WO2022096571A1 - Surveillance d'installation d'énergie solaire - Google Patents

Surveillance d'installation d'énergie solaire Download PDF

Info

Publication number
WO2022096571A1
WO2022096571A1 PCT/EP2021/080634 EP2021080634W WO2022096571A1 WO 2022096571 A1 WO2022096571 A1 WO 2022096571A1 EP 2021080634 W EP2021080634 W EP 2021080634W WO 2022096571 A1 WO2022096571 A1 WO 2022096571A1
Authority
WO
WIPO (PCT)
Prior art keywords
array
photovoltaic modules
numeric
performance
solar
Prior art date
Application number
PCT/EP2021/080634
Other languages
English (en)
Inventor
Ian Humphery-Smith
Original Assignee
Lannesolaire Sarl
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Lannesolaire Sarl filed Critical Lannesolaire Sarl
Publication of WO2022096571A1 publication Critical patent/WO2022096571A1/fr

Links

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02S—GENERATION OF ELECTRIC POWER BY CONVERSION OF INFRARED RADIATION, VISIBLE LIGHT OR ULTRAVIOLET LIGHT, e.g. USING PHOTOVOLTAIC [PV] MODULES
    • H02S50/00—Monitoring or testing of PV systems, e.g. load balancing or fault identification
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/50—Photovoltaic [PV] energy

