IoT, Risk and Resilience based Framework for Quality Control:

Application for Production in Plastic Machining

Khaled Bahloul

1 a

and Nejib Moalla

2 b

EPSF, 80000, Amiens, France

Universite Lumiere Lyon 2, DISP Laboratory, EA4570, 69676 Bron, France

Keywords:

IoT Network, Risk Management, Resilience, Quality Control, Machine Learning, Machining in Plastic

Industry.

Abstract:

The deﬁnition of defect prediction models in manufacturing emerges as an attractive alternative supported

by industry 4.0 concepts and solutions. We propose in this paper an IoT-based approach for a global quality

control mechanism in industry. We cover in this work the in-process quality control inspection, the produc-

tion machines as well as the production environment monitoring. Our framework addresses data analytics

algorithms using monitoring data, risk assessment models, resilience parameters and acceptance criteria for

prediction models. The proposed concepts are implemented to control the manufacturing processes of a plas-

tic product where the distinction between irregularity and nonconformity needs to be supported by a smart

decision system.

1 INTRODUCTION

The fourth industrial revolution brings the concepts of

digital transformation and connectivity between cyber

and physical assets (Zamiri et al., 2019). The massive

digitalisation of business activities in industry is sup-

ported by facilitating the adoption of new technical

enablers, open APIs, etc (Lade et al., 2017a). To for-

malise activities in collaborative networks. The con-

nectivity is addressed by the integration of IoT net-

works, cyber-physical systems, cyber-physical pro-

duction systems, etc. as solution to increase the reac-

tivity of the industrial assets to build more smart sys-

tems (Lade et al., 2017b). For product quality control

in manufacturing, the new technological solutions and

paradigms are challenged to support more preventive

approach for in-process quality control. We aim to in-

crease the sensing capabilities during the manufactur-

ing steps (in-process) to collect data about machines,

products and environment. The collected data need to

be contextualised by adding risk and resilience crite-

ria and then trained to propose predictive model for

product conformity assessment.

We analyse in the second section existing research

contributions dealing with product quality control so-

https://orcid.org/0000-0003-2385-1249

https://orcid.org/0000-0003-4806-0320

lutions and covering the concepts of IoT-based data

analytics, risk management and resilience. We detail

in the third section a functional view about the pro-

posed Framework with the main building blocks nec-

essary to design and develop a global quality control

system. The fourth section provides some technical

details about the implementation of the proposed con-

cepts and their application on the quality control of a

plastic product.

2 RELATED WORKS

To set up a resilient quality control system, we analyse

in this section: the connection between resilience and

quality control, the applicability of machine learning

techniques to deﬁne quality predictive models and the

deﬁnition of resilience requirements to improve the

efﬁciency of the quality control mechanism.

2.1 Resilience in Quality Control

The evolution of technology during the last decades

have forced us to remodel solutions used for prod-

uct quality control in industries and to improve

resilience systems, especially changes related to

human-computer interface along with the continu-

Bahloul, K. and Moalla, N.

IoT, Risk and Resilience based Framework for Quality Control: Application for Production in Plastic Machining.

DOI: 10.5220/0010608106050611

In Proceedings of the 16th International Conference on Software Technologies (ICSOFT 2021), pages 605-611

ISBN: 978-989-758-523-4

605

ous evolution of information technology from cloud

computing, big data, internet of thing (Tsang et al.,

2018), machine-to-machine communication (Pereira

et al., 2019), Cyber Physical System (Lee, 2015),

Social-CPS derived from fundamental artiﬁcial intel-

ligence algorithms about machine learning (Shahbazi

and Byun, 2021), data mining, knowledge engineer-

ing, sensor network, Software Deﬁned Network (Ali

et al., 2017) and so forth. In this section, we re-

fer to some works that have opted to use these tech-

nologies to propose improvements in control and re-

silience systems for product quality. (Herrero et al.,

2002) propose an integrated quality and safety man-

agement system based on a systematic approach for

quality and safety management. Enforcing the barri-

ers for improving quality, safety and reliability con-

cept of technical systems, equipment, and various

components which are the main ideas and which are

also represented and conﬁrmed in (Celik, 2009). As

well as, (Kaar et al., 2018) have proposed a Resilient

Ontology based on different standards, concepts, and

frameworks. This ontology ensures a higher quality

data integrity. Furthermore, it provides the environ-

ment with an improved resilience in order to obtain a

new higher quality data. (Reis and Gins, 2017) pro-

pose three major stages of industrial process moni-

toring in the context Big Data - Industry 4.0 (detec-

tion, diagnosis and prognosis) for root cause analy-

sis and important diagnosis obtained for quality im-

provement. In addition, (Al-Shammari et al., 2020)

