Introduction: The Role of Infrared Inspections and the Challenges They Face

Infrared inspection (thermography-based inspection) has played an important role in the maintenance of equipment and infrastructure. It can non-contact detect abnormalities that are difficult to catch by visual inspection, such as abnormal heating in electrical equipment or wear on mechanical parts, and is widely used as a safe and efficient method to pinpoint problem areas. Nevertheless, despite its usefulness, several challenges have been pointed out in the field.

First, infrared inspection tends to be a highly subjective task. Correctly interpreting the temperature distribution shown by a thermographic camera and discerning signs of anomalies requires knowledge and intuition backed by experience. For example, an experienced technician can judge “this level of temperature rise is within acceptable limits” or “the heating at this location is due to load conditions,” but a novice may find such judgments difficult, making the decision process dependent on the individual.

Second, there is the problem of insufficient record keeping. Even if many infrared images are taken in the field, data other than suspected abnormal locations are often summarized briefly in reports, and detailed images or measurement values may not be sufficiently stored or shared. This makes it impossible to compare with the previous inspection next time, and it becomes difficult to quantitatively track changes in equipment condition. As a result, there is a risk of overlooking gradual degradation or small temperature increases (missed changes), making it impossible to grasp serious signs. Because past signs are not recorded, incidents can appear as if they “failed suddenly without warning.”

Furthermore, while infrastructure and equipment are aging and the number of inspection targets is increasing, the shortage of experienced technicians is also becoming serious. If issues such as subjectivity and insufficient record keeping are left unaddressed, the burden on site will increase and the risk of troubles caused by oversights may rise. Therefore, there is growing attention on attempts to elevate infrared inspection from mere defect detection to “predictive maintenance” by combining it with digital technologies to accumulate and analyze data. This article explains how infrared inspection is evolving from a simple means of finding problems to a “predictive inspection” that leverages time-series data and location information, and the technologies that hold the key.

Note that while other methods such as vibration sensors or IoT temperature sensors can be considered for detecting precursors to equipment abnormalities, infrared inspection has the advantage of being non-contact and able to check a wide area at once, and can flexibly handle locations where sensor installation is difficult. By leveraging these strengths and using the data, condition monitoring that is easy for anyone to use becomes possible.

“Predictive Inspection” Born from Infrared Images × Location Information × Time-Series Data

So what is “predictive inspection”? Simply put, it is an inspection method that adds location information and time-series (temporal) perspective to image data obtained by infrared inspection, allowing you to capture changes in equipment condition in three dimensions. Traditionally, thermography-based inspections focused on finding abnormal locations on the spot. In predictive inspection, infrared images taken at each inspection are linked to when and where they were acquired and accumulated and managed as time-series data from past to present.

By accumulating data in this way, you can visualize subtle changes such as “a specific connection point inside the same distribution panel has risen by 5 ℃ compared to last year” or “the maximum temperature at a motor bearing has gradually increased over every six months.” Even locations that looked normal during a single inspection may reveal signs of deterioration when the time-series data are reviewed. Detecting such early signs of anomaly is precisely the aim of predictive inspection. It enables planning maintenance early before abnormalities develop into major failures or accidents.

An important element of the predictive inspection concept is “location information.” It is necessary not only to compare past data but to clearly identify which part of which equipment the data corresponds to. If precise location tags are attached to infrared images, data can be linked not only temporally but also spatially. This allows accurate tracking of the aging of the “same place and the same object,” and consistent comparisons even when inspectors change. For example, even if hundreds of infrared photos are taken in a large factory, if it is managed on a map or drawing where each photo was taken within the facility, it becomes easy later to determine “where the problem is” and “which part was photographed last time.”

In this way, predictive inspection that incorporates location information and a time axis into infrared inspection can be seen as a step beyond conventional preventive maintenance (periodic inspections to prevent failures) toward condition-based maintenance (maintenance based on changes in equipment condition, i.e., predictive maintenance). The important point is that you do not need to install special fixed sensors; by intelligently using data from human-conducted patrol inspections, you can visualize equipment condition and catch trends in degradation early. In other words, turning routine inspection work into data-driven processes and obtaining objective materials for decision-making that do not rely on experience and intuition is the core of predictive inspection.