Definitions

  • This disclosure relates to methods and systems for monitoring of photovoltaic modules (also referred to as “solar panels” or “solar energy generators”).
  • photovoltaic modules also referred to as “solar panels” or “solar energy generators”.
  • Solar energy facilities are designed to produce renewable energy by harnessing the inherent energy given off by the sun’s rays.
  • Arrays of photovoltaic modules are employed to convert this energy into electrical energy, which may then be supplied to the national grid, used to power a private region or enterprise, or stored by suitable means.
  • many present solar energy facilities, or solar energy generators 100 comprise collections of solar modules or panels 102, 104 connected in string series to a combiner box 106, 108, which is connected to a DC/AC inverter 110, 112, before being exported to a grid or to storage, typically via arrangements of transformers and/or switchgears 114.
  • This string series arrangement presents problems; for instance, if one panel in the string experiences a drop in power production, then the performance of each of the other panels in the same string will fall to a newly-established minimal energy production level, or “lowest common denominator”. Deploying panels in string series also necessarily involves working with higher voltages, increasing the risk to human operatives and the risk of arc fires occurring.
  • String or central inverter faults can lead to major system shutdowns and significant losses in power production until repaired or replaced by a technician.
  • inverter failure is by far the most commonly-recorded fault type.
  • string or central inverters must be replaced regularly (every 5-10 years) and have a cost which is high, relative to other operations and maintenance expenses during systems lifespans.
  • inverter tripping can be caused by any number of module and cabling issues within a solar energy facility, which in turn causes shutdown of energy production for varying periods. In spite of this, it is still conventional across the solar industry today to rely on such string series arrangements, especially for particularly large-scale Solar Energy Facilities.
  • Solar energy facility operators who desire to find and remedy cases of under-performance may visually inspect the panels (either by eye, or by using drone-mounted cameras flown over the facility, for example) to find faults, but often the faults which are discovered by this visual inspection are too far advanced to address easily and should desirably have been addressed much earlier. That is, by the time that a panel is so badly damaged that the damage is visible to an observer, it is often too late to fix it. There is little utility in being able to detect instances of under-performing panels long after the opportunity to correct those under-performing panels has passed (and/or after the manufacturer guarantee or warranty has expired). There is therefore a need for a method by which an operator may detect such issues early, and thereby expand the useful life of the solar energy facility.
  • a monitored statistic might be the power output from one string of panels.
  • An operator might notice that there is something wrong with production when the power produced by the facility (or a given string of panels) drops off in magnitude, but it is not always possible to determine precisely what has gone wrong, or whereabouts in the facility the fault(s) has occurred.
  • photographs including infra-red photographs or electro- fluorescent images
  • micro-inverters have emerged as a new technological means of transforming the direct current (DC) electricity produced by solar panels into alternating current (AC) at the output point of the panel itself, so that these AC outputs can be connected in parallel, rather than in series, before being exported to storage or to the grid.
  • DC direct current
  • AC alternating current
  • These microinverters are generally configured to each couple to a single solar panel; however, some models are known which can be coupled to more than one solar panel.
  • micro-inverters may be built into the solar panels themselves, rather than being fitted as independent pieces of equipment. Micro-inverters can help to optimise performance at the level of the individual panel (or panels) to which they are coupled, by employing maximum power point tracking algorithms on a per-panel basis.
  • Micro-inverters also have a secondary effect - namely, they are able to measure (or compute) and subsequently transmit one or more performance metrics, such as power, for their associated panel (or for each of their associated panels) as an output signal via a suitable medium.
  • performance metrics such as power
  • these microinverters have typically seen use in domestic or very small-scale solar installations.
  • Existing medium- or large-scale solar energy facilities continue to employ string series arrangements, and thus lack high-granularity monitoring capabilities (at best, the overall performance of a unitary “block” of panels can be monitored).
  • testing the performance of panels in a solar energy facility to locate and repair faults currently comprises sending technicians into the field to travel around a facility and test the panel outputs manually, which is an undesirably slow method; the panels may be spread around a very wide geographical area, particularly in facilities intended to power a national utility grid, so the ability to quickly determine the location of one or more faulty (or otherwise underperforming) panels matters hugely.
  • An individual string series can vary from 4 to 72 panels and occasionally many more, and, in turn, multiple strings can be joined in a string connector or combiner box; it is therefore not uncommon for several hundred or even several thousand solar panels to be interconnected in series in some facilities.
  • each individual solar panel may contain many (in some examples, between about 60 and about 96) solar cells and inter-cell junctions, each with the potential to fail and thus reduce that panel’s energy production capacity.
  • these large- scale string series arrangements represent a considerable potential for fault occurrence over the lifetime of a typical facility (i.e. two or three decades).
  • Finding the actual locations of faults (or impending faults) which may affect a particular solar panel in these agglomerations remains a major technical challenge.
  • the challenge in accurately locating solar panels having faults (or impending faults) is further exacerbated when multiple faults occur concurrently within the same series, or within the same facility.
  • Figure 2 depicts an example of an unstable substrate (such as a landfill site) undergoing differential settlement.
  • the solar energy facility 204 on the left of the figure represents the ideal orientation of solar panels for solar radiation 200 incident thereon, and the solar energy facility 206 on the right of the figure represents what may happen over time when the solar panels are arranged on an unstable substrate.
  • Landfill is unstable for several reasons - heterogeny of the waste material (horizontally, vertically and during the period of waste deposition), with different components of waste breaking down at different rates; the possibility of excess water percolation into the site; internal leachate pressures; engineering of landfill prior to waste deposition; pluviometry and resultant erosion; subterranean accumulations and flows of liquids and gases; different tensile properties and compressibilities of waste materials; variations in particle size of waste components - all of these factors play a part in affecting how the landfill settles. Substrates may remain highly unstable and unpredictable even after three to five decades of being left to settle; moreover, depending on the particular geotechnics at work, a substrate might settle down, well up, undergo twisting or shearing forces, or remain level.
  • This disclosure relates to a computer-implemented method for monitoring an array of photovoltaic modules, the method comprising: receiving a performance metric for each photovoltaic module in the array of photovoltaic modules; computing a numeric array based on the performance metrics, the numeric array comprising, at each array position, a relative performance score for one of the plurality of photovoltaic modules located at the corresponding position within the array of photovoltaic modules; wherein the relative performance score for the one of the plurality of photovoltaic modules is computed with respect to the performance metrics of a subset of the plurality of photovoltaic modules; and providing an output based on the numeric array.
  • providing an output based on the numeric array may comprise generating an image from the numeric array by using the relative performance score at each array position and adjacent array positions to assign a value for a pixel or group of pixels at the corresponding position in the image and displaying said image.
  • Displaying the image may comprise displaying the image as either a synthetic grey-scale or a synthetic colour pixelated image.
  • the method may comprise receiving a plurality of sequentially acquired performance metrics for each photovoltaic module in the array of photovoltaic modules and computing a plurality of numeric arrays based on the plurality of sequentially acquired performance metrics, wherein providing an output based on the numeric array comprises generating a plurality of images from the plurality of numeric arrays and displaying said plurality of images as an animation.
  • the probable cause for the one or more underperforming photovoltaic modules may be identified based on one or more of: a visual form of shapes detected within an image generated using the numeric array; a relative pixel intensity comprised of pixels corresponding to underperforming modules within an image generated using the numeric array; and the consistency or intermittent nature of shapes and spots present in an image generated using the numeric array.
  • the probable cause for the one or more underperforming photovoltaic modules may be identified based on the consistency or intermittent nature of shapes and spots present in a series of images generated over time using the numeric arrays.
