Radoslaw Zimroz

Wind Turbine Main Bearing Diagnosis - A Proposal of Data Processing and Decision Making Procedure under Non Stationary Load Condition

Condition Monitoring of bearings used in Wind Turbines (WT) is an important issue. In general, bearings diagnostics is well-recognized field; however it is not the case for machines working under non-stationary load. An additional difficulty is that the Main Bearing (MB) discussed here, it is used to support low speed shaft, so dynamic response of MB is not clear as for a high-speed shaft. In the case of varying load/speed a vibration signal acquired from bearings is affected by operation and makes the diagnosis difficult. These difficulties come from the variation of diagnostic features caused mostly by load/speed variation, low energy of sought features and high noise levels. In the paper a novel diagnostic approach is proposed for main rotor bearings used in wind turbines. From a commercial diagnostic system two kind of information have been acquired: peak-to-peak vibration acceleration and generator power that is related to the operating conditions. The received data cover the period of several months, when the bearing has been replaced due to its failure and the new one has been installed. Due to serious variability of the mentioned data, a decision-making regarding the condition of bearings is pretty difficult. Application of classical statistical pattern recognition for data from the period A (bad condition) and the period B (after replacement, good condition) is not sufficient because the probability density functions of features overlap each other (pdf of peak-to-peak feature for bad and good conditions). Proposed approach is based on an idea proposed earlier for planetary gearboxes, i.e. to analyse data for bad/good conditions in two-dimensional space, feature - load. It is shown that the final data presentation is a good basis to the very successful classification of data (i.e. recognition of damaged and undamaged bearings).

Advances in Condition Monitoring of Machinery in Non-stationary Operations

Part 1. Signal Processing -- Part 2. Data Mining -- Part 3. Condition Monitoring Techniques.

Bearings Fault Detection in Gas Compressor in Presence of High Level of Non-Gaussian Impulsive Noise

The paper deals with the local damage detection in rolling element bearings in presence of a high level non-Gaussian noise. In many theoretical signal processing papers and engineering application related to damage detection, a simple model of the vibration is assumed. Basically it consists of signal of interest (SOI) and some unwanted (deterministic and/or random) components masking SOI in acquired observation. So, damage detection problem has to concern filtering, decomposition or extraction issue. Unfortunately, in the most of the industrial systems mentioned unwanted sources are in fact not Gaussian, so many of de-noising techniques cannot be applied directly or at least might give unexpected results. In this paper an industrial example will be discussed and novel approach based on advanced cyclostationary-based technique will be proposed. In the paper disturbances include periodic impacts in reciprocating compressor on an oil rig. Existing classical detection techniques (statistics in time domain, analysis of envelope spectrum, time-frequency decompositions) are insufficient to perform the task due to high power of disturbance contribution in comparison to damage signature. In the proposed technique, the Spectral Coherence Density Map (SCDM) is estimated first. Next step is related to analysis of SCDM contents and selection of informative part. If informative and non-informative components lay in separate frequency regions, such selection should fix the problem immediately

Application of compound Poisson process for modelling of ore flow in a belt conveyor system with cyclic loading

Abstract This paper deals with the analysis of the random process of cyclic loading of a mining belt conveyor with portions of ore discharged by loaders or trucks. Such transfer of transported ore from a cyclic to a continuous transport is typical for the specific mining operations implemented in the underground copper ore mines with room and pillar mining. The conveyors in such systems are usually significantly oversized in order to match the peak loads of cumulated discharges of ore hauled from mining fields by loaders. Therefore, the actual loadings that occur in the mining transportation systems need to be analysed to provide the data for more accurate design and control of belt conveyors. The large data-set of actual loadings of the belt conveyors has been used for the stochastic modelling of the analysed process. The compound Poisson process has been proposed as mathematical tool to analyse/describe properties of ore flow. The discussion of the chosen distribution functions and results of the fitted model simulations compared with the examined measurement data are presented. In this paper, the case of a belt conveyor loaded only by a single feeding point where loaders are randomly discharged have been analysed. More complex cases (several loading points, mixed ore supply from cyclic and continuous ore mass flow from preceding conveyors or ore bunkers) are under investigation and will be presented in the future.

Technical condition change detection using Anderson–Darling statistic approach for LHD machines – engine overheating problem

Abstract In underground mine so called Load-Haul-Dump machines (LHD) plays a key role in horizontal transportation process. LHD machines execute ore haulage from mining faces to dumping points in a cyclic way. Time-varying and harsh environmental conditions determine high workload, so effectiveness demands are big challenges for maintenance staff. One of the most important issue is related to engine overheating, what is the main cause of unjustifiable loader stoppages and unwanted disturbances in production. Operator is obligated to react quickly and switch machine to idle operation until it cools down. Existing on-board monitoring systems dedicated for LHD machines provide data necessary to perform diagnostics of the engine as well as its cooling system. Understanding how load, wear level of machine and ambient temperature influence diagnostic data is the key in development of fault detection algorithms. In this paper, authors propose to use longterm temperature data. The Anderson–Darling statistic is applied in order to detect segments of different statistical properties which are related to different technical condition. Analysis of obtained two-dimensional data structure allows to find points of change of technical condition of the machine. It could be considered as training for diagnostic system that could be used for machine monitoring.

