Large Number Of Machines Research Articles

Dynamic management of periodicity between measurements in predictive maintenance

In practice, on a large number of machines in industrial plants, predictive maintenance relies on periodic measurements to diagnose the condition of the equipment, rather than continuous monitoring of vibrations. In those cases, choosing an appropriate period between measurements is the key to success. Setting a long period implies taking a serious risk of breakdown, while a very short time interval between measurements can unnecessarily increase the costs of the maintenance plan.This work shows a methodology to determine and manage this Time Interval Between Measurements (TIBeM) dynamically adapted to each machine and situation. Depending on the criticality of each machine and its reliability, more specifically, its present diagnosed functional condition and the history of failures and measurements, the most appropriate TIBeM is recalculated each time a new measurement and diagnostic is performed.The described method has been implemented and validated in a large process plant and has led to a considerable improvement in costs and the management of its predictive maintenance plan.

Dynamic load assessment of building structures

The questions of the theory and practice of testing structures with dynamic loads are presented. A concrete example of testing a structure on its physical model by dynamic load is clearly shown. Some information necessary for statistical processing of research results is given. Building structures are exposed to the intensive dynamic loads. The existing methods and regulatory requirements should consider the loss of strength subject to dynamic impacts. The dynamic load types will show what needs to be more focused on during the construction. Building structures should be designed to stay out of the resonance band during the use. Nowadays, the construction of new industrial complexes and modernization of factories is widely developing. Increased production, leads to increased loads, including load-bearing structures of buildings and structures, namely there are dynamic effects from a large number of machines and mechanisms. Dynamic loads are very varied in nature. They include impacts associated with natural phenomena, such as seismic shocks and wind gusts, as well as dynamic impacts of technological and accidental origin. The most important characteristic determine behavior of buildings and structures under the action of external dynamic load, and are frequency, and nature of vibration. All methods for calculating wind and seismic loads are based on determining these parameters. The purpose of this paper is to demonstrate the effects of dynamic loads on strength, durability, rigidity and crack resistance of building structures; to develop measures for mitigating oscillations; to verify the design performance of commercialized and exploited structures by the frequency and attenuation intensity of natural oscillations. The scientific novelty in this paper is that experiments were conducted to assess the dynamic stability and deficient dynamic stability using the “soil-to-building”.

A Study of Hadoop and Mapping Approach Techniques on Big Data Strategies

Research is an art of scientific examination. The advance learner’s vocabulary of current English lays down the meaning of research as “A careful exploration and enquiry especially through search for new facts in any branch of knowledge. Bradman and Morry define research as “A standardize efforts to increase new knowledge”. Research is, thus an original contribution to existing stock of knowledge making for its advancement. It is detection of truth with the help of study, observation, comparison, and experiments. The technologies that give support to the entire process of cost-effectively storing and processing data, and utilize internet technologies in a scattered way have arisen in the past few years. NoSQL and Cloud computing are the renowned ones that improve the potential offered by Big Data Technologies. Map Reduce is a software manufacture introduced by Google to act upon parallel processing on large datasets supercilious that large dataset storage is distributed over a large number of machines. Each machine computes data stored locally, which in turn contributes to distribute and parallel processing. This paper focuses on the Big data and Cloud services using impact of Map Reduce Algorithm and very advantageous for the researchers and corporate sectors who are using Map Reducing System technology.

A data-driven predictive maintenance framework for injection molding process

Injection molding is the most common process to produce a wide range of complex plastic parts for many different applications, and a large number of machines and devices used in the plastics industry are associated with this process. Maintenance instructions and procedures used in the majority of injection molding plants currently are based on reactive and/or preventive strategies such as replacing failed components and/or performing regularly scheduled maintenance. However, such strategies are not cost-efficient and only partially effective in preventing machine downtime or producing scraps. The emergence of Industry 4.0 related technologies, such as cyber-physical systems, Internet of Things (IoT), cloud and edge computing, new sensors, and vision-based systems, brings new opportunities for the plastics industry to enhance their production and enterprise systems. Developing data-driven, predictive maintenance systems is one such opportunity that can help injection molding companies significantly reduce their maintenance cost while increasing their product quality and production efficiency. Accordingly, in this work, we introduce a generalized framework for implementation of predictive maintenance in injection molding process by integrating a variety of different data sources available in this process and taking the advantage of both cloud and edge computing. To demonstrate this framework, a case study on monitoring of the cooling system in injection molding process is presented. The results show the effectiveness of this approach in detecting cooling issues by monitoring other process data that are not directly correlated to the mold temperature. The comparison of the predicted mold temperature with the respective sensor value demonstrates an average error of 3.29 %, which can gradually be improved by accumulating more training data in the cloud-based system.