  • Providing an output based on the numeric array may further comprise prompting a user to repair, maintain, replace or adjust a component of the solar energy facility in order to improve a performance of the one or more underperforming photovoltaic modules at known locations within a solar energy facility.
  • the method may comprise receiving an input from the user after providing said output; and adjusting how underperforming photovoltaic modules are detected and/or predicted based on the received input.
  • providing an output based on the numeric array may further comprise analysing the numeric array to predict future underperformance of one or more photovoltaic modules.
  • computing the relative performance scores may comprise, for each photovoltaic module, calculating an average value of the performance metric over a given sliding window of time. In embodiments where a plurality of numeric arrays are computed, this sliding window average may be calculated, for example, once a second, or several times per second.
  • the measured or computed performance metrics may be received for each photovoltaic module in the array of photovoltaic modules in parallel or within a predetermined time.
  • the subset of the plurality of photovoltaic modules may comprise photovoltaic modules having a performance metric less than or equal to a predetermined performance metric.
  • the predetermined performance metric may be determined based on a plurality of well-performing photovoltaic modules during a predetermined period of time and may restrict the effect of variation in the level of solar irradiation observed during said period of time.
  • the plurality of well-performing photovoltaic modules may have a predetermined variation in the level of solar irradiation.
  • the predetermined period of time might be a value of seconds in the range 1 to 60 seconds, a value of minutes in the range 1 to 60 minutes, or a value of hours in a range 1 to 24 hours.
  • the subset may consist of the worstperforming X% of the plurality of photovoltaic modules in terms of absolute performance metric over a given period, for some value of X.
  • This percentage may be in the range of 0 to 20%, 0 to 10%, O to 5%, or O to 2.5%, including, at least, 0.5%, 1.0%, 1.5%, 2.0%, 2.5%, 3.0%, 3.5%, 4.0%, 4.5%, 5.0%, 5.5%, 6.0%, 6.5%, 7.0%, 7.5%, 8.0%, 8.5%, 9.0%, 9.5%, and 10.0%.
  • Computing the relative scores may further comprise mapping each performance metric of the subset of the plurality of photovoltaic modules to an integer value on a negatively-skewed linear scale, wherein the values of the linear scale range uniformly from a highest integer, corresponding to the lowest performance metric value recorded for said subset, to a lowest integer, corresponding to the highest performance metric value recorded for said subset.
  • the method may comprise receiving a plurality of sequentially acquired performance metrics for each photovoltaic module in the array of photovoltaic modules; and computing a plurality of numeric arrays based on the plurality of sequentially acquired performance metrics; wherein computing the numeric array further comprises generating stacked arrays by averaging and/or summing a plurality of numeric arrays representing performance metric values measured or computed over a predetermined period of time.
  • each numeric array and/or each image may be associated with one or more items of metadata relating to the measured or computed performance metrics from which the numeric array or image is derived.
  • This disclosure also relates to a monitoring system for a solar energy generator, the system comprising: an array of photovoltaic modules; a plurality of measuring devices coupled to the photovoltaic modules and configured to measure or compute a performance metric for each photovoltaic module of the array of photovoltaic modules; and a processor configured to perform the method of any preceding embodiment.
  • the measuring devices may comprise one or more of a microinverter, a DC-current measuring device, and a temperature sensor. [024] In one embodiment, the measuring devices may be configured to continuously measure or compute the performance metrics at regular intervals.
  • the system may comprise a surge protection and/or automatic shutdown system.
  • the array of photovoltaic modules may comprise a plurality of solar energy generators geographically distributed, wherein each solar energy generator comprises a plurality of photovoltaic modules.
  • At least a portion of the plurality of photovoltaic modules may be situated on an unstable geological substrate, the unstable geological substrate comprising land which is prone to flooding, land which has a substantially high sand, loam and/or silt content, or a landfill site.
  • the measuring devices inclusive of micro-inverters, DC/DC converters, temperature sensors (i.e. entities capable of measuring electrical parameters or temperature), and contact-less measuring devices, may be contained within a solar panel junction box and may be coupled therein with one or more of: an aluminium electrolytic capacitor; a battery-less connectivity module; a Bluetooth low energy transmission module; and a 3G, 4G and/or 5G telecommunications module.
  • This disclosure also relates to a non-transitory computer-readable medium comprising instructions which, when executed on a computer, will cause the computer to carry out the method of any preceding embodiment.
  • This disclosure also relates to the production of a standardised linear comparator of relative energy production efficiency for individual solar panels that can be manipulated mathematically and function independently of sunshine levels that vary constantly due to weather, time of day and season and that can operate over short (for example, milliseconds) or extended (for example, decades) periods of time via the computer-implemented method of any of the appended claims.
  • the claimed systems and methods When deployed in conjunction with micro- inverters, the claimed systems and methods obviate the need for string and central inverters in solar installations and afford operational systems resilience to solar power production due to inverter redundancy at the level of individual solar panels.
  • Figure 1 illustrates a traditional configuration employed in some solar energy facilities, in which solar panels are mounted in a string series arrangement
  • Figure 2 illustrates an example of differential settlement in an unstable substrate negatively impacting the ability of panels in an array to harvest solar energy
  • Figure 3 illustrates an arrangement of photovoltaic modules and solar micro-inverters, DC/DC converters, temperature or other sensors in accordance with an embodiment of the invention
  • Figure 4 illustrates an example of a set of performance metrics being used to compute a set of relative performance scores in accordance with an embodiment of the invention
  • Figure 5 illustrates an example of a portion of a numeric array of performance scores in accordance with an embodiment of the invention
  • Figure 6 illustrates a sequence of steps of a computer-implemented method for monitoring an array of photovoltaic modules in accordance with an embodiment of the invention
  • Figure 7 illustrates a computer-generated image representing several faulty or underperforming solar panels in accordance with an embodiment of the invention, where the ‘Sea of White Squares’ corresponds to those modules that have performed well during the period of analysis or averaged analyses and the accompanying ‘grey-scale’ (bottom left) corresponds to degrees of underperformance or frequency of fault occurrence;
  • Figure 8 illustrates a computer-generated image representing a rectangular array of intermittently under-performing solar panels caused by abrasions and moisture infiltration into cables connecting the affected panels in accordance with an embodiment of the invention, where the ‘Sea of White Squares’ corresponds to those modules that have performed well during the period of analysis or averaged analyses and the accompanying ‘grey-scale’ (bottom left) corresponds to degrees of underperformance or frequency of fault occurrence;
  • Figure 9 illustrates a computer-generated image which may be indicative of a region in a solar energy facility having experienced differential settlement in accordance with an embodiment of the invention, where the ‘Sea of White Squares’ corresponds to those modules that have performed well during the period of analysis or averaged analyses, and the accompanying ‘grey-scale’ (bottom left) corresponds to degrees of underperformance or frequency of fault occurrence;
  • Figure 10 illustrates a computer-generated image which may be indicative of differential settlement affecting a set of panels mounted on a single panel support system in accordance with an embodiment of the invention, where the ‘Sea of White Squares’ correspond to those modules that have performed well during the period of analysis or averaged analyses, and the accompanying ‘grey-scale’ (bottom left) corresponds to degrees of underperformance or frequency of fault occurrence;
  • Figure 11 illustrates an example of a graphical user interface of a computer system generating an output comprising a prompt for a user to rectify a solar panel fault in accordance with an embodiment of the present invention
  • Figure 12 depicts a simplified network of interconnected solar energy facility sites to illustrates how analysis, data and trends conducted, gathered or identified at a particular solar energy facility (small-scale domestic or large-scale commercial) may be used to improve monitoring capabilities at other solar energy facilities in accordance with an embodiment of the invention.
  • Figure 13 illustrates a scree plot showing the results of principal component analysis of solar panel performance data in accordance with an embodiment of the invention.
  • This disclosure presents systems and methods for solar panel array performance monitoring within solar energy facilities which enable and facilitate the detection and/or prediction of faults and performance issues which may affect photovoltaic modules or other important equipment.
  • FIG. 3 illustrates a solar energy facility 300 comprising a plurality of solar panels 302 arranged in a 2-dimensional (2D) array in accordance with an embodiment of the invention.
  • the array may be square or rectangular, or it might be a more unconventional shape.
  • a solar energy facility may comprise only a single array of panels 302, or there might be multiple arrays in the same facility.