have proposed a new resilience technique for internet

of think networks dealing with, among other things,

resilient service embedding with sensor actuator node

redundancy. The aim is to reduce potential attacks and

data loss by enhancing reliability and accountability,

with real-time data collection. (Wang et al., 2018)

targets building an in-depth model to effectively de-

tect defects in products. This model is based on deep

convolutional neural networks. The results show that

the model is able to categorize the image sample in

the correct image class and indicate whether it con-

tains defective regions or not. Equally, the resilience

maturity model presented in (Marrella et al., 2019),

is based on a design-time, date-centric maturity. It is

an extension of the case management model and no-

tation. This model is a real tool to provide support for

process designers so that they may become aware of

how resilient their processes are. In fact, authors have

conﬁrmed that a proper analysis of involved data al-

lows the process designer to identify possible failures.

2.2 AI and Machine Learning for

Quality Control Decision Support

The use of AI and machine learning techniques for

quality control decision support is commonly ad-

dressed in the literature. (Dey et al., 2020) intro-

duce the key idea is to join a simulation approach

with machine learning models. This allows to deter-

mine the optimized parameter values of checkpoint

intervals and checkpoint counts for different conﬁgu-

rations to improve quality control. In (Tellaeche and

Arana, 2013), a process for quality control for plas-

tic injections was developed using machine learning

algorithms. (Benacchio et al., 2021) use the same

learning approach to present some recommendations

for resilience strategy based on performance, efﬁ-

ciency and effectiveness. These recommendations

have an impact on hardware developments. (Esco-

bar and Morales-Menendez, 2018) and (Escobar and

Morales-Menendez, 2017) have proposed a learning

process and pattern recognition strategy for quality

control based on machine learning techniques. They

have formulated the defect detection as a binary clas-

siﬁcation problem. They have also proposed an ap-

proach to detect rare quality events in manufactur-

ing systems and have identiﬁed the most relevant fea-

tures for product quality. The experimental results

conﬁrm that 100% of defects can be detected effec-

tively. For the monitoring of industrial machines and

structural health (Bhuiyan et al., 2017) and (Mohanty

et al., 2015), the authors have proposed a solution for

a low-complexity signal processing where the sensor

reduces a signiﬁcant amount of data without sacri-

ﬁcing the quality of data. As well, they present a

decision-making algorithm by which each sensor can

make a decision on its acquired data, so communi-

cation is reduced without using inconsequential data.

(Farahani et al., 2019) use a sensor network to col-

lect data about operations of injection and molding.

This data was then synchronized and integrated with

the machine data to have the maximum variety of on-

line time-based data sources. Thereafter, all data was

used in predicting variations on quality indices of the

ﬁnal product, namely weight, thickness, and diame-

ter. In the same way, (Berger et al., 2017) presents the

optimized operation of both 3D and 2D measurement

systems in line production, which are used to detect

defects that appear in the production of hybrid metal

components in the resin transfer moulding process.

2.3 Resilience Requirements

In the literature, a resiliency is deﬁned by ”the ability

of a system (here a manufacturing system) to recover

ICSOFT 2021 - 16th International Conference on Software Technologies

606

from an undesired state to its desired state” (Holl-

nagel et al., 2006), (Shefﬁ, 2007). The attributes of re-

siliency of a production system are deﬁned by persis-

tence, adaptability, agility, redundancy, learning ca-

pability, and decentralization (Schmitt et al., 2017).

Within the development of new technologies used

in information systems, resilience engineering began

with the study of safety systems. Also, resiliency

is one of the six characteristics of smart manufac-

turing which are data-driven, networked, connected,

resource sharing, resilient, and sustainable (Kusiak,

2019). We believe that the resilience of a system is

measured by its ability to be adapted to new organiza-

tional requirements and unforeseen changes that were

not initially deﬁned in the ﬁrst design of the existing

system. This means that a resilience system is able

to automatically adapt to such changes (M

uller et al.,

2013) and (Rosemann and Recker, 2006). In order

to apply this afﬁrmation and improve the resilience

of our system, the ﬁrst essential step was to deﬁne a

Requirement Resilience set. In many literatures, sev-

eral studies have focused on deﬁning the resilience re-

quirements, as illustrated in this paper (Marrella et al.,

2019), authors have proposed a method that helps the

process designers improve their process models by

considering their previous failures generated by un-

availability data. The authors have given a prereq-

uisite that needs to be satisﬁed in order to model a

resilient process for business. In addition, in these pa-

pers (Agarwal et al., 2014) and (Agarwal et al., 2014),

authors have identiﬁed the requirements for the re-

silient nuclear power plant outage control. These re-

quirements concern information of nuclear plant op-

eration, process automation and process of data col-

lection/processing techniques for the improvement of

a resilient nuclear power plant outage control.

