ES2533317A9 - Automatic detection method and diagnosis of operating faults in distributed photovoltaic solar installations, based on the comparison of their energy productions - Google Patents

Abstract

Method of automatic detection and diagnosis of operations failures in distributed photovoltaic solar installations. # The method only requires the measurement of the energy production data, so that the material required by the operator of the photovoltaic installation is based on equipment able to measure the energy produced by the facilities and whose equipment can be the own energy meter of a house or the meter incorporated in the inverter of the installation itself. In any case by the entity that performs the analysis, equipment is required that allows the acquisition and analysis of the measured energy data, to carry out the sending of that data from the energy meter to the server where it is they perform the analysis themselves, so that the information is transferred and stored either by text files, or by Web services, through which communication between user and server is established automatically, the data being stored in a database and processed by a calculation server.

Description

image 1

METHOD OF AUTOMATIC AND DIAGNOSTIC DETECTION OF OPERATION FAILURES IN SOLAR PHOTOVOLTAIC FACILITIES DISTRIBUTED, BASED ON THE COMPARISON OF YOUR ENERGY PRODUCTIONS

DESCRIPTION

Object of the invention

The present invention relates to a method of automatic detection and diagnosis of operating failures in distributed photovoltaic solar installations, based on the comparison of their energy productions, requiring only the measurement of the energy production data themselves as input data.

Technical sector

The proposed prediction methodology is part of the energy sector, specifically in the production of electricity by photovoltaic energy.

Background of the invention

The photovoltaic generation supposes a paradigm shift, from the centralized generation, to a distributed generation. In particular, residential photovoltaic systems are characterized by an important geographical dispersion, climatic conditions, a wide variety of components, models, technologies, orientations and inclinations. This forces to organize the operation and maintenance phase in a very different way. When the same operator has to take care of several thousand or tens of thousands of facilities distributed geographically throughout an entire country,

or even of the whole European continent, it is impossible to carry it out by going to each site to detect breakdowns or problems of low productivity in the facilities In this sense, the following references can be cited:

-J. Leloux, L. Narvarte, D. Trebosc, Review of the performance of residential PV systems in Belgium, Renewable and Sustainable Energy Reviews, 2012.

-J. Leloux, L. Narvarte, D. Trebosc, Review of the performance of residential PV systems in France, Renewable and Sustainable Energy Reviews, 2012.

image2

-J. Leloux, L. Narvarte, D. Trebosc, Performance analysis of 10,000 residential PV systems in France and Belgium, EUPVSEC26, 2011.

Additionally, due to costs, the vast majority of these photovoltaic installations do not have monitoring systems that allow to know the incident irradiance on the panels, nor the temperature of the air and the panels. Consequently, in order to be applicable to most domestic photovolastic installations, fault detection methods can only be based on production data taken by energy meters, making it impossible to detect some failure detection methods that are applied in the context of large centralized solar plants

In this regard, the following references can be cited:

-

US2012084027A1, System and method for monitoring performance of a

photovoltaic array, 2011.

-

Martínez-Moreno, F., Lorenzo, E., Muñoz, J. and Moretón, R. (2012), On the

testing of large PV arrays. Prog. Photovolt: Res. Appl., 20: 100-105.

doi: 10.1002 / pip.1102.

During the last two years, thanks to the latest advances in the field of telecommunications technology, companies specializing in the monitoring of photovoltaic systems by smart meters equipped with telematics have been created. These companies offer a service, usually based on GPRS or Internet, at a relatively low cost. Energy production data is read directly from production meters, at intervals of one minute to one hour, which opens the doors to a whole new set of performance analysis and detection of certain types of problems in a way It was almost impossible to imagine a few years ago. These companies have a core business and technical knowledge that belongs to the field of telecommunications. Therefore, they favor the acquisition of data from the counters of numerous photovoltaic systems, but they still lack advanced methods of system analysis, and as a consequence they are unable to deduce hidden problems from the monitored data. The analyzes they offer are poor and very few accurate. They basically consist of performing performance factors that relate the instantaneous energy produced, with the instantaneous solar radiation, generally obtained from an external data provider.