  • solar panels 302 are precisely arranged in an ordered grid of rows and columns, but it will be appreciated that they need not be in such an exact arrangement.
  • the number of solar panels 302 in an array is at least one hundred, and could be in the order of thousands, or hundreds of thousands.
  • the solar energy facility may comprise single-axis and/or dual-axis solar tracking systems for the solar panels 302.
  • the panels 302 are fixed-angle installations.
  • the panels 302 may be ground-mounted, or mounted on water or on roofing of any kind, including those deployed for shade creation.
  • FIG. 3 also illustrates a plurality of solar micro-inverters 304.
  • one microinverter is coupled to each solar panel.
  • the (alternating-current) electrical outputs from the micro-inverters are connected 306 in parallel and exported to the grid via a transformer, to storage, or to some other suitable use 310.
  • outputs from the micro-inverters 304 representative of a performance metric measured or computed by the micro-inverter for its associated solar panel are transmitted to a software system 312 of the solar energy facility 300.
  • the performance metric is a measurement of current and/or power, a metric which is frequently facilitated by solar micro-inverters.
  • the performance metric or metrics could be any one or several of voltage, resistance, power, maximum power point, current frequency/amplitude, waveform or leakage or any other relevant metric. It should be noted that not all of the metrics in the above list can be directly “measured” per se, for instance the maximum power point metric, which is why the term “measured or computed” will be employed herein.
  • Figure 3 depicts an embodiment in which there are wired connections 308 between the micro-inverters 304 and the software system 312, those skilled in the art will recognise that the data values measured or computed by the micro-inverters 304 may be transmitted to the software system 312 by any suitable means, including wired or wireless connections.
  • the micro-inverters 304 may be omitted and the one or more performance metrics transmitted to the software system for each of the solar panels may be measured or computed by other measuring means or devices, including DC-current measuring devices, DC/DC converters (DC power optimisations) and temperature sensors.
  • each micro-inverter is coupled to (and configured to measure and/or compute one or more performance metrics for) one solar panel, but in other embodiments micro-inverters may be coupled to 2, 4 or 6 solar panels each, by way of non-limiting examples.
  • the measuring devices may comprise any of the abovementioned measuring devices (i.e. entities capable of measuring electrical parameters or temperature), and contactless measuring devices, contained within a solar panel junction box (e.g. a single junction box, or a plurality of junction boxes), and the measuring device may be coupled in the solar panel junction box with one or more of: an aluminium electrolytic capacitor; a battery-less connectivity module; a Bluetooth low energy transmission module; and a 3G, 4G and/or 5G telecommunications module.
  • the measuring devices may also include a separation of measuring and transmission tasks.
  • the solar micro-inverters 304 may be built-in to the panels or solar panel junction box, or may be separate entities.
  • the solar micro-inverters may be configured to convert DC to AC on an individual panel basis.
  • the micro-inverters 304 are connected in parallel, such that the majority of electrical cabling throughout the solar energy facility 300 carries a significantly lower voltage (in some embodiments, less than 250V) than would be carried in the case of a DC string series connection, particularly for a series comprising a large number of solar panels. Power transmission in the present invention is therefore achieved with a much greater degree of safety than that achieved by string series arrangements.
  • the solar micro-inverters may use maximum power point tracking (MPPT) algorithms to optimise power production from solar energy on a per-individual-panel basis.
  • MPPT maximum power point tracking
  • MPPT MPPT-inverter
  • computations may be carried out by one or more processors in the solar micro-inverters; an MPPT algorithm is one example of such a computation.
  • absolute values of one or more performance metrics may be computed by one or more processors in the solar micro-inverters.
  • relative performance scores may be computed by one or more processors in the solar micro-inverters.
  • some or all of the above computations may be carried out by a software system in the solar energy facility, which will be discussed in greater detail below.
  • the array of solar panels, or at least a portion thereof may be situated on an unstable substrate.
  • This may be a disused landfill site, or a low-lying area of land which is susceptible to flooding, or an area having soil which is rich in sand, loam and/or silt.
  • the landfill may in some embodiments be a capped landfill site, i.e. one having a pollutioncontainment geomembrane placed on the surface.
  • the performance metric used in the computer-implemented method is computed by using a “sliding window” to calculate, for each solar panel, the average output signal over the previous X units of time for some fixed X.
  • the performance metric for any given panel may comprise the average value of that panel’s measured output over the last five minutes.
  • performance metrics are measured and/or transmitted at least once per second, optionally four or more times per second, and optionally many more than four times per second and may have a high level of sensitivity.
  • performance metric measurements for a solar panel might be based on sliding windows which overlap or partially overlap; embodiments are also contemplated in which performance metrics are computed based on the averages of measurements taken over non-overlapping windows of time. It will be appreciated by those skilled in the art that the benefit of the invention may still be realised even in the event that sliding window averages are not employed at all.
  • continuous performance monitoring of the solar panels may be carried over a long time period, consequently generating and storing large volumes of data for future analysis.
  • performance measurements (which may be cumulative performance measurements) are taken at least once a day, preferably at least once every 3 seconds, and most preferably many times per second (e.g. 10, 20, 30, 40, 50, 60 or more times a second).
  • the measurements are sensitive to several decimal places for milliamp and/or Watt measurements.
  • the measurements are sent to the software system in real time as they are recorded. Embodiments are contemplated wherein measurements are stored on the micro-inverter devices before transmission to the software system, and/or wherein the software system polls the micro-inverters.
  • the solar micro-inverters may be configured to continuously measure and/or compute their associated performance metric or metrics at regular intervals.
  • the measuring and/or computing of the performance metric or metrics by the measuring device may occur independently of their transmission to the software system, since there may exist energy-related advantages in separating signal recording from signal transmission for long-term systems durability.
  • the solar micro-inverters may be configured to transmit the measured and/or computed performance metric or metrics at regular intervals and / or as a function of availability of power generated by the SEF, as the latter undergoes significant diurnal variation.
  • the software system may be configured to continuously receive measured or computed performance metric data from the micro-inverters or other hardware. In some embodiments, the software system may collect this data by polling the micro-inverters.
  • the software system in some embodiments, is configured to store, recover, visualise and/or display (via a graphical user interface) data representing values of the measured or computed performance metric for every panel in an array or for every panel in the solar energy facility, optionally coupled with local measures of e.g. weather.
  • the absolute data values may be conveyed from the solar micro- inverters to the software system via a local area network employing Ethernet cables, internet-of-things and/or smartphone communication technologies such as 3G, 4G or 5G, by way of non-limiting examples.
  • the software system may use one or more hard disk drives for data collection and storage.
  • the software system may be linked to a standalone data collection and/or storage facility.
  • a relational database or any other kind of database may be employed in connection with some embodiments, as will be understood by those having ordinary skill in the art.
  • the absolute empirical values of the measured or computed performance metrics are processed as relative measurements with respect to the performance levels of the other solar panels.
  • the absolute panel performance data is transformed into a plurality of relative solar panel performance scores.
  • the relative performance scores are transformed onto a uniform scale, for example, the range of integer values from 0 to 255.
  • a subset of the solar panels in the array may be identified, and used to determine the performance scores.
  • the identified subset is a subset of the solar panels having the lowest measured or computed absolute performance values of all those in the array.
  • Such a subset may be identified by, for example, taking the worst-performing X% of panels in the array, for some value X, going by the measured or computed performance metric.
  • Figure 4 which represents an illustrative and non-limiting example, the worstperforming 2.5% of panels in the array have been identified as the subset, these panels being likely to be associated with statistical significance and distinct from a much larger population of well-performing panels.
  • this threshold could be greater or lower, depending upon a range of factors, without departing from the scope of the present invention.
  • the subset may comprise up to the worst-performing 15% of panels, such as the worst-performing 1 %, 2%, 3%, 4% or 5% of panels.
