Proceedings of 2009 4th International Conference on Computer Science & Education

Research and Implementation of Microcomputer

Online Fault Detection of Solar Array

Huang Zhiqiang Guo Li

School of Computer Science and Technology School of Computer Science and Technology
North China Electric Power University North China Electric Power University
Baoding, Hebei Province, P.R.China Baoding, Hebei Province, P.R.China
[email protected] [email protected]

Abstract—In this article, a way of fault detection for solar TABLE I. COMMON FAULTS OF SOLAR BATTERIES
array is proposed. Using microcomputer technique, the electric
current and voltage signal of each series branch is collected by Components Failure condition
Hall sensor. The difference of maximum current value and other
values is used as the basis for determine fault. Then calculate crack
cell rust by water permeation
each branch’s deviation, the faulty solar series branch can be
degradation by ultraviolet/thermal
identified. According to voltage signal, it can determine the fault open/short circuit between cells
section in the branch. At last, failure message will be produced module
degeneration of cover/sealant materials
for maintainer’s prompt attention. (Abstract) open/short circuit between modules
wiring mistake at install
array
Index Terms—solar array, fault detection, single chip(key performance mismatch between modules
words) partial shade
There is some common fault detection methods listed in
I. INTRODUCTION Table II.
Solar energy is a kind of clean and taintless renewable
energy. The solar energy technology is very mature, and has TABLE II. FAULT DETECTION METHODS FOR SOLAR ARRAY

got wide attention in the field of energy production. The JRC of Methods Descriptions
Europe predicted that solar energy’s share in the entire energy
dark I-V measurement
structure will account for 68% in 2100. At present, in the field electrical
high frequency reaction measurement
of using solar energy, the solar power station is most extraordinary heating
concerned. Its’ emergence and promotion is bound to relieve thermal
heating of by-pass diode
the problem of energy crisis, and promote the sound visual tarnish of cells/modules
development of energy structure. Photovoltaic power
generation, known as a kind of zero-emission clean energy, In these methods, the thermal and visual methods need to
captures the sun’s energy and converts it into usable electricity. look down the PV array and observe the thermal properties
Figures show that the world PV industry’s average annual such as hot spots and the color changes of the modules. So,
growth rate has reached as high as 31.2%, and become one of those methods need to install thermal cameras or other
the most concerned and fastest-growing energy industries in the equipments in front of the array. They are costly, restricted by
world. circumstances, and unsuitable for microcomputer automated
on-line detection. Among those electrical methods listed in
The solar array is an important part in the PV system, which Table II, the dark I-V measurement method can only find the
converted solar energy into electrical energy. It consists of fault of solar array from the I-V curve but not pinpointing the
series of parallel connected solar modules. Partial faults of failed module. On the other hand, the high frequency reaction
solar array is difficult to determine and sometimes even hard to measurement method needs analysis of frequency domain and
be noticed by people. So it is necessary to use microcomputer time domain, and it is complex to achieve.
technique to detect partial faults automatically and help
maintainer timely solve the problems. The method proposed in this paper only need to measure
the voltage and current of the solar array. It is able to realize
II. ANALYSIS OF FAULT DETECTION METHODS microcomputer automated on-line detection without complex
detecting equipment.
A. Types of Fault of and Detection Methods
B. Adoptive Detection Method
By investigation and research, the factors to reduce the
solar array performance at present are listed in Table I. 1) The I-V curve of solar array

978-1-4244-3519-7/09/$25.00 ©2009 IEEE ICCSE 2009

1052 978-1-4244-3521-0/09/$25.00 ©2009 IEEE

 Proceedings of 2009 4th International Conference on Computer Science & Education

I-V curve of solar array is shown in Figure 1. The curve whether the solar series branch is in failure based on two
indicates that the solar battery is non-linear DC power supply. principles as follows:Firstly, if any series branch has a
The output current is fairly stable in most of the operating relatively low output, it can be predicated in fault state.The
voltage range. After a certain height of voltage, the current currentÿs deviation ratio will be compared with each other.
quickly fell to zero. Based on the properties of current varying Secondly, if any solar series branch has a relatively large ratio,
in different state, a new fault detection approach of the solar it can be predicated in fault state.
array, which is based on current detection, is proposed. It is
consistent with the physical properties of solar batteries. b) Fault treatment: After locating the fault branch,
microcomputer will measure the voltage value of the branch
ISC &XUUHQWVRXUFH 9ROWDJHVRXUFH and analysis the detailed part. Then failure message will be
I(A)

