Fig 6 - uploaded by Wojciech Kacalak
Content may be subject to copyright.
Source publication
Authors propose a method of normalizing and analyzing structures, enabling automated cross-comparison of features observed in tested structures against predefined ones (both correct and incorrect reference elements). Application of this method, enables usage of numerical methods and neural networks for rapid classification of elements, while mainta...
Contexts in source publication
Context 1
... the purposes of a specific analysis, the values of a normalized, symbolic description have to be serialized in a consistent manner, so that the algorithms using it can make assumptions about the data's structure. In addition to the simple arrays and binary forms, we can present input data as a binary table (Fig. 6.) where rows list all processed features and the assignment of the active bit to one of the columns represent its value. Activated bit will depend on the magnitude of the normalized feature value. This approach can lead to precision loss that is dependent on the number of columns allowing differentiation between feature values, however a ...
Context 2
... in mind the significance of the calculation models enabled by neural networks and machine learning algorithms, we propose usage of such normalized, symbolic representation of structural features with a Hamming net [10] (Fig. 6.) and a probabilistic network [13] (Fig. 7.). The benefits of applying neural networks arise largely from being able to handle much larger amounts and variations of input data, resulting in a more flexible but possibly less precise qualitative calculation model. Applying a Hamming net to the feature data of the tested mechanical ...
Context 3
... the feature values is considered a bit, providing a value directly related to the differences observed in the structures. This means that as much as there is no direct valuation or grading between tested elements, the method provides a quick way of locating the most similar antipatterns. This can be observed as locating similarities in the tables (Fig. 6.) and making a quality decision based on the similarity to one of the antipatterns (possibly expanded by using Hamming's weight ...
Context 4
... approaches (Fig. 6., Fig. 7.) enable classification of structures basing on their features. In the Hamming net, the distance between two compared structures (being the bias of the neuron function) is calculated as a Hamming distance (number of differing bits, in our case rows) between two serialized values. The precision of both these approach can also be further ...
Similar publications
The rapid development of deep learning improves the detection and classification of attacks on intrusion detection systems. However, the unbalanced data issue increases the complexity of the architecture model. This study proposes a novel deep learning model to overcome the problem of classifying multi-class attacks. The deep learning model consist...
Citations
... Dla właściwego przeprowadzenia procesu szlifowania powierzchni śrubowych, niezbędne jest rozwiązanie nie tylko problemów geometrycznych, ale także wielu problemów technologicznych [29,30,31,32,33,34,36,38,41,54]. ...
Powierzchnie śrubowe znajdują szerokie zastosowanie w technice. Do nich należą między innymi powierzchnie zwojów gwintów, kół zębatych, ślimaków, ślimacznic, powierzchni rowków wierteł oraz innych elementów roboczych w układach mechanicznych. Mechanizmy śrubowo-toczne, zwane także śrubami pociągowymi tocznymi stosowane są powszechnie w układach napędów i pozycjonowania. Przekładnie ślimakowe są stosowane między innymi w mechanizmach podziałowych w obrabiarkach, podzielnicach, stołach obrotowych, w przekładniach kierowniczych samochodów, reduktorach ślimakowych, w układach pozycjonowania, systemach sterowania napędem statków. Najważniejsze wymagania stawiane procesom kształtowania powierzchni śrubowych dotyczą wysokiej dokładności skoku, gdyż od tego zależy dokładność pozycjonowania przemieszczanych elementów. Opracowanie zawiera omówienie konstrukcyjnych cech powierzchni śrubowych i problemów związanych z ich szlifowaniem. Procesy te bowiem mają uwarunkowania wynikające z kształtów narzędzi ściernych. Przedstawiono matematyczny opis powierzchni śrubowych, w szczególności zębów ślimaków walcowych. Przeprowadzono analizy dotyczące dokładności kształtowania zarysu szlifowanych powierzchni śrubowych. Wskazano sposoby minimalizowania odchyłek dokładności. Z kolei od dokładności zarysu powierzchni śrubowych współpracujących elementów przekładni oraz właściwości stereometrycznych powierzchni i właściwości fizycznych warstwy wierzchniej, zależy obciążalność i trwałość przekładni.
