Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
печать.docx
Скачиваний:
0
Добавлен:
01.07.2025
Размер:
122 Кб
Скачать
☆
  1. Condition Monitoring Tool for Electrical Equipment — Thermography

IRT is a method to calculate the temperature at multiple position of the surface of equipment with the effective measurement and visualization, with maximum precision. Due to the no physical contact with the surface, it is non-destructive in nature and non-interference with operation of the equipment or system. This non-intrusive nature is essential requirement in industrial applications that demands enclosed and highly influenced by external protrusion, It may reduce significant costs.

Infrared radiation measurements are utilized thermal condition and temperature variation which are then processed and renovated into convenient and communicative electrical signals. The measuring unit includes an infrared camera with high resolution, superficial temperature gradients and high sensitivity for temperature with capability of capture of high quality images [18].

The radiation calculated by the IR camera’s governed by the temperature of the object as well as its emissivity. The radiation approaching from the surrounding and reflected on the object also influences the measuring. Consequently, surrounding temperature due to effects of different sources is also noteworthy, moreover parameters like emissivity, separation between camera and test object and environment conditions should also be measured to capture an accurate and high-pitched image [20].With the advancement in applications of IR thermography, numerous international association and manufacturer published standards for test procedures [21,22], electrical and mechanical inspections [23] and temperature resolution [24].

One of the vital sources of winding insulation deterioration is thermal stress, in spite that the topic is not advances to efficiently monitor and process the winding insulation temperature in operation. A procedure is implement to condition monitor the equipment such as transformer and motors using thermal imaging of the motor, transformer applying thermal camera. Hotspots can therefore be mapped at different positions over equipment. Experimental setup includes110/33 kV transformer set of induction motor drive system and thermal camera. Thermal Images are captured using IRI thermal camera with great care, maximum resolution and be in line of sight. When thermal images are taken improperly, images show minimum unclear difference between the main objects and their backgrounds, dull in color, it must not be processed to determine the situations and monitoring report.

  1. Equipment monitoring for temperature related failures using thermography cameras

The factors that affect temperature accuracy include internal camera temperatures of different components, ambient conditions, and emissivity of target materials. Standard thermal cameras only need to display relative temperature differences in an image which do not need to take these factors into consideration.

The LWIR thermal camera in this application uses an un-cooled amorphous silicon (α - si) microbolometer. Until recently, thermal cameras were prohibitively costly, and so were limited primarily to military and government installations. Now the cost of thermal detectors has been decreasing rapidly, which makes thermal cameras a very attractive solution to industrial markets. The uncooled detectors make the thermal camera easy to maintain with much longer life cycles.

There are several layers of silicon in an uncooled microbolometer focal plane array (FPA) detector. The layers are in the following order: the IR absorbing material, the thermal isolation gap to avoid thermal conductance to the bottom layer, a reflector layer to reflect the light back to maximize absorption follow by the silicon substrate and readout circuit (ROIC). The ROIC contains of an electrical gain and an offset which vary with detector temperature.

The first stage process is for removing electrical noise generated from electrical gain and offset as well as achieving optical spatial uniformity correction. It is not sufficient to provide radiometrically accurate measurement of the target radiance. In order to achieve temperature accuracy, a more complex model which includes optical element and radiometric conversion will be described below.

Two important factors which relate to radiometric accuracy are emissivity and radiance transfer function.

Emissivity(ε): Emissivity determines the efficiency of how well a source surface material is able to transmit it's radiation energy. Emissivity numbers ranges from 0.0 to 1.0. The opposite of emissivity is reflectivity( r). The sum of ε and r equals to 1.0.

Radiance transfer function: The sensor response in ADC counts vs. target temperature is non-linear at a given FPA temperature [8], [9]. Due to this non-linearity property, it is difficult for temperature estimation to account for extra background radiation in ADC space. Although ADC vs. target temperature is non-linear, ADC vs. radiance has a linear relationship within a limited range which allows for simpler thermal modeling and cancelling of background radiation. The conversion between radiance to temperature can be calculated by integrating the spectral radiation band limited energy between 8 to 13 μm for each blackbody temperature.

The relationship between ADC and radiance is defined as radiance transfer function (RTF). It is linear within our temperature range of -20°C to 120°C and typically limited by analog to digital converter and sensor saturation. It is sensor/camera dependent and can be obtained by a secondary calibration. This calibration will be referred as radiometric calibration in the later context. The radiometric calibration is done after uniformity correction and offset noise removal from the first calibration. This is so that a single RTF function can be used for all pixels. RTF function is FPA temperature dependent.

