Articulo Académico / Academic Article

Recibido: 31-10-2024, Aprobado tras revisión: 08-01-2025

Forma sugerida de citación: Heredia, J.; Ayala, E. (2025). “Diseño de Sistema para la Generación de Mantenimiento Predictivo

Basado en IoT e Inteligencia Artificial para Talleres de Mecánica Exprés”. Revista Técnica “energía”. No. 21, Issue I, Pp. 81-86

ISSN On-line: 2602-8492 - ISSN Impreso: 1390-5074

Doi: https://doi.org/10.37116/revistaenergia.v21.n2.2025.678

No Comercial 4.0 (https://creativecommons.org/licenses/by-nc/4.0/)

IoT and AI-Based Predictive Maintenance System Design for Express Auto

Repair Shops

Diseño de Sistema para la Generación de Mantenimiento Predictivo Basado

en IoT e Inteligencia Artificial para Talleres de Mecánica Exprés

J.M. Heredia1

0009-0002-9431-9559

E.L. Ayala1

0000-0003-2528-4380

1 Universidad Politécnica Salesiana, Cuenca, Ecuador

E-mail: jherediap2@est.ups.edu.ec, eayala@ups.edu.ec

Abstract

The purpose of this document is to highlight the existing

issue caused by a lack of knowledge about the actual

condition of machinery and precise monitoring in an

express mechanic workshop. This workshop consists of

mechanical maintenance equipment such as vehicle

lifts, balancers, aligners, and other common machinery

in such work environments. The data collected from

these machines are classified and processed using

Artificial Intelligence, specifically Machine Learning,

by employing a tabulation and interpretation algorithm

alongside IoT (Internet of Things) through the

instrumentation of these machines with sensors

appropriate to their mechanical operation. This

facilitates and enables the creation of predictive

maintenance plans as well as operational schemes that

help reduce operational costs, maintenance expenses,

and energy consumption of the workshop equipment.

The system innovatively uses a modular approach

without requiring intervention or modification of the

machines, allowing their interconnectivity with a

computer that automatically manages the collected data.

This results in a clear view of the usage of each

component, providing critical information for

generating predictive maintenance strategies.

Resumen

El propósito de este documento es dar a conocer la

problemática existente por una falta de conocimiento

del estado real de la maquinaria y del monitoreo preciso

de un taller de mecánica exprés que se compone por

elementos de mantenimiento mecánico tales como

elevadores de vehículos, balanceadora, alineadora,

entre otras las maquinarias más comunes en este tipo de

centros de trabajo, estos datos son clasificados y

procesados utilizando inteligencia artificial como

Machine Learning al emplear un algoritmo de

tabulación e interpretación y conjuntamente IoT

(internet de las cosas) a través de la instrumentación de

dichas máquinas con sensores acordes a su tipo de

funcionamiento mecánico, esto facilita y permite

generar planes de mantenimiento predictivo al igual que

esquemas operativos que permitirán disminuir el costo

operacional, costos de mantenimiento y consumo

energético de la indumentaria del taller, utilizando de

manera innovadora un sistema modular sin necesidad de

intervenir o modificar las mismas, que permite la

interconectividad de la maquinaria con un ordenador

que gestione la data recogida de manera automática

dando como resultado una visión clara del uso de cada

componente, esta información es clave para la

generación de un mantenimiento predictivo.

Index terms— Optimization, IoT, Predictive

Maintenance, Energy Savings, Artificial Intelligence.

Palabras clave— Optimización, IoT, Mantenimiento

Predictivo, Ahorro Energético, Inteligencia Artificial.

Edición No. 21, Issue II, Enero 2025

1. INTRODUCTION

In express mechanical workshops, there is a lack of

control over various aspects of their operations, which

leads to excessive energy consumption. This is often due

to the absence of adequate maintenance plans, downtime,

and logistical issues when creating new work orders. This

situation is exacerbated by the lack of accurate

monitoring and recording of all tasks and processes

performed, so the system concept is to get all the data

about the machines usage, then tabulate, process and get

an clear vision about how these are running, in a period

of time heading to decisions to fix the issues if they

belong to machine errors or human problems.

Additionally, in the current context of an energy crisis,

many centers and workshops are forced to rely on

combustion-powered generators to meet their electricity

needs. This reliance increases costs associated with fuel,

acquisition, and maintenance of such generators,

highlighting the urgent need to optimize energy

consumption as much as possible.

