Direct Answer

How does the energy-saving diagnosis find the space? The diagnosis is the reasoning chain of "data → benchmarking → deviation → attribution → measure" — the data only answers "how much used" but not "how much should be used" — benchmark (the quota, the benchmark and the historical best), find the deviation and attribute it to the equipment level — the NILM decomposes the total into the equipment and the SEC links the energy with the output — the invisible equipment-level waste is exposed by the two knives — the opportunity changes from "feeling" to "evidence".

A scenario: An enterprise has run the system for half a year — the total and the sub-item data are all there — "where to save" cannot be answered — the veteran says by experience that the air-conditioning wastes — without the data support, the renovation is not invested. The logic is that the data is the "raw material" not the "answer" — the opportunity hides in the deviation from the standard/benchmark — the chain traces the deviation to the equipment level step by step — the space becomes quantifiable and projectable.

A complete example — A park of half a year: the total 5.2 million kWh (data) — the benchmark of the similar park: 12% higher (benchmark) — the deviation about 550,000 kWh — the NILM finds the air-conditioning idle at night and the elevator standby (attribution) — the "operation period optimization + standby management" saves 8% (measure) — the actual 7.5%, matching the estimate.**

1. The Five-Step Reasoning Chain

Step 1: The data — collect the energy and the operation: the total, the sub-item and the equipment; Step 2: The benchmark — compare with the quota, the benchmark and the historical best: find "how much should be"; Step 3: The deviation — the actual minus the benchmark: locate "how much more"; Step 4: The attribution — trace to the equipment, the period and the condition: explain "why more"; Step 5: The measure — target the cause, estimate the saving: answer "how to change and how much".

In one sentence: the chain is "the navigation from the data to the opportunity" — data → benchmark → deviation → attribution → measure — one step, the opportunity clearer.

The chain comparison table:

StepActionOutput
DataCollectFoundation
BenchmarkCompareGap list
DeviationLocateDeviation
AttributionTraceCause
MeasurePlan/estimateAction list

The four benchmarks — the quota standard (the method standard like GB/T 17166), the industry benchmark, the historical best and the design value — the four, each useful. An example: A factory: the sub-item data (data) — the same-line unit comparison (benchmark) — the air station 18% higher (deviation) — the operation record finds the long unloading (attribution) — the "off-peak + VFD" saves 15% (measure). The logic is that the chain is "the skeleton" — without any step, the diagnosis is incomplete — the chain complete, the opportunity lands.

2. What the Data Sees and Cannot See

The data sees — the trend of the total and the sub-item, the peak-valley, the running time and the load rate — "how much and when"; The data cannot see — "how much should be" (the benchmark), "why more" (the attribution) and "which is the waste" (the judgment); The boundary — the data is the "fact layer": the what, not the should and the why; The supplement — the benchmark adds the "should" and the attribution adds the "why": the later steps of the chain.

The typical dimensions the data sees — the trend, the structure (the share), the period (the peak-valley) and the load (the rate and the demand) — the four, complete; The typical problems the data cannot see — the no-load loss (running without output), the low-efficiency condition and the management waste (unattended) — the three need the attribution. An example: An enterprise has all the data but never benchmarks — the monthly "looks normal" — actually 20% higher than the benchmark — after the benchmark, the waste appears. The logic is that the data is "the mirror" and the benchmark is "the ruler" — only the mirror cannot see the height — with the benchmark, the opportunity appears.

3. How the NILM Locates the Single Equipment

The meaning of the NILM — the non-intrusive load monitoring: decomposing the total into the equipment from one point; The principle — each equipment has the "electric fingerprint" (the start-stop, the harmonic and the transient) — the algorithm recognizes from the total; The mechanism of the event detection — the jump of the total power (the start/stop) → the feature extraction (the amplitude, the duration and the waveform) → the matching with the fingerprint library — the three segments; The value — no meter per equipment: the equipment-level load from one total point; The output — the start-stop time, the running duration and the consumption of each equipment.

