Methods of fitting

To fit a sensor there are two different methods for permanent mounting, using adhesive or screwing it into the machine. There is also one method for temporary mounting, magnetic. Each method has its advantages and disadvantages, which we will explore here. In all cases, before installation, make sure that the surface is smooth, without any paint or dirt. Ensure the sensor is placed as close as possible to the vibration source.

Screwing

Many sensors have a threaded stud at the end. Using this by screwing it into the machine provides the best possible transmission to the sensors which enables monitoring of the widest possible frequency range (as long as the sensor supports it). The sensor must be torqued correctly. Under-torquing may lead to looseness, causing noise in the data, while over-torquing may damage the threads, sensor, or machine. It is advisable to apply some thread lock (Loctite 222 for example) to the threads during installation to prevent the sensor from unscrewing by the machine’s vibrations. Most sensors require a coupling fluid or coupling agent for the best transmission of high frequencies, please use what your sensor manufacturer prescribes for your sensors. Additionally, both disturbances cause changes in vibration metrics, making it harder to understand the machine’s development over time. Even a perfectly mounted sensor cannot provide accurate results above 30 kHz.

Adhesive

If drilling in the machine is not possible, adhesive mounting a disc and then screwing into the disc is the second best alternative not to damage the sensor. To make a good adhesive mount it is important to use the correct adhesive, many experts agree that epoxy or Loctite 454 is a good options that give good frequency responses. It is also important to keep the adhesive layer incredibly thin to not disturb the transmission of vibrations. Adhesive mounting cannot provide accurate results above 15 kHz.

Magnet

Using a magnet is an option mostly for handheld devices. However, this has a lot worse frequency response for high frequencies than a stud mount. If it should be used a completely flat, smooth, and ferromagnetic area close to the vibration source needs to be found. Magnet mounting will never yield good results over 10 kHz. However, this has improved over the last years with stronger magnets.

Mounting pads

Using a mounting pad is a clever option to get a good frequency response while still having the flexibility of using a magnet mount. Glue or screw the mounting pad to the machine and then stick the sensor to it by magnetic mounting. Since this option has a magnet both on the machine side and the sensor side the connection becomes significantly stronger and can therefore give a better frequency response than just using the sensor’s magnet to keep it to the machine.

Signs of poor fitting

The pictures below show the results of a sensor that was slowly loosened by the machine’s vibration over time. At the bottom, we see the blue spectra where we still see the machine vibration. Later the machine vibration disappears as the sensor gets looser and looser until the red where there is almost no signal left. The sensor is then reinstalled in the same place correctly and we see how the signal should look.

Bad mounting practice

There is in some ways the mounting position of the sensor can cause disturbances. For example, if the sensor is mounted on an irregularly moving element. Here are some examples:

This sensor are mounted on a surface that rotates 90 degrees periodically every 1.5 seconds. This results in low-frequency contributions to the spectra that make the signal harder to analyze.

This disturbance causes the vibration metrics to change, which makes it harder to understand the machine’s development over time. This would cause significant issues for a threshold method and is also unsuitable for MultiViz.

Conclusions

The mounting of sensors has a large impact on which frequencies are important. Magnet-mounted sensors cannot reliably capture signals over 5 kHz and should therefore not be used to collect data to produce spectra over 5 kHz. For optimal measurements, use stud-mounted or glued vibrational sensors or stud-mounted or glued magnetic mounting pads on smooth surfaces with the manufacturer-specified coupling fluid or agent.

The post Mounting vibration sensor for condition monitoring appeared first on Viking Analytics - MultiViz AI.

Ground looping

Effective management of cable grounding is crucial to ensure robust shielding and prevent ground loops. Ground loops occur when a shared conductor, such as the signal return or shield, is connected to different points with varying electrical potentials creating a small current and causing unwanted interference.

