Measurement of quality indicator of electrical insulator

Abstract

And a measuring device performing a measurement on a quality indicator for the electrical insulator (2), the quality indicator comprising a ratio of a first value of the electrical parameter at a first time to a second value at a second time. An output (6) from a circuit for measuring an electrical parameter of an electrical insulator is sampled to produce at least one sample, and the following steps are performed: sampling is performed on the output from the circuit to produce one or more additional samples in consecutive time increments, and processing at least some of the samples to generate an approximation function correlating the measured electrical parameter with time (8). A predicted value for the quality indicator is calculated (10) based on an electrical parameter comprising an indication of the second time by an approximation function, and an electrical signal indicative of a state of the measurement (11) is generated based at least on the predicted value of the quality indicator.

Description

Technical Field

[0001] The present invention generally relates to an improved method of operating a measuring device and an improved measuring device for measuring a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time to a second value of the electrical parameter at a second time. For example, the quality indicator may be a polarization index (PI) or a dielectric absorption ratio (DAR), and the electrical parameter may be resistance. Background Art

[0002] A quality indicator of an electrical insulator, such as a polarization index (PI) or a dielectric absorption ratio (DAR) comprises a ratio of a first value of an electrical parameter at a first time to a second value of the electrical parameter at a second time. The electrical parameter is typically electrical resistance. The quality indicator can be used to assess the condition of an electrical insulator. The extent to which the measured resistance changes between the first time and the second time can give useful information about whether the insulator is safe or hazardous. In one example, a first measurement of resistance can be made 1 minute after a voltage is applied to the insulator, and a second measurement of resistance can be made 10 minutes after the voltage is applied. It may be necessary to make multiple measurements of the quality indicator at the installation, for example for different phase windings of an electric motor, so the testing process can be particularly time consuming. Summary of the invention

[0003] According to a first aspect, there is provided a method of operating a measuring device for performing a measurement of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time and a second value of the electrical parameter at a second time, the method comprising:

[0004] sampling an output from a circuit for measuring an electrical parameter of an electrical insulator to produce at least one sample;

[0005] The steps of: sampling an output from the circuit to produce one or more additional samples at successive time increments, and processing at least some of the samples to generate an approximate function relating a measured electrical parameter to time;

[0006] calculating a predicted value for the quality indicator based on the electrical parameter indicated by the approximation function including for the second time; and

[0007] An electrical signal indicative of a status of the measurement is generated based at least on the predicted value of the quality indicator.

[0008] This feature allows shortening the measurement time by using the predicted value of the quality indicator instead of the measured value at the second time and allows adjusting the measurement time according to the predicted value.It has been found that the time taken to achieve a reliable estimate of the predicted value of the quality indicator depends on the predicted value.

[0009] In an example, the electrical parameter may be resistance or current.

[0010] In an example, the method includes generating a measure of confidence in a predicted value of the quality indicator.

[0011] Therein, the step of generating an electrical signal indicative of a status of the measurement is based at least on a measure of confidence in a predicted value of the quality indicator.

[0012] This feature reduces the measurement time by basing the measure of confidence in the predicted value of the quality indicator, thereby allowing for shorter measurement times.

[0013] In an example, the measure of confidence in the predicted value of the quality indicator is a confidence range for the predicted value of the quality indicator based on a measure of noise for at least some of the samples.

[0014] This feature provides an efficient method to generate a measure of confidence in the predicted value of a quality indicator.

[0015] In an example, the electrical signal indicating the status of the measurement indicates that the measurement is complete.

[0016] This feature allows measurements to be taken before the second time is reached.

[0017] In an example, the method includes stopping the measurement in response to the electrical signal indicating a status of the measurement.

[0018] This feature allows to automatically stop the measurement before the second time.

[0019] In an example, the method includes generating an electrical signal based on the electrical signal indicative of a state of the measurement, resulting in displaying a predicted value of the quality indicator.

[0020] In an example, processing some of the samples to generate an approximate function relating measured resistance to time includes discarding a group of samples.

[0021] This feature allows generating accurate predictions of the quality indicator, for example by discarding initial samples.

[0022] In an example, processing at least some of the samples to generate an approximate function relating the measured resistance to time includes least squares curve fitting, linear regression, and / or nonlinear regression.

[0023] This feature allows efficient and convenient generation of approximate functions.

[0024] A quality indicator of an electrical insulator may be, for example, the polarization index or the dielectric absorption ratio.

[0025] According to a second aspect, a measuring device is provided for performing a measurement of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time and a second value of the electrical parameter at a second time, the measuring device comprising at least one processor, the at least one processor being configured to cause the measuring device to:

[0026] sampling an output from a circuit for measuring an electrical parameter of an electrical insulator to produce at least one sample;

[0027] The steps of: sampling an output from the circuit to produce one or more additional samples at successive time increments, and processing at least some of the samples to generate an approximate function relating a measured electrical parameter to time;

[0028] calculating a predicted value for the quality indicator based on the value of the electrical parameter indicated by the approximation function for the second time; and

[0029] An electrical signal indicative of a status of the measurement is generated based at least on the predicted value of the quality indicator.

[0030] According to a third aspect, there is provided a computer readable storage medium holding instructions for causing one or more processors to cause a measurement device to perform a measurement of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time and a second value of the electrical parameter at a second time, to:

[0031] receiving a sample of the output from a circuit for measuring an electrical parameter of an electrical insulator to produce at least one sample;

[0032] receiving additional samples at successive time increments and processing at least some of the samples to generate an approximate function relating the measured electrical parameter to time;

[0033] calculating a predicted value for the quality indicator based on the electrical parameter indicated by the approximation function including for the second time; and

[0034] An electrical signal indicative of a status of the measurement is generated based at least on the predicted value of the quality indicator.

