Peter Martinek

Quality management and workflow control in printed circuit board production at the TU Budapest

Production of printed circuit boards (PCBs) is a very complicated process. There are many checkpoints and phases during the whole production, where application of quality management functions is necessary and the occurrence of faults is rather high. Following the path of a faultless work piece can be easy, but it is more difficult if a defect occurs at a given checkpoint. The appropriate action should be determined: the piece can be discarded or repaired. Investigating how repairing influences the quality of the final product is also required. Modeling of the process helps in making these decisions and gives the solution for the given worker, who is not even aware of printed circuit production technology and its planning. This paper presents a multi-leveled process model of printed circuit production. The model is created with the Architecture of Integrated Systems (ARIS) tool.

Justified performance measurement of schema matching solutions

Due to the heterogeneity of softwares used by companies, Enterprise Application Integration faces many challenges, one of which is the matching of schemas. In order to be able to launch and maintain communication among these applications, their interfaces should be semantically mapped. Many promising schema matching solutions have been proposed throughout the last decade and we could witness some convincing performance on their side. However, our research effort has proven that their performance may vary on a large scale if they are not set up properly for the given scenario. One potential source of inaccuracy could be the false classification of entity pairs bearing semantic distances close to the threshold value. In order to lessen this distortion effect, we propose a new approach to evaluate the accuracy of schema matching solutions and we also propose refinement to our existing accuracy measure maximization technique. Our new approach would then enable to gain justified accuracy values, while retaining the basics of the classic accuracy evaluation.

Semantically-enriched Service-Oriented Business Applications

Process planning and control has a key role at building a proper manufacturing both in theoretical and practical manner. Quality and effectiveness of completing end-user requirements depend on process control and quality assurance which is mostly supported by computer aided systems. However complex software solutions supporting manufacturing, process control and quality assurance are often placed separated at different participants of the supply-manufacturing-distribution process. Software solutions and systems used by the participants can be different in many ways like the platform they are built on. Re-integration of the whole process (processes might be partitioned into supply, production, distribution, etc. parts corresponding due to the specialization of the participants) indicates the need for integration of these separated systems. Therefore an integration solution must be conceptualized and used for ensuring proper co-working via well defined communication between these systems.

Representative subschema extraction method for schemas in technological applications

Data handled in Statistical Process Control (SPC) are collected from several different sources e.g. inspection machines. Schema matching is a technique which enables to connect technology applications of the machines and the process control system to avoid data redundancy and to help application cooperation. Schema matching procedures use several components combined by means of weights. Earlier on, we have provided a methodology to optimize these schema matcher parameters to achieve faster processing speed enabling real-time applicability as well. Our optimization methodology, however, requires a learning schema set. There are several considerations pertaining to the ideal learning set, e.g. it should be minimal and representative. In this paper, we introduce a representative subschema elicitation method. Our results were validated on the PCB manufacturing process at our department.

A comparison of schema matching threshold function and ANFIS generated membership function

Schema matching has the task of identifying semantically related entities in schemas. In the classic approach, a semantic distance is established among schema entities of the input schemas, based on which values the entity pairs are classified as match or non-match by means of a threshold value. This approach is, however, cumbersome in some schema matching related problem, like the accuracy measure maximization and the cutting threshold problem. In our earlier works, we proposed the concept of the schema matching threshold function for such cases. We assumed that the schema matching threshold function is, however, highly related to the concept of fuzzy membership functions. This assumed relation has encouraged us to perform a comparison between the schema matching threshold function and the fuzzy membership function which comparison is the topic of this paper. We used ANFIS for obtaining membership functions which were mapped to adequate threshold functions in order to be able to compare them. The outcome of our comparative analysis was that these mapped function pairs significantly resemble to each other.

