{"id":3803,"date":"2026-11-04T09:00:00","date_gmt":"2026-11-04T01:00:00","guid":{"rendered":"https:\/\/cnbygele.com\/?p=3803"},"modified":"2026-11-03T19:10:05","modified_gmt":"2026-11-03T11:10:05","slug":"meter-csv-locale-validation","status":"publish","type":"post","link":"https:\/\/cnbygele.com\/ar\/blog\/meter-csv-locale-validation\/","title":{"rendered":"Power Meter CSV Import: Validate Decimal Separators and Units"},"content":{"rendered":"

Power meter CSV import decimal separator errors occur when the exporting system and the importing application interpret the same characters differently. A value written as 1,25<\/code> may mean one and a quarter, two separate fields, or text that never becomes a number. Validate the delimiter, quotation rules, decimal convention, unit and timestamp together before importing a complete energy file. Preserve the original export, test a small known sample in a staging area, and reconcile both record counts and values before accepting the result.<\/p>\n

This is a data handling procedure for engineers receiving files from a meter gateway, historian or reporting service. It does not establish that every panel meter exports CSV or contains historical storage. Software capabilities and the exact instrument’s fitted options must be confirmed separately.<\/p>\n

Separate file structure from numerical meaning<\/h2>\n

A field delimiter divides columns; a decimal mark divides the integer and fractional parts of a number. They perform different jobs even when both are represented by a comma. Some regional exports therefore use semicolons between fields and commas inside numbers. Other exports use comma-separated fields with quoted decimal-comma values. Selecting a region in a spreadsheet does not automatically prove that every column has been interpreted correctly.<\/p>\n

RFC 4180’s CSV description<\/a> documents common comma-separated record and quotation conventions. It does not define a universal energy measurement schema or mandate a particular meaning for a number. A file called CSV may use a different dialect, which the exporter must explain. Treat the extension as a starting point and inspect actual characters before choosing import settings.<\/p>\n

For example, M01,\"1,25\",kWh<\/code> contains three fields under common CSV quoting rules. The middle field still needs an explicit numerical interpretation after the parser removes quotation marks. By contrast, M01;1,25;kWh<\/code> requires a semicolon delimiter. Splitting either line blindly on every comma damages the structure or changes the number.<\/p>\n

Preserve the original export and its context<\/h2>\n

Save the original bytes under a unique filename before opening the export in an application that might save changes automatically. Record the source system, export time, selected meters, reporting period, timezone, software version where available and export settings. A file checksum helps establish whether two copies are identical, but cannot establish that the values themselves are correct.<\/p>\n

Keep a working copy for testing. Record the importing application’s version and the explicit parsing settings used. If an operator later asks why a particular interval changed, the original file and the import record should explain the transformation without depending on somebody’s memory. Retain rejected rows separately with the reason for rejection.<\/p>\n

For the electrical source, the BY194 series product description<\/a> provides family context. Request the actual communication option, data map and software export specification for the supplied system. A front-panel image cannot establish a CSV exporter, decimal setting, memory capacity or downloadable interval format.<\/p>\n

\"Power
Reference-based illustration of a disconnected BY194 front beside external power meter csv import review materials. Documents and symbols illustrate a workflow; they do not establish installed product functions.<\/figcaption><\/figure>\n<\/p>\n

Build a column contract before importing<\/h2>\n

List each expected column by position and name. Define whether it contains an identifier, timestamp, numerical measurement, unit, status or comment. Preserve meter identifiers as text: leading zeroes can distinguish assets, and a long identifier must not silently become a rounded number. Document optional columns so that a software update cannot shift every subsequent field unnoticed.<\/p>\n

For numerical columns, specify decimal mark, allowed grouping characters, sign, exponent notation if supported, blank handling and permitted units. An empty cell is not automatically zero. A value such as 1,234<\/code> is ambiguous without context because it may represent a fraction or a grouped integer. Quarantine it when the documented convention and surrounding evidence cannot resolve the meaning.<\/p>\n

Do not remove all punctuation as a cleaning shortcut. That could change 12.50<\/code> into 1250<\/code>, erase a negative sign or convert two fields into one value. First parse the record with its documented dialect; then validate the numerical field under the approved locale. Preserve the original text alongside the interpreted value during acceptance.<\/p>\n

\n\n\n\n\n\n\n\n\n\n\n
Check<\/th>\nTest evidence<\/th>\nReject or investigate when<\/th>\n<\/tr>\n<\/thead>\n
Column structure<\/td>\nHeader and representative raw lines<\/td>\nField counts change unexpectedly<\/td>\n<\/tr>\n
Decimal convention<\/td>\nKnown fractional reference value<\/td>\nThe fraction changes or remains text<\/td>\n<\/tr>\n
Identifier<\/td>\nOriginal asset code and imported text<\/td>\nLeading zeroes disappear<\/td>\n<\/tr>\n
Unit and scale<\/td>\nExport schema and sample calculation<\/td>\nW, kW, Wh or kWh are mixed<\/td>\n<\/tr>\n
Time<\/td>\nExplicit offset and interval definition<\/td>\nTimezone or interval boundary is unknown<\/td>\n<\/tr>\n
Missing value<\/td>\nDocumented quality rule<\/td>\nBlank or invalid data becomes zero<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

This is a qualitative acceptance checklist, not a manufacturer file specification. Adapt the column contract to the actual exporter rather than renaming unknown fields until the import appears to work.<\/p>\n

Use a small sample that can expose silent errors<\/h2>\n

Choose records containing a fractional value, a negative value if the quantity permits one, a zero, a blank, an identifier with leading zeroes and a timestamp crossing a date boundary. Include a quoted comment containing a comma when such comments are possible. These examples test distinct failure modes that an all-integer sample would miss.<\/p>\n

For an illustrative energy sample, suppose three valid interval quantities are 1.25, 2.50 and 0.75 kWh. Their expected sum is 4.50 kWh. Import a deliberately small file and verify each value, all three identifiers, the row count and the sum. The figures are arithmetic examples, not BY194 performance data or a recommended sampling interval.<\/p>\n

A matching sum alone is insufficient: one value could increase while another decreases. Compare the row-level mapping as well. An unchanged row count is also insufficient because a column could remain text and be ignored by an aggregate. Acceptance requires the structure, individual values and aggregate to agree with the defined sample.<\/p>\n

Keep units separate from locale conversion<\/h2>\n

Locale interpretation answers what a numeric string means. Unit conversion answers what physical quantity that number represents. Perform and record these as separate steps. The NIST guidance on expressing quantity values<\/a> supports treating a quantity as a number together with its unit; a bare number is not enough for an engineering comparison.<\/p>\n

If an exporter records 1250 Wh, converting it to 1.250 kWh involves division by 1000. Reading 1,250<\/code> as an English thousands-grouped number could instead create 1250 kWh. That mistake combines an interpretation error with a unit error and can survive a quick visual check. Save both input unit and output unit in the transformation record.<\/p>\n

Confirm whether CT or voltage-transformer scaling has already been applied upstream. CSV cleaning must not become an opportunity to apply an undocumented ratio again. Use the separate meter unit normalization guide<\/a> for physical unit mapping; this article focuses on preserving meaning while parsing a file.<\/p>\n

Preserve timestamp and interval meaning<\/h2>\n

A timestamp can identify an interval start, interval end or reading instant. These are different concepts. Record the timezone or explicit UTC offset and the interval duration where relevant. RFC 3339<\/a> describes an Internet timestamp representation with date, time and offset information; it does not establish which interval an energy exporter assigns to that timestamp.<\/p>\n

Avoid relying on the computer’s regional date guess. A date such as 03\/04\/2026<\/code> is ambiguous without a declared convention. Preserve the raw field and reject an unresolved interpretation before processing the complete month. When daylight-saving transitions apply to the source site, retain enough timezone information to distinguish repeated local clock times.<\/p>\n

If timestamp validation uncovers missing coverage, follow the separate maintenance data gap procedure<\/a>. Do not fill a missing interval with zero merely to make an import complete. A format repair should preserve the distinction between an observation, an estimate and unavailable evidence.<\/p>\n

\"Power
Reference-based illustration of a disconnected BY194 front beside external power meter csv import review materials. Documents and symbols illustrate a workflow; they do not establish installed product functions.<\/figcaption><\/figure>\n<\/p>\n

Approve the transformation before replacing reports<\/h2>\n

Run the validated parser against a staging copy of the complete file. Compare source and accepted row counts, rejected rows, per-meter coverage and selected daily totals. Investigate outliers by returning to original rows instead of repeatedly changing import settings until the chart looks familiar. Keep the earlier report available until the replacement has been approved.<\/p>\n

The acceptance record should identify the file, parsing rules, unit conversions, time interpretation, software version, exceptions and reviewer. Where a scheduled integration imports files repeatedly, repeat the controlled sample test after an exporter or importer update. Preserve an explicit failure path so that a new dialect produces an alert instead of silently changing historical energy values.<\/p>\n

If communication data already look wrong before export, start with meter Modbus setup and decoding checks<\/a>. Correctly parsing an incorrect upstream value does not repair the measurement chain. Keep file handling, decoded data and electrical verification as separate acceptance responsibilities.<\/p>\n

Further learning: representation and interpretation<\/h2>\n

The following Computerphile lesson explains endianness: the same stored bytes can acquire a different meaning when their order is interpreted differently. It provides a useful analogy for explicit CSV parsing rules. It is not a CSV tutorial, product demonstration or evidence of a particular BY194 data format.<\/p>\n