09 Signal Processing, Condition and Process Monitoring Flashcards
Signal Sources
o Internal Data (e.g. motor current)
o External sensors mounted on machine parts (e.g. acceleration/force sensors, strain gages)
goal of process control
optimization of the process guidance
Control Loops - Basics
Work pieces are being measured periodically for the systematic process control and systematical errors by wear and thermal drift of the machine are being compensated.
Machines settings are chosen on the “safe side” due to unpredictable tool and material deviations Does not lead to a full exploit of the machine performance
-> Improvement can be achieved by the use of a machine intern control loop
-> Processing quality and safety can be improved by the collection of disturbing influences and feedback By closing the open loop control with a process controller
Quality Assurance Tasks
o Can be performed directly or indirectly -> Machine and monitoring parameter are evaluated locally and immediately, directly or indirectly by their effects on measureable variables
o Time grid for quality assuring tasks has to be created, depending on the product and process quality and quality assuring apparatus and maintained during the manufacturing process
Reasons for collisions
o Programmatic errors -> Wrong paths; wrong programmed tool
o Control errors -> Error in measurement system; error in workpiece; read error
o User errors -> Wrong movement commands/program used; set-up error; wrong tool
o Machine errors -> Tool change error; Positioning error
o Work piece errors -> Wrong over measure
Collision Detection Strategies
o Without sensors (calculation of geometry)
o With sensors
Collision Area
Reduction of the number of points to be considered for a possible collision -> Reduction of necessary computing power
o One point of the tool (collision point) should have a prescribed movement path (e.g. tool tip)
o Parts of the working space that can be involved in a collision are artificially enlarged in such a way that moving the collision point outside this artificially enlarged surface does not lead to a collision with any point of the moving machine part.
Look Ahead Function (Collision Avoidance)
Checks if next NC line is within boundaries
Application of Sensors in Process- and Condition Monitoring
o It is important, that the sensors do not affect the characteristics of the machine or that this influence is insignificant (e.g. the dynamometers influence on the stiffness)
o Machine internal signals (current of drives, positions etc.) could be used in addition to external sensors
Mounting Options for structure borne sound sensor
o The resonance frequency limits the sensors range of application -> Should only be used in the range, where it has a linear frequency response
o Screw and wax connections are to be preferred for measuring high-frequency vibrations
Fourier Transformation
o (Almost) each periodical function can be described as a sum of sine and cosine terms
o Transforms a signal s(t) from the time domain into the frequency domain S(f)
o In the frequency domain, it is easier to determine and identify (unwanted) oscillations
o The frequency often gives a hint towards its origin
What Filter do you add if you only want to analyze low frequencies
Low Pass filter
Digitalization of Signals
o Either first quantization and then sampling or the other way around
o Data must be converted into a binary system, in order to digitally transmit and store the values after sampling and quantization
o Then, the binary data can be transferred from the physical layer via technologies such as ethernet or radio technologies
Quantization
o Quantization transforms value continuous signals in value discrete ones
o The resolution determines, how exact the original signal values can be described
o A resolution of 3 Bit yields in a range of 2^3 = 8 values
o With 16 Bit there are 2^16 = 65.536 possible values
Quantization - Target
The best possible utilization of the range of values
Quantization Errors
Difference between the value continuous and value discrete amplitude
Sampling
o Only a countable number of signal values can be stored in the computer (reduction of data volume)
o Conversion of time-continuous signals into time-discrete signals
o The signal is sampled with the sampling rate 1/T. This means that only the signal values at multiples of the sampling rate 1/T are stored.
o Mathematical point of view: Anaesthetic signal is multiplied by a Dirac pulse signal
-> Dirac pulse signal is unequal to zero at one point Only those values remain after the multiplication which are represented at multiples of the sampling rate 1/T
Fourier Transformation of Time Discrete Signals
o Due to sampling in the time domain, the frequency spectrum is being repeated endless times
o Repetitions at multiples of the sample rate fa = 1/T
o Reconstruction of the original signal is possible by use of a low pass filter
Alias Effect
Through sampling the frequency spectrum changes -> New frequency components are formed in the spectrum by the periodic repetition of the analog signals spectrum with the period fa (sampling rate)
A sampling rate fa < 2fmax leads to overlapping frequency bands and therefore to a distortion of the signal (see picture) -> This effect is called aliasing
Calculation of the sliding mean value
o The sliding mean value is used e.g. to detect steps in the dressing process during grinding or to eliminate stochastic parts in structure borne sound signals during bearing monitoring.
o Corresponds to a low-pass filter with PT1-Member
o The adjustable mean value factor MWF indicates the weighting of the actual measured value in the new calculated mean value XGi. -> A high mean value factor results in a sliding mean value, that slowly follows the original signal
-> The larger the sliding mean factor, the bigger is the low-pass effect of the calculation of the sliding mean value
Tool Breakage Detection with dynamic threshholds
o Dynamic thresholds maintain the detection of sudden, damage caused signal variations, e.g. due to a collision or tool breakage.
o The dynamic thresholds are calculated during operation, based on the sliding mean value
o Upper threshold: A specific amount is added to the sliding mean value
o Lower threshold: The same amount is subtracted from the sliding mean value
o Advantage of dynamic thresholds over fixed thresholds: The adaption of the tolerance band to the signal, which changes due to variable process parameters, e.g. variations in allowances or tool wear during the tool breakage monitoring.
Controller Integrated Monitoring
o For the monitoring, instead of external sensors also controller integrated signals can be used (e.g. motor current or power of main spindle and feed drives)
o Additional information by rotational speed, command and current positions (and by derivation also accelerations)
Advantages of Controller Integrated Monitoring
No external sensors or hardware necessary
* Therefore cheaper
* Easy maintenance (e.g. no additional cables)
Analysis directly on numerical control
Direct access to controller internal functions (e.g. emergency stop)
