Power module detection method and system based on temperature measurement

Abstract

The invention relates to the technical field of power modules, and discloses a power module detection method and system based on temperature measurement, and the method comprises the steps: obtaining the temperature data of an air inlet outside a power module, in a heat dissipation channel, other units outside the heat dissipation channel, and beside the air inlet, and calculating an air inlet temperature value, a channel temperature value, a unit temperature value, and an inlet temperature value; and calculating a curve matching value of the two thermal difference temperature value change curves according to the thermal difference temperature values, the first thermal difference temperature value and the second thermal difference temperature value, comparing the curve matching value with a preset reference matching value, and if the curve matching value is smaller than the preset reference matching value, performing early warning prompt. The temperature conditions of different positions of the power module are comprehensively considered, the detection deviation is reduced, the accuracy of temperature state detection is improved, the temperature abnormal change trend is captured, potential faults are found in advance, and reliable guarantee is provided for stable operation of the power module.

Description

Technical Field

[0001] This application relates to the technical field of power modules, and in particular, to a detection method and system for power modules based on temperature measurement. Background Art

[0002] In the safe and stable operation of power systems, the status detection of power modules is of crucial significance. As the core unit for electric energy conversion, transmission, and distribution, power modules are widely used in scenarios such as substations, distribution cabinets, and new energy power generation equipment. Their performance directly affects the reliability and efficiency of power systems. With the improvement of the integration level of power systems, the operating environment of power modules has become increasingly complex. Being in a working condition of high voltage, large current, and strong electromagnetic interference for a long time, the performance degradation of internal components is often accompanied by abnormal temperature changes. Timely and accurately detecting the status of power modules, especially abnormal temperatures, can provide early warnings of potential faults.

[0003] Currently, detection schemes based on infrared temperature detection have been widely applied. Infrared temperature measurement technology captures the infrared radiation energy on the surface of an object to non-contactedly obtain the target temperature distribution, with advantages such as strong real-time performance, wide detection range, and the ability to achieve remote monitoring. In practical applications, for the circuit breaker module in a high-voltage distribution cabinet, the temperature of the contact point of the contact is monitored through an infrared sensor. When the temperature exceeds the preset threshold, an alarm is triggered to prompt the operation and maintenance personnel to check for problems such as poor contact or oxidation.

[0004] However, the existing detection schemes based on infrared temperature have problems with insufficient accuracy of detection results in practical applications. On the one hand, infrared temperature measurement is easily affected by environmental factors such as dust pollution and differences in the surface emissivity of the object to be measured, which may lead to deviations in the temperature acquisition data. On the other hand, infrared detection can only reflect the surface temperature of the power module. For faults such as poor contact and insulation aging of internal components (such as chip solder joints and winding wires) of the module, the resulting local temperature rise may be masked by the housing or heat dissipation structure, resulting in no significant abnormality in the surface temperature and causing missed detections. Summary of the Invention

[0005] In order to improve the accuracy of power module temperature detection, this application provides a detection method and system for power modules based on temperature measurement.

[0006] In a first aspect, this application provides a detection method for power modules based on temperature measurement, adopting the following technical solution:

[0007] A detection method for power modules based on temperature measurement includes the following steps:

[0008] Obtain the first temperature data outside the power module and at the air inlet;

[0009] Calculate the air inlet temperature value based on the first temperature data;

[0010] Obtain the second temperature data inside the power module and within the heat dissipation channel;

[0011] Calculate the channel temperature value based on the second temperature data;

[0012] Obtain the third temperature data of other units inside the power module and outside the heat dissipation channel;

[0013] Calculate the unit temperature value based on the third temperature data;

[0014] Obtain the fourth temperature data inside the power module and beside the air inlet;

[0015] Calculate the inlet temperature value based on the fourth temperature data;

[0016] Calculate the difference between the channel temperature value and the inlet air temperature value as the first thermal difference temperature value;

[0017] Calculate the difference between the unit temperature value and the inlet temperature value as the second thermal difference temperature value;

[0018] Within a preset time period, calculate the first change curve of the first thermal difference temperature value and the second change curve of the second thermal difference temperature value;

[0019] Calculate the curve matching value between the first change curve and the second change curve;

[0020] Compare the curve matching value with a preset reference matching value. If the curve matching value is less than the preset reference matching value, give a warning prompt.

[0021] By adopting the above technical solution, by separately collecting the temperature data at the air inlet outside the power module, inside the heat dissipation channel, other units outside the heat dissipation channel, and beside the air inlet, and calculating different temperature values and thermal difference temperature values, it comprehensively considers the temperature conditions at different positions of the power module. Compared with only relying on the temperature detection at a single position, it can more accurately reflect the true temperature state of the power module, reduce the detection deviation caused by being masked by environmental factors or local faults, and improve the accuracy of temperature detection. The warning method based on the temperature change curve and the curve matching value can capture the abnormal temperature change trend inside the power module caused by faults such as poor contact and insulation aging. It issues a warning at a stage when the fault has not yet developed to a significant abnormal surface temperature, discovers potential faults in advance, and buys more processing time for the operation and maintenance personnel to avoid the further expansion of the fault.

[0022] Optionally, the method for calculating the inlet air temperature value based on the first temperature data includes the following steps:

[0023] The first temperature data includes a plurality of temperature values, which are data measured at multiple positions distributed outside the power module and at the air inlet;

[0024] The inlet air temperature value is the average value of the plurality of temperature values;

[0025] The method for calculating the inlet temperature value based on the fourth temperature data includes the following steps:

[0026] The fourth temperature data includes a plurality of temperature values, which are data measured at multiple positions distributed inside the power module and beside the air inlet;

[0027] The inlet temperature value is the average value of the plurality of temperature values.

[0028] By adopting the above technical solution, through the averaging process of multiple data, the influence of environmental factors on individual temperature measurement points can be offset to a certain extent, comprehensively reflecting the temperature situation in this area, making the calculated temperature value have a certain representativeness and reliability.

[0029] Optionally, the method for calculating the unit temperature value based on the third temperature data includes the following steps:

[0030] The third temperature data includes a plurality of temperature values, which are data measured at multiple positions distributed inside the power module and at multiple positions of other units outside the heat dissipation channel;

[0031] The unit temperature value is the average value of the plurality of temperature values.

[0032] By adopting the above technical solution, the overall temperature situation of other units outside the heat dissipation channel can be more comprehensively reflected. Avoiding the deviation that may occur when only considering the temperature at individual positions makes the evaluation of the internal temperature state of the power module more accurate and reliable.

[0033] Optionally, the method for calculating the unit temperature value based on the third temperature data may also be the following steps:

[0034] According to the temperature data of each temperature measurement point and its corresponding weight, the unit temperature value is calculated by using the weighted average method.

[0035] By adopting the above technical solution, through the weighted average calculation, the temperature information of each temperature measurement point at the air inlet can be comprehensively considered, and a value that can more accurately reflect the true inlet air temperature can be obtained.

[0036] Optionally, the method for calculating the channel temperature value based on the second temperature data includes the following steps:

[0037] The second temperature data includes a plurality of temperature values, which are the data measured at multiple positions distributed within the power module and located in the heat dissipation channel;

[0038] The channel temperature value is the average value of the plurality of temperature values.

[0039] By adopting the above technical solution, there may be differences in the temperatures at different positions within the heat dissipation channel. Through the average calculation of the temperature values at multiple positions, the temperature distribution within the channel can be comprehensively considered, and a more representative channel temperature value can be obtained.

[0040] Optionally, the method for calculating the channel temperature value based on the second temperature data may also include the following steps:

[0041] Calculate the channel temperature value by using the weighted average method according to the temperature data of each temperature measurement point and its corresponding weight.

[0042] By adopting the above technical solution, the temperature in the area with a larger wind speed changes rapidly, and the thermal response is closer to the inlet air temperature, which is less representative of the overall channel temperature; while in the area with a smaller wind speed, heat accumulation is likely to occur, which can better reflect the actual heat generation situation inside the module.

[0043] Optionally, the method for calculating the curve matching value between the first change curve and the second change curve further includes the following steps:

[0044] Among them, the first change curve and the second change curve are the temperature change curves of the corresponding heat difference temperature values on the time axis;

[0045] Calculate the curvature change rate, inflection point position, and extreme point distribution of the two curves, and calculate the shape similarity according to the curvature change rate, inflection point position, and extreme point distribution;

[0046] Calculate the first derivative and second derivative of the two curves at multiple time points, and calculate the trend similarity according to the first derivative and second derivative;

[0047] Adopt the cross-correlation function to analyze the phase relationship between the two curves, and calculate the phase similarity;

[0048] Calculate the matching value according to the shape similarity, trend similarity, and phase similarity, and the matching value = α×shape similarity + β×trend similarity + γ×phase similarity; where α, β, and γ are weight coefficients;

[0049] Normalize the matching value to the interval [0, 1] as the curve matching value. The closer the curve matching value is to 1, the more similar the curves are, and the closer it is to 0, the greater the difference.

[0050] By adopting the above technical solution, through multi-dimensional feature fusion and weighted calculation, the first and second change curves are comprehensively analyzed from the three aspects of shape, trend and phase, and the subtle differences in the temperature change curves are accurately captured. It can keenly identify anomalies in the early stage of power module failure, and can adapt to different equipment and working conditions by adjusting the weights; at the same time, it has good anti-interference ability and computing efficiency, not only realizing quantitative evaluation of curve similarity, but also assisting fault location based on indicators of various dimensions, providing reliable technical support for power module status monitoring and fault warning.

[0051] Optionally, the method further comprises the following steps:

[0052] Obtain the data fluctuation amplitude of the temperature sensor at the set temperature measurement point;

[0053] The size of α is adjusted according to the anti-correlation of the data fluctuation range. The larger the data fluctuation range, the smaller α is; the smaller the data fluctuation range, the larger α is;

[0054] Get the load rate of the power module;

[0055] The value of β is adjusted according to the positive correlation of the load rate. The larger the load rate, the larger β is; the smaller the load rate, the smaller β is;

[0056] Obtain the electromagnetic interference intensity in the power module environment;

[0057] The value of γ is adjusted according to the anti-correlation of the electromagnetic interference intensity. The stronger the electromagnetic interference intensity, the smaller γ is; the weaker the electromagnetic interference intensity, the larger γ is.

[0058] By adopting the above technical solution, a dynamic weight adjustment mechanism is constructed by acquiring parameters such as the temperature sensor data fluctuation amplitude, the power module load rate, and the environmental electromagnetic interference intensity in real time. The shape similarity weight α is adjusted according to the anti-correlation of the data fluctuation amplitude to suppress noise interference; the trend similarity weight β is adjusted according to the positive correlation of the load rate to strengthen fault monitoring under high load; the phase similarity weight γ is adjusted based on the anti-correlation of the electromagnetic interference intensity to resist the influence of electromagnetic interference. The synergistic effect of various parameters enables the algorithm to adapt to the complex operating environment of the power module, while improving the accuracy of fault detection and reducing the risk of false alarms.

[0059] Optionally, a heat-deformable channel made of a heat-deformable material is provided in the heat dissipation channel, and the higher the temperature in the channel, the narrower the heat-deformable channel becomes after the heat-deformable material is bent; the lower the temperature in the channel, the wider the heat-deformable channel becomes after the heat-deformable material is bent;

[0060] The method further comprises the steps of:

[0061] Obtain channel width data at the temperature measurement point in the thermal deformation channel;

[0062] Set the distribution ratio of the corresponding weights in an anti-correlation manner according to the distribution ratio of the width data. The wider the channel width of the temperature measurement point, the smaller the weight ratio.

[0063] By adopting the above technical solution, the thermal deformation channel can automatically adjust the channel cross-sectional area according to the temperature, realizing "adaptive heat dissipation". When the temperature rises, the channel narrows, forcing the air flow to accelerate, enhancing the convective heat transfer effect and quickly taking away the heat; when the temperature drops, the channel widens to reduce the wind resistance and lower the energy consumption of the heat dissipation system. This dynamic adjustment mechanism enables the power module to maintain efficient heat dissipation under different working conditions, avoiding device aging and performance degradation caused by long-term high-temperature operation. Under the light load condition of the power module, the thermal deformation channel is wider. At this time, the system increases the weight of the low-sensitivity area, reduces the response to small temperature fluctuations, and lowers the energy consumption of the detection system; while under heavy load and high temperature, the channel shrinks, and the system automatically focuses on the temperature changes in the key area, improving the fault detection sensitivity.

[0064] In the second aspect, the present application provides a power module detection system based on temperature measurement, adopting the following technical solution:

[0065] A power module detection system based on temperature measurement includes a processor, and the processor executes the steps of the power module detection method based on temperature measurement described in any one of the above.

[0066] In summary, the present application includes at least one of the following beneficial technical effects:

[0067] By collecting the temperature data at multiple positions of the power module and calculating the different temperature values and the thermal difference temperature values, it comprehensively covers the key areas inside and outside the power module, effectively overcomes the limitations of single temperature measurement, greatly improves the accuracy of temperature detection, and reduces the detection deviation caused by environmental interference and fault hiding.

[0068] Based on the matching analysis of the thermal difference temperature value change curve, it can sensitively capture the temperature anomaly trend caused by early faults inside the power module, give early warnings before the faults cause significant surface temperature changes, gain valuable time for operation and maintenance, and prevent the faults from deteriorating and escalating.

[0069] Whether it is the weighted average in the temperature value calculation, the automatic adjustment of the thermal deformation channel, or the dynamic adjustment of the curve matching weight, it can optimize the detection strategy in real time according to the operating conditions of the power module and environmental factors, ensuring that the system maintains high performance under complex and changeable conditions.

[0070] The thermal deformation channel automatically adjusts the cross-sectional area of the heat dissipation channel according to the temperature, realizing the dynamic balance between the heat dissipation efficiency and the energy consumption. It can not only strengthen the heat dissipation to protect the device at high temperature, but also reduce the energy consumption at low temperature, prolonging the service life of the power module.

[0071] Curve matching adopts multi-dimensional feature fusion calculation and combines dynamic weight adjustment. It can not only accurately quantify the curve similarity, but also assist in fault location based on indicators such as shape, trend, and phase, providing a solid and reliable technical support for the state assessment of power modules and effectively reducing the risks of false alarms and missed reports. Description of the Drawings

[0072] Figure 1 It is a step diagram of a power module detection method based on temperature measurement.