3 RESILIENT QUALITY

CONTROL FRAMEWORK IN

MANUFACTURING

In the context of industrial manufacturing, the setup

of a global approach for in-process quality control re-

quires the deﬁnition of three control levels: the man-

ufacturing resources (machines, humans, etc.), the

product during the transformation processes and ﬁ-

nally the work-cell environment where external con-

ditions can impact the quality of the product. In addi-

tion, control results are collected from the traditional

quality control process after the manufacturing phase.

The following ﬁgure (Figure 1) presents an overview

about the proposed Framework.

3.1 Sensors Network Design

The design of the sensor network must ensure the ob-

servability of the variables, the detectability and isola-

bility of faults. In this step, it is necessary to analyse

the sensor location based on several indicators such as

redundancy, observability, precision, estimation and

reliability of the measurement system. In our case,

the sensor’s network design process is deﬁned using

the manufacturing machine-sensing capabilities, the

product dimensions and tolerances, the characteris-

tics of the manufacturing environment and ﬁnally the

history of the nonconformity problems and their ori-

gin. To ensure data reliability, each selected sensor

must be replicated (similar sensitivity, but different

brands), to ensure the reliability of collected data. The

ﬁrst round of harvested sensors is deﬁned so as to pro-

vide a direct set of usable data to treat the most com-

mon origins of quality problems and nonconformity.

3.2 Sensing Network Monitoring

A good sensor network monitoring must meet three

requirements. First, it must be responsive in order

to analyse the data quickly and correctly. Second, it

must evaluate the correct operation and the status of

the network. Third, it must make the data from the

sensor network available to the user and present this

information in the requested format to be processed

efﬁciently.

3.3 Risk Identiﬁcation and Assessment

The systematic use of all available information to

identify and to estimate the risk is required for risk

analysis . It is necessary to apply a procedure, based

on risk analysis to minimize product failures. This

approach is completed for a risk assessment step in

order to ensure a global approach. Among the differ-

ent nonconformity problems, a ﬁrst classiﬁcation is

provided (cost, delay, Return on Investment, environ-

mental, etc.) which have the most impact on quality

problems and, if available, their origins. We propose

to deﬁne the risk quotation according to the impact of

the observed nonconformity. Therefore, risks deﬁni-

tion depends on the product to control as well as the

production conditions and the work-cell environment.

3.4 Machine Learning for Decision

Support

This step is based on the previous results. The data

from the sensor network is used in the experimen-

tation of different ML algorithms in order to make

IoT, Risk and Resilience based Framework for Quality Control: Application for Production in Plastic Machining

607

Figure 1: Resilient Quality Control Framework in manufacturing.

their analysis and make the right decision according

to the deﬁned requirements. To build our quality con-

trol model, we consider the following inputs: the sen-

sor’s monitoring data (collected from machines and

control tools during the transformation operations),

the risk quotations related to the nonconformity prob-

lems (qualitative data to be normalised), and the qual-

ity control results evaluated after the manufacturing

stage. Using machine learning technic, we design a

prediction model for product defects predictions.

3.5 Evaluate Quality Control Model

To evaluate the proposed quality control model, we

consider the following criteria: the redundancy of de-

tected errors, the sensitivity of the model, and the

trade-off between prediction model precision and re-

call.

3.6 Resilience Assessment

For the resilience assessment, we adopt an indicators-

based approach. A resilience indicator is a descrip-

tion of information that is used to identify the state of

product quality. This information is qualitative and

measurable in accordance with the deﬁned require-

ments. A set of indicators can measure the resilience

characteristic. The resilience indicators can be the

outcome indicators or process indicators. In our case

study, we select the most adapted set of indicators

(common availability of raw data) from the one in-

troduced in (Jain et al., 2018):

• Phase I – Avoidance: Alarm rate, Unplanned

maintenance jobs, Unplanned shutdowns per

year;

• Phase II– Survival: Mechanical device shut-

down, Safety critical equipment (SCE) inspection,

Safety critical equipment (SCE) deﬁciency;

• Phase III – Recovery: Tests for emergency sys-

tems and procedures, Mock drills for emergency

situations;

• Common metrics: Process safety required train-

ing sessions completed, Required procedures re-

viewed/revised.

3.7 New Resilience Requirements

After the resilience assessment, in this step we investi-

gate the possibility of adding resilience requirements

to improve the quality control model of the product.

In fact, its requirements must comply with the condi-

tions of the step ”Sensing network monitoring” setup.

Therefore, our resilience requirements are related to

the:

• The number of sensors data considered in the pre-

diction model;

• The criticality of the risks considered in the pre-

diction model;

• The trade-off between precision and recall in the

prediction model.

3.8 Sensor’s Network Environment

Adaptation and Add New Sensors

According to the previous evaluation results of the

quality control model, we propose a feedback mech-

anism to ensure the minimal distribution of the sen-

sors network to maximize the robustness of the qual-

ity control model. A system is considered resilient if

its capabilities can be adapted to new organizational

requirements and changes that have not been explic-

itly incorporated into the design of the existing sys-

tem. To ensure this capability, our resilience require-

ments tend to maximize the number of sensors par-

ticipating in the deﬁnition of the quality assessment

model. For sensors environment adaptation, we pro-

ceeded with changing the sensors’ location or their

calibration. For sensors augmentation, we proceeded

by replacing sensors with low sensitivity. For con-

ﬂictual situations, we added new sensors in different

locations to better adapt our quality gates.

ICSOFT 2021 - 16th International Conference on Software Technologies

608

Table 1: Overview of monitoring data.

Scope in production Monitoring Perimeter Monitoring data from sensors Evaluated risks

Manufacturing Operation resource: Machine

Temperature, depressions,

throughput, time, position,

material ﬂuidity, etc

Breaking of manufacturing tools due to aging

Breaking of manufacturing tools due calibration problem

Wrong machining parameters or tools

Manufacturing In-Process Quality Control

Quotations, product centricity,

material thickness,

2D coordinates of the spots,

3D coordinates of the spots

Adaptability of machining speed with ongoing product

Wrong machining sequence

Level of light needed for laser measurement

Manufacturing Work-cell environment

Temperature

pressure

humidity

Cleanliness of the manufacturing environment

(presence of material fragments)

Quality Control

Quality Control for ﬁnished products

through 3D image processing

Quotations, product centricity,

material thickness,

2D coordinates of the spots,

3D coordinates of the spots

Position of the sensors

Level of light needed for laser measurement

Level of precision-time for control

Figure 2: Quality control after manufacturing.

4 IMPLEMENTATION AND CASE

STUDY

We implement the proposed concepts as detailed in

the previous sub-sections for the in-process quality

control of product machining in the plastic indus-

trial. Our main goal was to detect the sources of

non-quality at the earlier stages of the machining

process and then reduce manufacturing scraps. Re-

garding sensors network design, we collect sensors

related data covering machines behaviour (tempera-

ture, depressions, throughput, time, position, mate-

rial ﬂuidity, etc.); in-process product controls (quo-

tations, product centricity, material thickness, etc.);

and manufacturing environment (temperature, pres-

sure, humidity). At the end of the manufacturing

operations, additional quality control actions are per-

formed through a 3D image processing (Open CV)

(Figure 2 and Figure 3) to conﬁrm the acceptance

of the product quality comparing the theoretical tol-

erance values. The set of monitoring data are sum-

marised in the following table (Table 1).

For risk management, regarding the history of

Figure 3: Image processing for plastic items quality control.

nonconformity, we identify and classify the critical

transformation steps, speciﬁc machining tools, raw

material transformation processes and human ma-

chining operations. The risk gravity is considered for

the generation of quality control model. As veriﬁca-

tion rules for product quality are quite complex and

need to be frequently adapted, we propose in this re-

IoT, Risk and Resilience based Framework for Quality Control: Application for Production in Plastic Machining

609

Figure 4: Quality control report.

search to adopt a machine learning approach built to

automatically ingest new monitoring data related to

product, production machines, environment, control,

risk and resilience criteria. Our system follows a su-

pervised approach, learning incrementally on the ﬂy

and detecting patterns in the training data to build pre-

dictive models. Adapted learning algorithms (KNN,

LR, SVM, etc.) are experimented together to release

the most adapted predictive model helping to evalu-

ate the acceptance of detected defects. The follow-

ing ﬁgure (Figure 4) illustrates the quality control re-

port generated after each product inspection. The con-

trolled product was rejected as the number of controls

exceed the tolerance limit.

