Data informed seasonal assessment of a PV battery grid charging station for light electric vehicles with adaptive filter forecasting, state of charge and harmonic estimation

Abstract Photovoltaic (PV) charging stations for light electric vehicles are commonly sized with typical-year irradiance and verified over short simulation windows, which conceals the seasonal variability that sizes storage and grid support. This study presents a dual-timescale assessment of a grid-connected charging station with a 4 kWp PV array, a 21.6 kWh lithium iron phosphate battery and four 0.8 kW charging ports. Daily energy-meter records of an operating 325 kWp car-park canopy PV plant in northern India (January--September 2026, 270 recorded days) were screened with seven IEC 61724-1-informed and statistical rules, with co-located campus meters as a reference to separate meter faults from weather. Suspect days were excluded, not altered; 257 days were retained (95.2 % coverage), and the mean specific yield changed by only 0.8 %. The retained days were capacity-scaled to 4 kWp, converted into one-minute irradiance and simulated with a 1 s average-value model; a 1 \ensuremath{\mu}s switching model resolved the grid-current harmonics. Over the same months, the measured mean specific yield (4.13 kWh/kWp/day) agreed with the typical-year PVsyst estimate within 0.6 %, but monthly deviations ranged from \ensuremath{-}21 \% to +27 \%. Seasonal means were 3.49 (winter), 5.06 (pre-monsoon) and 3.72 kWh/kWp/day (monsoon). For an assumed demand of 11.2 e-bike sessions per valid day, PV supplied 73.7 %, the battery 22.9 % and the grid 3.3 % of the charging energy; winter and monsoon required 26--30 % battery support and accounted for nearly all grid import. Under deliberate model and sensor errors, the proposed recursive-least-squares (RLS) adaptive extended Kalman filter reduced the mean monthly state-of-charge RMSE to 0.58 %, versus 3.16 % for an extended Kalman filter. An RLS day-ahead PV forecast reduced the forecast RMSE by 10.5 % relative to persistence, yet lowered the operational energy cost by only 0.61 % under the assumed tariff. Under simulated grid-voltage distortion, a multi-resonant proportional-resonant controller achieved the lowest steady-state distortion, and RLS harmonic estimation reduced the current total harmonic distortion at 3 kW from 7.61 % to 1.74 %, with total demand distortion below 1.3 %; these are simulation-based comparisons, not an IEEE 519 compliance assessment.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-16
DOI
https://doi.org/10.21203/rs.3.rs-11399297/v1
Primary Topic
Photovoltaic System Optimization Techniques
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Data informed seasonal assessment of a PV battery grid charging station for light electric vehicles with adaptive filter forecasting, state of charge and harmonic estimation

Pallavi Gupta, ANKITA SINGH, Manisha Rajoriya, Sumanlata Dhar
2 links
Research Square
Photovoltaic System Optimization Techniques
preprint

Data informed seasonal assessment of a PV battery grid charging station for light electric vehicles with adaptive filter forecasting, state of charge and harmonic estimation

Pallavi Gupta, ANKITA SINGH, Manisha Rajoriya, Sumanlata Dhar
preprint en

Abstract

Abstract Photovoltaic (PV) charging stations for light electric vehicles are commonly sized with typical-year irradiance and verified over short simulation windows, which conceals the seasonal variability that sizes storage and grid support. This study presents a dual-timescale assessment of a grid-connected charging station with a 4 kWp PV array, a 21.6 kWh lithium iron phosphate battery and four 0.8 kW charging ports. Daily energy-meter records of an operating 325 kWp car-park canopy PV plant in northern India (January--September 2026, 270 recorded days) were screened with seven IEC 61724-1-informed and statistical rules, with co-located campus meters as a reference to separate meter faults from weather. Suspect days were excluded, not altered; 257 days were retained (95.2 % coverage), and the mean specific yield changed by only 0.8 %. The retained days were capacity-scaled to 4 kWp, converted into one-minute irradiance and simulated with a 1 s average-value model; a 1 \ensuremath{\mu}s switching model resolved the grid-current harmonics. Over the same months, the measured mean specific yield (4.13 kWh/kWp/day) agreed with the typical-year PVsyst estimate within 0.6 %, but monthly deviations ranged from \ensuremath{-}21 \% to +27 \%. Seasonal means were 3.49 (winter), 5.06 (pre-monsoon) and 3.72 kWh/kWp/day (monsoon). For an assumed demand of 11.2 e-bike sessions per valid day, PV supplied 73.7 %, the battery 22.9 % and the grid 3.3 % of the charging energy; winter and monsoon required 26--30 % battery support and accounted for nearly all grid import. Under deliberate model and sensor errors, the proposed recursive-least-squares (RLS) adaptive extended Kalman filter reduced the mean monthly state-of-charge RMSE to 0.58 %, versus 3.16 % for an extended Kalman filter. An RLS day-ahead PV forecast reduced the forecast RMSE by 10.5 % relative to persistence, yet lowered the operational energy cost by only 0.61 % under the assumed tariff. Under simulated grid-voltage distortion, a multi-resonant proportional-resonant controller achieved the lowest steady-state distortion, and RLS harmonic estimation reduced the current total harmonic distortion at 3 kW from 7.61 % to 1.74 %, with total demand distortion below 1.3 %; these are simulation-based comparisons, not an IEEE 519 compliance assessment.

Research Square
Sharda University (IN)
Affordable and clean energy, Industry, innovation and infrastructure, Climate action
Photovoltaic System Optimization Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon