Reliable high-rate bridge monitoring using dense wireless sensor arrays

Wireless sensor networks have been proposed extensively over the past several years as a means of alleviating instrumentation costs associated with structural health monitoring of civil infrastructure. However, low data throughput, unacceptable packet yield rates, and limited system resources have generally plagued many deployments by limiting the number of sensors and their sampling rate. It has only been recently that sensor network development has advanced to the state that such deployments can be realized as a feasible alternative to traditional cable-based systems. A Wireless Sensor Solution (WSS) developed specifically for diagnostic bridge monitoring provides independent conditioning for both accelerometers and strain transducers in addition to high-rate wireless data transmission capable of supporting large-scale sensor arrays. This design addresses the requirements of two methods at the forefront of condition assessment; characterizing modal properties through vibration measurements and approaches utilizing strain transducers for bridge load ratings. This paper presents the wireless bridge monitoring system developed within the Laboratory for Intelligent Infrastructure and Transportation Technologies (LIITT) at Clarkson University. The system interfaces with low-cost MEMS accelerometers using custom signal conditioning for amplification and filtering tailored to the spectrum of typical bridge vibrations, specifically from ambient excitation. Additionally, a signal conditioning and high resolution ADC interface is provided for Wheatstone bridge resistive sensors with digital temperature compensation enabled through the use of external thermistors. Embedded and host software applications permit flexible, user-friendly in-network control of sampling parameters, such as channel selections, amplification, sampling rates, digital filter coefficients, decimation ratios, and measurement duration. Additionally, the host software provides access to digital control of sub-circuit power, network status, real-time acquisition, and data logging. Analysis of dense field deployments as well as laboratory testing is presented to validate the performance of the instrumentation hardware and the design of the transmission protocol. ____________ Clarkson University, 8 Clarkson Ave, Potsdam, New York 13699-5712, U.S.A. INTRODUCTION The development of an application-oriented bridge management system for widespread utilization by national and state transportation agencies is a paramount task that currently necessitates further progress not only in large-scale data processing algorithms and statistical methods for damage detection, but also in wireless measurement hardware and sensor technologies. Since the advent of lowpower radio frequency (RF) chip transceivers, the commercial market has become flooded with wireless sensor network platforms and university research has produced continued investment in additional growth, while most wireless platforms share similar if not identical hardware components [1]. However, while the number of unique wireless sensor platforms has continued to rapidly expand, there has been limited success in replicating previous cable-based assessment test programs in regards to the number of deployed sensors and data rates. The system described in this paper has achieved both the higher sampling rates and large network size desired for structural health monitoring while maintaining reliable, high-yield communication. WIRELESS SENSOR SOLUTION (WSS) FOR BRIDGE MONITORING Traditionally, condition assessment of bridge structures is accomplished through the use of either vibration measurements or strain sensing. The Wireless Sensor Solution (WSS) developed specifically for diagnostic bridge monitoring provides a hybrid system that interfaces with both accelerometers and strain sensors to facilitate vibration-based bridge evaluation as well as load rating and static analysis on a universal platform (Fig. 1). Figure 1. Clarkson Wireless Sensor Solution (WSS) Wireless data transmission and network communication was accomplished through integration of the Tmote Sky wireless sensor network platform with a signal conditioning circuit designed in-house. The Tmote Sky mote is one of several wireless sensor platforms featuring the CC2420 2.4GHz chip transceiver for IEEE802.15.4 compliant operation with a 250kbps maximum data rate. Embedded software applications were written under the TinyOS operating system framework; however the vast majority of the code was written using low-level access to microcontroller registers and only basic Tinyos interface code was utilized. Packet yield and data throughput limitations experienced through use of the available TinyOS library prompted the development of application specific, optimized embedded software. Accelerometer Signal Conditioning To facilitate high-resolution acquisition of distributed acceleration measurements for modal