[0035] Further features and advantages thereof will become apparent from the following description of exemplary embodiments given by way of example only. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a schematic diagram illustrating a simplified electrical model of an insulator, showing the current flow due to leakage, capacitance, and absorption;

[0037] Figure 2is a graph showing typical current flow due to leakage, capacitance, and absorption as a function of time;

[0038] Figure 3 The calculation of the polarization index (PI) is shown;

[0039] Figure 4 An example of calculation of the polarization index for various insulation conditions is shown;

[0040] Figure 5 is an example of a table relating polarization index values ​​to insulation conditions;

[0041] Figure 6 is a schematic diagram showing an example of a meter connected to an insulator under test;

[0042] Figure 7 is a schematic diagram showing an example of a meter connected to an insulation under test, the meter being configured to automatically end the measurement based on a predicted value of a quality indicator;

[0043] Figure 8 is a schematic diagram showing an example of a meter connected to an insulation under test, the meter being configured to automatically end the measurement based on a measure of confidence in a predicted value of a quality indicator;

[0044] Fig. 9 is a schematic diagram showing a measuring device including a meter and a user equipment communicating with the meter.

[0045] Fig.10 is a flow chart illustrating an example of a method of operating a meter;

[0046] Fig.11 Predictions for quality indicators indicating "good" insulation condition (PI = 3-5) are shown;

[0047] Fig.12 shows that a quality indicator indicating a "bad" insulator condition (PI<2) is predicted in a reduced time compared to the time for predicting a quality indicator indicating a "good" insulator condition;

[0048] Fig.13 Predictions for quality indicators indicating a "brittle" insulator condition (PI>5) are shown;

[0049] Fig.14 An example of a meter display showing a prediction of a quality indicator from a noise sample is shown.

[0050] Fig.15 is a graph showing predicted values ​​of generated resistance at a second time;

[0051] Fig.16 Shown for Fig.11The instrument display showing the initial portion of the completed measurement is shown in;

[0052] Fig.17 Shown for Fig.11 The completed measurement is shown in the middle section of the instrument display;

[0053] Fig.18 A simplified prediction method is shown; and

[0054] Fig.19a , Fig.19b and Fig.19c Modeling of the absorption effect is shown, as well as modeling of an additional RC network for various component values. DETAILED DESCRIPTION

[0055] As an example, embodiments will now be described in the context of a digital meter capable of measuring the polarization index (PI) of an electrical insulator, but it will be understood that embodiments of the invention may relate to other electrical test equipment and that embodiments of the invention are not limited to the measurement of polarization index. Other quality indicators of an electrical insulator may be measured, where the quality indicator comprises a ratio of a first electrical parameter at a first time to a second electrical parameter at a second time.

[0056] Conventional meters for measuring polarization index measure the resistance of an insulator at a first time and also measure the resistance of the insulator at a second time, and then calculate the ratio of the measured values. According to this embodiment, the meter can be configured to reduce the measurement time by continuously collecting resistance samples during the measurement, and predicting rather than measuring the resistance at a second time based on the extrapolation of the samples already collected, and continuously updating the predicted value of the quality indicator as the measurement proceeds. The quality indicator can be predicted based on the ratio of the extrapolated resistance at the second time to the measured or extrapolated resistance at the first time. It has been found that the reliability of the predicted value of the quality indicator depends on the predicted value itself. If the predicted value of the quality indicator is found to be greater than or less than a predetermined limit, the measurement can be automatically stopped based on the fact that the predicted value is a sufficiently reliable indicator of the measurement result. The predicted value of the quality indicator can then be displayed as the measurement result.

[0057] In determining that the predicted value of the quality indicator is sufficiently reliable that the measurement can be stopped, a measure of the noise of the acquired samples may be used.

[0058] It should be understood that the measurement and prediction of resistance generally includes the measurement and prediction of current, provided that the voltage applied for each measurement (usually a constant voltage) is known or known to be the same. Therefore, references to the measurement and prediction of resistance should be interpreted as referring alternatively to the measurement and prediction of current. The polarization index can be calculated as the ratio of current to a given applied voltage.

[0059] To better explain the operation of the meter to predict the PI value, the process of calculating the polarization index will now be explained in more detail. The PI test is commonly used to test the quality of insulation systems, for example in rotating machines and high voltage (HV) cables. The insulation system can actually be considered as a capacitor, as it includes at least two conductors separated by an insulator, and can be represented as Figure 1 The equivalent electrical diagram shown in FIG. Capacitor 31 represents the main capacitance of the system, and leakage resistor 30 represents the leakage current I flowing therein. L Insulation resistance.

[0060] For example, by Figure 1 The insulation quality is measured by applying a DC voltage (that is, a voltage that is constant for the test) between the points indicated by + / - in the , and accurately measuring the resulting total applied voltage and total current. Because the test is made with a DC voltage, this voltage across the insulation system can be generated with very high stability, typically better than 1V of 1kV before the test begins, so that any AC current flowing through the capacitor (e.g. due to ripple from a switched mode power supply) is negligibly small compared to the DC current of interest.

[0061] The total current consists of several components, such as Figure 2 , which has a vertical scale showing current and a horizontal scale showing time in seconds. At the moment the test DC voltage is applied, the main capacitor 31 is being charged, which results in a very high charging current, as shown by curve 36. Once a stable DC voltage is reached when the capacitor 31 is charged, the capacitor charging current becomes negligible. For small capacitors, the voltage can be stabilized in less than 1 second.

