necessary so that it does not fail.<h2>Types of Equipment Maintenance and Their Advantages and Disadvantages</h2>Equipment maintenance includes breakdown maintenance and preventive maintenance. Preventive maintenance is further broken down into time-based maintenance and condition-based maintenance (Figure 1). We explain here the features and the advantages and disadvantages of each type.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714985901545-predictive-maintenance-img0001_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 1: Types of Equipment Maintenance<h3 id="isPasted">What Is Breakdown Maintenance?</h3>This is maintenance that is performed after a piece of equipment’s functions have declined or after it has stopped functioning (Fig. 2).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714985917150-predictive-maintenance-img0002_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 2: Breakdown MaintenanceAdvantages: The advantages of breakdown maintenance include the fact that the burden on maintenance work is low because the condition of the equipment is not checked until it fails or a malfunction is discovered. Another advantage is that it keeps the costs involved in maintenance low in the case of minor failures. Disadvantages: Failures and malfunctions occur unexpectedly. Therefore, there may be delays in obtaining repair components, and so it may take some time to carry out repairs. In addition, there are risks such as major disruptions to production plans if the failure or malfunction occurs in a core piece of equipment. Furthermore, if there is a delay in discovering the failure or malfunction in the equipment, it may lead to defective products and their flowing out of the factory. Dealing with such issues may require a considerable amount of expense and time.<h3>What is Preventive Maintenance?</h3>This is maintenance that is performed based on a piece of equipment’s service life, the number of times it is used, its condition, or other factors. It is also called “time-based maintenance” (Fig. 3).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714985934940-predictive-maintenance-img0003_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 3: Preventive MaintenanceAdvantages: The maintenance work is performed once a certain period has been reached. Therefore, it is possible to prevent failures and malfunctions from occurring in equipment. It is an effective maintenance method when the service life of each component used in equipment or the number of times those components are used can be grasped accurately. Disadvantages: It is necessary to establish criteria to ascertain as accurately as possible the life of components. Maintenance may be performed after a failure if the criteria is not accurate. Conversely, component and time losses may arise such as from the replacement of components that can still be used.<h3>What Is Predictive Maintenance?</h3>This is maintenance that is performed in line with extent of a piece of equipment’s deterioration by constantly monitoring its condition. It is also called “condition-based maintenance” (Fig. 4).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714985956264-predictive-maintenance-img0004_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 4: Predictive MaintenanceAdvantages: The maintenance work is performed when the equipment deteriorates. Therefore, it is possible to reduce component and time losses such as from the periodic stoppage of equipment for maintenance and the replacement of components that can still be used. Of course, it is also possible to prevent failures and malfunctions that would lead to the equipment stopping unexpectedly. Disadvantages: It is essential to have equipment to constantly monitor the condition of equipment and to store the collected data. Specifically, the constant monitoring of equipment requires a large number of various types of sensors. In addition, data loggers and similar recording devices are needed to store data. Furthermore, it is necessary to have equipment and other technologies to connect to those recording devices to communicate the data.<h2>Four Technologies to Turn Preventive Maintenance into Predictive Maintenance</h2>The most common method of maintenance currently is preventive maintenance whereby equipment is periodically maintained by determining the inspection timing based on the service life of each component used in equipment or the number of times those components can be used. Furthermore, there are an increasing number of cases in which semi-predictive maintenance is being adopted. This is when a large number of sensors that measure vibrations, temperature, current, voltage, and other elements are attached to equipment and thresholds are then set for each of them with maintenance being performed when those upper or lower limits are exceeded. In fact, a survey conducted on companies engaged in converting regular factories into smart factories produced results showing that over 96% of them cited “equipment maintenance utilizing sensors and AI” as the reason they are continuing to convert their factories into smart factories (Fig. 5).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714985974057-predictive-maintenance-img0005_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 5: Reasons Why Companies Want to Continue Converting Regular Factories into Smart Factories* About the survey:&nbsp; Surveying entity: Murata Manufacturing Co., Ltd. Survey focus: 11,084 workers in the manufacturing industry in Japan (screening survey) / 500 workers in the manufacturing industry involved in converting their company’s regular factories into smart factories (main survey) Survey method: Internet survey Survey period: 3 days from January 25 (Wed.) to 27 (Fri.), 2023 * The total values may not necessarily match due to rounding.In this way, equipment maintenance is currently undergoing a switch from preventive maintenance to predictive maintenance. We explain here the four technologies that can help with this switch from preventive maintenance to predictive maintenance.<h3>1. Introduction of Sensors into Equipment Maintenance</h3>Predictive maintenance begins with constant monitoring of the condition of equipment. There are individual differences in judgment when this monitoring is performed with the eyes and touch of people. This means it is not possible to respond accurately. We can quantify the condition of equipment by introducing sensors. Visualizing and analyzing the data collected from those sensors then enables us to respond more accurately. However, this requires installing many types and a large number of sensors in large factories. Accordingly, routing and connecting the sensor wiring requires many man-hours when changing the setup. Therefore, it is best to have noise-resistant sensors capable of wireless communication with a long transmission distance.Moreover, devices have been developed that can keep a log of the actions taken by workers on manufacturing lines in recent years. It is also possible to record how those workers respond when trouble occurs. Utilizing these records allows various matters to be saved as data, including the deterioration of equipment that occurs on the manufacturing line and the actions taken by workers in response. This data can be used to give hints to improving processes in addition to maintenance.