  • the threshold may be manually altered if the solar energy facility contains less than a particular quantity of panels (e.g. less than 1000).
  • the subset may be identified by selecting all the panels whose output is more than a predetermined number of standard deviations lower than the mean of the population.
  • those panels which are not members of the subset may be assigned “default” relative performance scores by the transformation, to ensure that it is well-defined over its whole domain.
  • FIG. 4 depicts, in the uppermost row 400, a collection of absolute values of measured or computed performance metrics. The row is shown as being truncated, but it can be assumed that it contains 320 entries, one for each of the 320 solar panels in an array (this relatively small number being chosen purely for ease of illustration).
  • the row below 402 depicts the result of extracting the lowest 2.5% of these absolute values (i.e. the values pertaining to the worst-performing 2.5% of the solar panels in the array).
  • each of the values in the second row is taken from somewhere within the first row, as this first step is merely a process of selection, filtering, stratification or extraction. The latter process dramatically reduces dataset complexity of otherwise similarly-behaving information (i.e. acceptable levels of module performance).
  • a linear transformation is then applied to values in the second row, so as to map the lowest of the eight absolute values to the highest possible relative score (in this case, 255), and to map the highest of the eight absolute values to the lowest possible relative score (in this case, 0).
  • the values in the third row 404 may in some embodiments be rounded in order to obtain the “final” relative performance scores. Whilst this is not necessary to achieve the benefit of the present invention, rounding the values (for instance, to integers) may be particularly useful when employed alongside other optional aspects of the present invention, such as the generation of images from arrays of the relative performance scores.
  • the values are being rounded to the closest or nearest integer; however, embodiments are contemplated in which values may be rounded up to the next integer (the mathematical “ceiling” function) or down to the next integer (the mathematical “floor” function).
  • the fourth row 406 illustrates final (round) values.
  • the other 312 absolute performance metric values from the first row 400 not appearing in the second row 402 will be assigned a default value of 0 by the transformation.
  • Embodiments of the present invention may comprise methods and systems which rely upon analysis of relative performance score data, either by computational techniques or by a user being provided with images or visualisations as output, to detect and predict actual and/or potential faults affecting solar panels and other equipment in and around the solar energy facility. Focusing on a particularly poorly-performing subset of the panels, and thereby limiting the analysis which must be carried out to examine only the scores from the very worst- performing panels, makes such analysis easier, more efficient, conceptually simpler, and vastly more effective. For instance, computer-implemented analysis based on artificial intelligence and machine learning techniques (such as those discussed below) will be computationally faster and more straightforward, because there is a smaller volume of data to work with, and also a lower quantity of noise in that data.
  • underperformance what is generally meant herein is a comparative drop in performance relative to the correctly-behaving population of panels (for instance, the best-performing 97.5% of panels in the array), which effectively operate as a kind of yardstick or control group against which performance may be tracked.
  • relative scores may be computed based on an ordered ranking of the solar panels’ performance metrics, to give just one example.
  • Comparisons of data can be made across different seasons, latitudes, levels of elevations above sea level, and so forth. In this way, sets of measurements can be compared numerically on the same unified performance scale even though their absolute values may vary greatly, and might otherwise not be meaningful when placed side-by-side. Were one only to consider absolute objective measurements, it may be more difficult to analytically identify evolving patterns over time, because some data (e.g. values representing solar panel performance early in the morning, late in the evening, during the winter, or during adverse weather) would give rise to a considerably different-looking set of values to those recorded at noon on a sunny day.
  • collections of relative performance scores for solar panels in the array may be stored or represented in the form of one or more numeric arrays, in order to facilitate analysis.
  • a numeric array may represent a “snapshot” of relative performance scores for each solar panel in the array; that is, it may comprise a collection of relative scores generated from absolute performance data values which were measured, computed and/or transmitted at the same or approximately the same point in time.
  • the relative scores in a numeric array may be derived from a set of absolute values which stem from sliding-window averages, as described above, where the start and end points of the sliding windows are the same or approximately the same.
  • numeric arrays may be generated by summing or averaging other numeric arrays, as described in more detail below.
  • array is used here in the general sense of a two-dimensional data structure, and is not intended to limit the invention to embodiments whose implementation involves the use of an “array” within the particular meaning used in the context of any specific programming language.
  • array may be employed in order to implement the present invention, without departing from its scope.
  • Positions within the one or more computed numeric arrays correspond to positions in the array of solar panels in the solar energy facility, and vice versa.
  • a mapping may be established between the actual physical locations on the ground of solar panels (noting that, since solar panels perform at their best when given an unobscured view of the sky, two coordinates are sufficient to represent panel positions) and each numeric array.
  • panels that are adjacent in real life will have their relative performance scores appear juxtaposed on each generated numeric array, and, likewise, scores at adjacent positions in each generated numeric array correspond to adjacent solar panels in the field.
  • the array of solar panels be arranged in a rectangular formation, or aligned in a perfectly accurate grid arrangement, provided that it can be established which solar panels in the physical array correspond to particular relative scores in the numeric array.
  • the mapping between solar panels and numeric array positions may be a logical or simplified mapping, rather than one based purely on geographical truth.
  • a plurality of different solar panel arrays (which might be located at different sites or sub-sites) may be represented in one single numeric array.
  • each numeric array may be associated with a set of contextual metadata relating to the performance metrics or snapshot in time from which the array or image is derived.
  • the metadata may include auxiliary details such as ambient temperature; wind strength; saturation deficit; rainfall; ages of solar panels; sub-contractor(s) responsible for installation of panel supports, solar panels, cabling or connector boxes; dates of installation; panel types; and/or panel geographical location data. Any or all of this metadata may be used to gain additional insight into the data reflected in the numeric arrays, and thereby to enable more accurate and reliable fault detection and prediction capabilities, as will be discussed below.
  • a “stacked” numeric array may be computed, by computing the element-wise sum of a plurality of other numeric arrays.
  • these other numeric arrays may each be associated with sliding windows of time, as previously discussed.
  • a numeric array may be computed by taking an element-wise average of a plurality of other numeric arrays. For example, a numeric array may be obtained by computing the average of all of the numeric arrays of relative performance scores for a particular solar array over the course of the previous five years, which would yield an array reflecting the “overall” long-term relative performances of the solar panels contained therein.
  • panels may be represented based on the percentage of the total sum (from the stacked image) that they represent. For instance, each solar module may be represented visually or numerically as a percentage of the total grey-scale intensities measured for each and every module over the exemplar five-year period and visualised in a stacked composite image. Stacked or averaged arrays may be used as input for computer- implemented analysis techniques for fault detection and/or for fault prediction (such as those described below).
  • stacked or averaged arrays may be output visually for display to a user (also described below) as a single composite image or an animation of each sliding window contained therein and over time.
  • individual arrays representing instantaneous “snapshots” or representing data pertaining to a narrow timeframe may be very useful for identifying instantaneous or intermittent faults
  • stacked or averaged numeric arrays are highly-suited for the identification of panels having a level of performance which is never catastrophic, but may be consistently and continuously sub-optimal (for instance, a panel which is partially misaligned as a consequence of differential settlement of the ground beneath it) or intermittently under-performing.
  • Stacking the numeric arrays can help to filter out or highlight the effect of brief “hiccups” over time to reveal the panels or groups of panels which are afflicted with the most serious long-term power production issues or those that underperform intermittently.
  • the subset of worst-performing solar panels in any given array is unlikely to remain constant over time, and faults may arise due to all manner of causes; these may be intermittent, occur over an instant and persist thereafter, or become evident over time due to gradual aging and degradation of module performance.