In sensors using coaxial cables, the inner conductor carries both the signal and power, while the outer braid acts as shielding and provides a return path for the signal. Maintaining isolation between the shield and the sensor housing is essential to prevent ground loops. When using non-isolated sensors, employing an isolated mounting pad is recommended to mitigate these issues.

Figure 1: If isolated mounting pads are not used and there is a potential difference between the machine and instrument a ground loop is created which creates noise.

For sensors utilizing two-conductor shielded cables, one wire transmits the signal and power, while the other serves as the signal return. The cable’s shield protects against electromagnetic interference (EMI) and electrostatic discharge (ESD). Grounding the shield at a single point, typically at the readout equipment is critical. Proper grounding of cable shields is essential to protect sensor electronics from potential damage, especially in environments with elevated EMI and ESD levels.

Below are three examples of the difference between before and after an isolated mounting pad was installed. In the yellow spectra, most of the signal is lost, and the noise from the ground looping is mostly visible. In the blue spectra, an undistorted picture is visible.

Electrical interference

Communication devices, power lines, and electrical discharges can all induce signal interference. To reduce errors caused by electromagnetic interference (EMI) and electrostatic discharge (ESD), it’s essential to use high-quality cables that are well-shielded. When splicing cables, ensure continuous shielding is maintained throughout.

Correctly routing cables is equally critical. Avoid placing sensor cables alongside AC power lines; if they must cross, do so perpendicularly. Whenever feasible, place sensor cables within a separate grounded conduit for added protection. Additionally, keep sensor cables clear of radio transmission equipment, electrical motors, generators, and transformers, as these can introduce interference. Minimize routing cables through areas prone to ESD. While sensors are typically safeguarded against ESD, severe cases can still briefly distort signals.

Conclusions

Following these guidelines can significantly reduce measurement errors and ensure more reliable sensor performance. There might be even more guidelines to follow to get the best results from your sensors. Make sure all of them are followed to ensure the best data quality, which will make it easier to achieve accurate analysis results for your machines.

The post Ground loops for wired sensors appeared first on Viking Analytics - MultiViz AI.

Set up

When setting up a sensor for a machine one needs to know approximately what magnitude of vibrations the machine will produce, for example, a fan might produce way less vibrations than a mixer. The sensors usually have a few predefined alternatives to choose from, select the one above what you think the machine might experience in a worst-case scenario.

What happens if set too high

If the dynamic range is set too high, the signal will not utilize the full range possible and will be forced into only a few possible values. This might happen faster for some sensors than others due to the number of bits used to store values. For example, some sensors only use 8 bits while others use 16 bits, the sensors that only use 8 bits will use less energy and therefore usually have a longer battery life but will need to be set up more accurately to gather good data. Setting the dynamic range too wide results in a distortion of the signal from the actual machine vibrations.

Figure 1: Here, we see the distortion. The top panel only has 3 bins, making it hard to discern the signal’s characteristics.
The middle one is better since it has 9 bins and the last one is the original signal. The signal has a 3.14 s duration and a sample rate of 955 Hz.

If we transform these three different representations of the signal to their respective spectrums, we see that we get significantly more noise when we have a higher dynamic range.

Figure 2: Here we see the consequences of having a too-high dynamic range, in the top panel which is the spectrum of the most distorted signal we see that there are lots more peaks and noise compared with the two others. However, it is not a large difference between the middle panel which is the spectrum of the distorted signal with 9 bins, and the bottom panel which is the spectrum of the original signal.

The first spectrum could be interpreted as something in the machine happening and causing a lot of noise and some small peaks. But since we know the waveform was distorted we know this is just an effect of the dynamic range settings.

What happens if set too low

If the Dynamic Range is set too low, the signal will be clipped at the Dynamic Range, causing distortion.

Figure 3: Here we see how it would look if the dynamic range were set to 1, 2, or 3. In the middle panel, it is set to two which clips the signal just a bit. The signal has a 3.14 s duration and a sample rate of 955 Hz.

