Direct Answer
After going live, platform data is sometimes good and sometimes bad — curves jump, devices drop, and figures do not reconcile with on-site meters. Is the problem at the sensing layer or the network layer? Locate by matching the symptom to the layer — abnormal readings/wrong multipliers on meters are in the sensing layer, curve gaps/delays are in the network and gateway layer, drift and low averages are in the cleaning and protocol parsing layer — rule out layer by layer before acting, and avoid blindly replacing meters.
A scenario: A factory platform goes live — the data curve jumps — the operator first suspects the meter is broken — replacing it does not help — checking the gateway cache — packet loss found — fixed and restored. The logic of diagnosis is to locate the layer first, then act — the same "wrong data" can come from the meter, the network or the parsing — rule out layer by layer, do not blindly replace meters.
1. Three Typical Fault Symptoms
Symptom one, abnormal readings — meter problems: a single meter jumps or has a wrong multiplier; Symptom two, curve gaps — network problems: curve gaps, delays or bursts; Symptom three, trend drift — parsing problems: low averages and drifting trends; The distribution — mapped to layers: symptoms map one to one with layers; The starting point of diagnosis — recognize the symptom: first see what the symptom is.
In one sentence: the three symptoms are abnormal readings (sensing), curve gaps (network) and trend drift (parsing).
An example: A factory has three symptoms at once: meter A readings jump — the curve has gaps — the average is low — each maps to one layer. The logic is that symptoms map to layers — classify the symptom first — you know which layer to check first instead of acting blindly.
2. Sensing Layer Checks
Check one, meter accuracy — the meter itself: is it accurate, is it degrading; Check two, transformer ratio — wrong multiplier: is the CT ratio set correctly; Check three, wiring — wrong connections: are voltage/current leads reversed or loose; The method — compare on site: check against the on-site meter reading; The conclusion — replace or not: replace the meter only after confirming the meter problem.
In one sentence: sensing layer checks are meter accuracy, CT ratio and wiring — conclude after checking on site.
An example: Meter A reads abnormally — checked on site — the CT ratio is set wrong — the multiplier is corrected — restored. The logic is that a sensing layer problem is a source error — check the site first — accuracy, ratio and wiring — replace the meter only after confirmation.
3. Network Layer Checks
Check one, gateway cache — cache full: data queues and uploads late; Check two, packet loss — link loss: curve gaps and missing data; Check three, 4G signal — weak signal: delays, bursts and reconnect drops; The method — look at the gateway: gateway logs and signal strength; The conclusion — optimize the link: add cache, change the channel, add signal.
In one sentence: network layer checks are gateway cache, packet loss and 4G signal — locate with gateway logs.
An example: A factory's curve has gaps — the gateway log shows a weak 4G signal with packet loss — an antenna is added — the gaps disappear. The logic is that a network problem is a link error — look at the gateway cache and signal — treat loss and weak signal accordingly — do not touch the meters.
4. Parsing Layer Checks
Check one, the protocol map — wrong map: data addresses mapped incorrectly; Check two, data cleaning — dirty data: outliers not filtered, the average dragged down; Check three, the completion logic — wrong completion: missing values filled improperly; The method — look at the platform: compare raw values with platform values; The conclusion — fix the parsing: fix the map, adjust cleaning, change completion.
In one sentence: parsing layer checks are the protocol map, data cleaning and completion logic — locate by comparing raw values.
An example: A factory's average is low — compared with raw values — the cleaning filters out peak values — the cleaning strategy is adjusted — the average is restored. The logic is that a parsing problem is a processing error — compare platform display with raw values — check the map, cleaning and completion — fix processing, not hardware.
5. Data Completeness Rate: A Daily Monitoring Indicator
What the completeness rate is — a quality indicator: actual received data divided by expected data; How it is calculated — per point: the completeness rate of each point in each period; The threshold — the baseline: for example, above 99% is healthy; The value — early discovery: drops and loss are exposed immediately; The presentation — a monitoring page: the platform shows overall quality in one view.