[0062] In the case of a stable voltage, the current through the leakage resistor 30 is constant, as shown by curve 34, and the capacitive current drops to a negligible level, as shown by curve 36. However, there is also an additional component called the absorption current, as shown by curve 35, which decays with a longer time constant, which can be several minutes, even after the voltage has stabilized. This is in Figure 1 In is modelled with an additional RC branch shown by a series capacitor and resistor 32 through which the absorption current flows, which gives a much longer RC time constant than the main capacitor. In practice, there may be multiple time constants involved, and this can be modelled as Figure 1 Further RC absorption branches other than the RC branch 32 shown in FIG. 4 . For example, additional RC branches may be connected in parallel.

[0063] Absorption current occurs because charges are trapped in the bulk of the insulator, but the applied voltage is able to slowly drag them and displace them locally. These charges behave as if they were suspended on springs and slowly move from their discharged position to the charged position to achieve equilibrium with the electric field in the bulk of the insulator. This phenomenon is called electrical induction or polarization / depolarization.

[0064] For example, if there is more than one insulating layer, more than one time constant may occur. Each layer behaves as a different capacitor, exhibiting slightly different behavior. As a result, the overall curve does not have a simple exponential shape, but may even have a negative curvature, with an accelerating slope instead of a decelerating slope, and may tend to plateau.

[0065] Because the applied voltage is constant, but the resulting current is not, the resistance calculated according to Ohm's law R = V / I varies during the measurement. Because the total current usually decreases during the measurement, such as Figure 2 As shown in curve 33 in the figure, the measured resistance R increases. A typical curve of insulation resistance (IR) over time is shown in Figure 3 as shown in .

[0066] like Figure 3 As shown in , the insulation resistance is conventionally measured for 10 minutes and two resistance values ​​are extracted: 1 minute (R 1min) 39 and 10 minutes (R10min) 40. The polarization index (PI) is calculated as the ratio between the two numbers, so that PI = R 10min / R1min (R10 / R1 in minutes, or R600 / R60 in seconds).

[0067] Figure 4 Curves 41 , 42 and 43 are shown, representing measured values ​​of PI values ​​>2, >1 and about 1, respectively.

[0068] The so-called dielectric absorption ratio (DAR) is another example of a quality indicator for an insulator. The DAR is the ratio of the 60-second to the 30-second resistance reading, such that DAR = R 60sec / R 30sec. The DAR is measured much faster (1 min), but the PI (10 min) gives more reliable information about the quality of the insulation system. Other quality indicators with other time ratios may also be used, such as R5min / R1min or R60sec / R15sec.

[0069] If the leakage current is smaller, the value of the PI ratio (R10 / R1) is larger, that is, the value of the leakage resistor 30 is higher. It should be noted that the PI ratio is an important parameter indicating the condition of the insulator, rather than the resistance value itself. A flat response curve 43 generally indicates a large leakage current, for example due to moisture trapped in the insulator. Moisture is generally regarded as a highly undesirable feature of an insulator, while a dry and clean insulator is generally desired and results in a higher resistance and a higher PI value.

[0070] Figure 5 is a table showing examples of recommended PI values. For some insulation materials, a value PI>5 can indicate a "brittle" insulation material that may be damaged by overheating. This insulation can be very dry and therefore have very small leakage currents, but its lifetime will be impaired, so this also conveys some useful diagnostic information. This is specified, for example, in standard IEEE 43.

[0071] The main disadvantage of the conventional PI test method involving measurements after 1 minute and 10 minutes is that it is very time consuming. For a three-phase system, each phase may need to be tested separately, which requires a test time of at least 30 minutes for the entire machine.

[0072] In some cases, an additional measurement of the depolarization current may be performed. This is done by short-circuiting the insulation system with an ammeter to discharge the main capacitor 30, e.g. Figure 1 . The voltage collapses very quickly, but the absorbing RC branch 32 keeps carrying some small current, as determined by its RC time constant. In a first approximation, the depolarization (discharge) time constant is the same as the polarization (charge) time constant. This can be used to detect problems with multi-layer insulation systems. Typically, depolarization takes as long as polarization. So if the PI measurement takes 10 minutes, depolarization will take an additional 10 minutes, effectively doubling the measurement time. Therefore, to fully test a three-phase machine, it will take at least 1 hour.

[0073] Figure 6A meter 1 according to an embodiment is shown, in this example a digital meter, which is connected to an insulator 2 under test. The meter is used to perform a measurement of a quality indicator, in this example a polarization index (PI) of an electrical insulator 2. The meter has at least a first terminal 3 and a second terminal 4 for connecting to an insulator under test, which may be, for example, an insulator of an electrical winding of an electric motor. In this case, the terminals of the meter will be connected to the electrical windings on each side of the insulator under test. The meter has a circuit for measuring resistance 5, which can typically generate a high voltage to be applied to the insulator and measure the resulting, usually very small, current. The circuit for measuring resistance is connected to the first terminal 3 and the second terminal 4. The circuit may not necessarily be directly connected to the first and second terminals, but may be connected via other components, such as an input protection circuit, which may include, for example, a fuse. Two terminals are shown, but some examples may be provided with an additional terminal, which may be referred to as a protection terminal provided to improve the measurement accuracy. The protection terminal may be maintained at the same potential as one of the terminals, but bypasses the internal ammeter of the meter. The protection terminal may be connected to the surface of the insulator under test so that unwanted surface leakage currents do not affect the test results.

[0074] like Figure 6 As shown in , the output 6 of the circuit for measuring resistance is connected to the sampler 7 to produce at least one sample of the measured resistance. The sampler can be part of the processor 13. For example, the processor 13 can include a microcontroller with a program memory containing program code, or can be part of another processor, or can be implemented by digital logic or a programmable gate array. The functions of the processor 13 for performing digital signal processing and the controller 12 for meter operation can be combined into one or more controller or processor circuits and can be performed by software or firmware configured to run on a programmable device.