If we transform this to its respective spectrum we will see how much they get distorted.

Figure 4: Here we see that the distortion caused by the wrong setting in the dynamic range significantly changes the spectrum even in the middle panel where it just clips the signal slightly.

In the spectrums above we see that when the dynamic range is set to 1 it gives a lot of small peaks that should not be there and lowers the peaks that should be there, making them more or less comparable. In the middle example with a dynamic range equal to 2, we see that by clipping the signal slightly, we introduce many small peaks without significantly lowering the signal’s amplitude. This also means that KPIs such as RMS, peak-to-peak, and peak will be bounded by the Dynamic Range and cannot exceed that value. Kurtosis will also be suppressed since the peaks in the signal get cut off.

Conclusions

Setting up the dynamic range can be a challenging task and it can give a false impression of what is happening in the machine. To set it up appropriately one needs to know some details of the machine and perhaps test a few different settings. It is also hard to tell by a spectrum if the dynamic range is set correctly or not since even a very distorted signal can create a spectrum that looks good. The only way to ensure it is set correctly is to check every waveform. Fortunately, MultiViz already checks each measurement and warns if it is too low or too high, allowing you to easily adjust.

The post How to set the dynamic range and what are the consequences of poor setting appeared first on Viking Analytics - MultiViz AI.

Duration

The frequency resolution will be inverse of the duration (frequency resolution = 1/duration), in other words, if we have a duration of 1 s, the distance between two points in the FFT will be 1 Hz away from each other, and if you have a duration of 2 s, the distance between two points in the FFT will be 1/2 Hz from each other.

To demonstrate the effects of varying duration, we constructed synthetic data with a rotational speed of 58.4 Hz and vibrations from the electrical network at 60 Hz, including harmonics.

Here, we observe a large, wide peak at 59 Hz and a double peak at 118 Hz and 120 Hz. Since we know the data we know that the first peak is two that have merged and the double peak is also two that have almost merged.

In the second case, we notice that the first peak is splitting up and the second peak is two different.

In the last case, the first peak has split up and the second is two different. We also observe that the signals stand out more from the noise, as the peaks become sharper with a longer duration.

Connection between Duration and Sample Rate

Many sensors have made a connection between duration and sample rate. This is to make sure users can not have high sample rates and duration at the same time since that would require significant memory for the sensor and deplete the battery faster. This usually means that if you increase the duration you will lower the sample rate thereby keeping the number of data points in the measurement constant.

Changing Duration and Sample Rate

If you compare the different spectrums from both this article and last week’s article you see that the picture you get of the machine vibrations is quite different, even though this is the same signal behind. This means that you cannot compare signals with different sample rates and durations. Therefore, it is important to set up these settings right from the beginning to avoid neglecting large amounts of data gathered, should you later change settings after discovering they were initially incorrect.

Trends will also be invalid for comparison across different sample rates, effectively negating the purpose of saving historical data for future reference.

Conclusions

Duration and Sample rate are two of the most central settings in vibrational monitoring and it is crucial to get them right from the beginning to not have to change them and discard historical data.

To be able to set them right it is important to answer a few questions

1. What is the expected rotational speed? If it is low or close to the electrical frequency, a longer duration is important, if it is high and far from the electricity grid frequency, a longer duration is not important.

2. What are the most likely failures I want to keep track of? If the primary faults manifest in low frequencies (e.g., unbalance, misalignment, looseness) – a longer duration is needed. If it is mostly faults manifesting in higher frequencies (bearing failures, lubrication failures, etc.), a higher sample rate is needed.

3. Do I expect peaks close to each other at low frequencies? If yes, a longer duration is needed.

The post How does duration impact the spectrum appeared first on Viking Analytics - MultiViz AI.