  • Figure 6 depicts a computer-implemented method 600 for monitoring an array of solar panels, in accordance with an embodiment of the present invention.
  • the computer has received the performance metrics for each of the solar panels 602, and computed one or more numeric arrays of relative scores based on the performance metrics 604, an output will be provided 606, based on the content of the one or more numeric arrays that have been computed.
  • providing this output may comprise generating one or more images or visualisations based on one or more of the computed numeric arrays, and showing these to a user via suitable hardware.
  • embodiments may include the generation of composite images by assigning greyscale intensity values to pixels or groups of pixels based on corresponding relative performance scores in a numeric array.
  • pixels corresponding to the highest relative score may be assigned black, and pixels corresponding to the lowest relative score may be assigned white, representing normal performance for a solar panel.
  • Intermediate greyscale intensity values may be assigned accordingly based on the remaining relative scores in the array.
  • relative scores in an array could be used to generate an image by using the scores to assign colour properties such as the hue, saturation, brightness or alpha values of pixels or groups of pixels. Other ways of mapping relative performance scores to pixel colour values will be apparent to those skilled in the art.
  • each relative performance score of the numeric array may be used to generate an intensity value for a single pixel.
  • a relative performance score for a solar module may correspond to a group of several pixels. Embodiments are contemplated in which, by way of illustrative and non-limiting examples, each solar panel may be represented by a square of 4, 9, 16, or 25 pixels, or may be represented by a rectangular array of pixels.
  • a user may be presented with an image, representing an instantaneous snapshot, or representing the result of stacking or averaging multiple arrays and/or sliding window averages for each solar module in an array gathered over a given period of time, so that they may visually perform analysis of its content.
  • the given period of time may span several minutes, or several decades, or any magnitude therebetween.
  • the user may be presented with a plurality of such images, either concurrently or in sequence. Displaying the images in sequence may comprise presenting the images to the user one after another in succession, like individual frames of an animated film.
  • Such an animated sequence of multiple images might assist a viewer with the task of intuitively predicting when and where particular performance issues will occur in the near future, by “following the pattern” shown in the succession of images.
  • the sequence may assist the user in the prediction of faults such as those associated with improperly oriented panels caused by differential settlement of land.
  • artificial intelligence techniques may be applied to such successions of images in order to further enhance the user’s ability to extract nuanced detection or predictions of underperformance for a given dataset, namely, well-beyond the level of data complexity (series of evolving image sets) when intuitive prediction I analysis defies human intelligence.
  • FIG. 7-10 exemplary illustrations of computergenerated greyscale images are depicted.
  • the present invention enables the detection of a wide variety of fault types, including in many cases the identification of the fault “type” itself. That is, the information present in the numeric arrays and/or the computer-generated images of the present invention is sufficient to allow a human user or a computer program (such as an adequately trained machine learning application) not only to infer that a fault has occurred, or will occur imminently, but also to make precise predictions about what it is likely to be that has gone wrong, or will go wrong.
  • a human user or a computer program such as an adequately trained machine learning application
  • Figure 7 depicts varying levels of fault occurrence either constantly or intermittently and affecting several solar panels; the pattern shown in Figure 8 comprises a set of multiple under-performing solar panels in a rectangular formation, which is likely to be indicative of a fault (abrasion of interconnecting cables caused by loose panel attachments, for example) somewhere in one of the facility’s cabling or connector arrangements and accompanied by moisture ingress - this is evidenced by the lower intensity of grey-scale shown for the affected panels.
  • a fault abbreviations of interconnecting cables caused by loose panel attachments, for example
  • FIG 9 features a small number of very poorly-performing panels in a tight cluster alongside a larger number of slightly under-performing panels in an amorphous “blob” formation, all surrounded by a sea of white; this kind of pattern is likely to be a strong indication to the user that the under-performing panels have been affected by differential settlement of the underlying land.
  • Figure 10 shows a gradient of underperforming panels sloping towards a lower-right corner, which a user or the computer software will be able to associate with differential settlement affecting one particular corner of a panel support system, for instance.
  • the output provided by the computer system may comprise an indication that a fault has been detected and/or a prediction that a fault may occur at some point in the future.
  • the indication and/or prediction may comprise additional information including but not limited to the location of the fault, the relative seriousness of the fault, the probable cause of the fault, and/or the degree of statistical probability with which the indication or prediction is made.
  • such an indication or prediction may be output in addition to one or more computer-generated images such as those described above.
  • the output may comprise only one or more images.
  • the output may comprise only indications and/or predictions of suspected or imminent faults. Other possible output combinations will be apparent to those skilled in the art.
  • the output comprises an indication that a fault has occurred or a prediction that a fault may occur imminently
  • such indications or predictions may be generated by analysis carried out by a computer program.
  • these computer-implemented methods may be used to determine which panels are performing most poorly, and where in the solar energy facility these are located. Importantly, these methods facilitate the detection of potentially a multitude of underperforming modules occurring simultaneously in parallel. Alternatively or additionally, computer-implemented methods may be employed to classify the numeric arrays of relative scores based upon which type of fault or performance problem they are deemed most likely to represent.
  • providing the output comprises calculating statistics for solar panel performance, either in a given instant, or over a fixed time frame; these statistics may include values for mean, median, standard deviation, standard error, population variance or any other relevant statistic, be that parametric or non-parametric.
  • classical statistical analysis tools may be employed in order to determine the statistical significance of a detected or predicted fault, in comparison with the null hypothesis - that is, the probability that the detection or prediction is the result of pure chance, caused by sensor noise, randomness, imperfections in equipment and/or bugs or inaccuracies in computer software or hardware.
  • a variety of statistical methods may be used to analyse the one or more numeric arrays, including but not limited to analysis of variance, hidden Markov models, or Monte Carlo methods.
  • providing the output comprises using established image analysis techniques to analyse the composite images and hence find shapes, spots and forms which might be indicative of actual faults, potential future faults, differential settlement of the land underlying the array, or other problems giving rise to instances of under-performing panels.
  • edge detection within in all directions may be used to identify shapes, spots and forms, namely, a difference from one pixel to any, or all, adjacent pixels is determined using Fourier transformed data, for example, based on the two-dimensional arrays of relative scores. This may be performed to avoid the identification of a false positive, and may be used, for example, to detect land subsidence and for the area over which the ground has subsided to be mapped more accurately.
  • “Adjacent”, with respect to pixels or array positions may refer to one or more of the eight pixels or positions surrounding a given pixel or position in a grid, i.e. the pixels or positions above, below, to the left of, to the right of, above and to the left of, above and to the right of, below and to the left of, or below and to the right of the given pixel or position.
  • Other valid embodiments in which the adjacent pixels or positions are only the four orthogonally adjacent pixels or positions (i.e. above, below, left and right) or some other set of pixels or positions (e.g. for alternative pixel or position layouts, such as triangular or hexagonal grids) will be known to those skilled in the art.
  • the analysis may comprise the application of one or more machine learning techniques to an array (or several arrays) of relative performance scores.
  • These machine learning techniques may include decision trees, support vector machines, regression or neural networks including convolutional neural networks.
  • a convolutional neural network may be trained on a dataset comprising a considerable number of two-dimensional arrays of relative scores, each array being labelled with data indicative of a particular type, cause or category of fault; the neural network will learn to accurately classify new arrays of relative scores and assigning each one a fault type or category based on its numeric content.
  • the computer analysis may additionally or alternatively be configured to identify the existence and/or the geographical location of a fault from a given numeric array or image.
  • a convolutional neural network may be trained to identify or predict faults by examining a plurality of different numeric arrays or images (or animated films or sliding window averages contained therein) which represent a set of evolving snapshots of the performance of an array of panels over time.