[0073] Figure 2 It is a step diagram for calculating the curve matching value between the first change curve and the second change curve.

[0074] Figure 3 It is a step diagram for dynamically adjusting the weight coefficients α, β, and γ. Detailed Implementation Modes

[0075] The following details the implementation modes of the present application, and the examples of the implementation modes are shown in the drawings.

[0076] In the description of this specification, the descriptions referring to the terms "certain implementation modes", "one implementation mode", "some implementation modes", "schematic implementation modes", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the implementation mode or example are included in at least one implementation mode or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same implementation mode or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more implementation modes or examples.

[0077] The embodiments of the present application disclose a power module detection method based on temperature measurement, referring to Figure 1 , and including the following steps:

[0078] Obtain the first temperature data outside the power module and at the air inlet, and calculate the air inlet temperature value according to the first temperature data. The first temperature data includes multiple temperature values, and the multiple temperature values are data measured at multiple positions distributed outside the power module and at the air inlet; the air inlet temperature value is the average value of the multiple temperature values. For example, in the air inlet area of a large high-voltage switchgear cabinet, 5 temperature sensors are arranged at intervals of 30 cm along the horizontal direction to obtain the ambient temperature data at different points of the air inlet in real time. The calculation of the air inlet temperature value adopts the average value method, that is, the arithmetic average of the multiple temperature values is taken to eliminate the single-point data deviation caused by local air flow disturbance or environmental differences, and a value that can reflect the overall temperature level of the air inlet is obtained.

[0079] Obtain the second temperature data inside the power module and within the heat dissipation channel, and calculate the channel temperature value based on the second temperature data. The second temperature data includes multiple temperature values, which are data measured at multiple positions distributed inside the power module and within the heat dissipation channel. In the heat dissipation channel, deploy a temperature sensor array at key positions near heat-generating components, bends, air outlets, etc., to collect temperature data at multiple positions. Taking a certain model of power converter as an example, 8 temperature measurement points are evenly distributed in its heat dissipation channel, covering different heights and cross-sectional positions of the channel.

[0080] There are two implementation methods for calculating the channel temperature value:

[0081] The first method: The channel temperature value is the average of multiple temperature values. Using the simple average method, directly average the temperature values of each temperature measurement point. This method is simple to calculate and can quickly obtain a comprehensive value reflecting the channel temperature distribution, and is suitable for scenarios where the air flow in the heat dissipation channel is relatively uniform and the temperature difference is small.

[0082] The second method: Calculate the channel temperature value using the weighted average method according to the temperature data of each temperature measurement point and its corresponding weight; among them, the weighted value is a set value corresponding to the power equipment, which is a value determined after experimental simulation. In addition, based on the weighted average method, the weight can be dynamically adjusted according to the wind speed magnitude at each temperature measurement point. For example, when the wind speed detected at a certain temperature measurement point reaches 8 m / s (high wind speed), its weight is set to 0.1; while for another temperature measurement point with a wind speed of only 2 m / s (low wind speed), the weight is set to 0.3. In this way, reduce the influence of areas with high wind speed, rapid temperature change but weak representativeness, and increase the weight ratio of areas with low wind speed and easy heat accumulation, so that the calculation result is more in line with the actual thermal state of the heat dissipation channel.

[0083] Obtain the third temperature data of other units inside the power module and outside the heat dissipation channel, and calculate the unit temperature value based on the third temperature data. The third temperature data includes multiple temperature values, which are data measured at multiple positions of other units distributed inside the power module and outside the heat dissipation channel. Arrange multiple temperature sensors near the core components such as chips and windings inside the module to collect temperature information at different unit positions. Taking a power transformer as an example, 6 temperature measurement points are set at key positions such as the upper, middle, and lower parts of its winding and the surface of the iron core.

[0084] There are two implementation methods for calculating the unit temperature value:

[0085] The first method: The unit temperature value is the average of multiple temperature values. Using the average value calculation can comprehensively reflect the overall temperature condition of each unit outside the heat dissipation channel and avoid misjudgment caused by abnormal individual measurement points.

[0086] The second method: Calculate the unit temperature value by using the weighted average method based on the temperature data of each temperature measurement point and its corresponding weight; where the weighted value is a set value corresponding to the power equipment and is a value determined after experimental simulation. In addition, based on the weighted average method, weights can be assigned according to factors such as the distance of each temperature measurement point from the core heat-generating component and the heat dissipation conditions. For example, a temperature measurement point closer to the transformer winding is given a higher weight because it can more directly reflect the winding heating situation; while a measurement point farther away and more affected by the environment has a correspondingly reduced weight, so as to more accurately obtain the true temperature of the unit.

[0087] Obtain the fourth temperature data inside the power module and beside the air inlet, and calculate the inlet temperature value according to the fourth temperature data. The fourth temperature data includes multiple temperature values, and the multiple temperature values are data measured at multiple positions distributed inside the power module and beside the air inlet. The inlet temperature value is the average of the multiple temperature values. The data is collected by multiple temperature sensors arranged around the air inlet, and the inlet temperature value is also calculated by the average value method. In an actual switchgear power module, a total of 3 sensors are set on the left and right sides and above the air inlet, and a stable inlet temperature value is obtained after average calculation.