5 CONCLUSIONS AND FUTURE

WORKS

We propose in this paper a global quality control ap-

proach covering the in-process and traditional qual-

ity control processes. We harvest a resilient network

of sensors at the manufacturing work cell perimeter,

and we propose a set of concepts helping to combine

technical data (from sensors) with identiﬁed risks and

resilience requirements. The proposed concepts are

experimented to predict defect and control the qual-

ity of machined products in plastic industry. For the

future work, we expect to increase the variety of the

controlled products. It’s about the consolidation of

the Quality Control models repository to be able to

upload the adequate quality control model when we

start product machining. Uploaded model will be

tuned through the new collected data and manufac-

turing events.

REFERENCES

Agarwal, V., Oxstrand, J. H., and Le Blanc, K. L. (2014).

Automated work packages architecture: An initial

set of human factors and instrumentation and con-

trols requirements. Technical report, Idaho National

Lab.(INL), Idaho Falls, ID (United States).

Al-Shammari, H. Q., Lawey, A. Q., El-Gorashi, T. E., and

Elmirghani, J. M. (2020). Resilient service embedding

in iot networks. IEEE Access, 8:123571–123584.

Ali, N. F., Said, A. M., Nisar, K., and Aziz, I. A. (2017).

A survey on software deﬁned network approaches for

achieving energy efﬁciency in wireless sensor net-

work. In 2017 IEEE Conference on Wireless Sensors

(ICWiSe), pages 1–6. IEEE.

Benacchio, T., Bonaventura, L., Altenbernd, M., Cantwell,

C. D., D

uben, P. D., Gillard, M., Giraud, L., G

oddeke,

D., Rafﬁn, E., Teranishi, K., et al. (2021). Resilience

and fault tolerance in high-performance computing for

numerical weather and climate prediction. The Inter-

national Journal of High Performance Computing Ap-

plications, page 1094342021990433.

Berger, D., Brabandt, D., Bakir, C., Hornung, T., Lanza, G.,

Summa, J., Schwarz, M., Herrmann, H.-G., Pohl, M.,

and Stommel, M. (2017). Effects of defects in series

production of hybrid cfrp lightweight components–

detection and evaluation of quality critical character-

istics. Measurement, 95:389–394.

Bhuiyan, M. Z. A., Wu, J., Wang, G., Chen, Z., Chen,

J., and Wang, T. (2017). Quality-guaranteed event-

sensitive data collection and monitoring in vibration

sensor networks. IEEE Transactions on Industrial In-

formatics, 13(2):572–583.

Celik, M. (2009). Designing of integrated quality and safety

management system (iqsms) for shipping operations.

Safety Science, 47(5):569–577.

Dey, T., Sato, K., Nicolae, B., Guo, J., Domke, J., Yu,

W., Cappello, F., and Mohror, K. (2020). Optimizing

asynchronous multi-level checkpoint/restart conﬁgu-

rations with machine learning. In 2020 IEEE Interna-

tional Parallel and Distributed Processing Symposium

Workshops (IPDPSW), pages 1036–1043. IEEE.

Escobar, C. A. and Morales-Menendez, R. (2017). Machine

learning and pattern recognition techniques for infor-

ICSOFT 2021 - 16th International Conference on Software Technologies

610

mation extraction to improve production control and

design decisions. In Industrial Conference on Data

Mining, pages 286–300. Springer.

Escobar, C. A. and Morales-Menendez, R. (2018). Machine

learning techniques for quality control in high confor-

mance manufacturing environment. Advances in Me-

chanical Engineering, 10(2):1687814018755519.

Farahani, S., Brown, N., Loftis, J., Krick, C., Pichl, F., Vac-

ulik, R., and Pilla, S. (2019). Evaluation of in-mold

sensors and machine data towards enhancing prod-

uct quality and process monitoring via industry 4.0.

The International Journal of Advanced Manufactur-

ing Technology, 105(1):1371–1389.

Herrero, S. G., Salda

na, M. A. M., del Campo, M. A. M.,

and Ritzel, D. O. (2002). From the traditional concept

of safety management to safety integrated with qual-

ity.

