The title says “photovoltaic energy system,” which sounds like hardware. US10554170B2, granted to Con Edison Battery Storage on February 4, 2020, is really a forecasting patent. Claim 1’s load-bearing limitation is the prediction of solar intensity feeding a control decision about storage.

Look at the CPC mix: H02S 50/00 is monitoring/testing of PV; H02J 3/385 is control of a PV source in a network; H02J 7/35 covers charging from a photovoltaic source. The combination describes a system that anticipates how much sun is coming and pre-positions the battery — charging before a cloud, holding before a peak. The cleverness is in the anticipation, not the silicon.

“A photovoltaic energy system includes a photovoltaic field configured to convert solar energy into electrical energy, one or more solar intensity sensors configured to measure solar intensity and detect a cloud approaching the photovoltaic field, and a controller.”— U.S. Patent No. 10,554,170 source

Read claim 1 in full and the mechanism is more specific than the summary quote suggests. The claim recites a PV field of multiple cells, one or more solar intensity sensors that both measure intensity and “detect a cloud approaching the PV field based on the measured solar intensity,” and — the limitation that does the real narrowing — a power output monitor watching the output of one individual cell. The controller then predicts a change in intensity across the field by predicting “a decrease in power output of a second individual PV cell…based on the output power of the first individual PV cell,” where the second cell sits “at a different location within the PV field than the first.” In plain terms: a cloud shadow that has already dimmed one corner of the array is used to forecast the dimming about to hit a corner that is still in sun, and the system “preemptively” adjusts power output before the shadow arrives.

That spatial inference is the engineering heart of the claim, and the dependent claims fill in how it is meant to work. Claim 5 has the controller computing “a position, a velocity, a size, and an opacity value of the cloud,” and claim 6 uses those four parameters to predict the field’s power output — a small cloud-tracking model running off the sensor and per-cell data. Claim 3 distributes the sensors deliberately: more of them along the long side of the array, fewer along the short side, so the field is instrumented in the direction a weather front is most likely to cross. Claim 4 adds that the controller uses the “known locations” of those sensors to predict when the change will reach a given point in the field. Claim 2 even allows the intensity sensors to sit outside the PV field entirely — sensing the approaching shade before it touches a single panel.

None of this is hardware novelty. A PV cell, an irradiance sensor, and a power monitor are all commodity parts. What claim 1 fences is the control loop that stitches them together: measure one cell, infer the future of another cell at a different location, and act on the inference before the irradiance actually drops. Stripped of the cloud-physics language, this is a feedforward control scheme — use an upstream measurement to anticipate a downstream disturbance — applied to a solar-plus-storage plant. The CPC placement into H02S 50/00 (PV monitoring) rather than a pure control class reflects that the patent office read it as a solar-system patent, but the inventive weight sits in H02J 3/385 and H02J 7/35: controlling the PV source and timing the battery charge.

The commercial implication: a utility-affiliated entity (Con Edison Battery Storage) patenting forecast-driven dispatch in 2020 is an early signal of solar-plus-storage being run as a predictive asset rather than a passive generator. The IP follows the operating model. A plant that can preemptively ramp before a cloud event smooths its output, reduces the ramp burden it pushes onto the rest of the grid, and times charging to capture energy that a reactive system would spill — the kind of behavior a utility values when it has to balance the feeder the plant sits on.

The standing caution: a claim on “predict, then dispatch” does not own solar-plus-storage. Dozens of energy-management systems forecast irradiance, and by 2020 generic predictive energy management had substantial prior art. What keeps claim 1 from reading on all of that is the specific per-cell inference — first cell’s measured output used to predict a second, differently located cell’s decrease. A competitor forecasting irradiance from sky cameras, satellite data, or a single field-wide sensor is using a different mechanism and arguably outside this fence; one inferring cell-to-cell shadow propagation from per-cell power monitoring is squarely inside it. The dependent claims (the cloud position/velocity/size/opacity model, the asymmetric sensor layout) are where the durable, harder-to-design-around protection lives. The patent fences off one articulation; how much daylight is left for competitors depends on how tightly that per-cell spatial-prediction limitation is read — pun fully intended.

It is worth dwelling on why the per-cell limitation is the difference between a strong and a weak fence, because the claim drafter clearly chose it deliberately. A claim that simply said “forecast irradiance and adjust the battery” would read on essentially every modern energy-management system and collapse under prior art. By anchoring the prediction to a measured decrease in one specific cell propagated to another specific cell at a different location, the inventors converted a generic forecasting idea into a concrete physical inference about how a shadow moves across a known array geometry. The dependent claims reinforce that this is a spatial model, not a clock: claim 4’s use of “known locations” of the sensors to predict when the change arrives, and claim 3’s asymmetric long-side/short-side sensor density, only make sense if the controller is reasoning about a disturbance travelling across a two-dimensional field. That is the kind of detail that is hard to invent around without abandoning the approach entirely — a competitor either reproduces the cell-to-cell propagation logic, in which case it likely infringes, or it forecasts some other way and gives up the specific advantage the claim captures. For a grid desk, the deeper point is that the patentable substance in solar-plus-storage has migrated almost entirely into the control and prediction layer: the panels, sensors, and battery are commodities, and the defensible IP is the software that decides, cell by cell and second by second, how to move energy ahead of the weather.