[0088] Calculate the difference between the channel temperature value and the inlet air temperature value as the first thermal difference temperature value, and the first thermal difference temperature value reflects the heat exchange efficiency of the heat dissipation channel.

[0089] Calculate the difference between the unit temperature value and the inlet temperature value as the second thermal difference temperature value, and the second thermal difference temperature value is used to evaluate the heating degree of the core unit inside the module.

[0090] Within a preset 1-hour time period, with a 1-minute time interval, respectively plot the first change curve of the first thermal difference temperature value and the second change curve of the second thermal difference temperature value. By calculating the curvature change rate, inflection point position, extreme point distribution and other characteristics of the two curves, analyze the shape similarity; compare the first derivative (temperature change rate) and the second derivative (change acceleration) at multiple time points to evaluate the trend similarity; use the cross-correlation function to explore the phase relationship of the curves to obtain the phase similarity. Finally, comprehensively calculate the curve matching value based on these indicators and compare it with the preset reference matching value.

[0091] If the curve matching value is less than the preset reference matching value, it indicates that there may be an abnormality inside the power module. For example, during the operation of a certain power converter, it is found through this detection method that the matching values of the first and second change curves continue to decline and are lower than the reference value. After investigation, it is determined that the local heating is caused by poor contact of the internal power devices, and the heat exchange efficiency of the heat dissipation channel and the temperature change trend of the core unit are abnormal. The system promptly triggers an alarm, and the maintenance personnel quickly handle it, avoiding the expansion of the fault and ensuring the safe and stable operation of the power module.

[0092] By separately collecting the temperature data at the external air inlet of the power module, inside the heat dissipation channel, other units outside the heat dissipation channel, and beside the air inlet, and calculating different temperature values and differential temperature values, this method can more accurately reflect the true temperature state of the power module compared to relying only on temperature detection at a single location, reduce the detection deviation caused by being masked by environmental factors or local faults, and improve the accuracy of temperature detection. The early warning method based on the temperature change curve and the curve matching value can capture the abnormal temperature change trend inside the power module caused by faults such as poor contact and insulation aging. It can issue an early warning at the stage when the fault has not yet developed to a significant abnormal surface temperature, discover potential faults in advance, gain more processing time for operation and maintenance personnel, avoid the further expansion of the fault, and effectively improve the reliability and stability of the power system operation.

[0093] Referring to Figure 2 , the method for calculating the curve matching value between the first change curve and the second change curve further includes the following steps:

[0094] Among them, the first change curve and the second change curve are the temperature change curves of the corresponding differential temperature values on the time axis.

[0095] Calculate the curvature change rate, inflection point position, and extreme point distribution of the two curves, and calculate the shape similarity based on the curvature change rate, inflection point position, and extreme point distribution. Based on differential geometry theory, perform point-by-point curvature calculation on the first and second change curves. Use the five-point difference method to fit the local shape of the curve, and dynamically solve the curvature value of each point through the formula where k represents the curvature value of the curve, x and y represent time and differential temperature respectively, and x′, y′ are the first derivatives, , are the second derivatives. On this basis, detect the area where the curvature change rate exceeds the threshold (such as 0.5 °C / min²) through the sliding window algorithm to locate the inflection point of the curve; use the local extreme value search algorithm to identify the maximum and minimum points.

[0096] Taking the fault data of a certain high-voltage frequency converter as an example, when operating normally, the fluctuation range of the curvature change rate of the two curves is within ±0.3 °C / min². When a power module aging fault occurs, the first curve has a curvature mutation (the change rate reaches 1.2 °C / min²) at the 30th minute, while the second curve remains stable. By comparing the position differences of the inflection points and extreme points, the quantified shape similarity is calculated to be 0.62.

[0097] Calculate the first derivatives and second derivatives of multiple time points of the two curves, and calculate the trend similarity based on the first derivatives and second derivatives. Perform time series differential processing on the curve, and use the central difference method to calculate the first derivative and the second derivative , where Δt is the sampling interval (set to 1 minute in this solution). Through the Dynamic Time Warping (DTW) algorithm, the time axes of the two curves are aligned to ensure the synchronization of derivative calculation. yt represents the thermal difference temperature value corresponding to the t-th sampling moment in the time series. It is a data point in the time series, reflecting the thermal difference temperature state at that moment. yt+1 represents the thermal difference temperature value corresponding to the (t + 1)-th sampling moment in the time series. That is, the thermal difference temperature value at the next sampling moment immediately following yt, which is used to cooperate with yt to calculate the change of the curve during this period.