DUHD DUHD generated. Specific implementation is as follows:

Firstly, compare each series current from I1~IM and the
Im deviation ratio to decide the fault series K. Secondly, read half
number of the batteries’ ending voltage UK and total voltage U.
Ideally under normal condition, the solar batteries have
equation:
U1=U2=....=UM= U/2
Vm VOC
When No.1 to No. (N/2) have open circuit, short circuit or
O
U (V)
other fault, No. (N/2) +1 to No. N was in light condition, then
U1>(U/2)˗When No. (N/2)+1 to No. N has open circuit, short
Figure 1. I-V curve of solar array circuit or other fault, then U1<(U/2); so we can use the
relationship between U1 and U to judge the location of failure
2) Fault detection based on electric parameters analysis
batteries. Details refer to Table III:

TABLE III. DETERMINE THE FAULT LOCATION

Condition Location
U1> U/2 No.1~No. N/2
U1< U/2 or U1=0 No.(N/2)+1~No.N
Generated fault information as follows:

TABLE IV. FAULT INFORMATION

Fault point
No.1~No.N/2 No.(N/2) +ˍ~No.N
Fault series
1 11 12
2 21 22
... ... ...
M M1 M2
Figure 2. Schematic of solar array and detection device
III. HARDWARE DESIGN OF DETECTING SYSTEM
As shown in Figure 2, a whole solar array is composed of
M parallel branches. Each parallel branch is wired in series A. Hardware Structure of the System
with N solar batteries. Current detection devices are set in the The hardware structure of the system can be seen in Figure
array to obtain each series branch’s current from I1 to IM. 3. The industrial microcomputer consists of the main chip,
Voltage detection is also set to obtain the total voltage U and signal acquisition unit and peripheral circuit articulated by I2C
half number of the batteries’ ending voltage UK(K=1...M). First, bus. According to the circumstances, industrial microcomputer
current I1 to IM is analyzed to identify the fault branch K. can get voltage and current from solar array to analyze.
Second, voltage UK and U is compared to determine the
specific fault location.
Ideally, when there is no fault in the array, the output
current of each series branch is equal Meanwhile, same
numbers of batteries have the same voltages. Therefore,
according to a change in output current and voltage, we can
calculate which series branch does not work fine.
Figure 3. Hardware structure of detection microcomputer
a) Principle of detection:A current detection device must
be wired in each series branch. It wakes up periodically to B. Current Detection Devices
detect M series branches’ output current signal from I1 to IM Hall sensor is used to detect current. It uses semiconductor
and sends them to microcomputer, which decides the fault Hall element’s Hall Effect to realize magnetoelectricity
series branch’s location. According to related information,

978-1-4244-3521-0/09/$25.00 ©2009 IEEE 1053

 Proceedings of 2009 4th International Conference on Computer Science & Education

conversion. Compared with low-resistance current divider, Hall B. Primary Module

sensor hardly consumes energy and can realize the isolation 1) Low Output Determination
between direct current detection circuit and circuit under test.
A Criterion-referenced value is needed to decide which
Its advantage is high sensitivity, good linearity, well stability,
small volume and high temperature resistance. So, the method series branch’s output current is low. We can either select the
of using Hall sensor to detect current is adopted. maximum in measured current or standard current value
calculated by solar battery parameters and present
A current detection module, which used the Hall sensorˈ environmental parameters. But the current is linearly
was set in the each series branches to detect the current. The proportional to the sunshine intensity, and change rapidly with
Hall sensor adopted UGN3501M element and its output is it. When some parts of solar array was blocked because of
voltage UH. The equation for the relationship between current
external factors, while other parts still expose under the sun, the
and magnetic field is as follows:
calculated standard value may not reflect the situation of each
B˙nI (1) series branch accurately. On the contrary, the maximum value
In the equation: B: magnetic induction; : permeability; n: can prevent such incidents and achieve higher accuracy.
coil turns; I: measured current. For the above reasons, the system uses maximum value as
standard value. The specific implementation process is as
The equation for Hall sensor’s output voltage is: follows Suppose the solar array consists of K series branches,
UH˙KHIHBcos+KeIH (2) now measure each branch’s current I1~IK.

