«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
28
normalized one due to geometrical properties or transforms, including size and pose. So, face is
normalized further using photometrical properties, like illumination.
After a face is finally normalized, the next step f feature extraction is performed to represent
exact data that is beneficial to distinguish faces of different persons. On the stage of face matching,
the vector of extracted feature of the input image is matched against the enrolled faces from the
database or catalog of human emotions. The output identify whether a match is found or indicates
that face is unknown.
Technical challenges
The results of many face recognition methods using artificial intelligence
deteriorate with several factors such as a bad quality of lighting or pose.
The key technical challenges are:
- Wide variety of face performance: Even if shape and reflectance are fundamental elements
of any face, the performance reflects on several other properties, like pose or facial
microexpressions.
- High dimensionality and small size of input image: Another challenge is the ability to
analyze face image that consists of, for instance, 110 × 90 resides in a high dimensional feature
space, like 10300. However, only few numbers of examples per person are available for analyzing
and the numerous variations are usually much smaller than the dimensionality of the image space.
Thus, system is not able to generalize effectively because of unseen parts of the face.
The only problem in usage of the technology could be the privacy concerns. If technology is
used by police, court of law or enforcement agencies during the interrogation, there can be posed a
danger due to participant’s dissent and misunderstood feelings.
Conclusion
As any innovation, emotion analyzing software brings valuable opportunities to expand the
abilities of artificial intelligence and influence of machines on the prosperity of the humanity. The
main process contains of four main stages starting from putting the image till the determination of
the face by identifier. Even if the calculations and functions are valued as complex ones, the
possibilities of machine learning and neural networks considered as a powerful tools to achieve
effective outcomes with exact information.
Nevertheless, similar to other modern technologies, the emotion detecting software raises
many types of privacy questions. Recently, there has been a heated discussion concerning to low
level of control over the third parties that use the software keeping the images and emotions for
personal purposes. Dr. Ekman says he hopes the government will step in and write rules to protect
people. He says that in public spaces, such as shopping malls, consumers should at least be
informed if their emotions are captured.
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
29
However, researchers point out that this tool has done more benefit than harm. Even though,
the human emotions are too complicated. If humans understand the hidden feelings, smart machines
can only recognize different patterns in data from database of emotions. Emotional artificial
intelligence is growing alongside with other techniques as the internet of everything. If computers
can monitor our every move and compile that data, then the line between humanity and machines
could disappear in the nearest future.
References:
1. Fundamentals of Face Recognition Techniques http://shodhganga. inflibnet.ac.in/ bitstream/
10603/3478/13/13_chapter%203.pdf
2.
Neal,
M.
(2013).
Emotionally
intelligent
machines
are
closer
than
ever.
http://motherboard.vice.com/blog/emotionally-intelligent-machines-are-closer-than-ever
Hutton, K. (2014). What is human analytics.
3. http://www.acl.com/2014/07/what-is-human-analytics/
UDC 004.89
BUKENOV S.A.
USING NEURAL NETWORKS TO IMPROVE EMOTIONAL STATE OF PERSON
(Kazakh-British Technical University, Almaty, Kazakhstan)
Abstract
Emotions is a big field of scientific research in psychology. There are a lot of scientific
papers and methods on how to improve your emotional state. Projects in field of neural networks
that are related to emotions usually try to simulate human emotions, but not so much done on using
neural networks to improve emotional state of human being.
1 Introduction and psychology background of paper
In this paper we will discuss human emotions and the ways to work with them by using
neural networks. Usually we don’t think about where do emotions come from. But psychology
gives basics for this paper.
As we know animals have emotions too. Emotions are part of our nature, because they
helped us to survive. We have emotional bonds with other people, because emotions keep us
together. In order to survive we need to be together. There are a lot of other emotions that help us
to survive. Fear activates our organisms to be very effective as heart rate goes up and reaction speed
increases. “Emotions have helped us to survive. When we lived in the wild with monkeys and
mastodons and tigers we needed emotions in order to react quickly to dangerous stimuli” Ilana
Simons Ph.D.. Negative emotions help us to stay away from situations and not to repeat them.
Positive emotions tend us to repeat such situations.
So our emotions are more part of our nature, organism than unique quality of human beings.
Emotions have a strong connection with our body. That’s why a lot of people have diseases that
came from their emotions. We don’t live in natural world anymore. Our worries are not predators or
other things that can be dangerous only for a short time. In modern world we have situations that
may make us anxious for a long time. And that is not what nature prepared us for. As negative
emotions influence our body in long time period – it can cause problems. “Chronic stress from
negative attitudes and feelings of helplessness and hopelessness can upset the body's hormone
balance and deplete the brain chemicals required for feelings of happiness, as well as have a
damaging impact on the immune system. New scientific understandings have also identified the
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
30
process by which chronic stress can actually decrease our life span by shortening our telomeres (the
“end caps” of our DNA strands, which play a big role in aging). Poorly managed or repressed anger
(hostility) is also related to a slew of health conditions, such as hypertension, cardiovascular
disease, digestive disorders, and infection” Karen Lawson, MD.