  • a convolutional neural network may be trained on a dataset comprising a considerable number of sets of two-dimensional arrays of relative scores, each set being based on a series of absolute performance values for each solar panel in the array obtained at a set of different times.
  • Each “set of snapshots” in the dataset may further be labelled with an indication of whether or not the evolving snapshots led to a fault or other cause of underperformance (such as differential settlement of the ground).
  • the network By training the network on such a dataset, it will learn to read in a set of numeric arrays (i.e. a three-dimensional input) representing the evolving state of the array, and use it to accurately predict whether a fault somewhere in the array is imminent.
  • the labelling data may be further augmented by the inclusion of metadata related to observed fault locations, times of occurrence, and/or fault types, such that, when trained, the neural network is able to accurately predict how long it will be until the fault occurs, what kind of fault will occur, and where in the physical array of panels the fault is most likely to occur.
  • Other suitable implementations of convolutional neural networks or other machine learning models will be known to those skilled in the art.
  • the computer analysis may conduct shape detection and/or spot detection via a range of methods in order to detect and/or predict potential faults based on the one or more numeric arrays of relative performance scores.
  • the computer analysis comprises applying centroid detection, followed by one or more edge detection and/or edge propagation techniques.
  • the computer analysis performs these techniques in multiple directions over the arrays (e.g. up, down, left and right).
  • the computer analysis performs these techniques in all eight orthogonal and diagonal directions (i.e. up, down, left, right, and diagonally therebetween). This enables features of interest associated with the individual solar panels to be more easily detected against the local background of their adjacent solar panels.
  • Shapes and outlines may be represented using Fourier descriptors and deployed against Fourier-transformed numeric arrays. Additionally or alternatively, the computer-implemented steps responsible for fault detection or prediction may comprise segmenting images into similar or dissimilar regions, to find areas of interest with respect to the local background region. Some embodiments may make use of region homology, contextual algorithms, thresholding techniques, or other similar methods known to those skilled in the art.
  • the thresholding techniques may include detection of Gaussian filtered maxima.
  • the computer-implemented method may, as previously discussed, use the form of the shape or spot to infer the probable cause of a fault or predicted fault, particularly in the case of inferring types of differential settlement of unstable areas of land or cabling defects that intermittently affect groups of interconnected modules.
  • the position within the array may be used to determine where the fault is, relative to the ground or to the rest of the panels.
  • the method may use the level of grey-scale intensity with respect to the summed grey-scale intensities detected for elements of the pixelated image over time to detect or predict faults and/or to infer their probable cause.
  • the output which is generated based on the one or more computed numeric arrays of relative performance scores may comprise a prompt for the user to perform or arrange repairs, to fix, re-orient or replace a panel or other piece of equipment when the system detects a fault having occurred, and/or to preemptively carry out preventative maintenance when the system predicts that a fault will or is likely to occur soon.
  • the user may be prompted to act on the information generated by the computer-implemented analysis, for example by tightening connections, moving panels, repairing or replacing cables, module connectors or combiner boxes, or taking any other appropriate steps in response to prompts which may be given as output.
  • a display of the computer system 1100 illustrates a visual representation of the solar energy facility in the top-left 1102, in addition to a computergenerated image indicating a position of a solar panel within a particular array in the solar energy facility 1104 (bottom-left). Also depicted on the display is a set of location data pertaining to an identified solar panel fault 1106 (top-right), above the output of a computer analysis intended to identify the probable type or cause of the detected solar panel fault 1108 (for example, based on the output of a convolutional neural network trained to classify numeric arrays of relative scores or animated image sets).
  • Embodiments may include either more or less information in the prompt than that described above in association with Figure 11 , and may differ in content, without departing from the scope of the present invention.
  • the present invention is able to give rise to improved levels of long-term energy production, particularly throughout the second and third decades of a solar energy facility project. More dependable power production may be achieved, thanks to the system’s provision of an ongoing ability for operators to fix deficiencies on an individual panel basis, to improve performance over the entire project lifetime. Details such as energy production levels, rate of replacement of panels and other equipment, and necessary maintenance to the system may be reliably forecasted in advance, in order to keep energy production going for longer than would otherwise be possible.
  • the present invention is able to inform the user of the geographical location of any detected or predicted fault, technical staff gain the ability to find the fault more quickly than would be the case if, for instance, the only known information was the identification of a fault having occurred somewhere within a string of panels. Moreover, because many embodiments of the present invention are able to output details relating to the probable nature of the fault, faults can also be rectified much more quickly than previously possible. By way of illustration, consider the opportunity to send one or more technical operatives into the field equipped with the necessary tools and equipment required to fix the specific identified fault type, in contrast with having to go into the array to locate the faulty panel(s), determine the nature of the fault, and then return later on with the appropriate equipment needed to rectify the fault.
  • Those skilled in the art refer to the relevant measures of Mean Time to Detect (MTD) and Mean Time to Repair (MTR) which each contributes to the extent of lost energy production.
  • MTD Mean Time to Detect
  • MTR Mean Time to Repair
  • the present invention is designed specifically to reduce both MTD and MTR.
  • prompts to users may additionally or alternatively utilise geographical and relative panel performance data, to generate and output an optimised panel cleaning programme in order to minimise wasted time, for example.
  • an empirically observed fault type may be logged in the software system and used automatically as training data for the convolutional neural network responsible for fault detection and/or prediction.
  • a user may manually update a model according to the detected fault and the previous output of the software system.
  • FIG. 12 depicts a simplified network 1200 of interconnected solar energy facility sites1202.
  • data, patterns or trends collected and/or identified from one solar energy facility site 1204, whether automatically by software, or resulting from manual information entry or selection by a human user might be used to detect faults at another site.
  • the present invention has been applied to a single array of photovoltaic modules.
  • the array referenced herein may be formed of several solar energy generators geographically distributed, each comprising a plurality of photovoltaic modules each varying in size and number of panels contained therein.
  • the present invention can be applied to a housing estate, county or state, or even country, by combining individual solar energy facilities or generators mounted on individual houses, for example, to form one large solar energy facility collectively comprising an array of photovoltaic modules, which can be monitored using the methods described herein.
  • PCA principal component analysis
  • This application of PCA helps to establish a statistical link between such faults and the circumstances which are associated with them.
  • the first four principle components may relate to circumstances such as (for example) high rainfall, high wind speed, a long-past installation date, and a particular sub-contractor who was responsible for the installation.
  • composition “comprising” encompasses “including” as well as “consisting” e.g. a composition “comprising” X may consist exclusively of X or may include something additional e.g. X + Y.
  • the methods described herein may be performed by software in machine readable form on a tangible storage medium e.g. in the form of a computer program comprising computer program code means adapted to perform all the steps of any of the methods described herein when the program is run on a computer and where the computer program may be embodied on a computer readable medium.
  • tangible (or non-transitory) storage media include disks, hard-drives, thumb drives, memory cards, etc. and do not include propagated signals.
  • the software can be suitable for execution on a parallel processor or a serial processor such that the method steps may be carried out in any suitable order, or simultaneously. This acknowledges that firmware and software can be valuable, separately tradable commodities.
  • a remote computer may store an example of the process described as software.
  • a local or terminal computer may access the remote computer and download a part or all of the software to run the program.
  • the local computer may download pieces of the software as needed, or execute some software instructions at the local terminal and some at the remote computer (or computer network).
  • a dedicated circuit such as a DSP (Digital Signal Processor), programmable logic array, or the like.

