1. Introduction

Recently, Prognostic Health Management (PHM) has caught the attention many researchers in a wide range of applications. PHM is one of the most promising disciplines of technologies and methods with potential of solving reliability, availability and maintainability problems [31]. Based on observed current condition of the system, the prognostic predicts system degradation and recommends proper maintenance at the right time. 

The essential requirements of PHM in preventing failures, extending operational life and improving current and future design with the end result of being more reliable and cost effective are considered as major achievements that have been sought after by the researchers. Several papers have focused on machine prognostics in condition based maintenance [19, 32], prognostic modeling options for remaining useful life estimation by industry for engineering assets [37] and manufacturing equipment prognostics [15]. 

For example in a vehicle, the prognostic and diagnostic information can be used by the driver to take appropriate action based on the component that is about to fall from the warning signal detected by the sensors such as low tire pressure and high vibration on the engine. Therefore, this system can predict component failure before it occurs and to avoid unwanted breakdowns. In addition, the manufacturer or dealer can receive the information through a telematics function on the performance of a particular vehicle model. This is used to prolong the usage of the vehicle by less maintenance and also to reduce unnecessary maintenance costs. Furthermore, the information obtained could be analyzed and used as design improvements for research and development. 

This paper investigates the recent research and development in automotive prognostics through international journals, conference proceedings and patents. This research also proposes on future trends of automotive prognostics which includes advanced sensor development, battery management system, artificial intelligence and knowledge based systems approach, and condition based maintenance for more potential work to be done in order to improve the current technologies and methods to be applied to more advanced applications. 

2. Methodology and Approach for Prognostics in Automobiles

Two approaches which are model-based and data driven have been used in the areas of automotive prognostics. Recently, integration between these two approaches is getting attention for more comprehensive, accurate and efficient prognostic techniques in determining system’s health. In <Table 1>, comparison of model-based and data driven approaches are described [17, 30]. 

Both system models and physics of failure modeling techniques are part of model-based approach. In system modeling, differential equation, statistical estimation, parity relations, Kalman filters and particle filters can be found as tools and techniques which are used to represent mathematical functions of the system [16]. In the physics of failure, many steps are involved which include identifying potential failures modes and mechanisms, analyzing the effects and determining damage models. Finally, the estimation of the degradation and RUL can be determined by understanding the nature and impact of the failures. 

<Table 1> Comparison between Model-Based and Data Driven Approach.

In data-driven approach, data collected during monitoring is analyzed for anomalies in trend or pattern to be detected to estimate the time to failure of the system and to provide valuable decision making information. Learning the behavior of the system through monitored data which can help to determine correlations, establish patterns and evaluate data trends leading to failure is one of the advantages of this approach. Markov chains [17], stochastic processes, time series analysis, Mahalanobis distance and principal components analysis [21] are some of the techniques in this approach which suits for monitoring a large number of parameters [28]. 

Currently, the integration of both approaches which could overcome their individual inadequacy has caught the interest for a more dynamic prognostics. Detecting intermittent failures in data driven approach and estimation of RUL in a model- based approach would enhance the prognostics of the system to a higher level. For more complex automotive applications, model-based and data driven approaches are required to tackle the variable nature of faults and modeling uncertainty so that all faults can be accurately diagnosed and predicted. 

3. Research Publications and Registered Patents

3.1 Journals and Conference Proceedings

In <Figure 1>, some of the parts that have been highlighted throughout the publications in the international journals and conference proceedings which have discussed about prognostics in automotive system. Many research efforts have been focusing on the battery and engine. Moreover, other parts such as suspension system, anti lock braking system and power steering wheel system are also discussed in the publications. Both model-based and data driven approach have been applied in the research. Integration of both approaches also could be seen in the anti lock braking system and power steering wheel system. 

<Figure 1> Application of Prognostics in Automotive Systems.

A : Battery System

Prognostics on battery management can be divided into two groups which are electric power generation and storage (EPGS), and traction battery. Electric power generation and storage is typically for cranking the engine, buffering electrical power for the vehicle during operation and providing electrical energy when the engine is off. It is usually recharged from an alternator driven by the engine of the vehicle [28]. On the other hand, traction battery is for vehicle propulsion typically for electric vehicle (EV) and also hybrid electric vehicle (HEV) that needs to be recharged at a recharge station. 

<Table 2> Research Papers on Prognostic for Automotive Systems.

Nowadays, the automotive system has become more complex and increasingly dependent on computer control and sensors to detect faults and to predict the remaining useful life of automotive parts. Thus, EPGS is extremely important to be a reliable supply of electrical energy for the automotive application. Most of the research papers have discussed about flooded and valve-regulated lead-acid battery systems improvement both performance and service life. 