The Significance of Attaching High-Precision Position and Orientation to Infrared Images with LRTK

To realize predictive inspection, it is necessary to link high-precision position and orientation information to infrared images. The technology that holds the key is “LRTK.” LRTK is a system that makes high-precision positioning technology (RTK-GNSS) easy to use with a smartphone; with a dedicated compact receiver and an app, it transforms a smartphone into a centimeter-class surveying instrument (cm level accuracy (half-inch accuracy)). In other words, you can obtain high-precision positioning and capture images with just a smartphone without carrying heavy surveying equipment. Ordinary GPS has an error on the order of several meters (several ft), but using LRTK you can obtain the current position outdoors on a map coordinate within a few centimeters (a few in). Combined with smartphone sensors, you can also record the camera’s direction (azimuth/tilt), so you can keep detailed information on “where the photo was taken and which direction the camera was pointing.”

If you apply this capability to infrared inspection, each thermography image can be clearly labeled with “when, where, and in which direction” it was taken. For example, an infrared image of piping taken in a vast plant could carry a position and orientation label such as “taken near Tank A in the southeast area of the plant, facing northwest.” Information that used to be handwritten as “near Tank A” will be automatically saved as precise coordinate data, preventing data omissions or incorrect entries. Especially in large-scale infrastructure where many similar pieces of equipment exist, it can be confusing which image corresponds to which equipment, but high-precision position tags clarify the linkage between photos and equipment and make searching and integrating past data easier.

Also, centimeter-level positioning (cm level accuracy (half-inch accuracy)) by LRTK makes it possible to overlay multiple data sets spatially. You can compare infrared images taken at different dates on geographic coordinates or plot them on site drawings and 3D models, making time-series change analysis intuitive. For example, you could overlay the previous and current thermography images and visually check the differences in temperature distribution.

Furthermore, the high-precision position information is highly compatible with AR (augmented reality) technology. Coordinates obtained with LRTK are based on absolute real-world coordinates (such as the World Geodetic System), so they can be displayed in AR systems without misalignment. While conventional AR relied on marker alignment or simple magnetic-sensor-based positioning, using LRTK’s precise positioning considerably improves the reliability of integrating the field and digital information in displays. I will explain AR utilization in more detail below, but in any case, LRTK endows infrared inspection data with “positional certainty,” which is an important foundation for predictive inspection data.

Inspection Style Evolving with Cloud and AR: Visualization, Comparison, Sharing

Location-tagged infrared data collected with LRTK can be centrally managed on a cloud platform. This allows the vast images and measurement results obtained in the field to be accumulated as a digital history and retrieved and utilized whenever needed. Inspection records that used to lie dormant on file servers or paper reports are organized on the cloud by equipment and location, making comparisons with the past and trend analysis dramatically easier. For example, opening an inspection page for a piece of equipment could display infrared images in chronological order, temperature trend graphs, and comments—data stored in a state where it is “visible.”

Moreover, having data on the cloud greatly smooths information sharing. Even remote offices or distant sites can instantly check the latest inspection results and easily seek expert opinions when necessary. If a field inspector uploads an infrared image, an experienced technician can later check it on the cloud and give advice, enabling remote collaboration. This allows less experienced on-site staff to proceed with inspections while receiving appropriate support, reducing subjectivity and aiding human resource development.

The use of AR further transforms inspection styles. When you view the site through a smartphone or tablet, past inspection data and information on abnormal locations stored in the cloud are overlaid in the real world. For example, past temperature anomaly locations for pipes behind a wall might appear as AR markers, or the position of a previously detected hot spot on the equipment in front of you could be shown as a colored silhouette. Inspectors can refer to history information while looking at the actual object, allowing them to intuitively compare current and past conditions and perform checks without overlooking anything. If AR arrows or guide lines direct you to the next inspection point to confirm, you could carry out patrol inspections efficiently along a route without omissions.

