Table of Contents
- 1 Background of the demonstration of the FAST-D Monitoring Edition, a noise detection solution
- 2 Yamagata Prefectural Enterprise Bureau, Waterworks Division Endorsement
- 3 Social Issues Related to Maintenance Operations
- 4 Basic features available in the FAST-D Monitoring Edition
- 5 What is noise detection?
- 6 What is the "FAST-D" noise detection platform?
- 7 About Hmcomm
- 8 About Yamagata Prefectural Enterprise Bureau
- 9 About National Institute of Advanced Industrial Science and Technology (AIST)
Background of the demonstration of the FAST-D Monitoring Edition, a noise detection solution
Hmcomm has proposed the FAST-D Monitoring Edition, which uses artificial intelligence (AI) to remotely detect and monitor abnormal noises in order to solve the issues that the Yamagata Prefectural Enterprise Bureau is working on to improve the efficiency of waterworks facility management operations.
Based on the actual results of detecting abnormal noises during the evaluation use of FAST-D Monitoring Edition from August 2022 to March 2023 with the Yamagata Prefectural Enterprise Bureau, we have decided to continue providing the FAST-D Monitoring Edition service in FY2023.
Yamagata Prefectural Enterprise Bureau, Waterworks Division Endorsement
In the monitoring of water pumps, we were able to detect the operating noise of drainage pumps in the water pump room as abnormal noise, and we feel that this could be utilized in solutions that help realize CBM, such as failure prediction and preliminary maintenance. We are looking forward to accumulating more results and know-how through the demonstration experiment in FY2023 and utilizing them in actual CBM.
This will lead to (1) labor saving and efficiency improvement in facility maintenance work, (2) early detection of abnormalities based on sound data, and (3) provision of water infrastructure maintenance services of stable quality and ensuring user safety.
Social Issues Related to Maintenance Operations
Labor shortage in the field
The shortage of field workers is one of the top concerns of field personnel, and the fact that 89.71 TP3T personnel responded that it is difficult to recruit field workers indicates that it is very difficult to solve the chronic shortage of manpower.
In addition, the ratio of workers aged 60 or older is approximately 37.21 TP3T or higher, which means that retirement needs to be addressed, but the person in charge of approximately 77.21 TP3T responded pessimistically that it is difficult to rejuvenate on-site employees, suggesting that it will take time until drastic improvements are made. (*1)
High cost of training and high losses
The turnover rate of workers is 18.71 TP3T, which indicates that the challenge related to labor shortage is not only limited to the difficulty of recruitment, but also the inability to fully retain human resources. 1 in 5 workers leave the company, which suggests that although sensory knowledge and experience are important in this work, the cost of education and training is a significant loss. The fact that one out of every five employees leaves the company suggests that, despite the importance of sensory knowledge and experience in this field, there is a significant loss in training costs. (*2)
Difficulty in abstract communication
The fact that 72.21 TP3T companies are positively considering accepting foreign technical interns into their building maintenance operations suggests that work teams will be composed of more diverse values than ever before, including cultural and social backgrounds.
While most inspection items related to "sound" are abstract expressions, it is expected that quantitative communication based on numerical information will become increasingly necessary to maintain maintenance quality. (*3)
(*1): Excerpt from the Building Maintenance Information Yearbook 2023 (53rd Fact-Finding Survey Report) published by the Japan Building Maintenance Association
(*2): Excerpt from the Ministry of Health, Labour and Welfare's Summary of Results of the 2021 Survey of Employment Trends
(*3): Building Maintenance Information Yearbook 2020 (50th Fact-Finding Survey Report), published by the Japan Building Maintenance Association - Intention to accept foreigners with "specified skills" residence status at establishments with monthly sales of 100 million yen or more (total of "considering on the premise of accepting" and "currently studying the surrounding situation")
Basic features available in the FAST-D Monitoring Edition
(1) Streamline inspection operations
Remote inspection efficiency
By combining with a SIM router or other network equipment, multiple locations can be monitored simultaneously by sound, regardless of where they are normally inaccessible or far away.
Easy to add or remove units
Initial costs are kept to a minimum, allowing for a more flexible business response than hiring workers. Installation and configuration can all be handled by on-site workers, allowing the site side to proceed with implementation at their own pace.
