This paper presents a state-of-the-art review on machine learning (ML) based intelligent diagnostics that have been applied for partial discharge (PD) detection, localization, and pattern recognition. ML techniques, particularly those developed in the last five years, are examined and classified as conventional ML or deep learning (DL). Important features of each method, such as types of input signal, sampling rate, core methodology, and accuracy, are summarized and compared in detail. Advantages and disadvantages of different ML algorithms are discussed. Moreover, technical roadblocks preventing intelligent PD diagnostics from being applied to industry are identified, such as insufficient/imbalanced dataset, data inconsistency, and difficulties in cost-effective real-time deployment. Finally, potential solutions are proposed, and future research directions are suggested.

Condition Monitoring Based on Partial Discharge Diagnostics Using Machine Learning Methods: A Comprehensive State-of-the-Art Review

To resolve the problem of the manual analytical method adopted in water quality detection with bad real-time character, this paper introduces a novel kind of Remote Water Quality Measuring and Monitoring System based on WSN. It has used wireless sensor network based on the ZigBee to realize the water quality parameter remote probing and the real-time monitoring function. Users can observe the current or historical water quality status easily, and it provide a reasonable basis for the further breeding plan. This system has a simple architecture, and isn't confined by the geographical position. According to the test results, this system can run stably. And its operation is convenient.

The Design of the Remote Water Quality Monitoring System Based on WSN

This paper suggests an IoT based smart farming system along with an efficient prediction method called WPART based on machine learning techniques to predict crop productivity and drought for proficient decision support making in IoT based smart farming systems. The crop productivity and drought predictions is very important to the farmers and agriculture’s executives, which greatly help agriculture-affected countries around the world. Drought prediction plays a significant role in drought early warning to mitigate its impacts on crop productivity, drought prediction research aims to enhance our understanding of the physical mechanism of drought and improve predictability skill by taking full advantage of sources of predictability. In this work, an intelligent method based on the blend of a wrapper feature selection approach, and PART classification technique is proposed for crop productivity and drought predicting. Five datasets are used for estimating the proposed method. The results indicated that the projected method is robust, accurate, and precise to classify and predict crop productivity and drought in comparison with the existing techniques. From the results, the proposed method proved to be most accurate in providing drought prediction as well as the productivity of crops like Bajra, Soybean, Jowar, and Sugarcane. The WPART method attains the maximum accuracy compared to the existing supreme standard algorithms, it is obtained up to 92.51%, 96.77%, 98.04%, 96.12%, and 98.15% for the five datasets for drought classification, and crop productivity respectively. Likewise, the proposed method outperforms existing algorithms with precision, sensitivity, and F Score metrics.

An efficient IoT based smart farming system using machine learning algorithms

We have theoretically studied how the presence of a small proportion of energetic beam electrons mixed to a bulk of Maxwellian electrons in a hot plasma affects the temperature-dependent intensity ratio G = (x + y + z)/w of the helium-like triplet intercombination (x, y) and forbidden (z) lines to the singlet resonance line (w). By modelling the electron distribution function as combination of a Maxwellian isotropic component and a monoenergetic beam component, detailed calculations of the G ratio of the Ne8+ lines have been performed for temperatures Te of the Maxwellian component and kinetic energies e0 of the beam component in the ranges 106−107 K and 1.5−25 keV, respectively. A magnetic sublevel-to-magnetic sublevel collisional-radiative model has been used for determining the populations of the upper magnetic sublevels of the four lines at an electron density below 1013 cm−3. Excitations from the ground 1s2 S0 and metastable 1s2s S1 magnetic sublevels to the 1snl (n=2−4) magnetic sublevels as well as inner-shell ionization of the lithiumlike ion in its ground level were taken into account. All basic atomic data, including the radiative transition probabilities and the collisional excitation and ionization cross sections, were computed using the Flexible Atomic Code. It is found that the contribution of a 5% fraction of the beam component can reduce the G ratio by a factor of 30 at Te = 106 K and of 2.4 at Te = 3 × 106 K. Our calculations also indicate that the effect of directionality of the beam component on G is negligible for e0 above ∼10 keV and that for a given Te, G is practically insensitive to variations in e0 above ∼7 keV.

/pdf/republique-algerienne-democratique-et-populaire-ministere-de-2b83ag39cw.pdf

Republique algerienne democratique et populaire ministere de l'enseignement superieur et de la recherche scientifique

Noise invariant partial discharge classification based on convolutional neural network

Partial discharge (PD) is regarded as a precursor to plant failure and therefore, an effective indication of plant condition. Locating the source of PD before failure is key to efficient maintenance and improving reliability of power systems. This paper presents a low cost, autonomous partial discharge radiolocation mechanism to improve PD localization precision. The proposed radio frequency-based technique uses the wavelet packet transform (WPT) and machine learning ensemble methods to locate PDs. More specifically, the received signals are decomposed by the WPT and analyzed in order to identify localized PD signal patterns in the presence of noise. The regression tree algorithm, bootstrap aggregating method, and regression random forest are used to develop PD localization models based on the WPT-based PD features. The proposed PD localization scheme has been found to successfully locate PD with negligible error. Additionally, the principle of the PD location scheme has been validated using a separate test dataset. Numerical results demonstrate that the WPT-random forest PD localization scheme produced superior performance as a result of its robustness against noise.