image3

operation of a photovoltaic installation. In fact, the world leaders in the monitoring of photovoltaic systems use it systematically, and the following references can be cited:

5

-SolarGIS (http://solargis.info/) -Rtone (http://rtone.fr/) -ISM Solar (http://www.ismsolar.com/) -SolarEdge (http: //www.solaredge. com/)

10 -SMA Solar Technology (http://www.sma.de/) -Green Power Monitoring (http://www.greenpowermonitor.com/)

This parameter is defined as the relationship between the energy that a photovoltaic system

15 really injects the network, and the one that would inject a hypothetical ideal photovoltaic system, understood as one whose solar cells always worked at the reference temperature (25 ° C) and, otherwise, were totally free of losses:

E (T) E (T) Pp

PR ==

Pp G (T)

G (t) dt

* ∫ *

G T G

20 Where:

-E is the actual production injected into the network; -G is the real radiation that affects the photovoltaic panel; -Pp is the peak power of the installation;

25 - G * is the reference irradiance whose value is 1000 W / m2.

As a quality index for fault detection, the PR has several important drawbacks:

• Because it is its excessively general reference, it does not allow differentiating between losses

30 unavoidable (temperature, DC / AC conversion, etc.) and avoidable (breakdowns, disconnections, etc.), making it difficult to contract;

•

The PR is a direct relationship between the energy production of the installation and the radiation that affects it. It has been proven that this factor is not stable enough to detect faults in the facilities, and that

due to the accumulation of heat caused by the roofs of the houses, and that causes a decrease in the value of the PR.

•

On the other hand, by definition, PR requires knowledge of solar radiation. The solar radiation data from most suppliers generally shows significant errors, which causes the ambiguity of the calculated PR. In particular, the quality of solar radiation data taken from satellites is greatly affected by high clouds and low solar positions on the horizon, the presence of snow, or in the presence of very high clouds.

•

The solar radiation data, when required in temporary hourly or infra-hourly resolutions, implies a relatively significant annual cost in relation to the budgets available to carry out the operation and maintenance of domestic photovoltaic installations.

image4

Description of the invention

The invention consists of a method for automatic fault detection in distributed photovoltaic systems that only requires the measurement of energy production data.

More specifically, the expected method is to compare the energy production of an installation with the energy production of its neighboring facilities, where the entity responsible for the installation performs an analysis using the appropriate equipment to obtain the measured energy data , by sending said data through GPRS or over the Internet, from a resulting energy meter to the respective server where the analyzes are executed.

For this, the necessary material by the operator of the photovoltaic installation is a device capable of measuring the energy produced by the installations. This equipment can be the own energy meter of a house, or the meter built into the inverter (DC / AC converter) of the installation itself. On the part of the entity that performs the analysis, equipment is required that allows the acquisition and analysis of the measured energy data. The data is sent by GPRS or by Internet from the counter to the server where the analyzes are executed. The information can be transferred and stored, either by text files, or by Web Services through which data communication is established, and are processed by a “cloud” type calculation server, in “cloud server”.

image5

The developed method is able to detect operating problems only in the identification of abnormal variations of a productivity indicator that characterizes the proper functioning of the installation, and whose construction does not require knowledge of its operating conditions. A particular requirement, therefore, for this indicator, is its stability during normal operation of the installation. This procedure makes extensive use of spatial and temporal correlations between the energy production data of neighboring photovoltaic systems, and the resulting indicator has been referred to as the "Relative Productivity Index" (IPR), since it is based on the comparison between neighboring and similar facilities, that is, similar.

The Relative Productivity Index (IPR) consists in the relationship between the normalized production of a given installation and the standardized productions of geographically nearby installations. The objective is to relate facilities that are in the same physical and environmental conditions. Therefore, the IPR is a factor that relates the standardized production of a facility with the standardized production of facilities that have similar conditions. It is expressed as:

fF IPR = Cref fC

Where:

F

-

Cf is the capacity factor of the “focus” installation, or installation that is required

analyze.

-

ref Cf it is the reference capacity factor for the “focus” installation, built to

from the energy productions of the "like" facilities.