(2) Information sharing based on analysis data
Visualization of analysis data
The change of AI scores can be checked with numerical information for each time series, and sound information on items that AI detects as "unusual" can be shared with team members in a visible form. Since information can be shared correctly while looking at the data, it can be used to quantitatively show tacit knowledge (know-how) about abnormal conditions and for time-series comparisons.
Improved efficiency of communication
For those with high AI scores, the recorded data can be downloaded so that the actual voice data can be used to explain the failure to all concerned, no matter when the failure occurs, 24 hours a day, 7 days a week.
Related Press Releases
Hmcomm and Yaskawa Electric Corporation begin joint development of "AI Noise Detection in Product Completion Inspection".
Hmcomm launches "FAST-D Monitoring Edition," an easy-to-start noise detection service - AI noise detection technology enables DXing of maintenance operations, including predictive maintenance and predictive detection.
Hmcomm and Eiraku Electric begin work on equipment noise detection
Hmcomm and Harada Sangyo begin collaboration on sales
What is noise detection?
AI noise detection is a technology that helps in stable monitoring, abnormality detection, and predictive detection by machine learning the sounds that machines, objects, and organisms make when they are operating normally and the sounds they make when they are in abnormal conditions.
When a person judges whether a sound is normal or abnormal by listening to it, there are issues such as ambiguous judgment criteria, cases where variations occur, and the need for skillful techniques. The detection of abnormal noise enables quantitative analysis that does not depend on human labor.
This is an initiative to have AI learn the findings that are determined by the ears of skilled craftsmen, and is based on the concept that "everything that can be heard by the human ear can be detected.
Noise detection by sound can be used in a wide range of industries and business categories, including abnormality detection in factory infrastructure, machine noise detection, footsteps, crime prevention, human emissions, and animal noises.
What is the "FAST-D" noise detection platform?
FAST-D (Flexible Anomaly Sound Training and Detection,) https://fast-d.hmcom.co.jp/(AI Noise Detection) is a subscription-based platform that allows the use of "AI Noise Detection".
AI learning models for noise detection commonly used in various industries are prepared as "standard learning models," making it possible to easily use the noise detection service through cloud computing. If you would like to use noise detection specific to your company's site or usage scenario, it is also possible to create an optimal AI learning model for noise detection while conducting a proof-of-concept experiment (PoC).
|name of company||Hmcomm Inc.|
|Establishment||July 24, 2012|
|Location||Head Office: 2F/5F Fuji Building, 2-11-1 Shiba Daimon, Minato-ku, Tokyo|
Kumamoto AI Lab: 1F, Mirai Meeting Room Sakura-machi, 1-25 Sakura-machi, Chuo-ku, Kumamoto City, Kumamoto Prefecture
|Business||As a venture company originating from AIST, we research and develop elemental technologies and provide solutions and services based on AIST's proprietary speech processing technologies.|
Based on the speech processing platform "The Voice" and the noise detection platform "FAST-D", our philosophy is "to contribute to society by creating value from sound and providing innovative services".
|Related patents||Patent 4604178 "Speech Recognition Equipment and Methods and Programs|
Patent 4997601 "Web site system for voice data retrieval
Patent 5366169 "Speech Recognition System and Program for Speech Recognition System
|uniform resouce locator||https:// hmcom.co.jp|
About Yamagata Prefectural Enterprise Bureau
|name of company||Yamagata Prefectural Enterprise Bureau|
|Location||8-1, Matsunami 2-chome, Yamagata City, Yamagata Prefecture, 990-8570, Japan|
|Business Overview||Electricity business, industrial water supply business, public enterprise asset management business, water supply business|
|uniform resouce locator||https://www.pref.yamagata.jp/kensei/shoukai/soshikiannai/kigyo/|
About National Institute of Advanced Industrial Science and Technology (AIST)
As one of the largest public research institutes in Japan, the institute focuses on the creation and practical application of technologies useful to Japanese industry and society, and on the "bridge" function for commercialization of innovative technology seeds.
Approximately 2,000 researchers at 10 research centers across Japan are conducting research and development as the core and pioneer of the national innovation system, based on changes in the environment surrounding innovation and national strategies formulated in light of these changes.