[0075] A processor 13 in the meter iteratively performs the steps of sampling the output from the circuit for measuring resistance, generating one or more additional samples at successive time increments, and processing at least some of the samples to generate an approximate function relating the measured resistance to time 8. The process for generating the approximate function may include least squares curve fitting, linear regression, and / or nonlinear regression.

[0076] The processor extrapolates the approximation function to include at least the second time 9 and calculates a predicted value of the quality indicator 10 based on the resistance indicated by the approximation function including the second time.

[0077] The processor then generates an electrical signal indicating the status of the measurement 11 based at least on the predicted value of the quality indicator, which may indicate that the measurement is complete. In one example, the controller 12 stops the measurement in response to the electrical signal indicating the status of the measurement, and in one example, the controller 12 stops the test and, for safety reasons, initiates an automatic discharge of the high voltage charge accumulated in the insulator under test. The display shows an indication of the completion of the measurement based on receiving the electrical signal.

[0078] As the measurement proceeds, the display 14 may show a graph of the predicted quality indicator (typically the polarization index) versus time. The processor may send a signal to the display based on the electrical signal indicating the measurement status of the measurement completion, so that the predicted value of the quality indicator is displayed as the measurement result at the end of the measurement.

[0079] Figure 7 An example of an instrument is shown in which a processor calculates a predicted value of a quality indicator 10 and compares the predicted value to at least one predetermined threshold value. Based on the comparison, the processor sends a signal 16 indicating that the measurement should end. The controller 12, typically a functional block executed by the processor 13, then stops the measurement. In one example, stopping the measurement initiates a series of events: communicating the termination of the test to the user via a display, shutting down the generation of the high voltage (HV), activating the discharge mechanism, and measuring the capacitance of the insulation under test. When the output voltage remains "dangerously live" due to the presence of high voltage (e.g., voltage greater than 50V), the instrument indicates the danger, for example, by a display with a flashing symbol or LED, and the test is fully completed after the output voltage is discharged to a safe level.

[0080] Thresholds may be used to determine that the predicted quality indicator falls within certain ranges, such as a range of polarization indices corresponding to "bad" or "brittle" insulation. The display then shows the predicted polarization index and / or its determined range as a measurement result.

[0081] Figure 8 An example of an instrument is shown in which a processor generates a measure of confidence in a predicted value of a quality indicator 17, and the generation of an electrical signal indicating the status of the measurement is based at least in part on the measure. In an example, the measure of confidence in the predicted value of the quality indicator is a confidence range for the predicted value of the quality indicator based on a measure of noise for at least some samples. If the noise level, as indicated by statistical variation between samples, is high, the measure of confidence will be lower, and so the processor will allow the measurement to continue until an acceptable confidence level is generated. The confidence level may depend on the predicted value of the quality indicator or the measure of noise, or both. In an example, a signal indicating the status of the measurement may be generated based on the measure of noise rather than on the predicted value of the quality indicator, and the measurement may be stopped based on the signal.

[0082] Fig. 9 A measuring arrangement is shown comprising a meter 1 and a user device 37 (such as a mobile phone or a computer) having one or more processors. The user device may be connected to the meter via a wireless link (such as Bluetooth or WiFi) or via a cable connection. The user device may be used to perform part of the functions of the meter, such as Figure 6 , Figure 7 and Figure 8 , such as providing a display, or in other cases providing a display and at least some of the functions of the processor 13. In other cases, at least some of the functions of the processor may be processed as an Internet-based cloud processing, where the meter and / or the user device has a connection to the Internet. For example, the extrapolation function may be performed remotely from the meter in an external processor.

[0083] Fig.10 is a flow chart showing an example of a method of operating a meter according to steps S10.1 to S10.5.

[0084] In an example, processing some samples to generate an approximate function that relates measured resistance to time includes discarding a set of samples. For example, by discarding initial samples. The initial samples may not be helpful in predicting the resistance value at a second time. Discarding samples allows for more efficient processing.

[0085] Fig.11 A prediction of a quality indicator indicating a good insulator condition (PI = 3-4) is shown. The vertical access shows the resistance in gigaohms and the horizontal access shows the time in seconds. In this example, the second time for the PI calculation is 600 seconds, at the right end of the graph. The measurements are plotted on the curve until the point indicated by reference numeral 22. In this example, the measurement stops at 300 seconds. The curve is extrapolated from point 22 to point 23. The predicted value of the resistance at the second time is at the point indicated by reference numeral 23, approximately 500 gigaohms. The resistance measured at the first time for the PI calculation at 60 seconds is approximately 150 gigaohms, so the predicted PI value is 500 / 150 = approximately 3.3. Figure 5 The diagram of shows an example of an instrument display which shows the PI value as a scale on the right side of the display. The display is automatically scaled by the processor so that the point 23 where the extrapolation curve intersects the PI scale can be used to read the predicted PI value.

[0086] Fig.12 It is shown that a quality indicator indicating a bad insulator condition (PI<2) is predicted in a reduced time compared to the time for predicting a quality indicator indicating a good insulator condition. The extrapolated curve gives a predicted PI value of <1.5 at point 25. The measurement is stopped at point 24 after 180 seconds.

[0087] Fig.13 A prediction of a quality indicator indicating a brittle insulator condition (PI>5) is shown. The extrapolated curve exceeds the top of the vertical scale at point 27, which if extrapolated to a second time at 600 seconds would correspond to a PI value much greater than 6. It can be seen that the measurement is stopped at point 26 after approximately 300 seconds. It can be seen that there is some noise in the measurement data. In the case shown, since the PI prediction is greater than a predefined threshold, the measurement should have been stopped earlier, for example at 180 seconds, but the measurement of noise greater than the threshold caused the measurement to be extended to 300 seconds to give sufficient confidence in the prediction result.