Given this issue, it is essential to develop a system

that integrates operational and administrative areas. This

paper presents a mechatronic solution to adapt machinery

into an ecosystem that maintains accurate and automated

records using IoT. These records, when analyzed with

artificial intelligence (AI), enable the generation of

predictive maintenance plans, which can prevent future

issues that could lead to unnecessary energy costs, as an

example on the automotive industry modern cars use IoT

for sending data about tire pressure, the state of the oil,

temperature of the motor and so forth, the data is used for

planning an predictive maintenance avoiding unexpected

stops and helping with the gas consumption.

Once a database of machinery operation and energy

consumption is available, an alternative energy system

can be better dimensioned to manage and administer

electricity supplied from external sources, bypassing the

public electrical grid.

As a result of implementing this system with tools

such as SVM, Random Forest, and Neural Networks [1],

these technologies demonstrate superior efficiency

compared to similar systems in generating predictive

maintenance plans [2]. The system operates by creating a

database that captures the behavior of an express

mechanical workshop, characterized by a high flow of

vehicles that require only minor repairs or maintenance.

Once this data is obtained, an AI-driven maintenance and

action plan is generated, significantly improving the

workshop’s performance and reducing material and

energy consumption, ultimately enhancing productivity.

2. EXPRESS MECHANICAL WORKSHOPS

An express auto repair shop is a specialized

establishment on maintenance, and automotive quick

repair services, the main characteristic is the kinds of

generally utilizing advanced technology to optimize

resources and time, that is, the vehicle does not remain in

the workshop for more than a day. These centers combine

traditional mechanical services with quality control

systems, digital diagnostic equipment, and, in many

cases, emerging technologies such as the Internet of

Things (IoT) and Artificial Intelligence (AI) tools for

predictive maintenance. The main objective of an express

auto repair shop is to provide efficient management that

ensures the safety, performance, and durability of

vehicles through professional administration of the

automotive maintenance life cycle.

Unlike conventional workshops, express auto repair

shops adopt preventive and predictive maintenance

practices to anticipate potential failures and optimize the

use of energy and material resources. These strategies are

based on data analysis and the integration of cyber-

physical systems, where the workshop tools can be

considered intelligent components within an

interconnected system [3]. In this way, the express auto

repair shop functions not only as a repair space but also

as an innovation hub within the automotive sector,

especially in the context of the Fourth Industrial

Revolution, where advanced diagnostic tools play a

crucial role [4].

Also known as an automotive service center, an

express auto repair shop is designed to perform

preventive maintenance on vehicles, meaning that

processes are carried out to prevent potential failures

rather than resolve them. It is also characterized by a

higher vehicle turnover compared to a traditional

mechanical workshop, as the work performed typically

takes no more than a couple of hours, ensuring that no

vehicle remains on the premises for an entire day.

3. ADQUISITION DATA SYSTEM

To generate the database needed for the

corresponding analysis, modules with sensors capable of

accurately transmitting signals like temperature, current

and proximity, that reflect the behavior and usage of the

various machinery used in an express auto repair shop are

utilized [5]. These modules consist of a sensor and a

microprocessor with Wi-Fi connectivity, enabling data to

be sent to the cloud, these devices use Wi-Fi 802.11

protocol that is the most common allowing a range of

data transmission velocities from 11 up to 150 Mbps. In

this way, real-time monitoring is possible, allowing data

acquisition regarding the usage and performance of the

machines.

In the following diagram, the operational scheme of

the monitoring system is represented in a simplified

manner, showing how sensors are integrated to connect

to a network where devices and machinery communicate.

This setup enables the display of real-time operational

variables, providing insights into productivity and

individual resource consumption for each device and

machine used in an express auto repair shop, thus

creating an intelligent system that after its application the

time , as is shown on the following diagram how the

process starts from the machines working, then the

sensors taking data and sending to the cloud, the data

being tabulated and interpreted by AI and shown on a

graphic display.

Figure 1: IoT-Based Data Flow from Machines to Cloud

Monitoring

3.1 IoT applied in machinery

Each machine used must have a system that generates

a usage and behavior history, as well as environmental

variables that can be recorded in a computerized system.

