Appunti Computer Science for Food industry

Computer science for food industry (Prof. Comba)

Introduction

Flows management in manufacturing process

Manufacturing industries produce goods. This is a general scheme that summarizes a lot of aspects. It is a justification of how management of data and, then, information, is really relevant in many situations. Let’s start from the “Physical process”. This part of the scheme represents the production plan in which a process is performed. It could be a thermal process, mixing and whatever you can imagine applied to the food industries.

There are many kinds of flows. In particular, there are materials (raw materials arrive in a company in which are stored and then processed to create something new from a mixing of different ingredients) as input of the process and, then, come out (processed products), ready for the next stage of the supply chain. In a process, we need energy, so there is also the incoming of it. The balance of energy should be always constant, because it can’t be created or destroyed. A portion of this energy can be stored in the product. If we have a pump, the product earn energy because, for example, it is moved up and, then, it will lose energy going down in another part of the process.

In every kind of plan, there is a “Control system”. We have a lot of different control systems. The simpler one is the manual control, but, even if the control is manual, we have both sensors and actuators. This scheme is valid in both the simplest and the most complex situation.

In every situation, there are sensors (that can be analogical for example). Every kind of measurement is made by a sensor (ex. visual sensors, alarm sensors, etc.). Sensors are our eyes on the system and actuators are the way we act on the plant. Sensors have become digital, linked with wires in a network, so that a machine can exploit the data provided by the sensor and perform automatically an operation we need. So, actuators are powered by motors, automatic valves, pumps (controlled by inverters). In this way, the system can react to some modification in an automatic way. Sensors can be wireless or controlled by a computer, a machine, PLC, etc. This kind of box can be of many technologies, even industrial computers.

The number of sensors, nowadays, is growing a lot, because they are cheaper and more and more data provided by sensors have been developed. Once, we only monitored some physical process in specific points of the plant, especially the more relevant ones. Nowadays, there are a lot of sensors, so many aspects are kept updated, from the monitoring point of view.

More and more sophisticated control systems have been developed. These control systems are nothing more than processes that have to collect data that are becoming a lot. Nowadays, many additional checks are performed on the products (ex. metal detectors, scanners, etc.). All these are sensors that increase safety of the process. Control system should be able to process a growing amount of data.

There is an additional arrow in the control system that is “incoming information”, not from the physical plant. The most challenging aspect nowadays is to exploit such a huge amount of data to extract valuable information. We can place a sensor on the plant (ex. add an optical sensor on a gate to check incoming and outcoming flow of products), however this kind of sensor provides us data with a quite high frame rate. This last is the speed with which we read the state of the sensor and is related to the dynamic behaviour of the system we are analysing. So, if we have something very fast in its change, we have to check it very often. A good and smart control system should be able to find the appropriate way of exploiting this kind of data. We should manage these data and detect hidden pattern. In few words, finding a hidden pattern means understanding the huge amount of data trying to find the valuable information for our decision (because, at the end, we have to make a decision).

Information systems (IS) – Sistema informativo aziendale (SIA)

Information Systems are the all set of things related to data management. In particular, integrated set of components for collecting, storing and processing data and for providing information and knowledge. There is a wide definition in which we can put many different technologies, people, procedures, so everything required to work in an efficient way with data and get information. An IS, in fact, includes software, hardware, data, people, and procedures.

There is a clear distinction between information system (IS) and informatic system (= hardware). There are hardwares such as servers, networks, but, in general, we are moving quite faster in the direction of not having hardware in the company anymore, but buy services, usually provided by external companies. The machine should be able to perform well in short time periods, in which we need a lot of resources.

Data

Data is a set of values of subjects with respect to qualitative or quantitative variables. Raw data is a term used to describe data in its most basic digital format. Big data require specific tools to be analysed, but also to be seen. Please note: data and information (or knowledge) are often used interchangeably; however, data becomes information when it is viewed in context or in post-analysis. Data is something rawer than information and, for example, provided by a sensor. The temperature itself do not provide many information if we don’t know the context from which this temperature is measured and what process is performed in that moment.

Levels of abstraction

• Physical level: hardware this term describe how data are stored and managed (which hardware, CPU, different kinds of connections, like protocol connections and so on – e.g., students) actually we don’t care about this
• Logical level: it is more relevant, because it describes data stored in database and the relationships among data. We consider some elements that describe the “students”. They are called attributes (ID, name, place of birth, etc.). The most important thing is the relationship among data that is also stored in this kind of level
• View level: it is what the user sees (e.g., web page in which you have to fill in your data, display of cash point, a social network on our smartphones is a user interface of a database, etc.). The independency of this level, since this is a complex system, is very important and allows to change the user interface (e.g., registration process, add passport ID, check-in, etc.). The information and the way you can interact with data could change the way you see the data, but all the other things that are hidden should and can remain the same. So, every level can be changed without affecting the others.