Towards a novel approach of structural schema matching

Modern enterprise system landscapes leverage heterogeneous applications. In order to launch and maintain communication, matching the schema of these diverse applications is of key importance. Since the manual matching of schemas is time-consuming and error prone, only the application of automated schema matchers constitute a viable solution to this problem. Several solutions are available to solve this problem, but an unwelcome common characteristic of them is that their accuracy may become mediocre under certain conditions; consequently an expensive follow-on human correctional step cannot be omitted. We would like to improve the performance of schema matchers so as to reduce costs. In this paper, we present our new schema matching solution, which is based on a unique structural matching approach which takes into account the similarity of neighboring nodes. Our schema matcher also makes use of our related terms set comparison approach, which we have separately tested and later integrated into a complex hybrid schema matcher.

Method for validating CT length measurement of cracks inside solder joints

Purpose – This paper aims to develop a method to measure the length of cracks inside solder joints, which enables the validation of computed tomography (CT) crack length measurements. Design/methodology/approach – Cracks were formed inside solder joints intentionally by aging solder joints of 0603 size resistors with thermal shock (TS) test (−40 to +140°C, 2,000 cycles), and CT images were captured about them with different rotational increment (1/4, 1/2 and 1°) of sample projection. The length of cracks was also measured with our method, which is based on capturing high-resolution radiography X-ray images about the cracks in two perpendicular projection planes. The radiography results were compared to the CT measurements. The percentage error for the different CT rotational increment settings was calculated, and the optimal CT settings have been determined. Findings – The results have proven that reducing the rotational increment increases the sharpness of the captured images and the accuracy of crack le...

Voting based fuzzy linguistic matching

Schema matching is an essential task in data integration scenarios. The proposed schema matchers can be classified into several categories, one of which is the category of linguistic matchers, which evaluate relatedness comparing the textual elements in the schemata. There are several linguistic matchers the performance of which may vary from scenario to scenario. In order to eliminate the incorrect similarity values, we have implemented a voting based approach. Another contribution presented in this paper is the proposed fuzzy linguistic matching, which involves the transformation of similarity values into membership functions.

Integrated enterprise resource planning systems

The last 20 years have seen a lot of changes in the life of organizations. The always strong competition to be on the edge increased due to the saturation of the consumer markets. The new way to improve everyday working is computer-based information systems, which have become essential for survival in the last couple of years. This paper presents the short history of ERP systems first. From computer support in separated areas, through understanding the importance of big information (data) storages to proving the unique advantages of integration we arrive to the most modern ERPs. The demands (new challenges) and capabilities are examined and compared. On the other hand the creation of some practical examples and solutions are also presented. The full life-cycle (planning, development, testing, installation and continuous expansion) of the printed circuit board support system at the Department of Electronic Technology of Budapest University of Technology and Economics (BUTE) is shortly described. Lastly a modern concept and a new model are presented based on the view of business and workflow processes

Optimising pin-in-paste technology using gradient boosted decision trees

Purpose This paper aims to present a robust prediction method for estimating the quality of electronic products assembled with pin-in-paste soldering technology. A specific board quality factor was also defined which describes the expected yield of the board assembly. Design/methodology/approach Experiments were performed to obtain the required input data for developing a prediction method based on decision tree learning techniques. A Type 4 lead-free solder paste (particle size 20–38 µm) was deposited by stencil printing with different printing speeds (from 20 mm/s to 70 mm/s) into the through-holes (0.8 mm, 1 mm, 1.1 mm, 1.4 mm) of an FR4 board. Hole-filling was investigated with X-ray analyses. Three test cases were evaluated. Findings The optimal parameters of the algorithm were determined as: subsample is 0.5, learning rate is 0.001, maximum tree depth is 6 and boosting iteration is 10,000. The mean absolute error, root mean square error and mean absolute percentage error resulted in 0.024, 0.03 and 3.5, respectively, on average for the prediction of the hole-filling value, based on the printing speed and hole-diameter after optimisation. Our method is able to predict the hole-filling in pin-in-paste technology for different through-hole diameters. Originality/value No research works are available in current literature regarding machine learning techniques for pin-in-paste technology. Therefore, we decided to develop a method using decision tree learning techniques for supporting the design of the stencil printing process for through-hole components and pin-in-paste technology. The first pass yield of the assembly can be enhanced, and the reflow soldering failures of pin-in-paste technology can be significantly reduced.