In this cloud-and-AR combined inspection style, visualization and sharing of inspection results are performed in real time and interactively. Observations or discovered signs felt on site are digitally recorded on the spot and clearly presented to everyone via AR. Information such as “how hot” or “where” something is, which was difficult to convey with paper reports, can be shared together with visual context so that all stakeholders can consider countermeasures with a common understanding. In short, by seamlessly connecting the field, data, and people, you can realize a highly productive and reliable inspection workflow.

Benefits of Introducing Predictive Inspection and the Changes It Brings to the Field

Introducing predictive inspection that incorporates the latest technologies into the field creates various positive changes in traditional inspection work. Here are four particularly important benefits, each presented with concrete examples.

• Precision improvement: The accuracy and consistency of inspection data improve dramatically. For example, differences in temperature measurement results caused by different inspectors or shooting angles used to occur, but by measuring the same location based on high-precision position and orientation information, data variability is reduced. As a result, there is less doubt about “is this really an anomaly, or is it due to measurement conditions,” enabling more reliable decision-making.

• Early detection of anomaly signs: You can capture small signs of anomalies without missing them. Because changes can be tracked as time-series data, you can grasp slight temperature rises or distribution changes before anomalies become apparent. For example, at a certain substation, it was discovered that the temperature of a connection terminal was increasing by 2-3 ℃ at each inspection, and parts were replaced before failure or smoking occurred. Predictive inspection enables planned preventive measures and reduces the risk of major accidents or sudden shutdowns.

• Labor savings and efficiency improvements: The efficiency of the inspection process itself increases, reducing the burden on personnel. Digitalization reduces unnecessary rework and duplication—such as AR navigation guiding inspection routes and automatic recording cutting report-creation time. For example, there are reports that patrol thermography inspections that used to require two people working together for half a day were completed in a short time by one person with just a tablet. Also, when an anomaly is found, internal communication and consultation can be shared and instructed immediately via the cloud, shortening the lead time to response.

• Easier human resource development and knowledge transfer: A digitalized inspection foundation contributes to human resource development. Newcomers can proceed with inspections by following on-site guidance provided by AR, accelerating proficiency compared to traditional on-the-job training where they learned by watching veterans. Furthermore, past inspection data accumulated on the cloud can serve as educational material for learning what kinds of abnormalities appear and how. By sharing veterans’ accumulated know-how as data, you can prevent knowledge gaps due to generational change and raise the overall technical level of the organization.

As described above, introducing predictive inspection brings tangible benefits to the field. You can feel the results of DX (digital transformation) across various aspects: improved accuracy and safety, operational efficiency, and human resource development. As inspection work shifts from “relying on intuition and experience” to “supported by data and technology,” equipment maintenance operations can achieve a higher level of reliability and productivity.

Conclusion: A Step Toward DX by Using LRTK in Daily Patrols

As a solution to realize predictive inspection, combining infrared inspection with LRTK, cloud, and AR strongly supports DX in equipment maintenance. In fact, efforts to advance maintenance through digital technology (so-called maintenance DX) are being promoted in the infrastructure maintenance field, and while DX often conjures images of large-scale system implementation or organizational reform, it can actually be started gradually from daily patrol inspections. For example, simply carrying a smartphone and an LRTK device during your usual rounds and quickly obtaining position and taking a photo when you find a piece of equipment of concern—that alone is a valuable accumulation of digital data. If you routinely digitize site information through such simple surveying and inspection, you build a foundation for later analyzing trends or sharing insights with other personnel.

What matters is introducing technology in a way that can be accepted on site without strain. With infrared inspection that includes position information using LRTK, field staff can intuitively use the tools and engage without burden. Even if you start with a limited set of equipment, once you feel the effect of data accumulation you can gradually expand the scope. In the process, the quality of inspection work improves, resistance to DX diminishes, and a culture in which data utilization is taken for granted will naturally take root.

Predictive inspection, an evolutionary form of infrared inspection, is a forward-looking approach achievable through the daily accumulation of on-site improvements. It may be a small step, but those steps add up to significant future results. For example, in the future you can expect developments such as AI analyzing accumulated data to automatically detect signs of anomalies or using the data as part of a digital twin of equipment. Why not take a step from familiar surroundings and begin predictive inspection using the power of infrared inspection × location information × time-series data?