The sampling rate or sampling frequency is crucial for acquiring high-fidelity measurements. It defines the interval at which vibration data is recorded, directly affecting the precision and reliability of signal analysis. Choosing the right sampling frequency is imperative for accurately reflecting the machine’s condition, facilitating early fault detection and effective maintenance.

Inadequate sampling frequency can result in the loss of critical details, leading to incomplete or erroneous analysis. Conversely, excessively high sampling rates may introduce high-frequency noise, complicating data interpretation and escalating costs due to the necessity of advanced hardware. Comprehending and implementing the optimal sampling frequency is vital for vibration experts to monitor machinery health and detect potential issues promptly.

Determining the appropriate sampling rate for a sensor involves balancing the expected vibration frequencies of the machinery (such as bearing characteristic frequencies) and the sensor’s limitations (including transmission, storage, mounting, and hardware constraints). Understanding the sensor’s maximum frequency (fmax) dictated by its hardware is essential when setting the sampling rate. The chosen sampling rate significantly impacts both vibration metrics and spectral analysis.

There are many different effects of different sample rates for example under or over sampling but this article will focus on how it affects the vibration metrics.

Effects on vibration metrics

If we look at a trend of different vibration metrics, we see that the metrics capturing amplitude are increasing with time. This could be signaling some issue with the machine. We will see something interesting if we look at a measurement of the trend’s beginning, middle, and end.

Figure 1: Four different vibration metrics over time.

However, the vibration metrics capturing how peaky the signal is have not changed significantly over time.

Conclusions

We see that the sample rate used affects both the spectrum and the vibration metrics capturing amplitude. Therefore measurements with different sample rates are unfit for comparison, both in terms of the vibration metrics measuring amplitude and in terms of spectrum. Even the parts under the Nyquist frequency can differ if the signal has components that are of higher frequency than the Nyquist frequency. This also means that, when setting thresholds, the sample rate must be considered, and the thresholds must be updated if the sample rate is changed.

The post How does the Sample rate affect the data? appeared first on Viking Analytics - MultiViz AI.

In today’s industrial landscape, sensor technology has revolutionized the way we monitor and maintain machinery, from manufacturing plants and power stations to oil refineries and transportation systems. These sensors generate vast amounts of data, capturing intricate details about machine performance, operational efficiency, and potential faults. As this data becomes increasingly integral to industrial operations, a fundamental question arises: who should own this data? Historically, practices have often seen data collected from devices like vibration data collectors being locked inside those devices, with no easy way to extract it. However, in today’s world and looking towards the future, we advocate for businesses to own machine sensor data. This paradigm shift empowers businesses to take control of their data destiny, with unrestricted access as the cornerstone principle for unlocking the true potential of machine sensor data.

Privacy and Autonomy

• Operational Data Protection: Machine sensor data often includes highly sensitive information, such as performance metrics, maintenance schedules, and usage patterns. Allowing businesses to own this data ensures they have control over who accesses their operational information and how it is used.

• Consent and Control: Ownership means businesses can decide how their data is shared and with whom, empowering them to consent to data use in a manner that aligns with their operational preferences.

Economic Benefits

• Value Realization: Machine sensor data can be valuable for various industries, such as manufacturing optimization, maintenance services, and operational efficiency studies. Allowing businesses to own and manage their data ensures they can realize the full value of the data they generate.

• Monetization Opportunities: By owning their machine sensor data, businesses can potentially monetize it by selling access to third parties or participating in data marketplaces. This creates a new revenue stream, shifting economic benefits from corporations to data generators.

Transparency and Trust

• Building Trust: Data ownership fosters transparency in how data is collected, stored, and used. When businesses have ownership, they are more likely to trust the devices and services they use, as they have assurance over the data handling processes.

• Reducing Exploitation: Companies may exploit machine sensor data for proprietary benefits or competitive insights without explicit consent. Ownership rights reduce the risk of such exploitation, as businesses can set boundaries on data usage.

