I need a low to medium cost setup to measure solar energy over a long period to evaluate a site for installing small solar or hybrid power generation systems.
Best is to measure global solar irradiance on site for a longer period of time (at least one year, better 5 or 10 years). Usually a pyranometer is best for that, you may also may use a cheaper solar sensor (if it is the same technology as your future solar power plant, e.g., crystalline Silicon). If you want to use concentrating systems you should measure direct (DNI) and diffuse irradiance sperately.
The following information may be useful for your research
While a variety of instruments have been used to collect solar radiation data, the most common and reliable are broadband thermal-sensing pyranometers and pyrheliometers, typically having measurement uncertainties of less than 5 percent. The pyranometer measures the sum of direct and diffuse radiation, while the pyrheliometer measures only the direct normal component. Stations using these instruments are generally designated as Class 1.
Because of their relative robustness, better response and lower cost, simple photosensors are currently the most commonly used type of instrument. Such instruments use photovoltaic cells and thus operate over a limited portion of the solar spectrum (300-1120 nm), introducing some uncertainty for broadband applications. A rotating shadow band (RSB) instrument uses a photosensor with a motorized rotating band that periodically blocks direct sunlight from the sensor. This single instrument measures both global (beam plus diffuse) and diffuse radiation and from these two quantities direct normal can be computed. RSBs are becoming a standard instrument and data from them are generally designated as Class 2.
The satellite-based models have been improved and verified against ground-based measurements and may be used to provide solar radiation estimates at any ground location where suitable satellite imagery is available. One result of these efforts is a 10 km gridded hourly solar database for the entire U.S.9 This gridded solar radiation data provides about 9 times better resolution than the approximately 90 km spacing provided by the 89 Texas ground-based stations included in the 2005 National Solar Radiation Database. Data are available for global horizontal, direct normal, and diffuse solar radiation on a temporal basis.
The other major attribute of satellite-derived data is that, being “recent” data, it has the potential to support forecasting local solar availability, an important consideration for a solar power generation facility feeding into the grid. Earth observation satellites circle the Earth on approximately 90 minute orbits so, with rapid processing of the data, information used in solar energy forecasting need be no older than about 1½ hours. Recently, extensive research has been done on improving solar energy forecasting and the results are very promising.
Pyranometer with PIC or similar micro controller based remote monitoring system thro TCP/IP protocol are available.Details of similar project reference available in http://www.kmitl.ac.th/~kswichit/.
You would need to collect data daily over a period of at least two years (for repeatability). It might be more convenient to determine from your local Meteorological Office if they have data on sunshine hours collected from a nearby site. Looking at data for a few years would give you an appreciation of the monthly or seasonal variation in insolation which could be useful in assisting your planning. This would probably be more cost-effective than DIY measurements over a sustained period.
Instead of measuring for two years, you could use historic satellite based irradiance measurements. The SoDa database provides free data and can be found here: http://www.soda-pro.com/. It covers Europe, Africa, the middle east and a part of South America.
Both pyrheliometers and photometers are used to measure solar irradiance (insolation) at a given spot. Unless you are going into business as a solar energy consultant, these could be too expensive for your purpose. A simple way is to buy a rated solar panel, set it up at the recommended angle for your latitude, and measure the voltage and current output with an Ammeter (about $135) multimeter. An electrician or electronics technician can show you how. The panel output will vary with the time of day, the season, and the local cloud cover. The NREL website has solar energy maps that can give you an approximate daily energy insolation for your area if you live in the USA.
You can use a reference cell for measuring the solar irradiance. it measure the Isc of the reference cell and using this Isc you can measure the solar irradiance. it is very low cost as compared to pyranometers.
you can also use lux meter which is cheaper option available with data-logging. you can convert lux into w/m2 using online tools or by equation in excel data sheet.
Your question could be divided in two parts: device for measurement and prediction.
For prediction you do not need any device, because there is so many tools to do that.
For Europe, and so many of rest World, you can use free tools PV GIS: http://re.jrc.ec.europa.eu/pvg_tools/en/tools.html, it is quite reliable. For almost all Earth you can use one of the best tool - SolarGis (https://solargis.com/) only disadvantage: it is not free.
For measurement, which can help you confirm prediction, there is a lot of possibility, but one things you need consider: direct and short term measurement could lead to different result as prediction will do. Prediction is based on long data and meteorological situation. Short measurement do not need follow historical long data.