Servos will warn but keep on going

May 2002 Feature story: Servos will warn but keep on going Within a year, servo drives will become available that can advise users of faults via mobile phones. A year after that, the same drives will detect developing problems before they become serious and advise of the need for maintenance. And shortly after that, they will be able to re-tune themselves for optimum performance and make every effort to keep going, while informing their users of exactly how much their performance might have degraded. Such capabilities are just becoming available for process valve actuators, as described in Eureka's February edition. However, as Dr David Brown of Portsmouth University explains: "Servo drives are very complex beasts, and giving them a similar capability is an order of magnitude more difficult." Application areas range from aerospace and defence to medical and the food industry Dr Brown, who has spent most of his career in industry, is currently running a TCS (formerly Teaching Company Scheme) project to achieve all these objectives, with support from Infranor, with the goal of turning his developments into commercial products in the very near future. The value of the grant and industrial contribution is put at £86,228. The citation states, "Artificial Intelligent algorithms with direct Internet access to system information will enable operators to work on many remote installations in parallel, providing significantly increased levels of customer support at reduced cost. Analysis of multiple installed systems will give increased levels of understanding of the performance of the company's equipment which will lead to improvements in the new product development process and an increased capability for new products." In a demonstration conducted for Eureka by research assistant James Hui, an Infranor servo drive was connected to a laptop via its RS232 port. Disconnecting the resolver immediately stopped the drive, sent a text message with the motor number and an error code to a mobile phone, and an e-mail message with more details. The e-mail message said: "Resolver error. Suggest: Check the resolver cable. Check the motor power cable. Restart the servo system." Mr Hui then allowed Eureka to use the phone to command the drive to make the motor stop, start and reverse, and go through some of the operations that a test engineer might wish to subject a drive to before giving it an 'OK'. The phone was a standard WAP mobile and control was by scroll down menu options and click procedure. Dr Brown then pointed out that, while this stage of the development is almost ready for commercialisation, the research project is aimed at making accurate assessments of condition and predicting future performance. If excessive vibration is predicted, for example, this too will be relayed to the mobile phone and by e-mail, with an assessment of likely consequences and advice on what needs doing. Compare the changes The approach is one which should be familiar to process engineers - to set up a Matlab computer- based model and compare the performance of the real motor and drive with the model. Changes in the real system will be reflected in changes in the constantly updated computer model, which will in turn be compared with a reference computer model. A neural network will be used to compare measured performance, particularly motor current, with that expected, and deduce inferences. In the case of sensors, it is expected to be able to catch malfunctioning sensors before they drop out. When they do fail, the system will be capable of looking for another system nearby which can take over, perhaps with some degradation of performance. If there is no alternative and no safe way of keeping going, it will issue a warning and initiate a safe shut down. "We want to get away from both routine preventive maintenance, which tends to be expensive, and also waiting for a breakdown, which tends to be even more expensive," says Dr Brown. "We want something like the light on a BMW dashboard which flashes and tells you that you need a service, but also with the additional capability of being able to tell you exactly what is going wrong." In order to do this, the system will monitor currents, voltages, vibration, the temperature of the motor, the temperature of the amplifier, and any other sensor inputs deemed necessary. An MSc student research project is devoted to looking at loads, particularly flimsy loads, and the effects of resonances. Part of the project is to be able to distinguish load induced vibration from vibration resulting from damage to bearings. It is expected that this facility will be included in new algorithms for set-up tuning. There is, Eureka was told, a special need for improved auto tuning of servo drives anyway, because of the lack of engineers capable of undertaking such tasks manually. Part of the research programme involves working with good servo motors and drives and seeing how they behave, and then working with motors and drives with known faults and seeing whether they can be identified correctly. The roll-out programme will involve the new drives going to 'selected' customers as the various enhancements are developed and verified as being reliable. General commercial release will come later. The food industry is a particular target because of its washdown procedures. Electrics may be sealed, but seals do become weak or may not properly be re-installed after maintenance. Servo drives are becoming increasingly ubiquitous and are often the preferred solution for packaging machines and other automated production units. Other applications are seen in process industries, defence, and aerospace.