• Limiting Unwanted Usage: When businesses own their data, they can prevent it from being used in ways they do not approve of, such as for competitive intelligence or unauthorized research.

• Ethical Use: Businesses can ensure their data is used ethically and in ways that align with their values.

Legal and Ethical Considerations

• Compliance with Regulations: Increasingly, laws and regulations around the world recognize the importance of data privacy and user control. Allowing businesses to own their machine sensor data aligns with these legal frameworks and helps companies stay compliant.

• Ethical Responsibility: Ethically, it is reasonable to argue that businesses should have rights over data derived from their operations and machinery. This respects their autonomy and inherent rights to operational data.

Unrestricted Access to Data

• User Empowerment: Owning their machine sensor data allows businesses unrestricted access to their information, enabling them to make informed decisions about maintenance schedules, operational efficiency, and potential upgrades. For instance, businesses can analyze their operational data to optimize performance and extend machinery lifespan.

• Historical Data Access: Having ownership means businesses can maintain access to their historical data indefinitely. This long-term access is crucial for tracking changes over time, such as performance metrics, which can be valuable for diagnosing issues or monitoring improvements.

Innovation and Competition

• Promoting Innovation: With businesses owning their data, new business models and services can emerge that prioritize user empowerment and data protection. This can drive innovation in how data is managed, analyzed, and utilized.

• Enhancing Competition: Data ownership can reduce the monopolistic control of big tech companies over operational data. This encourages competition and allows smaller companies to compete more fairly in the marketplace by offering services that prioritize user data rights.

Additional Reasons for Customer Ownership of Machine Sensor Data

Customization and Personalization

• Personalized Services: Ownership of machine sensor data enables businesses to share their data selectively with service providers who can then offer personalized and improved services. For example, maintenance providers can offer tailored service plans based on precise operational data shared by the business.

• Control Over Personalization: Businesses can decide how much personalization they want in their services. They can choose to share data to get personalized recommendations while maintaining the ability to stop sharing when they wish.

Interoperability and Portability

• Data Portability: Businesses can transfer their data between different service providers without losing valuable information. This promotes interoperability and allows for a seamless transition between services, enhancing user experience and reducing dependency on any single provider.

• Standards and Formats: Ownership can drive the adoption of open standards and formats for machine sensor data, making it easier for different systems to work together and for businesses to utilize their data across various platforms and applications.

Advocacy and Collective Bargaining

• Strengthened Advocacy: Individual data ownership can lead to collective bargaining power. Groups of businesses can form data cooperatives to negotiate better terms with service providers, ensuring more favorable and fair treatment of their data. Furthermore, when businesses own their data, they can advocate more effectively for policies that protect data rights and privacy.

Resilience Against Manipulation

• Resisting Manipulative Practices: Data ownership helps businesses resist manipulative practices by companies, such as price discrimination or operational manipulation through targeted content. Ownership ensures transparency and accountability in how data influences operational decisions.

Environmental and Social Impact

• Sustainability Initiatives: Businesses can choose to share their data with organizations that promote environmental sustainability or social impact projects. For example, operational data can be used to optimize energy usage and contribute to environmental conservation efforts.

• Community Projects: Ownership allows businesses to contribute their data to industry-based projects, such as operational efficiency studies or regulatory compliance initiatives, fostering industry development and collective well-being.

Innovation in Data Services

• New Market Opportunities: With business-owned data, new business models and services can emerge that prioritize data privacy and user control. Startups and innovators can develop products that cater specifically to users who demand high standards of data ownership and privacy.

Conclusion

Advocating for customer ownership of machine sensor data is both a logical and ethical stance that aligns with the principles of operational autonomy and data privacy. By granting businesses ownership of their data, we empower them to control how their information is used, ensuring that it serves their best interests rather than merely fueling corporate profit. This shift not only enhances transparency and trust between businesses and service providers but also fosters innovation. Ultimately, when businesses own their machine sensor data, it reinforces the fundamental right to data privacy in the industrial age, promoting a more equitable and respectful technological landscape.