Maintenance Management of Mining Belt Conveyor System Based on Data Fusion and Advanced Analytics

Belt conveyor network is an important transportation form used in the underground copper ore mines. Effective maintenance of this infrastructure is critical–serious failure of single conveyor might stop operation of several conveyors connected in series and finally might affect production volume. To achieve expected reliability, one should use appropriate tools for supporting maintenance management and decision-making process. Nowadays, predictive maintenance seems to be the most powerful approach for industrial applications. Deep understanding design and operational factors, knowledge about of repairs (number, type, reasons, etc.), and finally acquisition and processing of appropriate physical variables might provide suitable information for maintenance staff. However, mining industry, especially underground mine is a specific kind of factory. Harsh environmental conditions (high humidity, temperature, dust, etc.), varying load, specific damage scenarios make practical implementation of predictive maintenance difficult. Also a scale of transportation system plays an important role: >80 conveyors with different configurations (1–4 drives), diversity of dimensions (short . long conveyors), locations (environmental issues, operation on the slope), etc. All these facts required special approach based on diagnostic data acquisition, necessary processing and context-based reasoning. In this paper, we will discuss details related to development of analytical IT-based environment integrating data from many different sources and procedures supporting decision-making process. We will propose the concept of Decision Support System for maintenance of conveyors system. Because of the multidimensional nature of diagnostic data and diversified technical configurations of the facilities, it was necessary to develop and implement multivariate analytical models including data fusion and artificial intelligence techniques. Consequently, it allows to avoid failures, supports scheduling repairs, and finally provides reduction of repairs costs and production losses related to breakdowns.

Complementary View on Multivariate Data Structure Based on Kohonen’s SOM, Parallel Coordinates and t-SNE Methods

Nowadays, it is often required in modern condition monitoring applications, to describe acquired signal by set of parameters. It directly leads to mD diagnostic data. Before starting the proper analysis of the recorded data, it is advisable to look at the data globally to get an idea what really they are representing. Visualization of mD data is a challenging problem and probably it is not possible to find an ideal method that could take into account all aspects in case of high dimensional, nonlinear, redundant, etc., data. We propose to use for that goal jointly the triplet multivariate visualization methods: Self-organizing maps, Parallel coordinate plots and t-distributed Stochastic neighbor embedding. The methods use concepts of Machine Learning, simple Geometry and Probabilistic Modeling for finding indices of distances or similarities between the data vectors represented in the multivariate data space as data points. The methods permit to visualize the data points in a plane with possibly preserving their mutual between-point distances in the multidimensional data space. The three proposed methods are complementary, and they are supplementing each other. The considerations are illustrated using a data matrix X of size (\(1000 \times 15\)) containing gearbox diagnostic data structured into 4 (sub)groups. Indeed, the three applied (unsupervised) methods permit to get an insight into the 15-dimensional data space and to state that data points belonging to different subgroups of X have different geometrical location. However, the employed methods do not yield indications for reducing the dimensionality (number of variables) of the considered data.

Multidimensional Data Segmentation Based on Blind Source Separation and Statistical Analysis

Horizontal transport in underground copper ore mines mainly consists of LHD machines (loaders, haulers ) and belt conveyors. One of the most crucial mining issues for assessment of efficiency of production is identification of operation cycles of haulage machines. In the literature one can find procedure based on analyzing of pressure signal variability developed for loader (Polak et al Identification of loading process based on hydraulic pressure signal pp 459–466, 2016, Stefaniak et al An effectiveness indicator for a mining loader based on the pressure signal measured at a bucket’s hydraulic cylinder 15, pp 797–805 [6, 7]). The algorithm allows to identify partial operations of loader cycles like: loading, haulage and return to mining face. For haulers this task can seem to be very easy to solve—machines are driving from point A to point B. Nevertheless, when we take into account harsh and specific conditions of underground mine, the problem remains very hard to solve using classical methods based on single variable and if-then-else rules. In most cases, those methods are not robust enough due many random factors (logistical, human factors, work organisation with loaders etc.). In this paper, we propose some kind of data fusion approach to recognition of partial hauler operations. Our method is based on blind source separation approach with particular focus on independent component analysis technique that uses JADE algorithm based on joint approximate diagonalization of eigenmatrices. Obtained components allow for easy segmentation of the signals.

Unsupervised Anomaly Detection for Conveyor Temperature SCADA Data

Belt conveyor system is a crucial element of ore transport process in underground copper ore mine. Damage of single belt conveyor might cause stopping of huge part of underground transport network, especially when failure concerns the main haulage conveyor line. For that reason it is important to use SCADA monitoring system. The symptom of damage can be found in increasing temperature measured within the system. Unfortunately, operating belt conveyors can be considered as time-varying system and direct decision making using temperature value is difficult. Long-term analysis of time series enables to learn how to recognize alarming moment. Thus the removal of failure can be scheduled so as to minimize the losses in production. In this paper the clustering method was applied to the long-term observations of the temperature in order to gearbox fault detection. Moreover, the breaks in the activity of belt conveyors (no operation) caused by holidays will be determined. The clustering algorithm identifies also the specific character of the work at the beginning and end of week.

Application of Independent Component Analysis in Temperature Data Analysis for Gearbox Fault Detection

In the real multisource signal analysis one of the main problems is the fact that true information is divided partially among the individual signals and/or measured signal is a mixture of different sources. This comes from the fact that input channels are typically related, and carry information about different processes occurring during the measurement. Those processes can be thought of as independent sources of vaguely understood “information”. In many cases separation and extraction of those sources can be crucial. In this paper we present the usage of Independent Component Analysis as a tool for information extraction from real-life multichannel temperature data measured on heavy duty gearboxes used in mining industry. Original signals, due to operational factors reveal cyclic variability and detection of damage was difficult. Thanks to proposed procedure, from four channels acquired from 4 gearboxes driving belt conveyor we have extracted one of 4 components, that is related to change of condition of a single gearbox. For new signal visibility of change is clear and simple automatic detection rule can be applied.