vor 2 Jahren

O+P Fluidtechnik 10/2016

  • Text
  • Fluidtechnik
  • Sensor
  • Predictive
  • Maintenance
  • Industrie
  • Spool
  • Valve
  • Controller
  • Anwendungen
  • Produkte
O+P Fluidtechnik 10/2016


FORSCHUNG UND ENTWICKLUNG STEUERUNGEN UND REGELUNGEN FAIL OPERATIONAL CONTROLS FOR AN INDEPENDENT METERING VALVE Michael Rannow As intelligent hydraulic systems with embedded sensors become more ubiquitous, the real or perceived reliability challenge associated with sensors must be addressed to encourage their adoption. In this paper, a fault-tolerant control strategy for an intelligent independent metering valve that allows continued operation if a sensor fails is described. 70 O+P Fluidtechnik 10/2016

STEUERUNGEN UND REGELUNGEN 1 INTRODUCTION As the demand for intelligent hydraulic systems increases, there is an increasing demand for hydraulic components with integrated sensing capability. The addition of sensors can enable improved performance and increased functionality, particularly when combined with intelligent controllers. However, the reputation of electronics and sensors as less reliable than purely mechanical devices can be a significant barrier to the adoption of intelligent machines. This can be particularly true if feedback from sensors is relied upon for baseline or safe operation, and a faulty sensor could render a machine inoperable. The development of fail-operational controllers, which offer some level of functionality in the presence of a sensor failure, can reduce the real and perceived risk of downtime due to a failed sensor. The Eaton CMA valve, is an example of an intelligent hydraulic valve that relies on sensor feedback for operation. It is a two-stage independent metering valve with a position sensor and a pressure sensor for each control spool. The closed-loop control of the spool position enables fast performance and zero hysteresis, but it also creates a reliance on the position sensor for operation. Pressure sensor feedback is used to enable features such as digital pressure compensation, load direction detection, back pressure control, and electronic load sense, among others. In typical operation, all four sensors are used to make the valve function. However, there is a redundancy that exists when the two spools are being used to control two sides of the same actuator. Since there is a relationship between valve position, pressure drop, and flow, and there is a known relationship between the flow in and the flow out of an actuator, only three of the four sensors are needed to have complete information about the state of the valve. Thus, while purely redundant sensors are often cost prohibitive, the redundancy provided by knowledge of an actuator area ratio can be utilized to develop fail operational controllers. Detecting and diagnosing a faulty sensor is a critical component of creating fault tolerant systems. There are many methods for detecting faults in a system, some of which, such as sensor out-ofrange detection, can immediately identify a faulty sensor. In other methods, such as poor demand tracking or a mismatch between the estimated flow in and out of an actuator, they can detect a fault somewhere in this system, but cannot determine the root cause without additional tests. For the purposes of this paper, it is assumed that a sensor fault has been detected and correctly identified. Once a fault has been identified, reconfigured control modes can allow the valve to operate with any one of the four sensors failed. In the next section, the structure and operation of the controller for a fault in any of the four sensors is described. 2 FAIL OPERATIONAL MODES The basic principle of the fault tolerant control algorithms is described as cross-port pressure control, which essentially uses the faulty side of the valve to control the pressure on the non-faulty side of the valve. With a valve that has independent control of the meter-in and meter-out spools, there are numerous methods for controlling the two work ports. In most cases, an actuator is given a flow command, which is typically realized by controlling the flow across the spool that is holding the load. In a valve with digital pressure compensation, the measured pressure across the valve and the flow demand are used to determine the desired position of the metering spool: χ = f( ∆P, Q ) (1) des des The position of the main stage spool is controlled by the pilot stage spool and sensed by an LVDT. In this configuration, the pilot spool is a flow control spool, which means that there is a relationship bet- ween the input current and the velocity of the main stage spool. In order to stabilize the main stage at a fixed position, feedback from the LVDT is required. Thus, to accurately control the flow across a spool, feedback from both the LVDT and the pressure sensor are required. In a faulty condition, one of the sensors is not available, meaning that the spool on the faulty side cannot be used to control the flow. If that side of the spool is holding the load, the conventional controller structure must be changed so that the actuator speed can still be controlled by the operator. In the following sections, the structure of the controller for each of the four possible sensor faults is described. 2.1 FAILED METER-OUT PRESSURE SENSOR The function of the meter-out valve varies, depending on whether the load is passive or overrunning. For a passive load, the meter-out valve is often commanded to be fully open or control some low back pressure to minimize the pressure needed to move the load. When the load is overrunning, the meter-out valve is typically used to control the flow out of the actuator, with the pressure in the service determined by the load. Unfortunately, the typical method for determining the direction of the load is to compare the pressure on both sides of the actuator, but with a faulty sensor, this determination is not possible. Thus the control algorithm must function for either a or overrunning load. While the flow across the faulty meter-out side cannot be controlled, the flow into the meter-in side can. Using the pressure state equation on the non-faulty side and the fact that the flow into and out of an actuator are proportional, the concept of cross-port pressure control can be explained: Ain P& β in = ( Qin − Qout ) (2) V A out From this equation it is clear that if the pressure is held to be a constant (i. e. the derivative set to 0), then the output flow will be the same as the input flow, modified by the area ratio. Thus, if the meter-out valve can be manipulated to control the meter-in pressure to be a constant, and the flow in to the actuator is controlled to the desired output flow times the area ratio, then the flow out of the actuator will be controlled to the desired value. In order to control the meter-in pressure to a constant using the meter-out valve, the following equations can be used: χ = f ( P −P Q ) (3) out, des meter−out out tank, out, des P% = K ( P − P ) + K ∫( P − P ) dt+ K out p indes , in i indes , in d dP ( in, des − Pin ) dt (4) Where (3) uses the typical flow control function shown in (1), but instead of using the measured pressure, an estimated outlet pressure is used. The estimated meter-out pressure is determined by some control function that is driven by an error term between the measured pressure on the meter-in side and some constant set point. An example of a PID controller is given in (4). The integrator in (4) will adapt to the error between the estimated outlet pressure and the true value, which is unknown. Thus, the controller “learns” the true value of the missing sensor input. The state of the integrator in (4) can be initialized in a number of ways, but the most conservative approach for a case of a variable load pressure is to assume that the load is at its maximum possible value to avoid dropping a heavy load, then adapt the estimate to achieve the desired flow rate. With O+P Fluidtechnik 10/2016 71


© 2018 by Vereinigte Fachverlage GmbH. Alle Rechte vorbehalten.