Abstract: This paper discusses and analyzes the necessity, importance, and feasibility of establishing an autonomous motion control rule base for autonomous underwater vehicles (AUVs). It points out that the active mechanism is a crucial function of the active database, and that event-condition-action rules are used to realize its active function. A control rule database design scheme is proposed based on the event and action characteristics of the control system, solving the bottleneck problem of single control rules in autonomous control systems. Finally, the implementation technology of the active control rule base is discussed, and corresponding solutions are proposed. Keywords: Autonomous underwater vehicle; Autonomous control; ECA rules; Rule base; Active database 1 Overview In organizations operating in marine environments, the development and utilization of underwater robots and autonomous underwater vehicles (AUVs) is undoubtedly a promising direction. The uncertainty of the marine environment places higher demands on the intelligence and maneuverability of underwater robots and AUVs. For AUVs to achieve autonomous control under different operating conditions, a single control rule is clearly insufficient. We hope that the AUV's control system operates under optimal conditions to achieve the best efficiency. This presents certain challenges to the motion control of AUVs. Conventional solutions involve adaptive control of the entire system or on-site optimization of control parameters. However, these methods are highly targeted and relatively singular, costly, and inflexible. Therefore, it is necessary to establish a control rule library containing multiple control rules that can be switched and invoked on-site according to mission objectives and sea state changes. 2. Proactive Control Requirements of the AUV Control Rule Library Based on an understanding of the autonomous underwater vehicle (AUV) operation process and uncertainty analysis, the proactive control requirements of the AUV during power plant operation are as follows: 1) A real-time monitoring system collects operational data from the AUV power plant in real time to monitor the plant's operating status. This includes status monitoring, performance monitoring, functional monitoring, safety monitoring, and fault monitoring. Specific operational data is recorded in the database and compared with event parameters in the control rule library. Through monitoring and analysis of key data during operation, problems are fed back to the AUV's mission planning layer. 2) Timely handling of problems: The system proactively matches control parameters to the AUV controller based on mission requirements, then combines these parameters with control algorithms to form corresponding control rules. This satisfies the mission planning layer's scheduling and switching of control rules to the behavior layer, ensuring optimal operation of the power unit. 3) Adaptive and self-organizing functions: When a significant change in operating status occurs and no matching rule exists in the rule base, the control system can perform real-time optimization or self-learning to correct certain control rules. The rule base then reorganizes and edits the optimized or corrected control parameters, expanding the rule base's capacity. 4) User-added pre-tuned rules: When completing the control rule base, users can consult experts to pre-add pre-tuned control rules and parameters to enrich the rule base's capacity and reduce future system operating costs. Considering the importance of the control rule base in the structure of autonomous underwater vehicles (AUVs), although these proactive functions manifest in different forms, they can all be implemented using a unified mechanism provided by the proactive database. 3. AUV Motion Control Rule Base System The motion control system of an autonomous underwater vehicle (AUV) is event-driven. Utilizing proactive database technology and an event-driven rule base, various events and their corresponding actions are predefined according to application needs, allowing different events to trigger different actions. A proactive control rule base system functionally consists of a traditional database system, an event-driven knowledge base, and corresponding event monitors. The knowledge base is a collection of event-driven knowledge, called the "event base." Each piece of knowledge represents how to proactively execute the user-preset actions contained within it when the corresponding event occurs. The event monitor is a monitoring module that constantly monitors whether events in the knowledge base have occurred. Once an event is detected, it proactively triggers the system to execute the pre-set actions specified in the knowledge base. It is evident that different knowledge representations in a knowledge base will result in various proactive behaviors. Currently, event-driven "Event-Condition-Action" (ECA) rules are commonly used to represent this knowledge. 1) Design of ECA Rules The most common method for proactive databases is to combine rule systems with database technology to enhance the proactive functionality of the database. The meaning of an ECA rule is: when a specific event occurs, the system will check whether the specified conditions are met. If these conditions are met, the system will trigger the corresponding action to complete the pre-set work. E—When the event occurs, the corresponding rule reacts (it can be an atomic event or a combination of a series of atomic events); C—Condition checks the situation before and after the event occurs; A—Action describes the task to be executed through the rule if the relevant event occurs and the conditions are met. The working principle of the proactive control rule base is shown in Figure 1. 2) Rule Model ECA rules describe events, conditions, actions, and the coupling methods between them. The event part has been explained above (mainly atomic events or compound events), the condition is generally a set of database queries or a set of logical expressions, and the action part can be a set of database operations or a specific operation performed by the user. ECA rules can be coupled in two ways: E-C coupling and C-A coupling. The former describes when the rule's condition is evaluated, while the latter describes when the action is executed when the condition is true. These two coupling methods and their combinations bring great flexibility to ECA rules and enhance their expressive power. The event-driven "event-condition-action" has the following general form: RULE<rule name>[(<parameters>...)] WHEN(event expression) IF<condition 1> THEN<action 1>; ………… IF<condition n> THEN<action n>; (n 1) END RULE[<rule name>] [align=center] Figure 1. Working principle diagram of the active control rule base[/align] These active rules are monitored and controlled by an event monitor in the system, which actively monitors the fact base at all times. In this way, users can set various different event-driven rules to implement many active processing functions in a unified mechanism, thereby meeting various objective needs. 