<h3>2. Visualization of the Condition of Complex Equipment: Sensor Fusion</h3>Sensor fusion is a technology that realizes cognitive functions that cannot be obtained with a single sensor on an engineering basis. Specifically, sensor fusion has already been introduced into Advanced Driving Assistant Systems (ADAS). For instance, examples include reading the shapes of people, obstacles, and other objects with an image sensor and temperature sensor and then measuring the distance with a radar sensor utilizing milliwaves and microwaves to grasp the situation in the direction in which those objects are moving. It has also become possible to display warning messages on monitors in vehicles and to apply brakes automatically. This utilizes sensor fusion as a technology to, for example, automatically determine the situation with data collected from multiple sensors such as image sensors and radars and to then convey those results to the driver. When we use this technology in equipment maintenance, we can analyze the measurement data from vibration sensors, temperature sensors, current sensors, and voltage sensors on an integrated basis. For example, if a current or voltage value is abnormal despite the motor vibrations and temperature being normal, we can determine that there has been a problematic change to the load on the part where the motor is operating. In addition, if an abnormal value is seen only in a vibration sensor, it is likely to be bearing wear, shaft deformation, level abnormality, or similar (Fig. 6).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714986016706-predictive-maintenance-img0006_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 6: Introduction of Sensor Fusion into Equipment MaintenanceConventionally, such abnormalities were grasped and determined by workers reading each sensor value and then comparing those values with their past experiences.&nbsp; However, if we use the technology of sensor fusion, we can aggregate data collected by many sensors and display it on monitors or other devices as results that indicate the condition of equipment. This allows even inexperienced workers to accurately grasp the condition of equipment.<h3>3. Introduction of AI into Equipment Maintenance</h3>Sensor fusion is an outstanding technology. However, as the number of sensors increases and measurements become more precise, the amount of data increases. This leads to calculations becoming complex. It also becomes difficult to output data that indicates accurate results. Accordingly, the introduction of deep learning into equipment maintenance has come to be seen as important. Introducing the AI technology of deep learning into equipment maintenance enables data collected from multiple sensors to be integrated and analyzed by AI. If there is an error in the output results, it is corrected through machine learning. Furthermore, accurate results are produced. When we utilize AI in equipment maintenance, we become able to respond correctly to the condition of equipment that is constantly changing with self-learning such as by comparing the data collected during normal times and data collected from sensors to extract the difference in addition to simply detecting abnormalities that are determined by establishing a threshold (Fig. 7).<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714986035251-predictive-maintenance-img0007_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 7: Introduction of AI into Equipment MaintenanceIn this way, introducing AI into equipment maintenance doesn’t just reduce the differences in judgments due to differences in the proficiency level of the workers; we can also solve issues in conventional maintenance activities that rely on manpower, which involves investing a lot of personnel and hours in work, including monitoring and inspections.&nbsp;<h3>4. Introduction of Edge AI Cameras and Other Edge AI Devices</h3>Edge AI is a technology that applies edge computing to AI. In edge AI, AI processing is performed with various sensors, cameras, mobile terminals, and other edge devices at the end of networks (Fig. 8). On the other hand, in contrast to edge AI, there is a technology called “cloud AI.” This is a technology that applies cloud computing to AI. This means AI processing is performed on the cloud server connected to the network and the results are then returned to the edge devices. If we compare edge AI and cloud AI, we find that there are advantages to edge AI. These advantages include outstanding response speed and data confidentiality because the AI processing is performed on edge devices without involving a network. This also means edge AI is not impacted by the quality of the network. High data confidentiality is an absolute requirement especially for equipment maintenance that handles data in factories. Moreover, a high response speed is sought in vibration and acceleration calculations, which require data collection at high sampling cycles. Therefore, we can say that edge AI is an attractive technology for predictive maintenance involving the handling of large volumes of data.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714986052981-predictive-maintenance-img0008_en.png" style="width: 100%px;" class="fr-fic fr-dib">Fig. 8: Introduction to Edge AI DevicesHowever, edge devices equipped with edge AI are small. This means they have disadvantages such as inferior memory size and calculating capacity compared to cloud AI. Accordingly, it is necessary to select either cloud AI, edge AI, or to use both together according to your purpose.<h2>Summary</h2>The importance of equipment maintenance for the stable operation of manufacturing lines has long been recognized. Nevertheless, on the other hand, it is also true that too much emphasis on production has led to daily maintenance activities being neglected. Nowadays, with the globalization of manufacturing having long since progressed, manufacturing lines are continuing to constantly supply products to supply chains. In particular, supply chain management has been built based on production output during normal operations in regular factories that are being converted into smart factories. Under these circumstances, unexpected equipment failures on the manufacturing line can deal a major blow to the supply chain and cause damage to a company’s image. Moreover, the introduction of sensor fusion, AI, and other technologies into equipment management is not just about optimizing equipment maintenance, reducing personnel expenses, and solving labor shortages. It is a cutting-edge approach with many advantages including the continuing maintenance of high productivity, the optimization of energy usage, and the effective utilization of personnel. It is no exaggeration to say that realizing predictive maintenance by introducing sensor fusion, AI, and other technologies into equipment maintenance is the key to future corporate management for the manufacturing industry, which is surrounded by a complex economic environment such as the widening of supply chains.&nbsp;<hr><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1714986081243-1714986081243.png" class="fr-fic fr-dii">If you have any questions based on this article that you would like to discuss with an expert at Murata, don't hesitate to reach out. There's an entire team ready to support you with your question.<a href="https://www.murata.com/en-eu/contactform?&Detail=Wevolver%20request" target="_blank" rel="nofollow" class="button-main-action">GET IN TOUCH</a>