[0098] In a monitoring scenario of a certain photovoltaic inverter, when a fault occurs where the rotational speed of the cooling fan decreases, the first derivative of the first curve rises from the normal 0.1 °C / min to 0.4 °C / min, and the second derivative changes from 0 to 0.05 °C / min²; while the derivative change of the second curve lags by approximately 15 minutes. By calculating the Euclidean distance of the derivative curves, the trend similarity is obtained as 0.58.

[0099] The cross-correlation function is used to analyze the phase relationship between the two curves and calculate the phase similarity. The fast Fourier transform (FFT) is used to convert the time-domain curve into a frequency-domain signal to identify the dominant frequency components of the curve. Using the cross-correlation function Calculate the correlation at different delay times τ, where x(t) and y(t) are the sampling values of the two curves respectively. By finding the peak position of the cross-correlation function, the phase difference is determined.

[0100] Among them, Rxy(τ) is the cross-correlation function, which is used to analyze the correlation between the two curves x(t) and y(t) at different delay times τ;

[0101] y(t) is a function of time t, representing the value corresponding to the curve of the thermal difference temperature at time t, that is, the value of the curve at a specific moment;

[0102] The value of the function y at time t + τ, τ is the delay time. In the calculation of the cross-correlation function, it is multiplied by x(t) and summed up to measure the degree of association between the two curves at different delays.

[0103] In the detection of a certain wind turbine converter, when operating normally, the phase difference between the two curves remains within ±5 minutes; when harmonic interference causes abnormal heat dissipation, the phase difference expands to 12 minutes, and the phase similarity is calculated as 0.71 through normalization processing.

[0104] The matching value is calculated based on shape similarity, trend similarity, and phase similarity. The matching value = α × shape similarity + β × trend similarity + γ × phase similarity; where α, β, and γ are weight coefficients. An adaptive weight adjustment strategy is adopted to adjust α, β, and γ in real time according to the operating characteristics of the power module. For example, for a power transformer with a complex heat dissipation structure, set α = 0.4, β = 0.3, γ = 0.3, focusing on shape similarity analysis; while for a high-frequency switching power supply module, due to rapid temperature changes, adjust to α = 0.3, β = 0.4, γ = 0.3 to strengthen trend detection.

[0105] The matching value is normalized to the interval [0, 1] as the curve matching value. The closer the curve matching value is to 1, the more similar the curves are, and the closer it is to 0, the greater the difference. The three-dimensional similarity index is weighted and fused through the formula M = α × Ss + β × St + γ × Sp, where Ss, St, and Sp represent shape, trend, and phase similarity respectively. Finally, through normalization processing M′ = (M - Mmin) / (Mmax - Mmin), the matching value is mapped to the interval [0, 1].

[0106] In practical applications, in a 220 kV substation circuit breaker monitoring system, through this algorithm, at the initial stage of abnormal contact resistance of the contact (when the surface temperature only rises by 2°C), it is detected that the curve matching value drops from 0.92 to 0.78, triggering an early warning 2 hours in advance. Compared with the traditional threshold detection method, the early warning time is improved.

[0107] Through the above multi-dimensional feature fusion and dynamic weighted calculation, this solution realizes the in-depth analysis of the temperature change curve of the power module. It can not only accurately capture the subtle temperature difference at the 0.1°C level, but also effectively distinguish normal fluctuations from fault anomalies through three-dimensional feature cross-validation. The adaptive weight mechanism makes it suitable for the monitoring needs of different types of power equipment, and it performs excellently in anti-electromagnetic interference, sensor noise suppression, etc., providing a reliable technical guarantee for the intelligent operation and maintenance of the power system.

[0108] Refer to Figure 3 , the method further includes the following steps:

[0109] The system collects the original data of the temperature sensors at the set temperature measurement points in real time, and uses the sliding standard deviation algorithm to calculate the data fluctuation amplitude. Specifically, with a 5-minute sliding window, calculate the standard deviation of the temperature data within the window, which can intuitively reflect the degree of dispersion of the data. Let the data fluctuation amplitude be X (measured by the standard deviation), and the initial value of the shape similarity weight α be α0. According to the inverse correlation between the data fluctuation amplitude and the adjustment of α, a linear relationship α = α0 - k1 × X can be established. Where k1 is the adjustment coefficient, obtained by fitting a large amount of actual monitoring data, reflecting the sensitivity of the data fluctuation amplitude to the value of α.

[0110] For example, in a monitoring scenario of a high-speed rail traction converter, due to the continuous vibration during train operation, the standard deviation of the sensor data at a certain temperature measurement point suddenly rises from 0.3 °C in the normal state to 1.5 °C. The system immediately triggers the weight adjustment mechanism and lowers the shape similarity weight α from the initial value of 0.4 to 0.2. This adjustment enables the shape similarity calculation based on curvature and inflection point analysis to effectively avoid the noise interference caused by vibration, and the misjudgment rate drops from 18% before adjustment to 5%. When the train stops, the vibration weakens, the data standard deviation drops back to 0.4 °C, and the α value immediately returns to 0.35, enhancing the ability to capture subtle curve shape changes again.