RTF function is obtained by setting the camera inside a climate controlled chamber and varying the blackbody temperatures at each chamber temperature. Each RTF can be approximated with a linear function. A piecewise linear interpolation can be performed for RTFs for the missing FPA temperatures.

In order to achieve temperature accuracy, the attenuation and flare from the optical path need to be taken into consideration. These factors could be caused by heat from internal components or ambient atmospheric condition changes.

Several factors can affect temperature accuracy [6]. These factors include lens attenuation, flare uniformity, apparent reflected temperature (ART) and atmospheric attenuations.

The traditional temperature measurement method requires the user to draw the exact shape of the object to be measured. Inside each measurement zone, maximum and minimum temperature measurement points are reported. There are several downsides to this method. One, it is very difficult to draw the exact shape of the object. Two, a camera is typically mounted high up on a pole and if there is any vibration caused by the wind a false temperature could be reported with background temperature going in and out of the zone. Three, if there are any defective pixels in the zone, wrong temperatures could be reported based on the defective pixel.

It is not statistically accurate to use a temperature value based only one pixel to represent the overall temperature of the object.

In this paper, a histogram distribution method based on pixel temperature values was developed. The histogram distribution algorithm has the advantage of temperature measurement unaffected by the spatial location of the object inside the zone. This also allows for a simple rectangular zone being drawn around or on the target object without worrying about camera vibration. It provides a statistically representative value based on temperature distribution on the object of based on only a single pixel. The only two questions needed to be answered from the users are: 1) Is the target object the entire, warmest or coldest part of the measurement zone? 2) Should the warmest, coldest or average temperature of the target object be reported and tested against the alarm threshold? These two questions allow one of nine points on the histogram to be selected to represent the temperature of object.

Traditional temperature methods also require the usage of mathematical operations between each zone if temperatures need to be compared. It is cumbersome and sometimes impossible to achieve simple temperature comparisons. Three temperature measurement methods were developed to simplify this process. The three methods are:

1) Absolute Threshold: An absolute temperature threshold is set for either the object temperature being too high or too low. An alarm will be triggered based on this threshold.

2) Relative Threshold: This is a comparison method for all temperature zones of the same group. If the temperature range is too large as defined by the user, for example greater than 3ιC, an alarm will be triggered. This is especially useful for the case such as comparing bushing cap temperatures from the same phase on an electrical transformer at defferemt times and different electrical loads when the absolute temperature is not the main concern but rather the relative temperatures between them. This is also an effective method that can remove solar loading or ambient temperature effects to all object temperatures being compared.

3) Self Reference Threshold: A temperature threshold is set based on how much temperature has changed compare to its current temperature. For example, the threshold can be set to the temperature has changed more than 10 C. This is useful for if the absolute triggering temperature is not known but only interested in the temperature changes compared to its current temperature.

Tests were setup to verify the thermal model and its temperature accuracy under different conditions. A thermography camera was put inside an environmental chamber aiming at a blackbody target through an open port on the side. Temperature accuracy was tested while the chamber temperature was changed from high to low. The measured temperatures are close to the ideal temperatures set on the blackbody target at different ambient temperatures.

An outdoor temperature controlled plate with high emissivity paint was built for outdoor experiments. The emissivity of the outdoor temperature target is 0.93. The temperature is controlled through Ethernet with PID loops.

The camera was set up to point at the outdoor temperature target at 10 meters distance. A custom built weather station was used to automatically communicate the ambient information for atmospheric correction inside the camera. The temperature measurement was collected during 24 hours in Fort Collins, Colorado. The measured temperature accuracy is tracking fairly well to the ideal set target temperatures.

The mathematical thermal model, calibration, challenges and temperature measurement algorithms are described in this paper. The experimental results and use case showed the great value of the developed solution. There are many benefits for being able to remotely monitor temperature of equipment over a prolonged period of time. In addition to trigger temperature alarms, understanding and optimizing equipment performance are as equally important. Images and videos not only allow for detecting abnormal temperature points within field of view and easily configuring a measuring zone on the questionable points. It is also provides verification and feedback on what has happened in case of equipment failure. This is difficult to do with point measuring and hand held devices. The number of applications for this technology is broad. Any temperature failure related monitoring can be applied. Some of the applications [1] include but are not limited to electrical substations, radiators, rail and wheel temperatures, oil refinery flame detection, oil or LNG tank levels, automotive manufacturing robots, mining, food processing, etc...