[0088] Fig.14 An example of a meter display showing a prediction of the quality indicator from a noise sample is shown. In this example, the measurement is stopped at point 28 after approximately 420 seconds. The PI prediction at point 29 has a value of approximately 4.2, which corresponds to a classification as a good insulator condition. However, with Figure 5 The measurement time is extended compared to the measurement time in the example of because the noise level of the measurement, as shown by the statistical variation between samples, is above a predetermined threshold.

[0089] The following is a specific example of how to do it.

[0090] 1. Set the test auto-stop time to the initial value, such as 300 seconds (5 minutes).

[0091] 2. Check the noise level by calculating the standard deviation of the measured samples from the predicted curve.

[0092] 3. If the noise level is greater than or equal to the first threshold, the time for the test to automatically stop is extended, for example, to 360 seconds. If the noise level is greater than the second threshold, the test time is extended in larger increments, for example, to 420 seconds.

[0093] 4. If the noise value is less than the first threshold, check the predicted PI value by extrapolating to 10 minutes. If the predicted PI value is greater than an upper threshold, such as PI>5, and / or the predicted PI value is less than a lower threshold, such as PI<1.25, PI<1.5, or PI<2, reduce the test time, such as to 180 seconds.

[0094] Fig.14 It is shown that the comprehensibility of the display can be improved by shading or color coding areas of the graph to show the possible PI values ​​that will result from the measured values ​​in that area at different points in the measurement. In this example, the measurement results fall mainly in the shaded area, resulting in PI values ​​in the range of 3-5. This provides the operator with an indication of the possible results of the PI value as the measurement progresses.

[0095] From the above it can be seen that this example gives a method of reliably predicting PI and has been shown to have + / -10% accuracy over a range of curve shapes. For most curves the prediction can give a 10 minute result after 5 minutes of measurement. If the insulation is "bad" (e.g. PI < 1.5) then this can be detected immediately after the 3 minute test time and there is then no need to continue for the full 10 minutes. If the measurement is noisy then the test time can be automatically extended to avoid premature incorrect predictions.

[0096] With the new method, an industrially accurate assessment of the insulation condition can be achieved, but this can be done in a much shorter practical time, which is not possible with any other known method. As a result, the productivity of the tests is greatly increased. Bad or brittle insulation systems can be identified quickly.

[0097] The method using ratios (PI, DAR) is used instead of using absolute resistance values ​​because resistance changes significantly with temperature. It drops by about 50% for every 10 degrees Celsius increase. The ratio is affected much less than the absolute value, which is why this method is very popular in diagnostic testing, especially in large motors and generators.

[0098] Fig.15 The new PI prediction method is conceptually illustrated in FIG. Insulation resistance (IR) data is measured from time 0 seconds to 600 seconds (10 minutes). The measured IR data 17 is used to perform the calculation. In the example, the initial part of the data (e.g., the initial 30 seconds or 60 seconds) is ignored because there are still some fast processes occurring there and they provide very little information about the values ​​at later times. This is Fig.13 The reason why the measurement data 17 shown in FIG. 1 starts at 60 seconds is because in this example, any data before 60 seconds is not used when predicting the PI value. Fig.15 In the example of , the remainder of the measurement data from 60 seconds to 300 seconds is used to perform a curve fit by means of linear or nonlinear regression. An appropriate type of function, such as a power function, a polynomial function, an exponential function or a linear function, is selected to give the best possible fit to the measurement data. More than one function may be used for approximation, and the best fit may then be selected, for example, by means of the method of least squares, for prediction of the PI value. An average between two or more functions may be applied, for example by means of a weighted average depending on the quality of the curve fit, for example measured by the so-called residuals from the nonlinear / linear regression.

[0099] The final approximation function is then taken and its value is extrapolated to a final time, e.g. 600 seconds / 10 minutes, before the value at the final time is measured. This extrapolated / predicted value is given by Fig.15 The rectangle at 600 seconds on the extrapolated curve 19 in FIG. Fig.15It can be seen that the 10 minute readings can be predicted well in advance of the hardware measurements.

[0100] The measurement data can be processed in real time, so that the user can get visual feedback about the likely outcome of a measurement even before the prediction-enabled measurement is actually completed. The data is dynamically analyzed, for example the readings are updated every 1 second, so that if the noise is small, the measurement ends after 3-5 minutes, and if the noise is large, the algorithm will automatically ask the hardware to continue testing for more than 5 minutes, possibly all the way up to 10 minutes, so that even extremely noisy conditions can provide good measurements. This has been combined with Fig.14 Described.

[0101] Return to Fig.14 , which shows that the DAR (also called DA in the US) and PI scales can be color coded and placed on the same screen ( Fig.14 The arrows at the top indicate these two scales). In addition, the entire background can be segmented into colored areas, which provides very clear visual feedback to the operator as to where the predicted curve is likely to end when extrapolated to a 10 minute period. The scale can be either insulation resistance (left) or PI or DAR ratio (right). Fig.14 In the PR and DAR scales, they are in the same units because they are both ratios, so a "1" on one scale corresponds to a "1" on the other. However, smaller DAR values ​​correspond to larger PI values, for example, DAR>1.75 indicates "good", while "good" requires PI>3.

