曝光台 注意防骗
网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
run loop to complete. Because more than one error might occur in a single
cycle and each takes so long to print, a data structure is used to store the
starting addresses of each error's corresponding message. Up to 15 addresses
can be held in this circular queue. Figure 22 shows the way starting addresses
of messages are saved. It is a more detailed version of the boxed area in
figures 19 to 21.
After the digital-to-analog conversion of the simulator outputs in the
run loop, the program checks the error message queue. If no printing is in
progress and a message is waiting to be printed, the program will initiate
printing the message at the head of the queue. Otherwise the program returns
to the task which was interrupted by the current cycle of the run loop - elther
MINDS or printing a message. The message generation code, MESGEN (fig. 23), is
actually a portion of the multiplexer interrupt routine and is shown in the
boxed region of figure 17.
Figure 24 demonstrates the way the pointers move around the queue when
two errors occur in rapid succession and are then printed out to the terminal
device. The operation of the total noncatastrophlc error handling system can
be understood by tracing through the flowcharts in figures 17 and 19 to 23.
In general, the user would like to know the cause of any errors which
occur. To help him determine what happened, both the divide interrupt service
routine and the 8087 exception service routine save the instruction pointer and
ii_-!Iithe
code segment of the instruction after which the error occurred. These values
can be examined vla MINDS to determine which line of code prompted the
error. In addition, the 8087 exception handler stores the 8087 status word and
the address of the 8087 environment. Since this saved information would be
overwritten the next time a similar error occurs, these two service routines
each set a latch to prevent new information from being stored. After the data
have been examined, the user can use MINDS to reset the latches in preparation
for the next error should one occur.
The only catastrophic error which is handled by an interrupt is a system
bus timeout error (fig. 25). This error usually means that program control is
lost and that execution has ceased. The timeout interrupt brings control to
the service routine where it remains until the routine has executed and control
is returned to the previously running instruction address. If this error
occurs, a message is printed out immediately in the service routine. The message
contains the location of the instruction pointer and code segment of the
calling instruction which failed. This can be used to aid in reconstructing
what caused the timeout.
A list of the messages which can be printed appears in table VIII.
SIMULATION RESULTS
The steady-state accuracy of the model is excellent. This is because the
HYTESS-Iike model was based on the steady-state performance of a turbofan
engine and the base point calculations which define steady-state performance In
HYTESS were derived from steady-state data. Also, the steady-state and transient
accuracies of the actuator simulation are excellent. The full engine
translent performance for small perturbations about a given operating point is
also quite good. The full engine large perturbation transient performance is
qulte limited since the engine is modeled as a linear system in the run loop.
CONCLUSIONS
Tests conducted in conjunction with the FIO0 Hybrid Simulation evaluation
of the ADIA algorithm showed that the simulator works well as a real-time,
steady-state and small perturbation substitute for the full hybrid, nonlinear
simulation. The full-scale engine demonstration of the ADIA proved the capabilities
of the simulator as a real-time code verifier and as a full envelope,
real-time actuator slmulator for actuator fault detection. This real-time,
portable simulator capability will be valuable in future engine tests. With
the rapid increases in microprocessor capabilities that have occurred since the
FIO0 simulator was built, it is conceivable that full envelope, full engine
simulation can now be achieved in real-time.
APPENDIX A
USER'S MANUAL FOR FIO0 ENGINE SIMULATOR
I. Turn on all of the equipment, I.e., the chassis, the dlsk drive, and
the terminal.
2. Insert the system disk into drive a: and the program disk into drive
b:.
3. Boot the simulator by pressing the RESET button on the chassis.
4. Nhen the simulator has booted, load and start the program by typing
b:(program-name>(RETURN).
5. Thls causes the program to start executlng. It goes through the onetlme
Inltlallzation routines, MSET and MTRXST, and enters the InltlaIizatlon
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址：
A_real-time_simulator_of_a_turbofan_engine(6)