A Genetic Algorithm for the Double Row Layout Problem

The double row layout problem (DRLP) is an NP-hard and has many applications in the industry. The problem concerns on arranging the position of machines on the two rows so that the material handling cost is minimized. Although several mathematical programming models and local heuristics have been previously proposed, there is still a requirement to develop an approach that can solve the problem efficiently. Here, a genetic algorithm is proposed, which is aimed to solve the DRLP in a reasonable and applicable time. The performances of the proposed method, both its obtained objective values and computational time, are evaluated by comparing it with the existing mathematical programming model. The results demonstrate that the proposed GA can find relatively high-quality solutions in much shorter time than the mathematical programming model, especially in the problem with large number of machines.

A Hybrid Simulation Study to Determine an Optimal Maintenance Strategy

With the increasing complexity of the process industry, having excellent maintenance management is essential for manufacturing industries. Various parts that interact and interdependent with each other make a well-planned maintenance strategy is one of the major challenges facing by industry. The whole system could be interrupted just simply because of the failure of a component. Therefore, a review of a maintenance strategy must be done from a system perspective. It is suggested that the optimal preventive maintenance time interval is not only determined by the lowest maintenance cost of each machine but also its impact on the whole system. Two main indicators that can accommodate the system perspective are reliability and revenue. A large number of machines and the array of machines can be synthesized in the reliability indicator. Moreover, the creation of maximum revenue is always the main goal for a business. The best maintenance strategy will be determined from the revenue obtained by a process industry. The process industry discussed in this study is a flour mill which is very well known in Surabaya. This study applied a hybrid simulation to solve this problem. Monte Carlo simulation was used to observe the machine individually and the results are reviewed using the application of System Dynamics. Three improvement scenarios were proposed in this simulation study. Scenario 2 was chosen as the best scenario because it was able to generate the highest revenue at the end of the period. Scenario 2 recommends setting the preventive maintenance time interval considering resource availability.

Environment for Threat Intelligence Analysis and Generation using Open Sources

Analyzing attacks on computer networks is complex given the volume of data considered and the large number of machines, even in small networks. The volume of data is large and the time to process and analyze it is short. The goal is to extract and analyze information about network attacks that has been obtained from open sources. Using a robust, elastic and scalable architecture that makes use of processing techniques with the use of Hadoop so that the information is available in a timely manner. With the proposed architecture implemented all the desired characteristics were obtained allowing the processing of the data in near real time. The system provides intelligence information about large-scale attacks with agility and efficiency.

Makespan Map Reduce Architecture for Efficient Memory Utilization

Makespan is referred to the total execution time taken to process the tasks or the jobs to the completion time.The High performance infrastructure in cloud computing provides extensive applications. These applications are preferred in Big Data. The existing Hadoop Map Reduce network incurs the input output memory overhead. The parallel Map Reduce network provides a parallel scheme to reduce makespan times in computing environments. The outcome provides improvement in the coefficient correlation and makespan time. The various challenges in computing dataset is handling large dataset efficiently and providing large amount of datasets with ease. The comprehensive method is to enhance data analyzation techniques. Massive large amount of data which are spread across the large number of machines needs to be parallelized.

Two-dimensional indexing to provide one-integrated-memory view of distributed memory for a massively-parallel search engine

We propose two-dimensional indexing—a novel in-memory indexing architecture that operates over distributed memory of a massively-parallel search engine. The goal of two-dimensional indexing is to provide a one-integrated-memory view as in a single node system using one large integrated memory. In two-dimensional indexing, we partition the entire index into n× m fragments and distribute them over the memories of multiple nodes in such a way that each fragment is entirely stored in main memory of one node. The proposed architecture is not only scalable as it uses a scaled-out shared-nothing architecture but also is capable of achieving low query response time as it processes queries in main memory. We also propose the concept of the one-memory point, which is the amount of the memory space required to completely store the entire index in main memory providing a one-integrated-memory view. We first prove the effectiveness of two-dimensional indexing with single-keyword queries, and then, extend the notion so as to be able to handle multiple-keyword queries. To handle multiple-keyword queries, we adopt pre-join that materializes a multiple-keyword query a priori as well as a new notion of semi-memory join that obviates extensive communication overhead to perform join across multiple nodes. In experiments using the real-life search query set over a database consisting of 100 million Web documents crawled, we show that two-dimensional indexing can effectively provide a one-integrated-memory view without too much of additional memory compared with the single node system using one large integrated memory. We also show that, with a six-node prototype, in an ideal case, it significantly improves the query processing performance over a disk-based search engine with an equivalent amount of in-memory buffer but without two-dimensional indexing — by up to 535.54 times. This improvement is expected to get larger as the system is scaled-out with a larger number of machines.

Scanner

A growing number of visual computing applications depend on the analysis of large video collections. The challenge is that scaling applications to operate on these datasets requires efficient systems for pixel data access and parallel processing across large numbers of machines. Few programmers have the capability to operate efficiently at these scales, limiting the field's ability to explore new applications that leverage big video data. In response, we have created Scanner, a system for productive and efficient video analysis at scale. Scanner organizes video collections as tables in a data store optimized for sampling frames from compressed video, and executes pixel processing computations, expressed as dataflow graphs, on these frames. Scanner schedules video analysis applications expressed using these abstractions onto heterogeneous throughput computing hardware, such as multi-core CPUs, GPUs, and media processing ASICs, for high-throughput pixel processing. We demonstrate the productivity of Scanner by authoring a variety of video processing applications including the synthesis of stereo VR video streams from multi-camera rigs, markerless 3D human pose reconstruction from video, and data-mining big video datasets such as hundreds of feature-length films or over 70,000 hours of TV news. These applications achieve near-expert performance on a single machine and scale efficiently to hundreds of machines, enabling formerly long-running big video data analysis tasks to be carried out in minutes to hours.

A Comparative Study of Join Algorithms in Mapreduce

To analyze large volumes of data, a set of techniques is present in the IT community as the MapReduce paradigm, parallel RDBMS, column storage, and combinations of those technics. MapReduce is a parallel programming model introduced by Google, which enables an easy parallelization of tasks while hiding the details and the complexity of parallel computations on very large datasets across a large number of machines. Our study will be concerned about the MapReduce Data Analytics, the most important data analysis treatments in MapReduce is the treatment of the Logs files as in the case of web applications, which can be in the form of selection operation, aggregation, or filtering, the most useful and expensive operation is that of the join, the logs files will often need to be joined with one or more table references, however the MapReduce paradigm is not designed to process multiple inputs. While processing relational data is a common need, this limitation causes difficulties and inefficiency when MapReduce is applied on relational operations like joins. The aim of this paper is to compare a number of wellknown join strategies in MapReduce, analyze the costs associated to a MapReduce program in terms of I/O and CPU used, and present some techniques of optimizations from related works.

A survey on high-precision time synchronization techniques for optical datacenter networks and a zero-overhead microsecond-accuracy solution

A datacenter, which is a highly distributed multiprocessing system, needs to keep accurate track of time across a large number of machines. Precise time synchronization has become a critical component due to stringent requirements of several time critical applications such as real-time big data analytics, high-performance computing, and financial trading. Our study starts with a survey on the most relevant time synchronization techniques for datacenter networks. Then, we propose a zero-overhead microsecond-accuracy solution to synchronize a packet-switched optical network for datacenters. To achieve the desired time accuracy, we consider precision time protocol to synchronize the server clocks with a central controller clock. Zero-overhead is maintained by using data traffic to carry the time messages instead of a separate control channel. Through simulation, we show that microsecond level of time accuracy can be achieved. We also discuss the dependency of the accuracy on different traffic loads, traffic distributions, and packet lengths.