[0102] Fig.14 Shown for Figure 5 The instrument display shows the initial portion of the completed measurement, and Fig.15 Shown for Figure 5 The middle section of the instrument display shows the completed measurement. Fig.14 In the example above, only the raw data is displayed point by point until 180 seconds, after which the prediction starts. Fig.15 shows the measured state at 240 seconds when the prediction reaches 390 seconds, Figure 5 The measurements performed are shown, and the predictions made from 300 to 600 seconds are shown.

[0103] As mentioned before, for very poor insulation (PI < 1.5), the measurement can be completed after 3 minutes. Similarly, if the PI is very high, in the example, the prediction can be reported as PI>7 or PI>5.

[0104] The initial data in a measured IR curve contains less useful information than data further back in time, so that, for example, initial measurement data up to 90 seconds can be ignored due to the presence of rapidly changing currents. Throughout the measurement, the data at the end of a given measurement provides more information about the possible shape of the curve than the initial data. Therefore, the measured data can be dynamically processed so that the data is used with weighting factors that favor data toward the end of the curve. In extreme cases, the first portion of the data, such as the first half, can be ignored in the curve fitting process. Fig.16 and Fig.17 shows that the previously combined Fig.11 A sequence of predicted resistance values ​​for increasingly distant future times during the measurement period described. Fig.16 , Fig.17 and Fig.11 It can be viewed as an animation showing how the display will proceed during a measurement. As the measurement progresses, predictions of the resistance value at a given confidence level can be made further and further in the future until a prediction can be made for a second time (typically 10 minutes) at a given confidence level. At low noise levels, the time to predict the resistance may depend on the number of measurement samples of the electrical parameter (typically current or resistance) that have been collected. If the noise is above a predetermined threshold, the time to which the predicted resistance is reached may depend on the noise level, so that at higher levels of noise a prediction may be made at a given confidence level rather than at lower noise levels. The time to extrapolate the approximation level may depend on a measure of confidence in the predicted value of the quality indicator, which may be based on a measure of the noise of at least some of the samples. In Fig.16 , Fig.17 and then Fig.11 During the sequence shown, the data between 60 and 240 seconds will initially be used for curve fitting, that is, to generate an approximate function such as Fig.16 As shown in Fig.11 In the final stage shown in , in the example, only the data between 200-300 seconds will be used for the curve fitting. This can be automatically scaled by the software. The samples used to generate the approximate function are typically used to generate a measure of noise which is used to generate a measure of confidence in the predicted value of the quality indicator.

[0105] In the example, the curve approximation function is chosen to be a power curve such that:

[0106] (1)y=a*x b +c

[0107] Where: y represents the measured IR value, x is time, a, b, c are function coefficients, so:

[0108] (2)IR(t)=a*t b +c

[0109] The a, b, c coefficients can be found by statistical and mathematical methods defined using the well-known nonlinear regression model.

[0110] from Fig.17 As can be seen, the length of the forecast is limited, so the final value is not reported until the algorithm is satisfied with the quality of the forecast. Only then is the forecast function extrapolated all the way to 600 seconds. The way in which the extrapolation is limited is controlled by the function, which is further limited by a set of conditions. The extrapolated data starts from the last measured point (time_now). The intervals at which the forecast function is used, i.e., Fig.17 The point to which the graph is drawn (i.e., its "final" point, time t2, such that t2 <= 600) can be described by the following function:

[0111] (3)t2={t1+(600–end_at_time)*(time_now-180) / (end_at_time-180)}

[0112] Where: t1 is synonymous with time_now (but is related to the number of points in the data array used for curve fitting; here we assume that we get 1 data point per second), and the number 180 represents the number of points after the prediction starts (see Fig.16 ). The value end_at_time specifies the length of time after which the test is stopped (provided there is no noise and the PI value is not too low, e.g. end_at_time = 300).

[0113] The end_at_time value may be modified by an IF condition in the following manner: If noise in the data is detected to be above a first threshold, the test time (the value of end_at_time) is extended by a first amount (eg, from 300 to 360).

[0114] If noise in the data is detected to be above a second threshold, the test time is extended by a second amount (eg, from 300 to 420). Additionally, a check may be performed for "bad" PI values ​​(PI ratios below 1.25, 1.5, or 2 in the example).

[0115] For a "bad" PI, the measurement may be less noisy because the current is high. Therefore, if the noise is below a first threshold and the PI is below the "bad PI threshold", it is allowed to stop the test after only 180 seconds, which may be shortly after the prediction starts. The mathematical function used for approximation / extrapolation can be selected as a power function, exponential function, polynomial function or linear function, for example. It is also possible to perform curve fitting on the data of current instead of resistance, because these two values ​​are the reciprocal of each other.

[0116] You can use a combination of functions, such as Fig.15In this example, there is a power function 21 and a linear function 20, both of which are only fitted to the second half of the measurement data 17. The "residual" value (e.g., the quality of the curve fit represented by a least squares calculation) is taken into account and used as a weight for calculating the weighted average between the two functions. Fig.15 In the particular case of , due to the noise in the green data, the two functions have similar residuals. Therefore, the weighted average function 19 is very close to the point-by-point average of the two functions. Therefore, the extrapolation is effectively calculated as a weighted average of the two extrapolations towards 600 seconds.

[0117] Return to Fig.14 , which shows the generation of automatic scale in the display. Fig.14 There are 3 vertical scales shown. The first scale (left) is specified in Ω (Ohms). The second scale (right, PI) is at the end of the graph at 600 seconds, in other examples this could be 300 seconds or some other time. The PI scale is in units of ratio relative to the reading at 60 seconds or any reference point specified by the user. The third scale (DAR) is at the reference point (60 seconds, adjusted by the user).

