Low-Cost Wireless Structural Health Monitoring of Bridges · Komarizadehasl S, Lozano F, Lozano-Galant JA et al.

2. Triaxle Wireless Low-Cost Adaptable Reliable Accelerometer (LARA) with Post-Synchronization Capability

In this section, the main features and innovations of the Low-cost Adaptable Reliable Accelerometer (LARA) in terms of software and hardware are detailed.

LARA is an updated CHEAP version, a wireless triaxial accelerometer that can be controlled, monitored, and programmed wirelessly.

This new accelerometer has a frequency sampling of 333 Hz, an acceleration range of ±2.0 g, and effective bandwidth of 165 Hz. LARA is built into two parts: (1) A sensing part: this contains the aligned accelerometers and the multiplexor, which is shown in Figure 1a. It should be noted that, the same as CHEAP, LARA consists of five aligned low-cost MEMS accelerometers. It is due to the fact that when the results of a few accelerometers with unique inherent dynamic noises are averaged, the signal under study remains invariable. However, the dynamic noises of individual sensors are divided by the square root of the number of averaged sensors, and (2) an acquisition part: consisting of an Arduino and a Raspberry Pi. The sensing and acquisition parts of LARA are shown in Figure 1b. In addition to the CHEAP components, LARA includes:

A Raspberry Pi: Raspberry Pi is a small size Linux-based computer that can be connected to Arduino microcontrollers. This way, the operator can access the Arduino codes for modifying or upgrading purposes. To save and acquire provided data of the accelerometers, a python code was written to save the acquisition data on the Raspberry Pi memory card. For synchronizing different LARAs for future SHM applications, the Arduino’s acquired data were saved with an accurate timestamp. The timestamp is reported from the inner clock of the Raspberry Pi. For constantly synchronizing the inner clock of the Raspberry Pi with the accurate time of the Internet, the Raspberry Pi’s NTP protocol was activated. Figure 1c shows the connection between the Arduino and the Raspberry Pi. Arduino gets its needed power from a USB port of the Raspberry Pi. Raspberry Pi can be powered up by using an adaptor or a power bank. It is also noteworthy to mention that the time keeping accuracy when using a Raspberry Pi model 3b+ with a NTP server was already published in the literature [48]. In this work, an overall deviation of 0.01 s was measured after a 40 h test. This value corresponds to a time accuracy of 0.07 Parts Per Million (PPM).

A PCB board: a Printed Circuit Board (PCB) was used to place the sensors in the targeted location with all their axes aligned. The sensors used the shortest possible wire lengths for connections between the multiplexor the accelerometers. In Figure 1a, the adjustment of sensors on LARA by the PCB is shown. In fact, it can be seen in Figure 1a,c that the PCB board has aligned x, y, and z axes of the MPU9250 accelerometers.

An aluminum box: the sensors are placed inside a box to preserve the accelerometers from environmental conditions (such as humidity, dust, and environmental activities). A very rigid and stiff material was needed to reach the same input signal to all the accelerometers [55]. Aluminum material was chosen to hold the accelerometers because it is very stiff, but, at the same time, it is a very light and conductive material. The conductivity helps the sensor’s grounding. Besides, in Figure 1b, the boxing of LARA is illustrated. The dimensions of this element are 52 × 72 × 44 mm.

A USB 4G dongle: this device included a modem with a USB dongle with 4G Internet connection connected to the Raspberry Pi that was used for the following purposes: (1) providing the accurate time of the Internet for the data acquisition timestamp, (2) controlling remotely the data acquisition process, and (3) transferring wirelessly the acquired data from the memory card of the Raspberry Pi to another computer.

A plastic box for the acquisition equipment: a plastic box was used to preserve the Arduino, Raspberry Pi part, and the Internet dongle. The connection between two boxes is within four sets of wires. The red, black, green, and orange wires connect the 5-volt Voltage Common Controller (VCC), GrouND (GND), Serial Clock Line (SCL), and Serial Data Line (SDA) of the sensing part to the acquisition part.

Besides the extra hardware parts that LARA possesses compared with CHEAP, LARA uses a new code and a library code that makes it faster and triaxial. The software enhancement of LARA refers to:(1)Increasing sampling frequency: frequency sampling of LARA is increased by rewriting the old library code and using a faster communication clock along with the main code;(2)No coding error: increasing the frequency sampling of a system can result in error reporting, interruptions, data loss, or fluctuation in the frequency sampling speed. In an experiment, LARA worked for an entire week and saved data with no errors or interruptions or data loss;(3)Schedule data acquisition: a code has been prepared on python to schedule and end vibration acquisition. This scheduling has two benefits. Firstly, it makes wireless sensor synchronization possible with free software. Secondly, it can be used for OMA applications. It should be noticed that when a structure is heavily excited by ambient causes, the accelerometers can extract more valuable data. Therefore, an accelerometer with schedule capability can help acquire data only when the structure is under high traffic or extreme activity;(4)Internal sensor synchronization lag enhancement: the accelerometers inside each LARA are not 100% synchronized. In fact, the Arduino executes codes one line at a time. When the main code is executed, the Arduino opens the library code and uses the information to get the first sensor’s acceleration, and after the second one, and so on. This operation takes time. In the CHEAP, the lag between each sensor-print was about 2200 microseconds. By the hardware and software improvement, the corresponding lag of LARA is decreased to 210 microseconds, which is 10.47 less than the lag of CHEAP. In fact, a lower lag time contributes to a better sensitivity of LARA for acquiring higher frequencies;(5)Post-synchronization of several LARAs: to use LARA’s outputs in the OMA application, various accelerometers’ data must first be synchronized and have the same sampling frequency. Since LARA has access to the accurate time of the Internet, the acquired data can be stamped with the precise time of the Internet with microsecond resolution. This timestamp helps calculate the sampling frequency of each LARA precisely and, more importantly, the fluctuation of the vibration acquisition process. Even though the reported sampling frequency of each LARA is calculated by measuring the number of acquired data during the acquisition process, the fluctuation is calculated by measuring the needed time for saving 100 data. It is seen through laboratory experiments that when the input power is insufficient, the fluctuation is unsteady. This insufficient input power can be due to low speed of the used USB wire, the long length of the USB cable, the imperfection of the used power bank, or simply using a power source that cannot provide 2.5 A and 5 V [9].