In EPGS, the detection and isolation of a specific alternator fault, including belt slipping, rectifier fault and voltage regulator fault is a critical important element. Moreover, a specific set of battery faults including internal resistance and capacity loss can be detected and isolated to further improve the battery performance and service life [31]. In <Table 2>, some of the research papers that have been published regarding improvement of performance and service life for EPGS using model-based approaches for application of condition based maintenance (CBM) [31, 35, 41].
 

For a traction battery, accurately predicting the remaining useful life in an operating condition is crucial in order to guarantee that batteries are capable of providing the required power. Therefore, PHM is important to be able to deliver safe and reliable operation when the EV is running or is recharged. Maintenance decisions can be made on a conditional basis and users can be given sufficient forewarning before a failure happens so the risks can be mitigated [40]. Most of the research focuses on Lithium-ion batteries which are listed in <Table 2> [12, 18, 32]. 

B : Engine System

For a modern vehicle engine, it is a requirement to have on-board health monitoring capabilities for engine management to be more efficient. It should be equipped with applications of micro electro-mechanical sensors, microprocessorbased control systems and wireless mobile telecommunication, On-Board Monitoring and Diagnosis (OBMD) provide more efficient performance, fuel saving, emission controls and safety. Diesel and gasoline engines have sparked interest in the research community for engine condition monitoring and combustion control. Significant cost saving by scheduling preventive maintenance and preventing major downtime caused by extensive failure can be avoided by effective engine fault diagnostics and prognostics. 

Some of the challenges faced by the researchers in this area are unexpected faults, intermittent faults, rapid degradation and system complexity. These challenges have been addressed by Zhang et al who introduced the concept called Connected Vehicle Diagnostics and Prognostics (CVDP) [43]. Adopting the latest technologies in sensors and communications, CVDP has the potential to effectively diagnose and predict complex system faults. In <Table 2>, the model-based and data driven approaches are commonly applied in this area [1, 2, 38]. 

C : Other Automotive Systems

Suspension system, anti lock braking system and electric power steering system are some of the systems that adopted the application of PHM. <Table 2> summarizes methodologies that have been applied to forecast system degradation that leads to condition-based maintenance and increase availability of the systems [14, 24, 25]. Integrated model-based and data driven approaches have been applied in the anti lock braking and power steering systems for fault diagnostics and prognostics. 

3.2 Patent Development

The development of various PHM methodologies and techniques in the automotive area has led to patent registration. Besides focusing on the methods and systems, the patents also cover the development of sensors, hardware and software that can be applied to the prognostics function. Most of the patents and published articles have been noted over the last five years. Some of the academic research has been applied to real world application; therefore it must be patented to protect the intellectual property. In this paper, U.S patents are used due to most of the patents can be redundantly registered in other countries. 

More than 62 US patents related to automotive PHM which include vehicle on-board monitoring and communication, environment scanning systems and accident avoidance systems have been patented by S. Breed [4-9] from Automotive Technologies International Inc. These inventions have involved sensors for monitoring the components and environments, processing modules to receive and process data, and wireless communication to deliver derived information from the module. Most of his inventions patented are conceptual designs for application in futuristic automobiles. 

<Table 3> Patents for On Board Monitoring and Diagnosis.

In patent publications, OBMD systems have contributed in large numbers. <Table 3> shows some of the patents registered to automotive manufacturers [13, 17, 22, 42]. Nowadays, OBMD is considered as a standard component of every modern vehicle to evaluate and monitor the health of individual subsystems. Other automotive systems that have been patented with PHM methods such as suspension systems [29] and disc brake systems [37]. For battery management systems, GM Global Technology has patented some inventions on battery state-of health monitoring system [40] and method for abnormality prediction of battery parasitic load [27]. 

4. Sensor Development for Prognostics

Advanced electronics are now being developed to monitor and control many important functions in the automobile. Sensors are important in PHM as forefront devices for monitoring parameters such as performance conditions, operational and environmental loads to acquire continuing accurate in situ information which give abnormality detection, fault isolation and fast failure prediction [10]. In <Figure 2>, the integrated PHM sensor system basically consists of internal and also external devices which interlink via wire or wireless data transmission system. Power sources are needed to provide power for the entire sensor system which will run the sensor modules, microprocessor (analog to digital converter), memory and receivers [11]. 

<Figure 2> PHM Integrated Sensor System.

In addition, the development of sensors with multiple sensing abilities, high rate and long range data transmission, fast data processing and high reliability give more advantages to the PHM application. 