Landscapes

  • Photovoltaic Devices (AREA)

Abstract

L'invention concerne également un procédé mis en œuvre par ordinateur pour surveiller un réseau de modules photovoltaïques. Le procédé comprend la réception d'une mesure de performance pour chaque module photovoltaïque dans le réseau de modules photovoltaïques ; le calcul d'un réseau numérique sur la base des mesures de performance, le réseau numérique comprenant, à chaque position de réseau, une note de performance relative pour l'un de la pluralité de modules photovoltaïques situés à la position correspondante à l'intérieur du réseau de modules photovoltaïques ; le score de performance relative pour l'un de la pluralité de modules photovoltaïques étant calculé par rapport aux métriques de performance d'un sous-ensemble de la pluralité de modules photovoltaïques ; et la fourniture d'une sortie sur la base du réseau numérique. Suite à la stratification initiale de données et à la manipulation de données, des analyses de données séquentielles peuvent permettre de classer les niveaux de sous-performance. Ce procédé peut utiliser un comparateur capable d'englober des millisecondes à des décennies de données et fonctionnant indépendamment des niveaux à variation constante de l'irradiation solaire.
PCT/EP2021/080634 2020-11-04 2021-11-04 Surveillance d'installation d'énergie solaire WO2022096571A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
GB2017459.5 2020-11-04
GBGB2017459.5A GB202017459D0 (en) 2020-11-04 2020-11-04 Solar energy facility monitoring

Publications (1)

Publication Number Publication Date
WO2022096571A1 true WO2022096571A1 (fr) 2022-05-12

Family

ID=73776597

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2021/080634 WO2022096571A1 (fr) 2020-11-04 2021-11-04 Surveillance d'installation d'énergie solaire

Country Status (2)

Country Link
GB (1) GB202017459D0 (fr)
WO (1) WO2022096571A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024072019A1 (fr) * 2022-09-29 2024-04-04 김동완 Procédé de détection d'anomalies d'un système de conversion de puissance parallèle hybride à l'aide d'un modèle de réseau de neurones d'apprentissage profond basé sur un auto-encodeur pour augmenter la production d'énergie

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110518880A (zh) * 2016-11-03 2019-11-29 许继集团有限公司 一种光伏电站状态诊断方法及装置
US10797639B1 (en) * 2011-07-25 2020-10-06 Clean Power Research, L.L.C. System and method for performing power utility remote consumer energy auditing with the aid of a digital computer

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10797639B1 (en) * 2011-07-25 2020-10-06 Clean Power Research, L.L.C. System and method for performing power utility remote consumer energy auditing with the aid of a digital computer
CN110518880A (zh) * 2016-11-03 2019-11-29 许继集团有限公司 一种光伏电站状态诊断方法及装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024072019A1 (fr) * 2022-09-29 2024-04-04 김동완 Procédé de détection d'anomalies d'un système de conversion de puissance parallèle hybride à l'aide d'un modèle de réseau de neurones d'apprentissage profond basé sur un auto-encodeur pour augmenter la production d'énergie

Also Published As

Publication number Publication date
GB202017459D0 (en) 2020-12-16

Similar Documents

Publication Publication Date Title
Platon et al. Online fault detection in PV systems
Li et al. A fault diagnosis method for photovoltaic arrays based on fault parameters identification
Shihavuddin et al. Image based surface damage detection of renewable energy installations using a unified deep learning approach
Di Tommaso et al. A multi-stage model based on YOLOv3 for defect detection in PV panels based on IR and visible imaging by unmanned aerial vehicle
Yurtseven et al. Sensorless fault detection method for photovoltaic systems through mapping the inherent characteristics of PV plant site: Simple and practical
Taghezouit et al. A simple and effective detection strategy using double exponential scheme for photovoltaic systems monitoring
Wang et al. Performance assessment of photovoltaic modules based on daily energy generation estimation
Dhimish et al. Novel photovoltaic hot-spotting fault detection algorithm
Livera et al. On-line failure diagnosis of grid-connected photovoltaic systems based on fuzzy logic
JP2018207595A (ja) 太陽光発電システムの異常または異常の予兆を検出するための方法及び装置
Bognár et al. An unsupervised method for identifying local PV shading based on AC power and regional irradiance data
Leloux et al. Performance to Peers (P2P): A benchmark approach to fault detections applied to photovoltaic system fleets
Braid et al. Image analysis method for quantifying snow losses on PV systems
Serrano-Luján et al. Case of study: Photovoltaic faults recognition method based on data mining techniques
Livera et al. Advanced failure detection algorithms and performance decision classification for grid-connected PV systems
Livera et al. Intelligent cloud-based monitoring and control digital twin for photovoltaic power plants
WO2022096571A1 (fr) Surveillance d'installation d'énergie solaire
Nilsson Fault detection in photovoltaic systems
Zhou et al. A shadow fault diagnosis method based on the quantitative analysis of photovoltaic output prediction error
Livera et al. Advanced diagnostic approach of failures for grid-connected photovoltaic (PV) systems
Taghezouit et al. Model-based fault detection in photovoltaic systems: A comprehensive review and avenues for enhancement
Özer et al. An approach based on deep learning methods to detect the condition of solar panels in solar power plants
Deline et al. PV fleet performance data initiative program and methodology
CN117764547A (zh) 一种光伏组串故障诊断方法及系统
Ma et al. Anomaly Detection of Mountain Photovoltaic Power Plant Based on Spectral Clustering

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21806684

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21806684

Country of ref document: EP

Kind code of ref document: A1