The post Data ownership appeared first on Viking Analytics - MultiViz AI.

Wireless vibration sensors are increasingly critical for machinery maintenance and monitoring in the rapidly evolving industrial sector. These sensors provide real-time data essential for predictive maintenance, helping to prevent costly downtimes and enhance operational efficiency. Despite their benefits, sensors’ diversity and software systems present significant challenges and opportunities.

Today’s Fragmented Ecosystem: Challenges and Opportunities

Advantages of Diversity

The current landscape of wireless vibration sensor systems features diverse software solutions, each designed to meet specific customer requirements while helping to optimize operational requirements.

Utilizing multiple systems can enhance operational resilience while being able to cater to different processes and machinery. This setup also avoids vendor lock-in, mitigating risks associated with dependency on a single provider. This fragmentation also allows for cost-effective scaling across a diversity of processes and equipment.

Disadvantages of System Diversity

However, managing a fragmented array of systems increases operational complexity and the likelihood of integration issues, complicating data aggregation and analysis. This can lead to higher long-term costs, as maintaining several software systems is complicated and distracts from core business functions, potentially impacting productivity.

Separate systems also create data silos, obstructing a unified operational view and slowing down decision-making processes. Moreover, diverse systems increase the risk of data breaches.

Our Future-Ready Solution: A Unified Monitoring Platform

To address these challenges, we have developed an innovative, comprehensive, and sensor-agnostic platform specifically designed for wireless vibration sensors in industrial settings. This solution is crafted to streamline the monitoring process at every layer, ensuring that businesses can fully leverage the potential of IoT and predictive maintenance.

Key Features of Viking Analytics MultiWiz platform:

• Advanced Analytics and Threshold-Free Smart Alarms: Powered by AI, our platform predicts potential failures and triggers health-related warnings automatically, irrespective of the machine type or sensor brand. No manual setting of thresholds is needed.
• Seamless Sensor Integration: Our platform supports various sensors regardless of their manufacturer, facilitating easy integration of legacy systems while standardizing new deployments.
• Minimal Dependency on Machine Meta Information: Our models primarily rely on raw vibration data, not on the meta information of the machines.
• Centralized Data and Model Management: Our solution enables customers to centralize data management using standard solutions available in the market while allowing the development of predictive models based on collected data and utilizing third-party models from OEMs or other vendors.
• Scalability and Customization: The platform easily scales to accommodate growing data volumes and diverse operational needs, ensuring specific monitoring requirements are met.

Impact and Benefits

• Reduced Complexity: Our platform simplifies monitoring by eliminating the need to manage multiple software systems and data silos.
• Faster Response Times: Integrated data analytics enhance decision-making capabilities, reducing downtime and extending machinery lifespan.
• Cost Efficiency: The system improves maintenance schedules and reduces unnecessary repairs, thereby cutting down operational costs.

Conclusion

For industrial companies aiming to fully leverage IoT and predictive maintenance, our unified platform for wireless vibration sensors is not just a technological upgrade—it’s a strategic overhaul. By consolidating sensor integration and data analysis with minimal dependency on machine metadata and threshold settings, our solution empowers businesses to proactively manage their machinery, ensuring operational continuity and driving significant improvements in efficiency and productivity. This comprehensive approach not only streamlines operations but also sets a new standard for industrial monitoring systems.

The post Revolutionizing Industrial Monitoring appeared first on Viking Analytics - MultiViz AI.

Vibration analysis is a superhero in condition monitoring and predictive maintenance, but its true power lies beyond basic trends and vibration metrics. While these metrics offer a general sense of machine health, the real story unfolds in the depths of spectra and time waveforms. This article explores why evaluating these details is crucial for early fault detection and avoiding potential disasters, and why raw vibration data and spectra are not being utilized to their full potential.