Application programs hide details of data types. Views can also hide information (such as an employee’s salary) for privacy purposes. The distinction between physical, logical and view level is the clearer hardware and how we save and organise data. A little bit less between logical level and the view one because, in real applications, sometimes, modifications in the logical level can affect the view one, only in the part we add data.

Example information

Set of data to which a principle for interpretation has been associated. Resource needed to effectively plan and control business activities. From dictionary: “News, data or element allowing to have more or less exact knowledge of facts, situations, ways of being”. We have to make decision as correct and fast as possible. Decision making is one of the most difficult phases in the management of an activity.

Quantity measurement

Kilobyte (kB) 1024 byte = 2¹⁰ byte 2 kB ≈ one page text 100 kB ≈ very-low resolution image. Megabyte (MB) 1024 kB = 2²⁰ byte ≈ 10⁶ byte 1.44 MB = floppy disk storage capability 5 MB ≈ whole Shakespeare literary production 10 MB ≈ high resolution image 10 MB ≈ 1 minute audio track (CD quality) 680 MB = CD-ROM storage capability. Gigabyte (GB) 2³⁰ byte ≈ 10⁹ byte 1 GB ≈ a truck full of books 9.4 GB = double side DVD storage capability 20 GB ≈ whole Beethoven musical production. Terabyte (TB) ≈ 10¹² byte 2 TB ≈ entire university library 400 TB ≈ climatic data in the National Climatic Data Center (NOAA) database. Petabyte (PB) ≈ 10¹⁵ byte 4 PB ≈ data created on Facebook 27 PB ≈ monthly YouTube data traffic. Exabyte (EB) ≈ 10¹⁸ byte 5 EB ≈ whole world printed pages (until 2007). Zettabyte (ZB) ≈ 10²¹ byte 5.2 ZB ≈ whole world used data in one year (2019).

We have many examples of how the amount of data is growing during the time. Images and videos are growing in their weights, in sense of storage requirements. For example, the whole Shakespeare literary production can be saved in only 5 MB. When we put together a huge amount of data, this amount grows a lot and, for example, climatic data in the National Climatic Data Center (NOAA) reach 400 TB.

Data warehouse

Companies spend a lot of money for maintaining the safety level of their information systems. One positive aspect of the cloud is that the safety is demanded to another company. For example, if we use an external hard drive and keep it in a safe place, but if something happens to it, we lose both the original date and the copy.

Production comparison

Assuming that:

• Reading 3 books a week of 400 pages each (size in pdf electronic format ≈ 500 kB)
• Live 100 years, 5200 weeks → 5200 x 3 x 500 kB = 7.8 GB are required to store the 15600 books

All this material doesn’t occupy even a DVD, whose nominal capacity, in the case of double-sided ones, is 9.4 GB!

Applications

In food industry, there are many applications. In it, in addition to standard company operations, an appropriate DB-system allows to:

• Implement efficient traceability system
• Properly manage recalls
• Provide additional information to consumers (mobile app, etc.)

We interact with a database when we buy something in an online shop.

Traceability

We use traceability if we want to know, for example, which specific raw material has been used for the product. In the label, we can find the amount of ingredients, but the company knows that the sugar used for this lot of sweets was bought in a certain period and comes from that company. If there are some problems in this particular sugar and if a good traceability system is implemented, the company can make a recall and avoid that some products reach the market or remove them from the market, knowing, for example, where they have been sold. From smartphones, for example, we can open a web page of the company, in which we can find all the information of the lot.

Data processing systems for the automatic retrieval of information: barcodes

Databases

From data to information

This image is a sort of pyramid organisation, in which, on one axis, there is the amount of data (related to a specific level of organisation in a company) and, on the other one, the value of these data (that becomes more and more).

At the bottom of the scheme, there is a huge amount of data (they can relate to facilities in the company, so, for example, the recordings of incoming and outcoming from warehouse, plant information of flow materials, etc.). Currently, these data are collected in networks in a quite automatic way. They are collected on the plant level in groups of machines. There are sensors linked to this kind of devices that are in the plant. At the beginning, in a machine level, then in a group of machines that, usually, interact with each other. This low-value, but large-volume data is collected in one plant level. Then they start to deep process in order to get information for higher decision level.