analysis, a custom signal conditioning circuit was developed to interface Micro-Electro-Mechanical Systems (MEMS) accelerometers with the mote transceiver board. Though the circuitry was designed specifically for the requirements of acquiring low-amplitude vibration signals from bridge structures, the external connection permits acquisition of any ground-referenced sensor output below three volts. This section of the signal conditioning circuitry provides analog low-pass filtering, digitally-controlled correction of analog offset, and digitally programmable gain for up to two analog signals. This facilitates concise acquisition of vertical and lateral measurements of acceleration across the bridge using integrated dual-axis accelerometers. Low-noise, stable power is sourced to the filter sections and external sensors using a 3V voltage reference. The analog filtering provided has a 100Hz frequency cut-off with a 5-th order Butterworth response and are provided to prevent signal aliasing. Low-noise, low-power programmable gain amplifiers (PGA) maximize the resolution of the conversion specific to the signal range. In-network wireless commands enable adjustment of the gain from zero (low-power shutdown) up to 4096V/V in increments of binary multiples. Independent digital offset nulling for each channel is provided with accuracy of 1mV by low-noise programmable voltage references. This adjusts the sensor input range such that it is balanced in both the positive and negative directions and permits the use of high gain amplification without driving the signal to saturation. A 1.25V reference is provided to bias the amplifier output to the mid-span of the 2.5V full-scale range of the Tmote Sky ADC, which is established with an external voltage reference. Hardware shutdown of the PGA and voltage reference supply to the sensor and filter sections conserves limited battery resources during periods of inactivity. Strain Conditioning and Acquisition The core of this sub-circuit is the ZMD31050 application specific integrated circuit (ASIC) designed specifically for the conditioning and acquisition of fullbridge sensor measurements. The signal conditioning chip features 13 stages of programmable gain up to 420V/V, digitally programmable analog offset nulling, and a 15-bit internal ADC with adjustable input ranges to maximize the resolution of the signal over the ADC range. In addition, signal inputs are provided for either temperature diode or thermistor-based temperature measurement. An internal microcontroller introduces a digital conditioning algorithm for linearization of the sensor signal as well as providing up to 2 nd -order polynomial temperature compensation using the external sensor. Temperature compensation is a critical correction for long duration strain measurements using strain transducers, as the temperature-induced expansion of bridge elements often differs with temperatureinduced transducer expansion, thereby resulting in a strain output in the absence of applied deck load. This signal conditioning ASIC communicates digitally with the mote’s microcontroller through the use of the 2-wire serial (I2C) interface. The ASIC can be programmed to output the digitally compensated sensor measurement as an analog voltage for conversion by the Tmote Sky ADC. While this permits higher sampling rates of the strain sensor when sampled in conjunction with the accelerometers, the 15-bit ADC is utilized for high-resolution measurements during field testing. A voltage regulator is used to provide a low-noise, regulated supply to the external transducer and signal conditioning chip as well as to enable digitally controlled hardware shutdown for power conservation during periods of inactivity. Sensors The WSS bridge monitoring system offers signal connections for simultaneous acquisition of up to two single-ended sensors, one differential sensor, three temperature sensors, and a humidity sensor. To date, the laboratory and field testing have focused on validating the single-ended and differential sensor signal conditioned inputs using accelerometers and full-bridge strain transducers. Future work will incorporate the temperature sensing and compensation as well as humidity measurement. Additionally, application of the system to complementary monitoring tasks using diverse sensors will be performed to demonstrate the flexibility of the system hardware and software. MEMS Accelerometers have afforded a new generation of large-scale dynamic testing and in-service monitoring of civil structures due to their low cost. The WSS system has currently been deployed using both ADXL203 accelerometers, developed by Analog Devices, and LIS2L02AL accelerometers, manufactured by ST Microelectronics. Both accelerometers are low-power, dual-axis sensors in pincompatible leadless chip carrier (LCC) packages. The accelerometers are mounted on custom printed circuit boards containing the real-pole section of the analog filter along with a buffer amplifier to reduce the signal impedan