In the equation:KH: Hall sensitivity; : angle between plane Im㧩max{I1,̖̖,IK},

normal and B; B: magnetic induction; IH: drive current of Hall I㧩80% Im
sensor; Ke: unbalanced factor.
Comparing each branch’s value with I respectively, if less
Make =KHIHcos,=KeIHˈwe can get from (1)(2) that: than I, the branch has problems; if more than I, then this branch
is normal. Flow chart as follows:
I= U H − β (3)
αμ n
According to the linear relation between measured current
and output voltage showed in (3), we can get the measured
current’s value.
IV. SOFTWARE DESIGN OF DETECTING SYSTEM

A. Main Module

Figure 5. Operating flow chart

Figure 4. Flow chart of main module 2) Deviation Ratio Determination

Main module’s function: complete system initialization, In a certain period of time, each battery’s discharge current
gather each series branch’s current and execute subprogram of is consistent, so the current changes ƸI in this period can
concern, decide whether the current is normal. If it is in a fault reflect the performance of the battery. The ƸI smaller the
state, then turn to determine the location of fault and generate better, usually.Each branch’s current is Ink(n is measure times,
failure message to report. The current value, which is an n=1,2,3,4,5, k is number of solar series branch, a total of k
average value using one hour testing from 10:00 to 14:00, is series), measure the current value of the period 10:00 to 14:30
compared. Usually at this time, sunlight exposure evenly to the every day. Measuring five groups, each hour measured a set of
solar array, so we are able to compare series branches more data. For example,10:00 measuring each branch ÿ s current
accurately. The flow chart of main module is shown in Figure 4.

1054 978-1-4244-3521-0/09/$25.00 ©2009 IEEE

 Proceedings of 2009 4th International Conference on Computer Science & Education

I11~I1K; 10:30 measuring each branchÿs current I11ÿ~I1Kÿ,etc. When 0İSkİ5ˈthe branch is in working order;
Calculate ƸI11= I11 ÿ -I11,....,ƸI1k= I1k ÿ -I1k. It is provided in When Sk>5ˈidentified as branch failure.
Table V :
V. CONCLUSIONS
TABLE V. RECORDED CURRENT In this paper, a solar array fault detection method is
Serial number proposed. By analyzing current and voltage signal of each
1 2 ... K series branch, the faulty unit of array can be found. It is an
Time period
10:00~10:30 ƸI11 ƸI12 ƸI1K implementation of the on-line detection without additional
11:00~11:30 ƸI21 ƸI22 ƸI2K complex detection device. At the same time, it can carry out the
12:00~12:30 ƸI31 ƸI32 ƸI3K analysis of solar array so that the maintainer could timely
13:00~13:30 ƸI41 ƸI42 ƸI4K master the solar array’s operating conditions.
14:00~14:30 ƸI51 ƸI52 ƸI5K
REFERENCES
Do ˄ƸI1 ˅min=MIN{ƸI11,ĂĂ,ƸI1k } and so on. The
current evaluation function˖
5 [1] Henrik Lund. Renewable energy strategies for sustainable
Fm= ¦[Inm−(ΔIn) min˄n=1,Ă,5˗m=1,Ă,k˅
n=1 [2]
development.Energy,2007,32(6):912-919.
Goswami DY, Vijayaraghavan S, Lu S, Tamm G., “New and emerging
developments in solar energy”. Solar Energy, Vol 76, January-March
Using the minimum nonzero value Fm in the current 2004.No.1. pp 33-43.
evaluation function as Fmin, and calculating Sk by ratio of Fmin [3] POPOVIE R S,RANDJELOVIE Z. Integrated Hall-effect magnetic
sensors. Sensors and Actuators,2001(A91):46-50.
and Fk. The method is as follows:
Fmin=MIN{F1, ……,Fk},Sk=Fk/(Fmin)
The smaller Sk, the better. The ideal battery can be zero.

978-1-4244-3521-0/09/$25.00 ©2009 IEEE 1055