ALBACETE, ENERO DE 2021
7. EVALUACIÓN DE RIESGOS
The deployment of vibration sensors and supporting instrumentation in rotating machinery requires to guarantee its safe and long-term operation. Therefore, the harsh conditions inside an engine need to be considered. More importantly, the deployment of traditional wired sensors on rotating machinery is not the best solu- tion. The wear of the wiring after continuous machinery rotation may cause wires to break and cause a short circuit, damage to the machinery or personnel. Besides, the cabling is used to interconnect devices which lead to costly installation and maintenance, added weight, high failure rate of connectors, and so on. An early solution to solve the communication problem for on-shaft vibration measurement was to use slip rings [64]. A slip ring enables a stationary wire or set of conductors to transmit power or data signals to one that is rotating [65]. They are widely used in applications with wind turbines, brushed DC motors, assembly line machines, and so on [65-66]. However, slip rings are costly and the noise associated encouraged their replacement by miniaturized WSNs [67-68]. Hence, wireless sensor technology opens the possibility to mount sensors directly onto rotating components without needing these expensive and electrically noisy slip-rings.
The availability of WSNs bring a series of advantages over wired solutions including easier deployment of sensor nodes, this brings an alternative over the use of wired sensors which are expensive, complex and usually hard to install. Moreover, the use of wireless sensors brings the opportunity to place them in critical places, to lower the costs of operation in industrial environments, to deploy in large scale, etc. Addi- tionally, the capabilities of self-configuration and self-organization in WSNs assures efficiency in energy services and reliable management [25, 26]. The environment within a GTE presents harsh conditions. This type of atmosphere may be subject to noise from machinery rotation, metallic reflections and frictions, noise generated from nearby equipment, engine vibrations, variations in temperature, channel interference and obstacles among others [59]. Thus, these severe situations may cause wired system solutions to be unsuitable in some applications and require cabling isolation. Reliability in wired systems could be enhanced by adding redundant wires although it would result in added complexity, weight and cost. Additionally, it is not a suitable solution if the machinery or equipment connected by these wires need to be relocated. As a result, using wireless alternatives appears as a viable solution to be
used in atmospheres under these conditions [23].
The use of wireless sensors for machine health monitoring eliminates physical con- nectors and wiring that could be exposed to hazardous environments. Hence, they enable to monitor systems that could not be monitored previously due to high main- tenance cost or personnel risk exposure to repair these systems. Also, wireless sensors allow monitoring of systems where external connections are impossible or impractical [69]. However, this technique requires the collection of a large amount of vibration data from the machinery or accessory of interest to facilitate accurate fault diagnosis. This situation is not always practical in a wireless environment due to sensor node limitations such as limited memory size, lengthy transmission time caused by low transmission rate and data packet retransmissions derived from random packet loss when installed in noisy environments. Hence, energy consumption and wireless transmissions may be minimised through local digital signal processing and data compression.
In wireless MHM, Fault Diagnosis may occur at the base station after the vibration signal is acquired, conditioned, processed and transmitted from the sensor node. It is desirable that signal postprocessing occurs at a dedicated BS because there are typically no constraints on power consumption to carry out computationally intensive algorithms and methods to diagnose faults. There exist two main approaches for this purpose: signal-based and model-based. In the case of the model-based method, a fault can be detected from continuously comparing the difference between the actual machine response and the model. The reliability of this method is dependent on the conformance or match between the real machine and the model, boundary conditions and the accuracy of materials parameters [70], more details in [71-72].
In the signal-based approach, a vibration signal is used to diagnose faults without the need to model the machine dynamics. The overall level of vibration is a robust indicator to determine machinery condition and to decide if a check-up is required [73]. Later, to determine a fault type, the signal may be analysed in the frequency domain. Detailed tables about common faults linked to their spectral representation can be found in [73-74].
Although the fault detection methods are out of the scope of this research, they were introduced for the interest of the reader with references to literature that cover these
topics in more detail. The focus of this research work is on energy efficient encoding procedures for vibration signals acquired from sensors prior to wireless transmission to increase robustness to random packet loss and data compression to minimise the number of transmitted samples. Also, the focus is to recover the signal with high accuracy and speed.
Opportunities and limitations in wireless sensor networks
Wireless Sensor Networks (WSN) have been extensively considered one of the most important new technologies of the present century [33]. The current evolution in MEMS and wireless communication technology have allowed the deployment of small inexpensive smart sensors in a physical area, these sensors networked using wireless links and Internet have opened opportunities for a variety of military and civilian applications such as environmental monitoring, battlefield surveillance and process control in industries [34].
WSNs have received great interest from both industry and academia around the world. However, WSNs present unique characteristics and constraints that should be considered when building a model for a specific application, those include limited power source, small memory space and constrained processing power, those char- acteristics including higher unreliability on sensor nodes differentiate WSNs from traditional wireless communication networks such as cellular systems and ad hoc networks [35]. Extensive research activities have been conducted to attempt to solve design and application issues, resulting in considerable advances in the deployment and development of WSNs. It is predicted that in a near future WSNs will be globally used in diverse civilian and military applications, revolutionizing the way we interact with the physical world and our quality of life [36].
WSN Network Characteristics
The sensor nodes communicate the sensed data over a short distance through the wireless medium and cooperate to achieve a common task such as battlefield sur- veillance, environmental monitoring and industrial process monitoring and control in remote locations [95-97].
A WSN typically consists of a collection of low cost, low power and multifunctional spatially distributed sensor nodes (SNs) that monitor and collect data from the
environment or area of interest in which they are deployed [94]. The SNs are small in size but incorporate sensors, embedded microprocessors and radio transceivers. Consequently, WSNs include sensing, data processing and communication capabil- ities. The sensor nodes communicate the sensed data over a short distance through the wireless medium and cooperate to achieve a common task such as battlefield surveillance, environmental monitoring and industrial process monitoring and control in remote locations [95-97]. A WSN system also includes a manager node or gateway which provides wireless connectivity back to distributed nodes and the wired world such as the Internet, a personal computer to display the data on a graphic user interface (GUI) or an Engine Monitoring Unit (EMU) in the case of a Gas Turbine Engine. The selected wireless protocol depends on the application requirements [98].
WSN Applications
Wireless sensor networks are typically distributed over a region of interest where they perform sensing, processing and communication tasks. The sensors can be used to monitor physical or environmental conditions such as temperature, pressure, sound, light, and vibration among others. The advantage of not requiring cabling to communicate and report sensed data within the existing network and the low cost of available sensors have allowed the creation of a variety of applications such as environmental monitoring [96], military applications and outdoor surveillance [95], healthcare monitoring [99], habitat monitoring [100], home automation and indoor surveillance [101], industrial process control [97] and aircraft health monitoring [102] among others. The previous applications, as well as many others, benefit from wireless data acquisition capabilities and deployment in many environments such as battlefields, outer space, oceans and machinery in remote locations.