Data from the sensor can be transmitted either by wire or wireless. Wireless monitoring has positive impact on PHM applications in which it can monitor the system from remote and inaccessible area of the automobile. 

The latest development in sensor technology has already considered operating under harsh environments such as strong mechanical stress, vibration, high temperature, humidity and high pollution level. Besides that, physical characteristics such lighter weight, smaller size, shape, packaging and mounting are also important. In table 4, wireless and magnetic sensors are presented in the publications and patents [11, 22, 34, 44]. 

<Table 4> Sensor Development for Automotive Prognostics.

5. Discussion on the Technological Trends

OBMD supports for CBM

Forecasting system degradation through prognostic has benefited in significant time saving and increased availability. Conditional Based Maintenance (CBM) recent progress has put OBMD systems as a major source of information to continually evaluate the health of the subsystems on board over time in order to identify any dormant degradation and to forecast the remaining useful life based on the monitored data. The information provided can be used for the further improvement current design or development of new models. The new generation automotive trend is toward CBM rather than preventive maintenance in which more complex embedded diagnostic and prognostic systems could be found. Developing intelligent remote with real-time diagnostic and prognostic technology for OBMD is the trend of research in the automotive area. 

EV and HEV Battery Management

In the past couple of years, traction battery has gained lots of interest due to the EV and HEV commercialization by the car manufacturers around the globe. Customer’s demands on the more cleaner and sustainable type of energy for automobiles have put pressure on the manufacturers to put more efforts on research and development on more reliable and efficient battery management system. This will boost up the number publications and also patents in this area. More robust and generic type of battery management system which can be used for different types of battery for the same vehicle platform will gain advantages for the marketability of EV and HEV in the future. Advanced SOH battery prognosis with a meter to indicate accurate RUL of battery is the goal that all the researchers are trying to achieve. 

Artificial Intelligent and Knowledge Based Approach

Application of artificial intelligence and knowledge based approach such as fuzzy logic, machine learning, and knowledge representation in OBMD systems for fault diagnostics and prognostics can be further expanded to accommodate the more complexity and more intelligent of the next generation automobiles. Expert system has been implemented in energy plant to monitor and control the health of equipment. For example CASSANDRA [23] and PROMISE [3], the online expert system for fault prognostics which generates real- time information upon the severity and the existence of faults, forecast faults in the future time and suggests on how to control the problems. This concept can be further expanded for automotive application in which both customers and automotive manufacturers would obtain numerous benefits from using intelligent systems and technologies. 

Advanced Sensors

Sensor development is becoming more rapid due to demanding complexity of automotive electronics in which most of the systems becoming smaller and lighter to incorporate into Micro Electronics Mechanical Systems (MEMS) or Nano Electronics Mechanical Systems (NEMS). Intelligent wireless and battery-free sensors are being considered for potential development of sensors. Multi-sensor systems with complex signal processing and sensor data fusion approaches are significant in the enhancement of the prognostic process. Intelligent wireless network, battery-free power or ultra low power consumption and miniaturization are some requirements for the latest sensor technology in automotive prognostics. Energy harvesting concept in which collecting possible energy sources from surroundings such as sunlight, vibration, thermal gradient, wind and magnetic coupling to be converted into electrical energy can be applied to battery free power sensors. 

More works need to be done by utilizing prognostic models in research by academic literature to real world applications in realistic operating environments to verify that all models are applicable and useful. Although some of the academic literatures illustrated several case studies, more real world problem applications need to be established to seek more information on the compatibility and limitations of the proposed systems. 

6. Conclusion

Integration of model-based and data driven approaches, application of knowledge based and artificial intelligence would be further enhanced in automotive prognostics. By dealing with the challenges in integrating more approaches in research could help in building more robust prognostic systems. Advanced prognostics in redefining customer’s expectation in the automotive performance such as less carbon emission, safety enhancement, fuel saving and eco-friendliness can be further considered in future research. In addition, battery management system with more accurate and reliable prediction of available power and energy is a challenging mission for EV and HEV performance to gain customer’s attention to be more competitive in the global automotive market. 

As the complication of automotive electronic system’s increases, the need for intelligent PHM becomes more important due to the complexity which involved electronic components and software applications to perform control functions. Maintenance on the basis of current condition practices is becoming more significant compared to scheduled or breakdown maintenance. The concept of being able to identify a problem before it occurs thus elimination of the risk of failure is part of the customer’s expectation nowadays. This has raised the barometer among the manufacturers to catch up with these trends to be more competitive and sustainable in the global automotive market. 

T1.JPG87.1KB