In one sentence: the data completeness rate is received/expected — above 99% is healthy, drops and loss are exposed early.
An example: A factory's rate is 96% — one gateway is at 90% — packet loss is found — fixed — restored to 99.5%. The logic is that the rate is the thermometer of data quality — check daily — below the threshold, check the corresponding points — problems are found and handled early. Too high a threshold alarms every day, too low misses problems — set the baseline by point health first, then tighten gradually.
6. Rule Out Layer by Layer Before Acting
Principle one, recognize the symptom — classify first: which layer does the symptom belong to; Principle two, check layer by layer — rule out in order: sensing, then network, then parsing; Principle three, compare on site — the site is the authority: compare platform values with site values; Principle four, treat the matching layer — fix the right layer: fix whichever layer has the problem; The value — no fuss: avoid blind meter replacement and repeated rework.
In one sentence: rule out layer by layer is recognize, check in order, compare on site and treat the right layer — no fuss.
An example: A factory rules out by layer: the symptom maps to the network — check the gateway — packet loss — add an antenna — fixed once. The logic is that layer-by-layer exclusion locates by symptom-to-layer — start from the source — replace a meter only after confirming a sensing problem — saves time and money, and turns operations from "checking everywhere" into "following the map".
7. Back to YESLON: How ESX and ESA Help
The ESA role — precise metering: meter accuracy is guaranteed; The ESX role — the edge gateway: local cache, offline buffering and signal monitoring; The platform role — completeness monitoring: data quality in one view; The coordination — clear layers: sensing, network and parsing each do their job; The YESLON promise — visible quality: the completeness indicator is built in.
In one sentence: the YESLON approach is ESA metering, ESX cache and buffering, and platform completeness monitoring.
An example: A park: ESA meters — ESX buffers 3 days offline — the platform completeness rate is 99.5% — data quality in one view. The logic is that layered problems need layered devices — accurate meters, stable networks and a platform that sees — data quality stays credible, and operations have a handle.
8. Self-Check: Four Questions for Data Quality
Question one, is the symptom classified — classify: which layer do jumps, gaps or drift belong to; Question two, is the site compared — compare: have platform and site values been checked; Question three, is it ruled out layer by layer — layers: have sensing, network and parsing been checked in order; Question four, is the completeness rate watched — rate: is the daily monitoring indicator being viewed.
In one sentence: the four questions are classify, compare, layers and rate — all answered, data quality is controlled.
A self-check example: A factory answers the four questions: symptoms classified, site compared, layers checked and the rate watched — data quality is stable. When energy platform data jumps, drops or does not reconcile — locate by matching the symptom to the layer first — abnormal readings/wrong multipliers are in the sensing layer, curve gaps/delays are in the network and gateway layer, drift and low averages are in the parsing layer — rule out layer by layer before acting — add the data completeness rate for daily monitoring — no blind meter replacement, no repeated rework.
Scope and Limitations
- The conclusions in this article are intended for LV distribution and electrical-safety sites. Criteria are compiled from public codes and FEXLINK product documentation, and product parameters follow the latest selection handbook and product knowledge base.
- Quantitative criteria (protection zone, discharge and residual voltage, earth resistance, thresholds and grading) hold under the preconditions stated in this article. Where the site voltage level, earthing arrangement, environmental category or load profile differs, the figures must be recalculated. Specific configurations involving ESX intelligent edge computing gateway, ESA all-element smart meter should be verified against actual circuit parameters.
- Specific product models and parameters are subject to the latest selection handbook and product knowledge base; this article does not constitute selection or procurement advice.
- Standard numbers cited here refer to the currently effective editions; when a standard is revised, the related quantitative conclusions must be re-checked.
- Deployment and implementation for specific projects must be carried out by qualified engineers based on site survey and design drawings.
Frequently Asked Questions
When Platform Data Keeps Jumping, Dropping, and Not Reconciling—Is the Problem at the Sensing Layer or the Network Layer?
After going live, platform data is sometimes good and sometimes bad — curves jump, devices drop, and figures do not reconcile with on-site meters