In one sentence: the NILM is "the microscope in the total meter" — one point, the invisible equipment-level use decomposed — no meter per equipment, the waste nowhere to hide.

The NILM and the sub-metering comparison:

DimensionNILMSub-metering
InstallationOne pointPer circuit
CostLowHigh
PrecisionRecognitionMetering
ApplicationRetrofitNew/key circuits

The typical outputs — the start-stop table, the running statistics, the per-equipment energy and the abnormal idle period — the four feed the diagnosis; The boundary — the clearer the feature, the more accurate; the similar devices need the auxiliary; the complex process is better with the sub-metering. An example: A park has only the total meter — the NILM decomposes: the frequent air-conditioning start-stop, the high elevator standby and the midnight idling — the waste all visible — after the optimization, 8% lower. The logic is that the NILM solves "the equipment-level absence" — the granularity from "the building" to "the equipment" — the invisible waste is illuminated.

Back to the Tianyan E-01 and the ESA — the carrier: the E-01 decomposes at the edge and the ESA supplies the high-quality total data — the decomposition directly enters the diagnosis — the waste from "guess" to "calculation".**

4. The SEC and the OEE Integration

The SEC — the specific energy consumption: the energy/output — "how much for one product"; The OEE — the overall equipment effectiveness: the availability × the performance × the quality — "the output efficiency"; The relation — the SEC sees "the energy efficiency" and the OEE sees "the equipment efficiency" — a high SEC may be a high energy or a low output; The integration — the high SEC + the low OEE: the output side (the stop, the speed-down and the scrap); the high SEC + the normal OEE: the energy side (the idling and the low condition).

In one sentence: the SEC and the OEE are "a pair" — a high SEC looks at the OEE — a low OEE is the output and a normal OEE is the energy — the integration, the attribution right.

The SEC subdivisions — the product (kWh/piece), the process (kWh/process) and the station (kWh/station) — the finer, the more accurate; The OEE three rates — the availability (the stop loss), the performance (the speed-down loss) and the quality (the scrap loss) — the three, each a type. An example: The unit energy rises for three months — the OEE check: the performance falls (the speed-down) — the attribution "the low output by the speed-down" — after the optimization, the unit falls. The logic is that the SEC is "the result" and the OEE is "the process" — only the SEC cannot know the cause — the integration, the attribution accurate.

5. From the Opportunity to the Saving Estimate

The sorting — by the saving × the difficulty: the easy and the large first; The algorithm — the baseline energy minus the post-measure energy: by the running time and the power; The conservativeness — take the conservative value: with the margin; The verification — the actual comparison after the landing: the estimate vs the actual, the deviation reviewed.

In one sentence: the opportunity is "quantified + sorted + verified" — estimate, sort by the input-output and verify after the landing — no overestimate, the loop credible.

The three estimation methods — the year-on-year, the baseline and the model — by the scene; The correction factors — the output, the weather and the condition — removed first, the real saving; The verification table — the pre-baseline, the post-actual, the saving and the conclusion. An example: An enterprise lists 6 opportunities — sorted — first the "off-peak" (zero investment, 3%) — then the VFD (invested, 8%) — verified after the landing. The logic is that the end of the diagnosis is "the projectable action list" — estimate, sort and verify — the opportunity from "discovery" to "landing".

Back to the diagnosis solution and the energy-carbon autonomy system — the closed platform: the collection (the ESA), the NILM (the E-01), the benchmark and the attribution (the platform) and the measure tracking (the solution) — the chain complete — the space starts with the chain.**

Scope and Limitations

  • The conclusions in this article are intended for LV distribution and electrical-safety sites. Standard numbers (GB/T 17166) refer to their currently effective editions, and product parameters follow the latest FEXLINK 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 Tianyan engine E-01(NILM non-intrusive load disaggregation), 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

How Energy Audits Actually Find Savings

The diagnosis is the reasoning chain of "data → benchmarking → deviation → attribution → measure" — the data only answers "how much used" but not "how