The importance of equipment maintenance is clear. In recent years, equipment maintenance techniques have also advanced from breakdown maintenance to preventive maintenance and now focus on predictive maintenance.

The Technology That Realizes Predictive Maintenance

Explore the transformative impact of predictive maintenance in industrial automation. Discover how Infineon’s advanced predictive maintenance solution enables operational efficiency, reduces downtime, and fosters sustainable industrial growth through innovative edge-to-cloud integration.


Industrial Efficiency with Predictive Maintenance Solution

Edge Impulse-Fri Jun 30 2023 23:59:59 GMT+0200 (Central European Summer Time)

Edge Impulse-Mon Jul 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

Edge Impulse-Thu Aug 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

Edge Impulse-Thu Aug 31 2023 23:59:59 GMT+0200 (Central European Summer Time)

Edge Impulse-Sat Sep 30 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

Edge Impulse-Mon Oct 30 2023 14:34:39 GMT+0200 (Eastern European Standard Time)

Edge Impulse-Tue Oct 31 2023 23:59:59 GMT+0200 (Eastern European Standard Time)

Unitree Robotics

Canonical-Fri Jun 30 2023 23:59:59 GMT+0200 (Central European Summer Time)

Canonical-Mon Jul 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

Canonical-Thu Aug 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

Canonical-Thu Aug 31 2023 23:59:59 GMT+0200 (Central European Summer Time)

Canonical-Sat Sep 30 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

READY Robotics-Fri Jun 30 2023 23:59:59 GMT+0200 (Central European Summer Time)

READY Robotics-Mon Jul 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

READY Robotics-Thu Aug 31 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

READY Robotics-Thu Aug 31 2023 23:59:59 GMT+0200 (Central European Summer Time)

READY Robotics-Sat Sep 30 2023 23:59:59 GMT+0300 (Eastern European Summer Time)

predictive maintenance

predictive maintenance

Profiles