The capacity factor of a photovoltaic system i during a time interval δt is defined by:

i Ei (δt)

fC (δt) = Ppi ΔT

Where:

Ei (δt)

-is the energy measured in the counter during the time interval δt; -is the peak power of the system;

image6

-ΔT is the temporal resolution with which the energy produced is measured.

Once these capacity factors have been calculated, the ratios between capacity factors of a Focus installation and a similar installation can be calculated:

Focus

i Focus −Peer fC (δt)

i

ρ (δt) = ρ (δt) =

C C Peer

i

fC (δt)

On the set of available historical values, the median and standard deviation of these ratios are calculated respectively as:

-Median:

i)

µ1 / 2 (ρC

-Standard deviation:

ii image7 1 ii 2

σ = σ (ρC) = ∑ [ρC (t) −µ1 / 2 (ρC)]

N

These standard deviations are used as weighting factors to generate a reference capacity factor from the capacity factors of all similar ones, by a method that belongs to the family of geostatistical methods called "kriging":

i Focus −Peer i

λ = λ = N

∑

i

Once these weighting factors (lambda) have been calculated, the reference capacity factor is calculated by kriging using the capacity factors of each of the similar ones and the lambda weighting factors corresponding to each similar:

ref i

fC (δt) = Kriging (fC (δt), λi)

The IPR is then obtained by making the ratio of the capacity factors of the focus facility with its corresponding reference:

image8

image9

image10

IPR (δt) = C f ref (δt)

C

Once this IPR quality factor has been calculated, the automatic fault detection is carried out by establishing a fault detection threshold that corresponds to a certain variation in the instantaneous value of the IPR with respect to its median value. For this, we calculate the median and the standard deviation of the IPR, analogously to the calculations made previously on similar facilities:

-Median:

µ1 / 2 (IPRk (δt))

Standard deviation:

σ (IPR (δt)) = image11 1 ∑ [IPR (δt) −µ1 / 2 (IPR)] 2

N

The threshold set for fault detection is set in such a way that the probability of issuing a false alarm to the operator of the photovoltaic system (signal a fault when it does not exist) is only 0.3%. According to statistical theory, assuming that the dispersion of the IPR values resembles a Gaussian distribution, this threshold corresponds to a value three times the standard deviation below the median value:

IPR θ = µ1 / 2 (IPR) - 3σ (IPR)

An IPR value below this threshold results in the issuance of a fault detection warning.

Once a fault i has been detected, an energy loss ∆Ei can be attributed by the difference between the instantaneous IPR value and its median value. It is quantified by the sum of the losses corresponding to each failure i detected, ∆Ei, as:

ΔE = FP ⋅ E *

i ii

Where: -FPi is the loss factor associated with failure i, given by:

- IPR

IPR medium i

FP = i IPR

median

image12

one

* = E

Ii

1− FPi

5 As mentioned, the method of the invention is based on the comparison of the energy production of the installation with the energy productions of its neighboring facilities; making extensive use of spatial and temporal corrections belonging to the family of geostatic procedures called “Kriging”, among the energy production data of neighboring photovoltaic systems.

10 The resulting indicator has been called the “relative productivity index” (IPR), since it is based on the comparison between neighboring and similar facilities, that is, similar.

The method does not require knowledge of the incident solar radiation on the panels

15 photovoltaic, nor does it require calculating the quality index commonly used and called the Performance Ratio (PR).

Furthermore, by means of the method of the invention, the automatic establishment of fault detection thresholds is carried out by simple calculation of the standard deviation of the IPR;

20 in addition to being able to quantify the energy loss associated with each fault detected, based only on the difference between the instantaneous IPR value and its average value.

Finally, by means of the method of the invention, it is possible to measure the energy production data by incorporating smart meters equipped with a system of

25 telematics, allowing to measure the energy production data of the photovoltaic installations themselves and transmit them by GPRS from said smart meters to an Internet server, as well as store them in a database and process them by a calculation engine executed from a server Internet, the results being provided to the client by a Web server.