[0111] By collecting the real-time current and voltage data of the power module and combining with the rated parameters of the equipment, the current load rate is calculated. Let the load rate be Y, and the initial value of the trend similarity weight β be β0. The value of β is adjusted positively correlated with the load rate, and the linear relationship can be expressed as β = β0 + k2×(Y - Y0). Where k2 is the adjustment coefficient and Y0 is the initial load rate. During the operation of a box-type transformer in a wind farm, when the wind speed suddenly increases and the fan generates electricity at full power, the transformer load rate jumps from 60% to 95%. After the system detects the change in the load rate, it automatically increases the trend similarity weight β from 0.3 to 0.5, enhancing the sensitivity to the temperature change trend. At this time, the algorithm can accurately capture the abnormal trend that the winding temperature change rate accelerates from 0.2 °C / min to 0.8 °C / min and issues an overheat warning 30 minutes in advance. During the low-wind speed period at night, the load rate drops to 30%, and the β value is lowered to 0.2, avoiding false alarms triggered by normal temperature fluctuations and effectively balancing the detection sensitivity and stability.

[0112] Use a dedicated electromagnetic environment monitoring device to obtain the electromagnetic interference intensity (unit: dBμV / m) around the power module in real time. Let the electromagnetic interference intensity be z (unit: dBμV / m), and the initial value of the phase similarity weight γ be γ0. The value of γ is adjusted anti-correlated with the electromagnetic interference intensity, and the linear relationship is γ = γ0 - k3×(z - z0). Where k3 is the adjustment coefficient and z0 is the initial electromagnetic interference intensity.

[0113] In the GIS equipment area of a 500 kV substation, when a high-voltage switch operation generates a strong electromagnetic pulse, the interference intensity suddenly soars from the background value of 45 dBμV / m to 85 dBμV / m. The system quickly reduces the phase similarity weight γ from 0.3 to 0.1 to prevent misjudgment caused by the phase shift of temperature data due to electromagnetic interference. At the same time, the system automatically enables a filtering algorithm to preprocess the original data. When the interference weakens, the γ value gradually rises back to 0.25, and the phase information is used again to detect the periodic speed abnormality of the cooling fan. In practical applications, this mechanism reduces the number of false alarms caused by electromagnetic interference by 70%.

[0114] In a complex industrial environment, the above three-parameter adjustment mechanism presents a significant synergistic effect. For example, when the power module supporting an electric arc furnace in a steel plant is operating, the high load operation of the equipment (load rate of 90%) causes the β value to increase to 0.5; at the same time, the strong electromagnetic interference generated by the electric arc (78 dBμV / m) reduces the γ value to 0.15; in addition, the high temperature environment leads to an increase in the fluctuation of sensor data (standard deviation of 1.2 °C), and the α value is adjusted down to 0.25. This dynamic weight combination enables the algorithm to accurately identify the abnormal temperature curve caused by the aging of power devices in a high-noise and strong-interference environment, successfully improving the fault warning accuracy rate from 72% of the traditional method to 91%, while reducing the false alarm rate to less than 3%.

[0115] By real-time sensing of multi-dimensional operating parameters such as data fluctuation, load rate, and electromagnetic interference intensity, the dynamic weight adjustment mechanism constructed by this detection method realizes the adaptive optimization of the detection algorithm. Each weight coefficient is adjusted in real time according to the equipment operating state and environmental conditions, effectively balancing the detection sensitivity and anti-interference ability, significantly improving the accuracy and reliability of power module fault diagnosis, and providing strong technical support for the intelligent operation and maintenance of industrial power equipment.

[0116] A thermally deformable channel made of thermally deformable materials (such as shape memory alloys, thermally actuated polymers, etc.) is innovatively arranged in the heat dissipation channel. This channel has unique thermal response characteristics: when the temperature in the channel rises, the thermally deformable material expands and bends due to heat, resulting in a decrease in the cross-sectional area of the channel and the narrowing of the channel; conversely, when the temperature drops, the material contracts and recovers, and the channel widens accordingly.

[0117] Based on the improvement of the above structure, the method includes the following steps:

[0118] In the data acquisition link, the channel width data at multiple temperature measurement points in the thermally deformable channel is obtained in real time through high-precision displacement sensors (such as laser rangefinders, capacitive displacement sensors). Taking the heat dissipation channel of a high-voltage inverter as an example, 5 monitoring points are arranged at equal intervals along the axial direction of the channel, and each point is equipped with a micro laser ranging device, which can accurately measure the distance between the inner walls of the channel, and the measurement accuracy reaches the micron level. The system analyzes the collected width data in real time and calculates the distribution ratio of the width values of each measurement point in the total width data. For example, when the power module is in normal operation, the width distribution of each measurement point is relatively uniform, and the corresponding weights of each measurement point are relatively balanced; once local overheating occurs inside the module, the thermally deformable channel near the overheated area shrinks significantly due to the increase in temperature, and the proportion of the width value of this measurement point in the overall distribution decreases significantly.