For this purpose, IoT is applied to each machine in

such a way that it integrates into the system, enabling

real-time monitoring through sensors and connected

networks. This allows real-time communication and

monitoring of the different machines within the shop,

bringing the current operation closer to an Industry 4.0

environment, where human presence is increasingly less

necessary for process supervision because there is no

need for a person to check on the status of the machines

due to the fact that the sensors are taking care the

important metrics on the devices and this is possible via

IoT, all the information is process and given to the

administration department with a clear guide of what is

the action protocol to be done, so the only decision to be

made is when to execute this plan.

If IIoT (Industrial Internet of Things) is integrated

with AI, we can create a system that not only monitors

the machines but, through the analysis and processing of

IoT-generated data, also develops tools that facilitate

decision-making and operational efficiency within the

shop.

3.2 Data Acquisition Methodology

Since continuous machinery monitoring is necessary,

specific types of sensors will be adapted to withstand the

constant and intense use characteristic of work in an

automotive workshop. The operational environment is

also assessed, as it directly impacts the wear and tear on

both the machinery and data acquisition devices, the

system will evaluate the work done by timing it, then

comparing with an optimal range of time database for

each machine and each case like size of tire. Therefore,

for each specific case, appropriate sensors are selected

like for example the MAX6675 with a K-type

thermocouple for temperature, the ADXL345

accelerometer for vibration monitoring, the LJ12A3-4-

Z/BX inductive sensor for proximity detection, and the

Honeywell Micro Switch V7 for mechanical position

sensing, along with the necessary microprocessor or

embedded chip, to enable unidirectional communication

between the devices and the control and monitoring

system using an Wi-Fi protocol to send all the data to a

private cloud.

4. DATA MANAGEMENT AND ANALYSIS

The data received from the cloud will undergo a series

of mathematical algorithms that tabulate, classify, and

process the data to be presented in a graphical

environment, facilitating understanding and analysis.

AI is used for post-processing this data, analyzing the

results to determine a methodology or guideline for

correctly and specifically applying predictive

maintenance in this case study. This approach enables a

complete analysis of the workflow without the need for

direct personnel intervention and without invasive access

to the machinery, thus avoiding downtime and saving

administrative costs [6], reducing the response time and

planning time that takes to generate a predictive

maintenance plan, so this result translates on higher

productivity because the repair stops are planned ahead,

so all the process reflects on a better stat of the number

of services given to the customers.

5. PREDICTIVE MAINTENANCE USING AI

Artificial Intelligence (AI) is a highly important

technology in engineering, as it enables various

approaches to solving specific issues. This solution

framework is built on tools such as SVM, Random

Forest, and Neural Networks, which can be used through

IoT. By employing this methodology for data

management and analysis, predictive maintenance can be

applied. Through continuous analysis of machinery

behavior, a history is generated. By setting thresholds,

potential failures can be accurately predicted. With

knowledge of the impending fault and its cause, a

predictive maintenance plan can be created, saving both

energy consumption and various costs that a failure

would incur.

Edición No. 21, Issue II, Enero 2025

Allowing AI to generate responses and make

decisions by considering all variables improves

operational dependency, as quick responses and an

efficient action plan are obtained to address all aspects

involved in a potential failure [7].

This process of analysis and decision-making based

on all established and input variables is managed under a

scheme where data is first standardized. Then, the data is

processed under specific guidelines, such as legal

requirements, operating regulations, AI predictions, and

simulation tools. This approach generates operational

messages, which are the results of AI after analyzing the

database and creates an automated operation program

that includes the predictive maintenance plan and the

allocation of material and energy resources. This scheme,

in general terms, is represented in the following diagram.

Figure 2: Workflow Scheme in the Software Component [6]

Since this algorithm is followed, one of the three tools

with the best performance for generating predictive

maintenance plans—SVM, Random Forest, and Neural

Networks—is used. The table below shows their superior

effectiveness compared to other Machine Learning tools.

Table 1: The performance of machine learning models in

predictive maintenance [1]

Model

Precision

Recall

Fi-

Score

Accuracy

AUC-

ROC

Support Vector

Machine

(SVM)

0,91

0,89

0.90

0,92

0,93

Random forest

0,94

0,93

0,94

0,95

Naural

networks

0,92

0,91

0,93

0,94

Logistic

regression

0,88

0.86

0,87

0,89

0.90

Gradient

boosting

0,93

0,92

0,93

0,94

This table allows us to assess the accuracy and

effectiveness based on relevant indicators regarding the

performance of each Machine Learning tool when

creating logistics and planning for efficient predictive

maintenance.