Lower-level decisions are related to the control of the production process (e.g., if the temperature is not high enough, we have to increase the heating of a cooker). Other relevant information should be exploited, for example, for the management of the warehouse, in order to always have raw materials to be processed, or the facilities of packaging and shipment of products.

Often the availability of excessive amount of data makes to extrapolate significant information very difficult. Methods of selection and progressive synthesis are required. There are different methods:

• Statistics
• Data mining system based on analysis of data (fonti informative primarie), that are selected to produce a report and, in conclusion, a decision (indicazione strategica). This decision is originated from a huge amount of data

All information about production can be further analysed to get information for business planning. The exponential increase in the volume of operational data has made the calculator the only support suitable for decision making. We have to automatise simple actions that we have to do many times (e.g., packaging machines). As we grow in the level of values of information, we need external information to make decisions (e.g., marketing, environment in which we have to act).

When we have to analyse huge amount of data, we need to organise them in an effective way and this is why, in this graph, we have reports. They are tools that summarise relevant information of data, reaching strategic layout for decision making. Given the shape of this graph, the quantity of information should decrease compared to raw data. There is also programmable automation. This kind of operation deeply modified manufacturing industries. The most modern approach of automation is the flexible automation. It is similar to the programmable one in a certain way, because you can change your production that is performed automatically, but the important thing is that here you can modify, in an automatic way, the production that is automatic. For these reasons, it is more dynamic compared to other programmable approaches. In the flexible automation, raised in the last 10-15 years, so many different technologies and solutions help to reach goals.

Why information become more and more effective? The production is becoming more flexible in general. It is difficult that a company produces one item for 15 years. The market is quite fast nowadays. Flexible production should follow, in a dynamic way, the market. Dynamic production is becoming more challenging, so it is important to be able to change fast the variables. The information system of the company could be linked to the website of the company, so customers can act with the company.

Data mining

The term “data mining” refers loosely to the process of semi-automatically analysing large databases to find useful patterns. Like knowledge discovery in artificial intelligence (also called machine learning) or statistical analysis, data mining attempts to discover rules and patterns from data. However, data mining differs from machine learning and statistics in that it deals with large volumes of data, stored primarily on disk. That is, data mining deals with “knowledge discovery in databases.” Businesses have begun to exploit the burgeoning data online to make better decisions about their activities, such as what items to stock and how best to target customers to increase sales.

The term “mining” is like “minare” in the sense of “miniera”, to get something. Data mining is strictly related to databases and it is becoming more and more relevant in industries. Please note: data mining is not the operation of getting data, so we have not to dig to find them, but it is the operation of explore data. For this reason, data mining is to get information from data (so, it should be “information mining” and not “data mining”).

Information systems (IS) - Sistema informativo aziendale (SIA)

Integrated set of components for collecting, storing and processing data and for providing information and knowledge to:

• Operational activities (service information)
• Management activities (management information)
• The organisation's planning, control and evaluation activities (government information or strategic level)

An IS includes software, hardware, data, people, and procedures.

Informatic systems

Set of computer tools used to automatically process information from an organization (e.g., something related to hardware, computer, networks, etc.).

Information systems life cycle

• Preliminary analysis system analysis, requirements definition
• System design
• Development, integration and testing
• Installation and deployment
• Maintenance evaluation disposal preliminary analysis …

The design phase (systems design) on which we focus on is only a small step in the overall life of information systems. Seeing the cycle, it seems to be a quite linear process; for some aspects it is true. The most relevant thing is that the cycle is performed many times during the normal life of the system for what concerned the evaluation and maintenance activities, as the company grows and modifies its activities. The IS follows these modifications, so there is a continuous process of updating these important activities of a company.

The only thing we have to remember is the aspect related to the cycle. For this reason, we have to make an operation of maintaining and updating a database because it is not something static, but dynamic during the time with a cycle, in which we have to see if the IS fulfilled the requirements of the company. If a database and an IS are well defined, there is an independency between the design (how data are organised) and how data are physically saved on the hard drive.

• Preliminary analysis: Decisions are made about the business area that must be the subject of automation
• Systems analysis, requirements definition: Define project goals into defined functions and operations of the intended application. This involves the process of gathering and interpreting facts, diagnosing problems and recommending improvements to the system. Project goals will be further aided by analysis of end-user information needs and the removal of any inconsistencies and incompleteness in these requirements