... The quality grading of the machine elements based on the variation from the set of antipatterns [14,15,17], can be used both to avoid repeatable design errors, and to automatically compare construction elements based on their similarity to incorrect ones. As long as this method allows to detect design errors that were known earlier, its results should not be treated as exhaustive. ...
... Creating a description of a construction starts with its function based decomposition into base elements, to which we later assign attributes with concrete feature values. Such structured description can be compared to another one, but not in a trivial manner due to variations in the constructions and their parameters [15]. Such approach makes it easier to compare elements of mechanical constructions through similarities in their structures and feature values. ...
... To enable automatic detection of mechanical design errors through the similarity to antipatterns (Fig. 3) [14,15,17], the classification algorithms have to enable concrete definition of the antipattern set. Such data set should contain data samples focusing on demonstrating the incorrect nature of the design. ...
Proposed method allows for early detection of mistakes in designs of mechanical constructions. It is based on a numerical classification of a symbolic representation of construction’s features against a set of defined antipatterns (known, incorrect, repeatable data patterns). We present an approach to identify antipatterns described using a symbolic language KXML and a method of intelligent quality assessment enabling calculation of the similarity of the tested element with the antipattern data set. Additionally, we highlight selected properties of numerical models directly supporting analysis of the structure of mechanical constructions.
... Analysis of antipatterns has been first conceptualized and identified in computer science [1] and since the concept has been slowly adopted by other science disciplines. We provide concrete examples of applications of antipatterns in machine construction and operations [3,4,9]. Symbolic representation of erroneous data patterns, for the purpose of detecting similarity, can be described with regards to the origin of the error (incorrect feature value, incorrect relation between feature values, incorrect structure with feature values) and the phase in its lifetime (design, production, operations, recall). ...
... In each phase of the machine's lifetime it is possible to identify and meaningfully apply the method of detecting similarities to known antipatterns [3,4,9] to increase the overall quality of the process. As in other phases, the antipatterns can be identified in designs of incorrect feature values (Fig. 4), relations between feature values (Fig. 5), and structural errors (Fig. 6). ...
Authors propose a method to classify the antipatterns, allowing their further definition for manual and automatic usage. We apply antipatterns in quality assessment as a negative quality reference. This enables counting and measuring similarity to predictable errors (antipatterns), automatic identification of erroneous structures found in mechanical designs, or anomalies in machine operations. We present approaches in which innovative use of antipatterns allows technical employees to improve their existing processes. In essence, we have found that antipatterns are useful in design, monitoring, and quality assurance of machine elements during their design, operations, and maintenance.
... Additionally, the low computation cost of classifying a structure by a pre-trained model is an attractive addition, compared to simpler iterative models (i.e. Antipattern Matching Factor [2]). This paper presents a method for quality assessment [3] of a worm gearbox using Convolutional Neural Network as a classification mechanism. ...
Taking gearbox as a reference structure, authors apply a method for grading the quality of mechanical structures using a convolutional neural network trained with antipatterns found in gearbox constructions. Antipatterns are used as a quality reference embodied in a neural network, which is used for classifying tested structures to match the antipatterns taught to it.
... We aim to reduce the occurrence of mechanical design errors through early detection of design mistakes by looking for similarities between the designed mechanical construction and known design errors represented as antipatterns. To enable detection of such errors we perform the classification of the tested element (when possible its decomposed parts) against the library of incorrect (antipattern) designs [7]. Such comparison can be performed using various numerical models allowing for quantifying the similarity of mechanical constructions (Fig. 1). ...
... Following outlined above argumentation and to enable classification of mechanical constructions based on their function and features, we have created KXML -an object oriented, symbolic language to represent the structure and feature values describing the construction and avoiding pitfalls of existing computer data formats [7]. Attempting to process as much information from the KXML symbolic representation (e.g. ...
... The additive (i.e. Antipattern matching factor [7]) and multiplicative models we've tested require onedimensional input, and so provide a very limited ability to process tree-like datasets to detect similarities. Such ability is not allowing for differentiation between tree's nodes and hence becomes useless if the input data's structure is not static (impractical for real life structures). ...