Hollnagel, E., Woods, D. D., and Leveson, N. (2006). Re-

silience engineering: Concepts and precepts. Ashgate

Publishing, Ltd.

Jain, P., Mentzer, R., and Mannan, M. S. (2018). Re-

silience metrics for improved process-risk decision

making: Survey, analysis and application. Safety sci-

ence, 108:13–28.

Kaar, C., Frysak, J., Stary, C., Kannengiesser, U., and

uller, H. (2018). Resilient ontology support facil-

itating multi-perspective process integration in indus-

try 4.0. In Proceedings of the 10th International Con-

ference on Subject-Oriented Business Process Man-

agement, pages 1–10.

Kusiak, A. (2019). Fundamentals of smart manufacturing:

A multi-thread perspective. Annual Reviews in Con-

trol, 47:214–220.

Lade, P., Ghosh, R., and Srinivasan, S. (2017a). Manufac-

turing analytics and industrial internet of things. IEEE

Intelligent Systems, 32(3):74–79.

Lade, P., Ghosh, R., and Srinivasan, S. (2017b). Manufac-

turing analytics and industrial internet of things. IEEE

Intelligent Systems, 32(3):74–79.

Lee, E. A. (2015). The past, present and future of cyber-

physical systems: A focus on models. Sensors,

15(3):4837–4869.

Marrella, A., Mecella, M., Pernici, B., and Plebani, P.

(2019). A design-time data-centric maturity model for

assessing resilience in multi-party business processes.

Information Systems, 86:62–78.

Mohanty, S., Gupta, K. K., and Raju, K. S. (2015). Vibra-

tion feature extraction and analysis of industrial ball

mill using mems accelerometer sensor and synchro-

nized data analysis technique. Procedia Computer

Science, 58:217–224.

uller, G., Koslowski, T. G., and Accorsi, R. (2013).

Resilience-a new research ﬁeld in business informa-

tion systems? In International Conference on Busi-

ness Information Systems, pages 3–14. Springer.

Pereira, E., Pinto, R., Reis, J., and Gonc¸alves, G. (2019).

Mqtt-rd: A mqtt based resource discovery for machine

to machine communication. In IoTBDS, pages 115–

124.

Reis, M. S. and Gins, G. (2017). Industrial process moni-

toring in the big data/industry 4.0 era: From detection,

to diagnosis, to prognosis. Processes, 5(3):35.

Rosemann, M. and Recker, J. (2006). Context-aware pro-

cess design: Exploring the extrinsic drivers for pro-

cess ﬂexibility. In Proceedings of the Workshops and

Doctoral Consortium, pages 149–158. Namur Univer-

sity Press.

Schmitt, R., Permin, E., Kerkhoff, J., Plutz, M., and

ockmann, M. G. (2017). Enhancing resiliency

in production facilities through cyber physical sys-

tems. In Industrial internet of things, pages 287–313.

Springer.

Shahbazi, Z. and Byun, Y.-C. (2021). Integration of

blockchain, iot and machine learning for multistage

quality control and enhancing security in smart manu-

facturing. Sensors, 21(4):1467.

Shefﬁ, Y. (2007). The resilient enterprise: overcoming vul-

nerability for competitive advantage. Zone Books.

Tellaeche, A. and Arana, R. (2013). Machine learning al-

gorithms for quality control in plastic molding indus-

try. In 2013 IEEE 18th Conference on Emerging Tech-

nologies & Factory Automation (ETFA), pages 1–4.

IEEE.

Tsang, Y. P., Choy, K. L., Wu, C.-H., Ho, G. T., Lam,

C. H., and Koo, P. (2018). An internet of things (iot)-

based risk monitoring system for managing cold sup-

ply chain risks. Industrial Management & Data Sys-

tems.

Wang, T., Chen, Y., Qiao, M., and Snoussi, H. (2018). A fast

and robust convolutional neural network-based defect

detection model in product quality control. The In-

ternational Journal of Advanced Manufacturing Tech-

nology, 94(9):3465–3471.

Zamiri, M., Marcelino-Jesus, E., Calado, J., Sarraipa, J.,

and Goncalves, R. J. (2019). Knowledge management

in research collaboration networks. In 2019 Interna-

tional Conference on Industrial Engineering and Sys-

tems Management (IESM), pages 1–6. IEEE.

IoT, Risk and Resilience based Framework for Quality Control: Application for Production in Plastic Machining

611