So if our emotions have such influence on our bodies and modern world created a lot of
situations where people may have worries we need something to understand. As we know there are
people that have more positive emotions in particular situation than others. For example for some
people financial problems can make them work harder as a result of love for close ones but for the
other it can make person surrender to situation. Our emotions have strong relation to our
effectiveness too. Depending on emotions that arises in us in particular situation we can have
different effectiveness.
Why the same situations can create different emotions in different people? It is very
important question. But there is a simple answer. It all depends on previous experience of a human.
“People respond with different emotions to the same situations depending on how they appraise the
situation” Matthias Siemer, James J. Gross, Iris Mauss. We will have a brief look at their
experiment in this field later.When a human being comes to a situation for the first time where there
is no emotions associated with it he just lives through it. But when some emotion arises during or
after the situation then he has an emotion that is associated to such situations. For example, if you
decided to go by bus for the first time and there are a lot of people there that are arguing and driver
drives bus like crazy you will have stressful emotion that will be associated with bus. And our
emotional association with situation is a sum of all experiences. But each next situation is affected
by previous emotional experiences so that human will see this situation through all previous
emotions. That’s why people are so dependent on emotions and not objective when emotions come
into play. But emotions are everywhere so there is a very little possibility to be objective for a
human.
Almost all situations we faced in our lives were faced by others. Almost in all cases we have
something to learn from others and we can find better ways to deal with it. Such neural network will
help people to do it.
2 What is the problem we’re trying to solve?
We want to change appraisals of human beings depending on his situation. If we can change
appraisal of human being then we will be able to change emotions related to specific situation.
Neural network will provide stories that will change the appraisal of user and then he will feel better
in his situation.
3 How neural network will work in this case?
So if we look at emotions from perspective of neural networks then our neural network gives
certain emotions associated with certain situation. But the most effective way for a human is when
his neural network gives good emotions on all situations. So the people that have positive emotions
in same situations where other people have negative emotions are just programmed in different
way, they have different neural network. So in order to change neural network of a human that has
bad experiences with certain situation we need to provide experiences of people that have good
emotions associated with same situations. In order to do it we need to create a neural network that
takes the situation and returns the experiences and stories of other people that had good results and
emotions associated with this situation.
Bad emotions are also needed for human beings but most of situations that create anxieties
in people nowadays shouldn’t have bad emotions associated with them. Once human being is in this
situation he should be effective. Bad emotions are useful to avoid situations, not to being in a
situation.
Most of modern papers that are related to neural networks and emotions are thinking about
the ways and use of creating neural network with emotions. There is a big question whether it is
useful or not, because in modern world bad emotions are not helpful in many situations.
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
31
But the aim of this paper is to create a neural network that will help people that have bad
emotions associated with their situation to have a more positive look on it. By reading the
experiences and ways other people dealt with same situations it can show the right and effective
ways to deal with it and also ways to have good emotions from same situations. That’s why Dale
Carnegie’s book gives such a big impression on people, because it provides tons of examples of
people going through hard situations and dealing with them in a good way. By reading such
experiences people can change their neural network and begin to associate their life situations with
positive emotions. It can change life of people and affect quality of their life. It can help people deal
with hard situations. It can help to share your own experiences and ways to deal with your
emotions.
Our neural network will consist of 3 layers. Input layer, one hidden layer and the output
layer.
This network will take words as an input. To obtain words from input text of a user we take
roots of all words and take n most popular words. After this process we have a vector of words,
associated with the frequency of its' occurrence in the text, normalized to a value from 0 (didn't
appear in text) to 1 (most popular word in text). Each input of net has an index that is associated
with a word. Number of inputs of neural net is a total number of words that net knows. Neural
network is not totally connected. When vector of n words comes to neural network only n inputs of
network are activated.
Hidden layer has same number of neurons as input layer. Each neuron of output layer
represents a text that neural network knows. Each neuron in hidden layer is connected to neurons in
output layer if this word occur in this text. Weight of connection between neuron of output layer
(text) and neuron of hidden layer (word) is frequency of occurrence if this word in the text,
generated the same way as with input words. So the more likeliness we have in popular words of
input text and popular words of output text - the more the value of output neuron will be.
Neural networks have different uses. In this case we'll use it's ability of pattern recognition.
When user wants to read the stories from neural network first he has to describe his situation. The
more details he will give – the more accurate will be the output of neural network. Pattern
recognition here is used to find to which topic does his situation apply. After system recognizes the
topic it will give the stories that are related to this topic that hopefully will help the person. All
stories should also have ratings. Single story can be related to several topics at the same time.
4 Conclusion
As a conclusion we want to say that such system can give connection of most wise people
with people that have emotional problems. In may help emotionally and also practically, because it
gives information about how to cope with situation. It can have great impact on worldview of
people and make them more effective and happy.
References:
1. Same Situation—Different Emotions: How Appraisals Shape Our Emotions - Matthias Siemer,
James J. Gross, Iris Mauss.
