Vibration Data Analysis with R

R is a statistical program used to analyze large amounts of data efficiently. The language is pretty simple and well-documented online.

For example, we can graph temperature and humidity data very easily over the course of an entire experiment.

Below we have a snippet of code to load all the data files from a folder on a computer.

     ## Change this to whatever subfolder of the main directory contains the files to analyze
     folder <- "Vibration Ground Test 2-24"
     
     ## Keep character vector of all files in subfolder
     fileNames <- list.files(path = folder, pattern="MAD.....JPG")
         
      ## Create an "Images" directory if one does not already exist in the working directory.
     if(!dir.exists(paste(folder,"/Images",sep=""))){
          
          dir.create(paste(folder,"/Images",sep=""))
          
     }

And here we have code to create matrix-like objects called data frames eventually used to store vibration data.

     ## Create 2 data frames to store data from files.
     allData <- as.data.frame((1:1024)/100) ## Will store raw data values converted to integers.
     freqData <- data.frame(1) ## Will store peak frequency data for each file
     
     ## Changes column titles in  data frame.
     names(allData) <- c("Time")

Below are a graph of vibrational data with frequency analysis, as well as a table of the peak frequencies represented in a series of data files. It’s easy in R!

Driving Frequency by File Name

	MAD0044	MAD0045	MAD0046	MAD0047	MAD0048	MAD0049	MAD0050	MAD0051	MAD0052	MAD0053
Peak Freq (Hz)	10.15625	14.45312	14.45312	19.92188	19.92188	25	25	0	5.078125	5.078125

And here is the code to make a whole series of these graphs for all the data files in our data folder. As you can see, comments are preceded by ##, and <- indicates an assignment.

     ## Go file by file and ...
     for(fileName in fileNames){
          ## Convert raw values to integers and store in a dataframe
          raw <- readRaw(paste(folder,"/",fileName,sep=""))
          data <- strtoi(raw$fileRaw,16L)
          data <- (as.numeric(data) - 127.5)/127.5
          df <- as.data.frame(cbind((1:length(data))/100.0, data))
         
          ## Append newly analyzed raw data to allData data frame
          allData <- cbind(allData,as.numeric(data[1:1024]))
          names(allData)[length(names(allData))] = sub(".JPG","",fileName)
       
          ## Append new frequency data to freqData data frame and generate PNG File
          spec <- meanspec(allData[,names(allData)==sub(".JPG","",fileName)],f=200, plot=FALSE) ## Performs FFT assuming 200 Hz was the recording frequency of the data
      
          ## Create images for each data file of data plotted over time and FFT information
          png(paste(paste(folder,"/Images/",sep=""),sub(".JPG","",fileName),".png",sep=""), height = 480, width = 480, bg = "white") ## Create png file
          par(mfrow = c(2,1))
          plot(allData$Time,data[1:1024], type = "n",xlab = "Time (s)", ylab = paste(sub(".JPG","",fileName)," ACCELERATION (m/s/s)",sep = "")) ## Create plot of data
          lines(allData$Time,data[1:1024])
          
          minAmplitude <- .7
          
          peaks <- fpeaks(spec, nmax = 6, plot=TRUE, labels = FALSE, legend = FALSE, title = FALSE, main = paste("Frequencies present in ", fileName, sep="")) ## Finds peak frequencies of the data and plot them
          
          
         if(all(peaks[,2] < minAmplitude)){
               
               drivingFreqKilo = 0
               drivingFreqAmp = .95
          }
          
          else{
               
               drivingFreqKilo <- peaks[peaks[,2] == max(peaks[,2]),1]
               
               if(any(5 == round(1000*peaks[,1],0))){
                    
                    drivingFreqKilo <- peaks[5 == round(1000*peaks[,1],0),1]
                    drivingFreqAmp <- .5
               }
               
               else{
                    
                    drivingFreqAmp <- peaks[peaks[,1] == drivingFreqKilo,2]
                    
               }
               
          }
          
          
          text(.05, .85, cex = 1.2, col = "red",paste("Driving Frequency = ",round(1000*drivingFreqKilo,2), " Hz", sep = ""))
          
          dev.off()
          
          freqData <- cbind(freqData,1000*drivingFreqAmp) ## 1000*peaks is used because frequencies are given in kHz.
          names(freqData)[length(names(freqData))] = sub(".JPG","",fileName) ## Change column names of data frame to file names
     }

… and here they are!