A Machine-to-Machine protocol benchmark for eHealth applications – Use case: Respiratory rehabilitation

BackgroundM2M (Machine-to-Machine) communications represent one of the main pillars of the new paradigm of the Internet of Things (IoT), and is making possible new opportunities for the eHealth business. Nevertheless, the large number of M2M protocols currently available hinders the election of a suitable solution that satisfies the requirements that can demand eHealth applications. ObjectivesIn the first place, to develop a tool that provides a benchmarking analysis in order to objectively select among the most relevant M2M protocols for eHealth solutions. In the second place, to validate the tool with a particular use case: the respiratory rehabilitation. MethodsA software tool, called Distributed Computing Framework (DFC), has been designed and developed to execute the benchmarking tests and facilitate the deployment in environments with a large number of machines, with independence of the protocol and performance metrics selected. ResultsDDS, MQTT, CoAP, JMS, AMQP and XMPP protocols were evaluated considering different specific performance metrics, including CPU usage, memory usage, bandwidth consumption, latency and jitter. The results obtained allowed to validate a case of use: respiratory rehabilitation of chronic obstructive pulmonary disease (COPD) patients in two scenarios with different types of requirement: Home-Based and Ambulatory. ConclusionsThe results of the benchmark comparison can guide eHealth developers in the choice of M2M technologies. In this regard, the framework presented is a simple and powerful tool for the deployment of benchmark tests under specific environments and conditions.

Bayes and big data: the consensus Monte Carlo algorithm

A useful definition of ‘big data’ is data that is too big to process comfortably on a single machine, either because of processor, memory, or disk bottlenecks. Graphics processing units can alleviate the processor bottleneck, but memory or disk bottlenecks can only be eliminated by splitting data across multiple machines. Communication between large numbers of machines is expensive (regardless of the amount of data being communicated), so there is a need for algorithms that perform distributed approximate Bayesian analyses with minimal communication. Consensus Monte Carlo operates by running a separate Monte Carlo algorithm on each machine, and then averaging individual Monte Carlo draws across machines. Depending on the model, the resulting draws can be nearly indistinguishable from the draws that would have been obtained by running a single-machine algorithm for a very long time. Examples of consensus Monte Carlo are shown for simple models where single-machine solutions are available, for large single-layer hierarchical models, and for Bayesian additive regression trees (BART).

PRISM: Fine-Grained Resource-Aware Scheduling for MapReduce

MapReduce has become a popular model for data-intensive computation in recent years. By breaking down each job into small map and reduce tasks and executing them in parallel across a large number of machines, MapReduce can significantly reduce the running time of data-intensive jobs. However, despite recent efforts toward designing resource-efficient MapReduce schedulers, existing solutions that focus on scheduling at the task-level still offer sub-optimal job performance. This is because tasks can have highly varying resource requirements during their lifetime, which makes it difficult for task-level schedulers to effectively utilize available resources to reduce job execution time. To address this limitation, we introduce PRISM, a fine-grained resource-aware MapReduce scheduler that divides tasks into phases, where each phase has a constant resource usage profile, and performs scheduling at the phase level. We first demonstrate the importance of phase-level scheduling by showing the resource usage variability within the lifetime of a task using a wide-range of MapReduce jobs. We then present a phase-level scheduling algorithm that improves execution parallelism and resource utilization without introducing stragglers. In a 10-node Hadoop cluster running standard benchmarks, PRISM offers high resource utilization and provides 1.3× improvement in job running time compared to the current Hadoop schedulers.

A scalable file based data store for forensic analysis

In the field of remote forensics, the GRR Response Rig has been used to access and store data from thousands of enterprise machines. Handling large numbers of machines requires efficient and scalable storage mechanisms that allow concurrent data operations and efficient data access, independent of the size of the stored data and the number of machines in the network. We studied the available GRR storage mechanisms and found them lacking in both speed and scalability. In this paper, we propose a new distributed data store that partitions data into database files that can be accessed independently so that distributed forensic analysis can be done in a scalable fashion. We also show how to use the NSRL software reference database in our scalable data store to avoid wasting resources when collecting harmless files from enterprise machines.

A queuing approach for a tri-objective manufacturing problem with defects: a tuned Pareto-based genetic algorithm

In this research, a manufacturing facility with independent workstations to remanufacture nonconforming products is investigated. Each workstation is first modeled as an M/M/m queuing system with m being a decision variable. Then, a tri-objective integer nonlinear programming models is developed to formulate the problem. The first objective tries to minimize the waiting times of products, while the second one tries to maximize the minimum reliability of machines at the workstations. Since minimization of the waiting times results in using a large number of machines with higher idle times, the third objective is considered to minimize the mean idle time of the machines. The aim is to determine optimal number of machines at each workstation. Since the problem belongs to the class of NP-hard problems, the non-dominated sorting genetic algorithm-II (NSGA-II) is utilized to find Pareto fronts. Because there is no benchmark available in the literature to validate the results obtained, the non-dominated ranked genetic algorithm (NRGA) is used as well. In both algorithms, not only the best operators are selected but also all of their important parameters are calibrated using statistical analysis. The performances of the algorithms are statistically compared using the t test. Besides, the multiple attribute decision-making method of TOPSIS is used to determine the better algorithm. The applicability of the proposed model and the solution algorithms is demonstrated via some illustrative examples.

