,allowExpansion)
Lavazza chose Amazon Web Services as its cloud platform and Reply, AWS Premier Consulting Partner, to support them in the adoption of machine learning models on AWS.
Manufacturing industries are putting an increasing effort to improve the efficiency of their production processes, in order to have a positive impact on many aspects, from waste reduction to a higher level of quality and customer satisfaction. Quality is therefore a major concern, and the related data collected plays an important role in the definition of a good outcome of the production line. Data gathered from sensors and unstructured sources may be used to build Artificial Intelligence tools which predict the quality of the next batch and reduce the quantity of waste products, as well as suggest improvements in the settings of the machines.
The production process of Lavazza is based on high value raw material and the quality of the final product plays an important role for the company to stay ahead of the competitors.
Tracking the product quality levels and monitoring the machine working state in real-time allows the production managers to identify potential anomalies in advance, preventing low production quality.
Among the challenges to keep high quality standards there are many variables that impact the quality of the output, like time and temperature to be used in the process. In addition, the production line needs to manage different types of products in order to guarantee production efficiency, especially in a market where the launch of a new blend of coffee in short time could guarantee a great advantage in winning the consumers.
The data produced by the machines had to be integrated with those built by the operators. The highly-skilled operators produce additional data which are injected in the system, along with the data produced by the machines.
To be effective, the response on the prediction of quality had to be in near real time, which is to say that the algorithm had to be optimized even when processing large quantities of data.
Lavazza worked with Reply to design a product which could fit their needs to predict the results of the tests performed on the production line to guide their operator’s activities.
To do so, the data are collected from heterogeneous sources in a single AWS Data Lake and processed using Python Reply used a Random Forest model to forecast the quality of the output based on both machine settings and working environment data, performed the analysis of the features importance to identify the parameters related to quality flaws and built a real-time custom web app to visualize results of the model and to interact with machine PLC letting the Operator to send back optimal working parameters according to the model.
Moreover, it was developed a mobile app to display alerts related to the process (e.g. start/end, excessive flaws) and abnormal values of the machine. The outcome of the prediction was compared with the quality data which is normally collected after the tests execution.
To do so, the data are collected from heterogeneous sources in a single AWS Data Lake and processed using Python Reply used a Random Forest model to forecast the quality of the output based on both machine settings and working environment data, performed the analysis of the features importance to identify the parameters related to quality flaws and built a real-time custom web app to visualize results of the model and to interact with machine PLC letting the Operator to send back optimal working parameters according to the model. Moreover, it was developed a mobile app to display alerts related to the process (e.g. start/end, excessive flaws) and abnormal values of the machine.
The outcome of the prediction was compared with the quality data which is normally collected after the tests execution.
Data Reply worked with Lavazza due to its high skills in data analytics, in cooperation with Hermes Reply and their deep knowledge of production systems. This organization provided the necessary know-how on the industry topics and the data processing. After the successful completion of a Proof-Of-Concept which demonstrated the quality of the algorithm, Lavazza asked Reply to proceed with the implementation and was able to understand that the prediction is very near to the final result from the production line data, resulting in an efficient tool to predict quality without waiting for the whole production process to complete.
Established in 1895 in Turin, the Italian roaster has been owned by the Lavazza family for four generations. In sixth place in the world ranking of roasters, the Group currently operates in more than 90 countries through subsidiaries and distributors, exporting 60% of its production. Lavazza employs a total of about 3,000 people — after the Carte Noire acquisition — with a turnover of more than €1.9 billion in 2016. Lavazza invented the concept of blending — or in other words the art of combining different types of coffee from different geographical areas — in its early years and this continues to be a distinctive feature of most of its products.
As part of the Reply group, Data Reply offers a wide range of services that support customers in becoming data driven. We operate in various industries and business areas and work intensively with our customers so that they can achieve meaningful results through the effective use of data. Data Reply offers many years of experience in transformation projects to achieve “data-driven companies”. We focus on the development of data platforms, machine learning solutions and streaming applications - automated, efficient, and scalable - without making any compromises in IT security.
Hermes Reply is a company within the Reply Group that specializes in architectural and technological solutions, application maintenance, business processes, and consulting for the digital transformation of manufacturing clients. The company acts as a system integrator between ERP, PLM, and CRM levels and shop floor solutions. Hermes Reply leverages Industrial IoT (I-IoT) technologies for the horizontal integration of production processes by implementing Edge, Fog, and Cloud computing solutions, as well as Industry 4.0 concepts to optimize customer processes.