/pdf/improving-rf-based-partial-discharge-localization-via-1o5s4yawyg.pdf

Improving RF-Based Partial Discharge Localization via Machine Learning Ensemble Method

In recent years, there has been significant increase in utilisation of embedded-microcontrollers in broad range of applications extending from commercial products to industrial process system monitoring. Furthermore, improvements in speed, size and power consumption of microcontrollers with added wireless capabilities has provided new generation of applications. These include versatile and
low cost solutions in wireless sensor networking applications such as wireless system monitoring and control.
In many applications, there are situations where multiple identical devices form a wireless network and work together towards achieving a common goal. Each individual device or node is controlled by a microcontroller and the whole network is controlled by an upper-level microcontroller to facilitate data distribution and supervisory tasks. In order to maximise the network performance the nodes are designed to control their local process in an intelligent manner so that they can adapt to their
environment and set-point changes. The knowledge gained by individual nodes is then made available
to other nodes on the network.
The aim of this research project is to implement Fuzzy Logic Controller (FLC) as an artificial intelligent technique to devise a microcontroller-based self-learning algorithm that enables real-time online adaptation to the new environment situations. The research to date has focused on machine-based FLC with extended recourses such as CPU power and storage in contrast with microcontrollers with limited resources.
In order to improve FLC performance, several techniques are available for defining its parameters such as scaling the universe of discourse (known as scaling gain), tuning membership functions and rules. Design simulations have been carried out and the latest results have been promising which indicate certain degree of improvement in system performance.

Intelligent Wireless Sensor Network

In some industrial process control systems it is desirable not to allow an overshoot beyond the setpoint or a threshold, this could be a safety constraint or the requirement of the system. This paper outlines our work in designing a fuzzy PID controller to achieve a step-response with zero overshoot while improving the output transient response. Our designed fuzzy PID controller is applied to stable, marginally stable and unstable systems and their step responses are compared with a tuned conventional PID controller. A comparative case study shows that the proposed fuzzy controller is highly effective and outperforms the PID controller in achieving a zero overshoot response and enhancing the output transient response.

Zero overshoot and fast transient response using a fuzzy logic controller

Since the first application of fuzzy logic in the field of control engineering, it has been extensively employed in controlling a wide range of applications. The human knowledge on controlling complex and non-linear processes can be incorporated into a controller in the form of linguistic terms. However, with the lack of analytical design study it is becoming more difficult to auto-tune controller parameters. Fuzzy logic controller has several parameters that can be adjusted, such as: membership functions, rule-base and scaling gains. Furthermore, it is not always easy to find the relation between the type of membership functions or rule-base and the controller performance. This study proposes a new systematic auto-tuning algorithm to fine tune fuzzy logic controller gains. A fuzzy PID controller is proposed and applied to several second order systems. The relationship between the closed-loop response and the controller parameters is analysed to devise an auto-tuning method. The results show that the proposed method is highly effective and produces zero overshoot with enhanced transient response. In addition, the robustness of the controller is investigated in the case of parameter changes and the results show a satisfactory performance.

/pdf/design-of-an-iterative-auto-tuning-algorithm-for-a-fuzzy-pid-35b1bjrdhy.pdf

Design of an iterative auto-tuning algorithm for a fuzzy PID controller

The measurement of partial discharge provides a means of monitoring insulation health in high-voltage equipment Traditional partial discharge measurements require separate installation for each item of plant to physically connect sensors with specific items Wireless measurement methods provide an attractive and scalable alternative Existing wireless monitoring technologies which use time-difference-of-arrival of a partial discharge signal at multiple, spatially separated, sensors place high demands on power consumption and cost due to a requirement for rapid sampling A recently proposed partial discharge monitoring system using a wireless sensor network and measuring received signal strength only, has potential cost and scalability advantages An incoherent wireless sensor incorporating a transistor-reset integrator has been developed that reduces the measurement bandwidth of the PD events and alleviates the need for high-speed sampling It is based on composite amplifier techniques to reduce the power requirements by a factor of approximately four without compromising precision The accuracy of the proposed sensor is compared to that obtained using a high-speed digital sampling oscilloscope Received energies were measured over a 10 m distance in 1 m increments and produced an error within 1 dB beyond 4 m and 32 dB at shorter distances, resulting in a measurement accuracy within 1 m

/pdf/low-power-radiometric-partial-discharge-sensor-using-2ft6t1zxxl.pdf

Bakhtiar I. Saeed

Papers

Improving RF-Based Partial Discharge Localization via Machine Learning Ensemble Method

Intelligent Wireless Sensor Network

Zero overshoot and fast transient response using a fuzzy logic controller

Design of an iterative auto-tuning algorithm for a fuzzy PID controller

Low power radiometric partial discharge sensor using composite transistor-reset integrator