Predictive Maintenance: Spotting Equipment Degradation in Position Data

Predictive maintenance in displacement measurement has a premise that is often overlooked: degradation that is to be predicted must first be observable. The value of position data is that it is a direct record of the actuator's action — internal cylinder leakage, guide-way wear, seal ageing and proportional-valve drift all leave traces on "the position curve under the same command". The trend in industry is that predictive maintenance is expanding from "single-point vibration/temperature monitoring" to "comparison of motion-process data". Magnetostrictive displacement sensors output absolute position, measure without contact and introduce no wear of their own, so they are well suited to serving as this kind of long-term baseline data source. Be clear, though: position data can indicate signs of degradation; it does not directly give remaining life. Any claim that displacement data can predict the exact time of failure should be treated with caution.

What position data can observe

Taking the position–time curve of a complete motion as the observation object, extractable features usually include: time to target, steady-state position error, position drift during pressure hold, repeatability of a round-trip stroke, and stick-slip jitter in the start-up section. These features are relatively stable when the machine is healthy and drift slowly as it degrades. The key is the baseline: a set of reference curves must first be captured while the machine is in good condition. All later judgements are relative comparisons, not absolute thresholds.

Common signs of degradation and how they appear in position data

The last row is especially important: a step change is not degradation. Gradual change indicates a wear trend; a jump usually points to earthing, shielding, the magnet or the fieldbus. Handle that first per the signal-jump troubleshooting flow and fieldbus data loss: termination / address conflicts / water ingress, or an electrical fault will be misread as mechanical degradation and trigger a useless overhaul.

Three hard constraints on the data-acquisition side

• Time base: curve comparison depends on time alignment. Analog sampled through a PLC scan has timestamp accuracy limited by the scan cycle; fieldbus types (especially EtherCAT with distributed clocks) can provide a more stable sampling beat; see EtherCAT real-time performance: DC synchronisation and jitter.
• Update-rate matching: if the interest is low-speed stick-slip or pressure-hold drift, the update-rate demand is modest; if shock-section features are to be captured, confirm that the update rate matches the control cycle; see matching response time / refresh rate to the control cycle.
• Comparability of figures: typical repeatability of ±0.002 mm and typical non-linearity of <0.02%FS are the capability boundary on the sensor side. If the degradation magnitude to be monitored is close to or smaller than that boundary, the observed "trend" may only be measurement noise; see the difference among resolution, repeatability and non-linearity.

From data to a maintenance action: layered judgement

In real projects, L1 and L2 usually have the highest return on investment; L3/L4 suit high-value single machines whose downtime cost is large. Data layering and local processing architecture are in IIoT and edge computing: an architecture for getting position data to the cloud; the networking foundation is in how to network displacement sensors under Industry 4.0.

The sensor's own maintainability must also be considered

Predictive maintenance assumes that the data source itself is reliable. A magnetostrictive displacement sensor measures without contact; there is no mechanical contact between the position magnet and the waveguide, and none of the sliding wear of a potentiometer, which makes it suitable as a long-term baseline source. Cable, connector seals and mounting fasteners still age and must be watched. Smart sensors with diagnostic capability can report their own status; see smart sensors: the trend towards self-awareness through self-diagnostics. Where downtime cost is extremely high, dual-channel redundancy can provide mutual checking; see redundant output and dual-channel safety design and redundant installation: how to compare two sensors.

On the product side, the cylinder-integrated Series 16 and the redundant 16R redundant cylinder-integrated are common on hydraulic machines that need long-term on-line monitoring; mobile and outdoor duty more often uses the Series 13 mobile hydraulics; when fieldbus diagnostics and high-rate sampling are needed, choose the Series 197 EtherCAT or the Series 194 CANopen.