Using these tools, a specific operational scheme is

developed that incorporates the predictive maintenance

plan based on each machine’s behavior data. The

integration of AI and databases revolutionizes data

management, enabling automated decision-making and

efficient processing of large datasets. AI analyzes

patterns and predicts outcomes, while databases provide

robust infrastructure for data storage and access. This

synergy drives innovation, improves operational

efficiency, enhances automation, and optimizes decision-

making across industries and diverse applications. The

interaction between the software and this data is as

follows:

Figure 3: Operational Steps of the Prediction Model [8]

In this way, the generated database is continuously

processed, resulting in real-time monitoring of the

express auto repair shop’s performance. This allows for

the determination of the need to apply predictive

maintenance, when necessary, thanks to the anticipation

of events provided by the IoT system combined with

Machine Learning and the data acquisition ecosystem

integrated within the shop.

6. ENERGY CONSUMPTION OPTIMIZATION

The results of predictive maintenance and monitoring

in an express auto repair shop are reflected in energy

savings achieved through process optimization and the

prediction of potential issues, such as failures,

unexpected shutdowns, and downtime. This ensures a

constant workflow, generating higher profits with

reduced energy consumption.

As shown in the table below, a case study on a vehicle

mechanic workshop illustrates the results in terms of

energy savings when applying Machine Learning to

maintenance and process management and various

metrics that are evidence of the improvement that means

applicate a system like this [9].

Table 2: Comparison of performance in terms of energy

consumption [1]

Metric

Before

Implementation

After

implementation

Percentage

reduction

(%)

Average

equipment

downtime

(hours/month)

15,40

8,20

0,47

Emergency

repair

instances (per

year)

25,00

10,00

0,60

Energy

consumption

for repairs

(MWH/year)

150,00

90,00

0,40

Maintenance

cists

(USD/year)

500,00

320,00

0,36

Operational

efficiency (%)

82,50

91,20

0,08

This allows us to confirm the effectiveness of

using Machine Learning and IoT in an integrated

system to optimize the operations of an express auto

repair shop, reducing its energy consumption. The

energy results derivates from a better management,

for example after applying an system of this kind the

uptime increases on a 90% to 95%, the failure rate

gets from 5 to 2 cases monthly and also the Mean

Time Between Failures (MTBF) increases from 200

hours to 400 hours [10], these parameters are

evidence of the interaction between IoT and ML on a

system, the dynamic that involve these two tools is

described in five stages:

1- Data recollection.

2- Feature engineering

3- Algorithm selection and training

4- Deployment and integration

5- Continuous Learning

7. SYSTEM INTEGRATION IN AN EXPRESS

AUTO REPAIR SHOP

The synergy between all parts of the proposed system

and the elements (machinery and workshop staff) that

make up the express auto repair shop is achieved through

administrative actions based on the results of the AI-

implemented algorithm. Once the analysis is generated,

the administrative department creates an action plan that

includes specific work protocols and a maintenance plan

for each machine. Additionally, training sessions or new

operational methods are organized to improve and

coordinate the workshop staff [11].

Monitoring each station in the express auto repair

shop allows for behavior analysis over time, revealing

types of failures—whether operator errors or machine

faults. The model is trained after entering a complete

database about the acceptable timing taken for a worker

to complete an specific activity, for example there is a

range of time for taking out a pneumatic from a rim, so

when a machine is activated a timer will star and then this

timer will be compared with an optimal time range so in

that way can be determinate if the operation was done

correctly and efficient. Once identified, a record is

created, and a learning system is applied to automatically

recognize failure trends and predict when they may

occur. This enables time for planning and proactive

action, resulting in a continuous workflow without

unexpected stoppages or temporary interruptions in the

shop’s service.

8. CONCLUSIONS AND RECOMMENDATIONS

As a result, implementing a system that enables real-

time monitoring and, at set intervals, generates an

automatic analysis of the management and performance

of an express auto repair shop provides a comprehensive

positive outcome like reducing the respond times and

inactivity, increasing the overall productivity and

reducing the maintenance general costs. This system

impacts both the administrative and workshop areas,

creating continuous communication channels that

facilitate decision-making backed by an AI system that

accurately predicts upcoming failures and creates

predictive maintenance plans, the benefits proved by

Easy-Manit is the reduction of repair costs by predicting

problems and planning with time the necessary stops.