Comfort Zones and Missed Opportunities

Many vibration analysts, especially those without a deep understanding of machinery or extensive experience, rely heavily on pre-defined “diagnostic wall charts.” This approach is akin to playing a game of reaction—reacting only when a pre-determined threshold is crossed. Imagine a doctor relying solely on a fever chart to diagnose an illness. Valuable clues, like specific rashes or coughs, could be missed. The problem? Subtle warning signs and pre-failure patterns often appear in the spectra and time waveforms well before a vibration metric triggers. Spectra and time waveforms reveal unique signatures of developing faults, allowing for early intervention and preventing breakdowns. While the severity of the condition may not immediately warrant action, identifying early symptoms of potential failures is critical.

The Upskilling Challenge

Vibration analysis is a complex art; interpreting raw data can be daunting. Training offers a solid foundation, but true mastery requires a commitment to lifelong learning. Analyzing real-world scenarios with spectra and time waveforms is crucial for applying theoretical knowledge to practical situations. Unfortunately, access to diverse case studies can be limited.

Moreover, the field is constantly evolving, with new fault signatures emerging. Analysts need the right tools and data skills to navigate this ever-expanding landscape, especially in today’s world where the amount of data per machine is exponentially greater and the number of monitored machines has increased significantly. Targeted training in spectrum and time waveform analysis, load balancing, root cause analysis using raw data, and continuous knowledge exchange are essential but not always readily promoted or available.

Unlocking the Full Potential

Vibration analysis is a powerful tool, but its effectiveness hinges on the analyst’s expertise. Instead of solely relying on wall charts, investing in in-depth analysis of spectra and time waveforms through case studies, advanced training, and continuous learning programs is paramount.

Imagine an analyst not just seeing a blurry picture (the trend) but a high-resolution image (the spectrum) revealing minute details. This empowers them to make informed decisions about machine health and recommend the most effective course of action.

By embracing the full spectrum of vibration analysis techniques, organizations can unlock the true potential of this practice, optimizing machine health, uptime, and ultimately, their bottom line.

The post Beyond trends and vibration metric analysis of vibration data appeared first on Viking Analytics - MultiViz AI.

In the last article, we reviewed one of the important fundamental topics in vibrational condition monitoring, vibration metrics. This week we are going to explain the limitations to what you can see in the spectrum, explain Ailiasing, and why some hardware designs have been made.

Aliasing and how sensors have solved it

How well we can capture a signal depends on how often we sample it. If we sample it more often we can capture it more accurately. However, a higher sample rate comes at a cost which we will discuss in our upcoming articles. The Nyquist theorem states that an analog signal can be digitized without distortion if and only if the sampling rate is greater than or equal to twice the highest frequency component in a given signal. So to be able to tell which frequencies compose a signal we need to sample it twice in each period. Half the sample rate is also known as the Nyquist frequency. The Nyquist theorem has a rigorous mathematical foundation but what is important for us to know is that to digitalize analog data fully and without corruption, one needs to sample it at least twice its highest frequency. The distortion caused by a sampling rate of less than half of the highest frequency of the signal is called Aliasing.

Let us look at a practical example; you see a signal with one 1 Hz oscillation and one 5 Hz oscillation combined. As the highest frequency component for the signal is 5 Hz, we want the Nyquist frequency to be 5 Hz. This means we must at least sample his signal at 10Hz to avoid distortion. Below is the original signal sampled at 30 Hz and less frequently sampled versions sampled at 15Hz and 7.5 Hz.

Figure 1: The same signal captured at three different sample rates.

We see that the original signal sampled at 30 Hz displays both frequency components clearly. In the second example, the sampling rate is 15 Hz yet both 1Hz and 5Hz components of the data are captured in the sampled data. In the third example, the sampling rate is reduced to 7.5 Hz which gives the Nyquist frequency to be less than 5 Hz. We can no longer reconstruct that original signal and the data has been distorted. This is because the signal is sampled just under twice per period. The spectrum of the third example is shown in Figure 2.

