【正文】 effect on remote diagnosis performance can better direct the system design. The rest of this paper is anized as follows: Section 2 discusses some existing remote diagnosis architectures and summarizes the needs. Section 3 discusses about the automated system used in the experiments and the experimental variables. Section 4 presents the results and discussion. Section 5 ends with conclusion and future work. 2． Literature review The ability of different levels of operator machine interface to assist operators in the local fault diagnosis of discrete automated systems was discussed in [8]. Experiments were performed wherein operators used three hierarchical levels of interfaces with increasing capabilities to diagnose three different failures in an automated manufacturing system. The purpose of the tests was to empirically evaluate the three types of operator interfaces and expose the drawbacks in some of the monly used user interfaces in terms of their inefficiency to facilitate human troubleshooting performance in fault diagnosis tasks. The time taken to diagnose the fault, the number of information screens viewed, and the number of diagnostic tests performed were identified as the measures of performance. The impact of confounding variables: type of interface, nature of failure, and the order of experiments were also considered. Experimental evaluation of the effectiveness of functionally abstracted information in fault diagnosis was done in [11] in order to design for visual information display for process control. Improving human problem solving performance is the objective of the process interface considered for fault diagnosis in nuclear power plants. The ability of the hierarchical abstraction to improve troubleshooting performance was tested by implementing three levels of interface with increasing capabilities. The impact of the plexity of the diagnosis problem on the performance was also considered. It was seen that certain binations of level of information and type of display exist that generate optimum performance. Remendations regarding the integration of information level with display type were made to improve the effectiveness of any given display. An empirical study to test the ability of ecological interfaces to help diagnose faults in petrochemical processes was performed in [12] with professional operators in realistic plant settings. Three types of display interfaces: one contemporarily used and two hierarchical levels of ecological interfaces (one traditional and another augmented with additional taskbased information) were used. It was seen that the ecological interface with additional taskbased information facilitates the operator to a greater extent to troubleshoot failures and the contemporary interface was least helpful. The time taken to plete the task, the number of control actions, and the diagnosis accuracy were used to determine the performance score. Lower task pletion time, lower number of control actions, and better diagnosis accuracy were seen as the desired characteristics of an effective interface. In [13] was proposed the experimental investigation of the patibility of information types with diagnostic strategy. The application was related to building decisionaiding systems for fault diagnosis in nuclear power plants. Experiments were performed using four different types of information aids that are representative of mon operator support systems for diagnosis tasks in nuclear power plants in order to determine what information type would be effective for a particular strategy and facilitate the operator during diagnosis. Conclusions were made regarding the suitability of information aids for operator strategy and that the effectiveness of information aids was dependent on the strategy employed. The effects of hierarchical display on human problem solving performance was studied in [14]. Faults were introduced in puter simulations of logic circuits which were diagnosed by subjects with different levels of technical petence. It was seen that with subjects less petent in diagnosis, the hierarchical display interface was more helpful where as petent troubleshooters found both types of interfaces similarly patible. Thus, they established that the ability of an interface to facilitate diagnosis was also dependent on the skill of the user. SEMATECH [2] laid down standards for ediagnostics for the semiconductor manufacturing industry. According to SEMATECH, ediagnostics capabilities can be described within four levels (0– 3), each level building on the previous with increased capability. The level numbers increase according to a blend of many factors: the sequence of support tasks performed, the ease of implementing the necessary infrastructure to execute the diagnostic tasks, decreasing human assistance, and increasing automation. While level0 begins with basic remote connectivity and remote collaboration, level1 builds on level0 and allows remote control and monitoring and storage of operational and exceptions data. Level2 supports automated failure alarms and processing of historical data while including the capabilities from level1. Level3 is the most advanced type of architectures with features such as automated decision support, self diagnostics predictive maintenance, etc. Representative of the capabilities of the different levels across the standards