In its application to a water hydraulic system, the use of neuromechanics allows mechanically inferior components and systems to be applied in an efficient and competitive way. For example, instead of using a hydraulic valve with the exact required dynamic characteristics that is expensive to produce, a cheap and inferior valve can be used with its dynamic characteristics altered using electronic control so that it effectively operates in the same way. This can be achieved through the implementation of an adaptive control system such as a fuzzy logic based controller [5], or a neural network based control strategy [4].
By adding a degree of intelligence to the system, using of a self-learning algorithm for example, the controller can compensate and adapt more effectively. A computer simulation of the hydraulic system can be written and used as a learning tool for the intelligent control system to prevent damage during the initial learning process. Intelligent control systems can be taught to recognise the effects of leakage and to take effective counter measures. The trained system can then be used to control the hydraulic system and further learning can be implemented on-line so that the control system can be fine-tuned.
Flow effects and leakage can be modelled and used to train the control system, while measures to compensate for bio-mass build up can be learnt on-line. The control system can be trained so that optimum and efficient control action is taken. In theory the non-complex water hydraulic system can be operated as efficiently as its complex precision engineered counterpart, the strain on the system being supported by the control algorithms.
A neuromechanical system can be assessed in terms of its cost, efficiency, and complexity. There are limits to the design simplicity of the hydraulic hardware in order to obtain suitable component operational lifetime, and there are also limits to the cost and complexity of the control system. A balance must therefore be struck between mechanical simplicity and control complexity to produce an efficient and economic system.
4. SYSTEM COMPONENTS AND CONFIGURATION
The principal design objective of this research programme was to design a water hydraulic actuation system that is less complex in its manufacture than conventional designs and has minimal maintenance requirements. The selection of modern engineering materials and the application of advanced control techniques can overcome the resultant problems of leakage and corrosion associated with a relatively simple design. The net result being an economically viable system with similar operating characteristics to those of an oil hydraulic system but with the advantages of water as an operating fluid.
The developed sub-sea actuation system hardware consists of a novel piston type actuator, shown in Figure 1 and four poppet type flow control valves, of the type shown in Figure 2. System pressure and actuator rotary positional feedback is achieved using dual pressure transducers and a rotary potentiometer respectively. The hardware design and specification is detailed in Dobson and Roskilly [6].
The configuration of system hardware governs the long-term success of the water hydraulic system. There are two main decisions to be made when considering the controllability of the hydraulic system. Firstly, the choice of hardware configuration, i.e. open or closed hydraulic system, and secondly, the type of control architecture.
Most current water hydraulic systems are based upon a closed system similar to its oil hydraulic counterpart. The hydraulic fluid constantly circulates around a closed hydraulic circuit. For certain applications, it would be beneficial to operate as an open hydraulic circuit that ingests filtered fluid from its ambient environment then exhausts the fluid back into its surroundings. This approach will remove the need for hydraulic reservoirs and tight component tolerances. Applications in the marine and nuclear industries lend themselves to such a system as both have a need for submerged hydraulic actuation systems that can utilise their ambient fluid.
Figure 3 shows the overall system configuration, based upon an 'H' valve formation open system. Valves A and D control the outflow of the fluid from the system and valves B and C control the supply of the fluid to the system. A computer using logic commands controls the valves; a logic high (1) closes the valve and a logic low (0) opens the valve. The four valves have sixteen distinct control patterns based upon the sixteen possible logic strings. For example, a logic string of 0101 (the valve configuration shown in Figure 3) will produce a positive movement of the actuator (clockwise pinion rotation).
Table 1 shows the sixteen possible logic control commands. The main commands highlighted in black are stationary, positive, and negative. The other minor commands can be used to implement corrective action to control the actuator in situations such as when valve internal leakage becomes a concern. For example, if the integrity of valve C starts to break down and leakage occurs, the pressure can be eased off the system by implementing valve control strategy 1011 during the negative stroke. Control resolution of the hydraulic system is further improved with the use of a pulse width modulation (PWM).
5. CONTROL SYSTEM
The type of control system chosen and implemented is important for the long-term success of the water hydraulic system. A self-organizing fuzzy logic control system (SOFLC) was chosen for its rapid learning capabilities.
Adaptive control algorithms can deal with unknown or drifting system dynamics and coupled with the ability of an intelligent control design, identifying a systems changing state and instigate corrective action, provides a powerful control architecture. Conventional algorithms such as PI and PID are the most common in use in industry and the performance of such systems is adequate for most modern control applications.