[0118] The behavior of the automatic scale can be as follows. At the beginning of the test, the value of the reference point is unknown, so it is impossible to locate the ratio scale (PI, DAR). Therefore, the first scale (Ω) can be automatically scaled only, while the other two scales are suppressed (not shown, or grayed out, or similar). Alternatively, the first axis (Ω) can always be scaled so that in the interval from 0 seconds to 60 seconds, the current measurement value (the last value of the real data on the graph) is always scaled to correspond to ratio=1.

[0119] If the DAR and PI scales are configured to have the same reference point (e.g. 60 seconds), then once that reference point is reached, both scales can be displayed and the entire graph is scaled so that the Ω value at 60 seconds corresponds exactly to a ratio = 1 (for both DAR and PI). From this point on, the scales of the axis remain fixed and the animation is similar to Fig.16 and Fig.17 Proceed in the manner shown in .

[0120] The second scale (DAR) may have a different reference point than the third scale (PI). For example, the DAR may be calculated as 30 seconds / 15 seconds, and the PI may be calculated as 300 seconds / 60 seconds. The DAR and PI scales may then differ in their ratio ranges (e.g., rather than both extending from 0 to 6), as dictated by the measured values. Both the DAR and PI scales may have similar color coding, and be joined by "color zones" to facilitate interpretation, e.g., similar to Fig.14 , but the DAR is rescaled to different values.

[0121] The PI can be predicted using a simplified method of extrapolation of the data. The curvature of a plot of the measured data tends to decrease over time, so that changes occur at a slower rate and the curve tends to be more and more "stationary". This can be approximated by a straight line. The straight line can be derived with the help of a linear regression, which is based on a statistical treatment of the data, for example on a mean value. However, the straight line can also be obtained by a simplified algorithm, which does not require the storage of the entire data array, but is based on recursive calculations. Such an approach will greatly reduce the requirements on digital memory, since only a few different variables need to be stored to perform the prediction / extrapolation, rather than the complete data array. Such methods do not have to be mathematically rigorous, but they can provide sufficient prediction "quality" that in practice they may be found to work well enough to be useful.

[0122] This simple linear model is Fig.18 In this case, the first 60 seconds of data are ignored. Starting from 60 seconds, the data are processed to calculate the running average (moving average) by means of recursive calculation, so that only a single variable is stored in the memory. The way in which this variable changes is determined by Fig.18 The curve 50 in FIG. 4 is represented. In the presented case, at time 240 seconds, the prediction / extrapolation is to 600 seconds. The value stored in the moving average variable is taken and simply assigned as if it would belong to the 60 second position (the peak of curve 50 at 240 seconds becomes the first point on the slope of curve 48 at 60 seconds). A straight line is then extended from this new starting point, through the actual value of the currently measured IR at 240 seconds, at which point line 48 intersects the measured point 48, and extrapolated to 600 seconds.

[0123] This can be calculated by the two-point method using the function y = a*x + b. Fig.18 It can be seen that the extrapolation of curve 48 to the 600 second point provides a useful estimate of the predicted value of PI. In the case presented, the actual value is PI = 2.09, while the simplified model returns PI = 2.22. If only one digit of resolution is used, then the ideal value would be PI = 2.1, while the predicted PI+ = 2.2.

[0124] As already mentioned, the absorption effect can be modeled by more than one RC branch. It can be assumed that all such processes should occur exponentially, so if the value of the RC constant of each branch can be estimated from the measured data, a very good fit can be obtained. For example, two RC branches can be modeled and the resulting total current can be calculated, from which the resistance can then be calculated. This simple model, in which the main capacitor 31 is neglected and one main resistor and two parallel RC branches are provided, can produce a whole family with very different characteristics, which can simulate the real behavior with the same model and only changing the RC parameters. Fig.19a , Fig.19b and Fig.19c A typical example is shown in , showing different curves 51, 52 and 53. Using this R+2RC model, the current components can be analyzed and extracted from the total current curve. All components should be pure exponential curves, which should theoretically be extractable from the main curve. Fourier transform techniques for non-periodic signals can be used. In other examples, Prony, Laplace, S and Z transform techniques can be used.

[0125] As already mentioned, the measuring device may include one or more processors for causing the device to perform the method as described. The one or more processors may include a computer-readable storage medium, or communicate with a computer-readable storage medium, such as a memory chip or other data storage device. At least one of the one or more processors may be located in a user device 37 (such as a smart phone) connected to the meter 1. The computer-readable storage medium may hold instructions for causing the one or more processors to cause the measuring device to receive samples of the output from a circuit for measuring an electrical parameter of an electrical insulator to produce at least one sample, and to receive additional samples at successive time increments, and to process at least some of the samples to generate an approximate function that relates the measured electrical parameter to time. The samples may be received from the meter 1 by the user device 37. The instructions may cause the measuring device to calculate a predicted value of a quality indicator based on the electrical parameter indicated by the approximate function including a second time, and to generate an electrical signal indicating the state of the measurement based at least on the predicted value of the quality indicator. The instructions may cause the one or more processors to cause the measuring device to sample the output from the circuit for measuring the electrical parameter of the electrical insulator to produce at least one sample and send the sample to at least one of the one or more processors; and to sample the output from the circuit to produce one or more additional samples at successive time increments and send the one or more additional samples to at least one of the one or more processors. The samples may be sent via a wired connection, or via a radio link such as Bluetooth or WiFi, or by any other method. Some of the instructions may be executed by a processor at the meter, and some of the instructions may be executed by a processor at the user device.

[0126] The above embodiments are to be understood as illustrative examples of the present invention. It should be understood that any feature described in connection with any embodiment may be used alone or in combination with other features described, and may also be used in combination with one or more features of any other embodiment, or in any combination of any other embodiment. In addition, equivalents and modifications not described above may also be employed without departing from the scope of the present invention as defined by the appended claims.