Task Scheduling to Extend Platform Useful Life using Prognostics

In this paper, we aim at maximizing the useful life of a heterogeneous distributed platform which has to deliver a given production. The machines perform independent tasks and may be configured with different profiles (one nominal mode and several degraded ones). Depending on the profile, a machine reaches a given throughput. At each time the sum of the machine throughputs that are currently running determines the global throughput. Moreover, each machine is supposed to be monitored and a prognostic module gives its remaining useful life depending on both its past and future usage (profile). The objective is to configure the platform so as to reach the demand as long as possible. We propose to discretize the time into periods and to choose a configuration for each period. We propose an Integer Linear Programming (ILP) model to find such configurations for a fixed time horizon. Due to the number of variables and constraints in the ILP, the largest horizon can be computed for small instances of the problem. For larger ones, we propose polynomial time heuristics to maximize the useful life. Exhaustive simulations show that the heuristic solutions are close to the optimal (5% in average) in the case where the optimal horizon can be computed. For other platforms with a very large number of machines, simulations assess the efficiency of our heuristics. The distance to the theoretical maximal value is about 8% in average.

Analyzing Massive Machine Maintenance Data in a Computing Cloud

We present a novel framework, CloudView, for storage, processing and analysis of massive machine maintenance data, collected from a large number of sensors embedded in industrial machines, in a cloud computing environment. This paper describes the architecture, design, and implementation of CloudView, and how the proposed framework leverages the parallel computing capability of a computing cloud based on a large-scale distributed batch processing infrastructure that is built of commodity hardware. A case-based reasoning (CBR) approach is adopted for machine fault prediction, where the past cases of failure from a large number of machines are collected in a cloud. A case-base of past cases of failure is created using the global information obtained from a large number of machines. CloudView facilitates organization of sensor data and creation of case-base with global information. Case-base creation jobs are formulated using the MapReduce parallel data processing model. CloudView captures the failure cases across a large number of machines and shares the failure information with a number of local nodes in the form of case-base updates that occur in a time scale of every few hours. At local nodes, the real-time sensor data from a group of machines in the same facility/plant is continuously matched to the cases from the case-base for predicting the incipient faults-this local processing takes a much shorter time of a few seconds. The case-base is updated regularly (in the time scale of a few hours) on the cloud to include new cases of failure, and these case-base updates are pushed from CloudView to the local nodes. Experimental measurements show that fault predictions can be done in real-time (on a timescale of seconds) at the local nodes and massive machine data analysis for case-base creation and updating can be done on a timescale of minutes in the cloud. Our approach, in addition to being the first reported use of the cloud architecture for maintenance data storage, processing and analysis, also evaluates several possible cloud-based architectures that leverage the advantages of the parallel computing capabilities of the cloud to make local decisions with global information efficiently, while avoiding potential data bottlenecks that can occur in getting the maintenance data in and out of the cloud.

A Hierarchy-Based Distributed Algorithm for Layout Geometry Operations

This paper introduces a novel distributed algorithm for performing the layout geometry operations usually found in design rule checking, layout verification, and mask synthesis. A large number of machines are typically available to the user during the mask synthesis flow. As multiple machines or cores become more ubiquitous, even designers using layout verification tools will have access to a large set of machines. Therefore, an efficient and scalable distributed algorithm for performing sequences of layout geometry operations will be of great value to both designers and mask synthesis engineers. Given a layout and a sequence of layout geometry operations, the proposed algorithm divides the layout into several partitions. The given sequence of layout geometry operations is executed in parallel on different partitions. New partitions are derived from the original set of partitions and the sequence of geometry operations is repeated on larger partitions with much fewer polygons. This process continues until it produces a partition that covers the entire layout area. A key feature of the proposed algorithm is that it is correct-by-construction, i.e., each partition is guaranteed to generate a subset of the correct results. Complete and correct results are generated for each layout geometry operation for the entire layout when the operation completes execution on all the partitions. The proposed algorithm was implemented in Gearman, an open-source distributed framework. Results on large industrial layouts show good performance and scalability.