3) Implementation of the active database: The active functions of the system are implemented using an embedded programming language. In the traditional way, an interface is designed to embed the database operations into the application. The operation of the rule base system is to embed SQL language in the main program to operate the database. Since it is relatively easy to establish an interface between the system running program and the database, the key to implementing an active database system is to implement an effective event monitor. On the one hand, it should effectively detect the occurrence of various events, and on the other hand, the execution speed of the original application should not change too much due to the addition of the event monitor. This often requires a combination of hardware and software to solve the problem, especially when the events are interrupt-type events, there must be corresponding hardware support. The following are some implementation mechanisms to choose from. (1) In a single-processor system, the event monitor can be implemented by a high-priority independent process under the control of the operating system to ensure that it can run frequently and play the role of actively monitoring the occurrence of various events. When the event library is divided into blocks, a pointer should be used to indicate which event library to monitor. (2) In a multiprocessor system, it can be independently allocated to a processor to complete the task. At this time, the communication and synchronization problems between multiple processors need to be solved. (3) Another way to implement the event monitor is: in the database management system, when execution reaches any place where an event may occur or at a pre-set checkpoint, an interrupt is generated, forcing the machine to switch to the event monitor to work, so as to verify whether the event that occurred at that time has been set in the event database by the user. Then, the execution of the following rule or rule group is triggered, otherwise, it returns to continue execution. The above implementation mechanisms each have their advantages and disadvantages. Generally speaking, the latter hardware and software combination approach may be quite attractive for systems that are simpler to implement on a single processor. Allocating a dedicated processor to implement the event monitoring system is naturally more ideal for some large systems. 4 Specific Implementation Examples The database management system uses Microsoft's SQL Server, the main program of the control system is implemented in C++, and the database operations are implemented using ADO technology. ADO (ActiveX Data Objects) is a new object-oriented database access technology of Microsoft. This technology is based on COM components and has many advantages, such as cross-language, cross-platform, and cross-operating system, and is suitable for all non-Visual C++ OLE DB access. ADO is a database programming module that allows OLE DB to be accessed from multiple languages. It is easy to use, fast, has low memory consumption, and small disk footprint. ADO technology is described in many resources, so it will not be detailed here. Taking the depth-holding PID multi-rule shear control of an autonomous underwater vehicle (AUV) as an example, the implementation diagram is shown in Figure 2. When performing depth-holding maneuvers, the AUV will perform different attitudes depending on the mission and sea conditions. At high speeds, the elevator can usually be used easily; at low speeds, due to the insignificant rudder effect, the vertical channel thruster is usually used to pull the AUV down to make it submerge; sometimes, to complete a specific mission, the AUV needs to perform rapid depth-holding, constant depth-holding, or more complex depth-holding maneuvers, etc. When encountering these situations, the AUV will monitor the mission events and sea conditions at the decision-making level through an event monitor, query the control rule base, and retrieve the PID depth-holding control rule or control parameters that match the event. Then, it will encapsulate the corresponding control algorithm into a function and pass it to the controller execution layer for AUV depth-holding maneuver control. PID control has three attributes: Kp, Ki, and Kd. This allows a single control algorithm to select multiple control parameter groups, thus forming various control strategies to meet the switching transformation calls of AUV constant-depth motion control for multiple constant-depth rules. Once the constant-depth PID rule base is built, functions such as adding, modifying, deleting, and querying rules can be performed on the rule base parameters for different constant-depth rules. [align=center] Figure 2. Example of AUV constant-depth PID control[/align] 5 Conclusion Active functions are crucial for various process control systems. Utilizing object-oriented and AI technologies enhances the knowledge representation capabilities of rules, making them more powerful in expressing complex events and operations, while also improving the efficiency of the system's rule inference engine. This is because many control systems require their application software, data acquisition, and control feedback to have high real-time performance, security, reliability, fault tolerance, and interactivity. Although the manifestations of active functions vary, they can be implemented using a unified mechanism through the active mechanism provided by the active database. This paper fully considers the active needs of the system in the design of the autonomous underwater vehicle (AUV) system underwater working autonomous control system, and uses active database technology to realize the active function of the AUV control system. Each rule name in the rule base is a professional term, which avoids the occurrence of rule name conflicts. The retrieval and calling speed is fast and easy to operate, thus meeting the real-time needs of the controller for multiple control rules. It avoids the need to write a large number of different control rule algorithm subroutines for calling, and reduces the resource consumption of the entire system. It solves the problem of single control rules in the past, making the AUV control system more reasonable for on-site online control, and providing a good working environment for on-site operations. References: [1] Sa Shixuan, Wang Shan. Introduction to Database Systems (3rd Edition). Beijing: Higher Education Press, 2000.2 [2] Zhou Zhida, Wu Heling, Lin Guozhang, Gao Feng. Research on Active Rules in Active Object-Oriented Database Systems. Journal of Beijing Institute of Technology, December 1998, Vol.38, No.6