Figure 2: The spectrum of the signal captured with the lowest sample rate.

Figure 2 gives a false impression that we have a signal that has both 1 Hz components and 2.5 Hz components I.e. the 5 Hz component appears wrongly at 2.5 Hz. This is caused by Aliasing which is when parts of the signal have higher frequencies than the Nyquist frequency ending up distorting the sampled signal.

To solve this the vibration sensors have a Low Pass filter filtering out everything above the Nyquist frequency, reducing the problem of aliasing. This removes the parts of the signal that have frequencies higher than the Nyquist frequency just leaving the parts the sensor can capture. If we use a low pass filter that removes everything over 4 Hz on the signal above we get this.

Figure 3: The signal after passing through a low pass filter in different sample rates.

Here we can still extract the same information from each figure even though it is sampled at different rates.

Conclusions

Aliasing and the Nyquist frequency are known and well-understood phenomena in signal processing. However, we will also feel their effects when we do vibration condition monitoring. It is therefore important to understand the concepts and the solutions that have been made to minimize these effects.

The post Vibrational condition monitoring fundamentals – Aliasing and the solution appeared first on Viking Analytics - MultiViz AI.

A new agreement has been signed between Viking Analytics and Bharat Forge Kilsta (BFK) from Karlskoga. The agreement, which is for three years, provides BFK with the AI-based optimization tool “Smartforge” after a 10-month implementation phase. Smartforge optimizes the forging process, primarily in the critical heat keeping process where the problems with scrap are greatest. The goal is to reduce discarded products by 50% and contribute to energy savings and a more environmentally friendly production.

Niclas Undén, CFO of Bharat Forge Kilsta, comments on the deal: “Through AI technology, a difficult step in the forging process is simplified. The result is lower scrap, lower energy consumption and reduced need for manual work. In SmartForge, Swedish heavy automotive industry meets world-leading AI technology from Viking Analytics. Bharat Forge Kilsta is very pleased with the collaboration with Viking Analytics, and we look forward to a deeper collaboration in the coming years.”

The majority of Bharat Forge’s customers are in the automotive industry and the value of this agreement exceeds SEK 4 million for both Viking Analytics and Bharat Forge.

Stefan Lagerkvist, COO at Viking Analytics: “This agreement is much more than a single business opportunity. Bharat forge has a lot of expertise in steel and forging, which contributes strongly to the solution. Their knowledge has been captured and translated into algorithms for better control of the process. This collaboration confirms everything we so long have been fighting for and gives us a great opportunity in the future to offer an environmentally friendly AI-powered solution to more factories within the Bharat Forge Group as well as to other players in the industry!”.

Viking Analytics
Strong in predictive maintenance and smart industrial optimization

Since 2017 the Swedish company Viking Analytics has been at the forefront of revolutionizing the maintenance process for OEMs, maintenance companies and industries. Their commitment to predictive maintenance, smart automation, optimization, and data analytics is evident in their specialized software tool MultiViz, which enables industries to operate, monitor, and understand their machines with unparalleled precision and efficiency. Vibration analysis is a major focus area, but a lot of customized AI solutions are also provided.

Bharat Forge Kilsta
Forgings for the automotive industry

Bharat Forge Kilsta manufactures forged and machined components for the automotive industry. The company’s most important customers are truck manufacturers in Sweden and internationally. Bharat Forge Kilsta is part of the Bharat Forge Group, which is the world’s largest forging group and is headquartered in India. Bharat Forge Kilsta has an annual turnover of SEK 1.3 billion and 320 employees. The Swedish company is located in Karlskoga – the city in Eastern Värmland that is known for its high-tech, and internationally oriented, industrial companies.

The post Viking Analytics & Bharat Forge Kilsta (BFK) appeared first on Viking Analytics - MultiViz AI.