Claims

1. A method of operating a measuring device for performing measurements of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time to a second value of the electrical parameter at a second time, the method comprising: sampling an output from a circuit for measuring the electrical parameter of the electrical insulator to produce at least one sample; performing the steps of: sampling an output from the circuit to produce one or more additional samples at successive time increments, and processing at least some of the samples to generate an approximate function relating a measured electrical parameter to time; calculating a predicted value for the quality indicator based on the electrical parameter indicated by the approximation function for the second time; and An electrical signal indicative of a status of a measurement is generated based at least on the predicted value of the quality indicator.

2. The method according to claim 1, wherein: The electrical parameter is electric current.

3. The method according to claim 1, wherein: The electrical parameter is resistance.

4. A method according to any preceding claim, comprising generating a measure of confidence in the predicted value of the quality indicator, in, The step of generating the electrical signal indicative of a status of the measurement is based at least on a measure of confidence in the predicted value of the quality indicator.

5. The method according to claim 4, wherein: The measure of confidence in the predicted value of the quality indicator is a confidence range for the predicted value of the quality indicator based on a measure of noise of at least some of the samples.

6. The method according to any one of claims 1 to 5, wherein: The electrical signal indicative of a status of the measurement indicates that the measurement is complete.

7. The method of claim 6, comprising ceasing measurement in response to the electrical signal indicating a status of the measurement.

8. A method according to any preceding claim, comprising generating an electrical signal from the electrical signal indicative of the status of the measurement such that the predicted value of the quality indicator is displayed.

9. A method according to any one of the preceding claims, wherein: Processing some of the samples to generate an approximate function relating measured resistance to time includes discarding a group of samples.

10. A method according to any one of the preceding claims, wherein: Processing at least some of the samples to generate an approximate function relating measured resistance to time includes least squares curve fitting.

11. The method according to any one of claims 1 to 9, wherein: Processing at least some of the samples to generate an approximate function relating measured resistance to time includes linear regression.

12. The method according to any one of claims 1 to 9, wherein: Processing at least some of the samples to generate an approximate function relating measured resistance to time includes nonlinear regression.

13. A method according to any one of the preceding claims, wherein: The quality indicator of the electrical insulator is a polarization index.

14. The method according to any one of claims 1 to 12, wherein: The quality indicator of the electrical insulator is the dielectric absorption ratio.

15. A measuring device for performing a measurement of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time and a second value of the electrical parameter at a second time, the measuring device comprising at least one processor, the at least one processor being configured to cause the measuring device to: sampling an output from a circuit for measuring the electrical parameter of the electrical insulator to produce at least one sample; performing the steps of: sampling an output from the circuit to produce one or more additional samples at successive time increments, and processing at least some of the samples to generate an approximate function relating a measured electrical parameter to time; calculating a predicted value for the quality indicator based on a value of the electrical parameter indicated by the approximation function for the second time; and An electrical signal indicative of a status of a measurement is generated based at least on the predicted value of the quality indicator.

16. The measuring device according to claim 15, wherein: The electrical parameter is electric current.

17. The measuring device according to claim 15, wherein: The electrical parameter is resistance.

18. The measuring device according to any one of claims 15 to 17, wherein: The at least one processor is configured to cause the measuring device to: generate a measure of confidence in the predicted value of the quality indicator, Therein, the step of causing the measuring device to generate the electrical signal indicative of a status of the measurement is based at least on a measure of confidence in the predicted value of the quality indicator.

19. The measuring device according to claim 18, wherein: The measure of confidence in the predicted value of the quality indicator is a confidence range for the predicted value of the quality indicator based on a measure of noise of at least some of the samples.

20. The measuring device according to any one of claims 15 to 19, wherein: The electrical signal indicative of the status of the measurement indicates that the measurement is complete.

21. The measuring device according to claim 20, wherein: The at least one processor is configured to cause the measuring device to stop measuring in response to the electrical signal indicating a status of the measurement.

22. The measuring device according to any one of claims 15 to 21, wherein: The at least one processor is configured to cause the measuring device to generate an electrical signal in dependence on the electrical signal indicative of a status of the measurement such that the predicted value of the quality indicator is displayed.

23. The measuring device according to any one of claims 15 to 22, wherein: Processing some of the samples to generate an approximate function relating measured resistance to time includes discarding a group of samples.

24. The measuring device according to any one of claims 15 to 23, wherein: The measuring device is a meter.

25. The measuring device according to any one of claims 15 to 23, wherein: The measuring apparatus comprises a meter and a user device comprising a processor in communication with a processor in the meter.

26. A computer readable storage medium retaining instructions for causing one or more processors to cause a measurement device to perform a measurement of a quality indicator of an electrical insulator, the quality indicator comprising a ratio of a first value of an electrical parameter at a first time to a second value of the electrical parameter at a second time, to: receiving a sample of the output from a circuit for measuring the electrical parameter of the electrical insulator to produce at least one sample; receiving additional samples at successive time increments and processing at least some of said samples to generate an approximate function relating the measured electrical parameter to time; calculating a predicted value for the quality indicator based on the electrical parameter indicated by the approximation function for the second time; and An electrical signal indicative of a status of the measurement is generated based at least on the predicted value of the quality indicator.

27. The computer-readable storage medium of claim 26, wherein the instructions retained therein are configured to cause one or more processors to cause the measuring device to: sampling an output from a circuit for measuring the electrical parameter of the electrical insulator to produce at least one sample and sending the sample to at least one of the one or more processors; and An output from the circuit is sampled to produce one or more further samples at successive time increments and the one or more further samples are sent to at least one of the one or more processors.

Images