30

Description of the drawings

To complement the description that will then be made and in order to help a better understanding of the characteristics of the invention, the present is accompanied

The descriptive report, forming an integral part of it, a plan where the following is illustrated and not limited to:

image13

Figure 1.- Shows a graph corresponding to the calculated value of the IPR, PR and fault detection threshold with hourly data for a period of one year in a specific domestic photovoltaic installation.

Preferred embodiment of the invention

As an example, this method of automatic fault detection has been applied over a period of one year, on a photovoltaic installation located in Belgium, whose energy production data has been measured every hour. This facility has 3,944 nearby facilities that can be potential similar facilities, and of which the corresponding energy production data has also been acquired. The energy production of photovoltaic installations has been measured by smart meters equipped with a telematics system. The data has been transmitted by GPRS from the counters to an Internet server, and has been stored in a MySQL database. The data has been processed by a calculation engine programmed in PHP and the results have been provided to the client by an XML Web service.

The IPR values have been calculated at the hourly level and for one year of operation, corresponding to the energy production data of this installation, taken as a focus, with respect to its neighbors, taken as similar. Figure 1 shows the comparison between the calculated IPR values and the PR values calculated by the classical methods that characterize the state of the art. As can be seen in this figure 1, the PR shows significant fluctuations during the whole year, making it very difficult to detect and quantify malfunctions that require an accurate quality indicator. The IPR shows a much higher stability at most times of the year, and a drop in value at specific times associated with failures that can be easily identified. In the present case, it has been assessed that the energy importance of the failures detected has represented annual energy losses of 2.8%.

Claims (3)

image 1 1.-Method of automatic detection and diagnosis of operating faults in a distributed photovoltaic solar system, characterized in that it comprises: -Calculate a relative productivity index, IPR, which relates the normalized production of the distributed photovoltaic solar system to be analyzed and the standardized productions of a plurality of neighboring distributed photovoltaic solar systems, according to the following formula: Focus f (δt) IPR (δt) = C ref fC (δt) being fCF is the capacity factor of the photovoltaic solar system distributed to analyze and fCref a reference capacity factor constructed from the energy productions of neighboring distributed photovoltaic solar systems; where the capacity factor of a distributed photovoltaic solar system i during a time interval δt is defined by: Ei (δt) fCi (δt) = i Pp ΔT Ei (δt) the energy being measured in said distributed photovoltaic solar system i Ppi during the time interval δt, the peak power of said solar system distributed photovoltaic i and ΔT the temporal resolution with which energy is measured produced; and where the reference capacity factor fCref is calculated by kriging using the Peeri capacity factors fC (δt) of each of the photovoltaic solar systems distributed neighbors and the lambda weighting factors corresponding to each of them according to the following formula: ref i f (δt) = Kriging (f (δt), λi). DC - Perform automatic fault detection by establishing a fault detection threshold that corresponds to a certain variation of the instantaneous value of said relative productivity index, IPR, with respect to its median value. 2. Method of automatic detection and diagnosis of operating faults in a distributed photovoltaic solar system according to claim 1, characterized in that the fault detection threshold is set based on the standard deviation of the relative productivity index. -eleven image2 distributed photovoltaic according to any of the preceding claims, characterized by that the energy loss ΔEi associated with each fault i detected is quantified from the 5 difference between the value of the instantaneous relative productivity index IPR and its value medium IPR, according to the following formula: median ΔE = FP ⋅ E * i ii where FPi is the loss factor associated with failure i, given by: IPR - IPR medium i FP = i IPR median 10 and Ei * is the energy that would be produced by the distributed photovoltaic solar system in the absence of failure i: *one Ei = Ei .1− FPi 4.-Method of automatic detection and diagnosis of operating faults in a distributed photovoltaic solar system according to any of the preceding claims, characterized in that it comprises: -measure the energy production data of the distributed solar photovoltaic system to be analyzed and the neighboring distributed solar photovoltaic systems by means of smart meters; 20 -transmit said data from said smart meters to an Internet server; - store said data in a database and process it by a calculation engine executed from an Internet server; -Facilitate the results to the client by a Web server. -12