2. Topic Detection, Tracking and Trend Analysis Using Self-organizing Neural Networks, K.
Rajaraman and Ah-Hwee Tan.
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
32
UDC 004.514.64
KAPSALYAMOV A.B., NURGALIYEV K., SERIMBETOV B.A.
DESIGN OF PLC BASED SMART ELEVATOR CONTROL
(Kazakh University of Technology and Business, Astana, Kazakhstan)
Abstract— This project is about the design and control of smart elevator system using
Siemens PLC S7-1200. SIMATIC HMI panel has been used for the interface between user and
system itself. Laboratory sized elevator prototype has been developed which consists of two cabins.
Each floor of the elevator has the HMI panel together with the limit switches installed. The relay
outputs of Programmable logic controller have been applied to drive 2-wire DC motors forward and
reverse directions. Moreover this article presents the Ladder logic diagram which describes what
algorithm has been applied to control the system. Specific descriptions of the Hardware and
Software parts are analyzed further in this paper. Keywords—Automation, Smart Elevator,
Hardware design, PLC (Programmable Logic Controller), HMI (Human machine interface).
I. Introduction
It is difficult to imagine the high skyscrapers without elevators installed on it in the modern
world. Lots of buildings would have been limited to a very few floors and the architecture of the
current world would not have existed in the way as it is today. Elisha Graves Otis was one who
invented the safety elevator in 1862 and provided rise for the modern skyscrapers [1]. Modern
elevators are responsible for transferring passengers, heavy hardware items from one floor of the
building to another one without making lots of effort. Due to modern technologies development,
skyscrapers are becoming higher, which made control system of the elevators more complex.
Previously, the control of the elevators was made by using chips based controllers [2]. The
drawback of using chip based controller is that they cannot operate and withstand all the harsh
conditions which you can face in industries and hospitals. In this project PLC has been applied for
controlling the system. PLC is widespread in industrial automation field. It has lots of advantages. It
is easy to control outputs based on what kind of inputs you have. PLCs are able to withstand harsh
conditions like high temperatures. They can operate for 5-10 years depending on different
conditions. Moreover due to the communication abilities PLC can be located in a far distance from
the system to perform various control instructions [3]. For programming, PLC uses Ladder logic
diagram. So the purpose of the project is to design smart algorithm that controls two cabins of the
elevator in an efficient way. And for the user interface HMI has been applied because it can meet
various complex processes of systems and it is easy to integrate to the PLC and system itself [4].
II. Control system of the elevator
A. Shortcomings of standard elevator control
Elevators are considered by many to be purely functional machines. Users only want to get
to their destination floor as fast as possible and wait for the elevator the shortest amount of time.
Technically, the goal of the generic elevator control algorithm is to minimize elevators 2 basic
variables: users waiting time and travel time. The oldest and most popular elevator control logic is
the one that minimizes the waiting time [5]. When the user requests an elevator cabin to a certain
floor the closest car to the users floor is called. This greatly decreases the waiting time, but
increases the travel time. Number of empty elevator travel is high. Moreover, the situation occurs
when the elevator is full, but it stops at the requested floors. In this project the logic was developed
to satisfy users basic needs of short waiting time and fewer stops. The algorithm also minimizes the
number of empty car travel thus reducing the energy consumption.
B. Control setup
Elevator control system consists of Siemens S7-1200 PLC and Simatic Human Machine
Interface and limit switches to detect the elevator car position. Inputs to the PLC are 8 limit
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
33
switches, car-call buttons and buttons which indicate the number of passengers requesting the car.
Outputs are 2 DC motors which move the car-counterweight pair up and down. Below is the
diagram of the inputs and outputs.
Figure 1 - PLC Inputs and Outputs
C. Software design
The program was written in Ladder Logic language. The program consists of three modules:
module that registers the elevator car current position, left car movement control module and right
car movement control module
D. Car position Control
Figure 2 - Car Position
The ladder program above registers the floor number at which the car is located. For
instance when the car is located on the first floor it closes the normally open limit switch activating
the circuit. Move block is then enabled and stores the car position in respecting variable. The same
principle works for the 2nd car.
E. Light load car control logic
«ҚОҒАМДЫ АҚПАРАТТАНДЫРУ» V ХАЛЫҚАРАЛЫҚ ҒЫЛЫМИ-ПРАКТИКАЛЫҚ КОНФЕРЕНЦИЯ
34
Figure 3 - Right Car Control Logic
If the number of passengers is less than three, this logic is activated. It compares current car
position with floor number where the car is called and moves up or down accordingly. When the car
reaches the request floor timer is activated which holds the elevator for 3 seconds. Then move block
assigns desired floor number to Request Floor variable to run the logic again in order to get the car
to the desired floor.
F. Heavy load car control
For the heavy load the algorithm is the same except it works for number of passengers equal
to or greater than 3. The HMI is located outside the car.
Figure 4 - Left Car Control Logic
Достарыңызбен бөлісу: |