valve. In this paper, a technique i

valve. In this paper, a technique is proposed to detect the fault in a pneumatic control valve using an accelerometer to acquire the vibration data. The novelty of this process is that the methods generally used for vibration data analysis are being used to detect control valve faults. By placing the accelerometer on the wall of the pipe, the flflow rate can be estimated as described in [28]. It can be inferred that the fault in the control valve will lead to a change in flflow rate, which can be detected using vibration data from an accelerometer. Frequency domain data are used as the feature space for classifification. The power spectrum density (PSD) of the time domain vibration data is the aspectElectronics 2019, 8, x FOR PEER REVIEW 3 of 16 as compared to fast Fourier transform (FFT) data. These PSD values are computed for both normal and faulty systems, which in turn are combined to form the training data set. A support vector machine is used to classify between the normal and faulty data from the training set. Since this method involves a non-contact type of measurement and the detection is done using a machine learning algorithm, this gives us a method to update the detection process to include more faults or to enhance the performance later on. To the best of our knowledge this type of sensing in combination with artificial intelligence has not been carried out on a real-time system regarding control valves. After the introduction in the first section, the experimental setup used to perform the desired work is discussed, followed by a problem statement, problem solution, results and discussion and finally a conclusion. 2. Experimental Setup A flow process is considered to demonstrate the proposed algorithm. The process diagram is shown in Figure 1. In the above process, liquid (water) is pumped from the reservoir at a constant rate (Q), a control valve is placed in the path to vary flow rate at the output of the valve (Q’). A bypass path is given at the inlet of the pneumatic control valve for reverse flow. A flow valve is also placed to control the reverse flow. A control signal (4–20 mA) from the controller is given to the pneumatic control valve through an I/P (current to pressure) converter. A rotameter is placed at the outlet of a control valve to display the flow rate (Q’). By varying the control signal, the pressure from the compressor to the I/P converter can be varied and the valve position can be changed which results in a change in flow rate (Q’) at the output of the valve. Furthermore, the flow is directed back to the reservoir, and an orifice flow meter exists to monitor the flow rate via a differential pressure transmitter. Figure 2 shows a picture of the actual process. Figure 1. Flow process workflow. Before going into the working of the proposed system a brief introduction to control valves and its faults are discussed here.