The administrative department gets the advantage to get

directly the information of the status of the workshop, not

only regarding the machinery component also includes

the human resources, facilitating the analysis, and

management of resources in general based on trustful

automatically generated information. This reduces

operational costs and promotes efficient management of

energy resources, making it an innovative project that

enhances the shop's productivity and provides a

sustainable response to the country's current energy

crisis.

As a recommendation is imperative to create an own

database with the timings of the workers on an specific

workshop because these metrics will variate depending

on the training and expertise of the human resources of

each worshop.

REFERENCIAS BIBLIOGRÁFICAS

[1] J. C. Cruz y A. M. Garcia, "Machine Learning for

Predictive Maintenance to Enhance Energy

Efficiency in Industrial Operations," Information

Technology Engineering Journals (ITEJ), vol. 9, no.

1, pp. 15–22, 2024.

Edición No. 21, Issue II, Enero 2025

[2] F. Artkin, "Applications of Artificial Intelligence in

Mechanical Engineering," European Journal of

Science and Technology, vol. 45, pp. 159–163,

2022.

[3] J. H. Rodríguez Ovalle y L. M. Dávila García,

“Gestión del mantenimiento automotriz: Un

acercamiento al mantenimiento preventivo 4.0 y los

carros conectados,” Escuela Colombiana de

Ingeniería Julio Garavito, 2020. Disponible en:

https://catalogo.escuelaing.edu.co.

[4] “CDPA: Centro de Desarrollo Productivo de

Mantenimiento Automotriz Estrategia para la

competitividad de la microempresa,” Revista

Científica de la Universidad del Norte, 2020.

Disponible en: https://rcientificas.uninorte.edu.co.

[5] K. D. P. Mariano, F. L. N. Almada, y M. A. Dutra,

"Smart Air Quality Monitoring for Automotive

Workshop Environments," Institute of Informatics –

Federal University of Goiás (UFG), Goiânia – GO,

Brazil, 2024.

[6] A. Petrov y I. Novak, "Optimization of Industrial

Energy Efficiency Through the Application of

Advanced Process Control, Monitoring

Technologies, and Predictive Maintenance,"

Eigenpub Review of Science and Technology, vol.

6, no. 1, pp. 1–10, 2022.

[7] M. Soori, F. K. Ghaleh Jough, R. Dastres, y B.

Arezoo, "AI-Based Decision Support Systems in

Industry 4.0, A Review," Journal of Economy and

Technology, 2024.

[8] T. Yamamoto, H. Hayama, T. Hayashi, y T. Mori,

"Automatic Energy-Saving Operations System

Using Robotic Process Automation," Energies, vol.

13, no. 2342, pp. 1–14, 2020.

[9] S. Thomas, A. O. Philip, y N. Vishwanath, "ML

Based Data Driven Energy Centered Predictive

Maintenance," Proceedings of the Second

International Conference on Edge Computing and

Applications (ICECAA 2023), vol. 2, pp. 994–1003,

2023.

[10] A. P., N. Chouhan, G. A. Chandhok, D. Sugumaran,

U. Aswal, and S. A., “Empowering IoT Devices with

Energy-Efficient AI and Machine Learning,” 2024

International Conference on Circuit Power and

Computing Technologies (ICCPCT), DOI:

10.1109/ICCPCT61902.2024.10672916.

[11] S. Gennitsaris et al., "Energy Efficiency

Management in Small and Medium-Sized

Enterprises: Current Situation, Case Studies and

Best Practices," Sustainability, vol. 15, no. 3727, pp.

1–26, 2023. [11] J. O. Williams, “Narrow-band

analyzer,” Ph.D. dissertation, Dept. Elect. Eng.,

Harvard Univ., Cambridge, MA, 1993.

Johnny Mateo Heredia.- Born in

Cuenca, Ecuador on 2001.

Mechatronic engineering last

semester student, IEEE member.

Edy Ayala. - Born in Cuenca,

Ecuador, in 1987. He holds a Ph.D.

in Engineering Sciences from the

Faculty of Mechatronics at the

University of Ferrara in Italy, a

master’s in engineering sciences

from Swinburne University in

Australia, a Bachelor's degree in

Electronic Engineering from Universidad Politécnica

Salesiana, and a Diploma in Industrial Electronics from

Universidad Politécnica Salesiana. He is a Research

Professor with experience in research projects in the

fields of Energy, Assistive Technologies, and

Electronics. In 2010, he received the José Peralta Award

from the Prefecture of Azuay for his development of

automated irrigation projects.