Authors present the comparison of properties observed in numerical models, that were used to classify features of mechanical constructions. We use these models to detect similarities between mechanical designs and compare them on how well they quantify the description of the mechanical elements, and how they deal with arising challenges: multi-dimensionality of feature values, multiple classes of relations, data incompleteness and variability of format. The key conclusion here, is the ability of modern numerical models (i.e. ConvNet, CapsNet) to process information describing mechanical constructions, while including meaningful structural data in the calculations. Application of these models can provide direct assistance in the mechanical design process.
... Additionally, the low computation cost of classifying a structure by a pre-trained model is an attrac- tive addition, compared to simpler iterative models (i.e. Antipattern Matching Factor [2]). This paper presents a method for quality assessment [3] of a worm gear- box using Convolutional Neural Network as a classification mechanism. ...
This paper presents a new concept of intelligent e-learning systems with intelligent two-way speech communication between an e-learning system and the user. The system uses intelligent methods for analysis, evaluation and assessment of user knowledge and skills as well as e-learning process control, supervision and optimization. Developed as a prototype for mobile technologies, the communication system by speech and a natural language between the intelligent e-learning system and external users consists of intelligent mechanisms for biometric user identification, speech recognition, word and sentence recognition, sentence meaning analysis, and user reaction assessment. Also discussed are selected problems of the new concept of intelligent e-learning systems using intelligent speech communication. The discussion focuses on recognition and evaluation of spoken natural language sentences with use of hybrid neural networks.
Konferencja KSNwIM, Gdańsk, 25-27 września 2024 r.
A summary of scientific achievements
ODWOŁANIA Człowiek prawdy nie tworzy, tylko ją odkrywa-Św. Jan Paweł II. Tylko to dzieło czegoś jest warte, z którego człowiek może się poprawić i mądrości nauczyć –Adam Mickiewicz. Wyobraźnia jest ważniejsza od wiedzy, ponieważ wiedza jest ograniczona. (Phantasie ist wichtiger als Wissen, denn Wissen ist begrenzt) – Albert Einstein. Innowacje, same w sobie, nie są celem działalności twórczej. Innowacje są tylko cennym środ-kiem w rozwoju i zwiększaniu efektywności procesów, kształtowaniu nowych i lepszych wła-ściwości wytworów materialnych i tworzeniu sprawniejszych metod działania, także w ulep-szaniu umiejętności i możliwości ludzi, a w przyszłości również, pomocnych ludziom, zroboty-zowanych systemów o zaawansowanej sztucznej inteligencji. O wartości innowacji, o tym na ile jest cenna dla nowych, a także odległych zastosowań, decy-dują głównie dwa czynniki. Pierwszy czynnik ma aspekt wywodzący się ze stopnia nieoczywistości nowych rozwiązań w stosunku do stanu wiedzy i jej zastosowań oraz z poziomu wykorzystania kreatywności i nie-znanych dotąd metod tworzenia. Istota drugiego czynnika wywodzi się ze znaczenia aplikacyjnego opracowanych innowacji, ich zasięgu, znaczenia oraz efektów, a także z możliwości kreowania kierunków dalszych analiz i zastosowań. W tworzeniu innowacji wykorzystuje się wiele zaawansowanych metod rozwiązywania pro-blemów i realizacji zadań, w których wykorzystuje się wiedzę jawną i niejawną oraz umiejęt-ności twórcy, do określania cech wytworu materialnego (lub metodyki działania), który do-piero powstanie, a jego właściwości będą skutkiem wielu złożonych procesów, pracy wielu wy-konawców i będą przez użytkowników oceniane w różnych warunkach eksploatacji i różnym stanie technicznym. Nie ma nic bardziej praktycznego od teorii (często niepotrzebnie dodaje się " dobrej " , a przecież inna nie istnieje), bo metoda prób i błędów jest kosztowną metodą poszukiwań innowacji. Tworzenie przenosi człowieka do odmiennego stanu aktywnej kreatywności, najlepiej, gdy to-warzyszy temu świadomość, iż osiąganie tego stanu nie jest stacją, do której zmierzamy, lecz sposobem ciągłego podróżowania.