[0119] Based on the above-mentioned width data distribution characteristics, the system uses an anti-correlation strategy to set the distribution ratio of the corresponding weights, that is, the larger the channel width of the temperature measurement point, the smaller its weight ratio. This strategy deeply fits the physical characteristics of the thermal deformation channel: areas with larger widths are usually in low-temperature environments, with relatively gentle temperature changes, and are less representative of the overall thermal state; while areas with smaller widths often correspond to high-temperature hotspots, which are high-incidence areas of potential fault hazards and need to be given higher weights for key monitoring. For example, when it is detected that the channel width of a certain measurement point has shrunk to 60% of the initial value, its weight ratio is increased from the default 20% to 40%, allowing the system to focus more on temperature changes in that area.

[0120] From the perspective of heat dissipation optimization, the thermal deformation channel builds a closed-loop adaptive system of "temperature-structure-heat dissipation". When the power module is running under heavy load, the internal components generate a lot of heat, causing the temperature of the thermal deformation channel to rise sharply, and the channel becomes narrower. The narrow channel forces the air flow rate to increase. According to the principle of convective heat transfer, the increase in flow rate can significantly enhance the heat exchange efficiency between the air and the heating components. Taking a certain power transformer as an example, when running at full load, the contraction of the thermal deformation channel increases the internal wind speed from 3m / s to 6m / s, and the convective heat transfer coefficient of the winding surface increases by about 40%, effectively suppressing the continuous rise in temperature. When the module enters a light-load condition, the heat generation is reduced, the channel automatically expands, and the wind resistance is reduced. At this time, the cooling fan can operate at a lower speed, which reduces energy consumption by about 30% compared with the traditional constant wind speed cooling solution.

[0121] At the fault monitoring level, this mechanism achieves a dynamic balance between detection sensitivity and energy consumption. Under light load conditions, the thermal deformation channel is in a relatively wide state, and the system actively increases the weight of the low-sensitivity area, lowers the response threshold to normal temperature fluctuations, reduces unnecessary data collection and analysis, and reduces the energy consumption of the detection system by about 25%; in a heavy load and high temperature environment, the channel shrinkage prompts the system to automatically focus on the key high-temperature area. By increasing the weight of the high-sensitivity area, it can capture small temperature changes of 0.5°C. Compared with the traditional average weighted monitoring method, the fault warning time is about 40% earlier.

[0122] In addition, the physical structural characteristics of the thermal deformation channel give it a natural fault indication function. When serious faults such as cooling fan stalling and cooling fin blockage occur inside the power module, the thermal deformation channel will undergo irreversible over-contraction due to continued high temperature. Operation and maintenance personnel can directly observe the abnormal channel morphology through the visualization window and quickly locate the fault area by combining the weight distribution data of the monitoring system. This combination of physical and digital monitoring significantly improves the intuitiveness and accuracy of power module fault diagnosis.

[0123] The embodiment of the present application also discloses a power module detection system based on temperature measurement, including a processor, and the processor executes the steps of the power module detection method and system based on temperature measurement as described in any one of the above.

[0124] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A power module detection method based on temperature measurement, characterized in that, It includes the following steps: Obtain the first temperature data outside the power module and at the air inlet; Calculate the air inlet temperature value according to the first temperature data; Obtain the second temperature data inside the power module and within the heat dissipation channel; Calculate the channel temperature value according to the second temperature data; Obtain the third temperature data of other units inside the power module and outside the heat dissipation channel; Calculate the unit temperature value according to the third temperature data; Obtain the fourth temperature data inside the power module and beside the air inlet; Calculate the inlet temperature value according to the fourth temperature data; Calculate the difference between the channel temperature value and the air inlet temperature value as the first thermal difference temperature value; Calculate the difference between the unit temperature value and the inlet temperature value as the second thermal difference temperature value; Within a preset time period, calculate the first change curve of the first thermal difference temperature value and the second change curve of the second thermal difference temperature value; Calculate the curve matching value between the first change curve and the second change curve; Compare the curve matching value with a preset reference matching value. If the curve matching value is less than the preset reference matching value, give a warning prompt.

2. The power module detection method based on temperature measurement according to claim 1, wherein The method for calculating the air inlet temperature value according to the first temperature data includes the following steps: The first temperature data includes multiple temperature values, which are data measured at multiple positions outside the power module and at the air inlet; The air inlet temperature value is the average of the multiple temperature values; The method for calculating the inlet temperature value according to the fourth temperature data includes the following steps: The fourth temperature data includes multiple temperature values, which are data measured at multiple positions inside the power module and beside the air inlet; The inlet temperature value is the average of the multiple temperature values.

3. The method for detecting a power module based on temperature measurement according to claim 2, wherein The method for calculating the unit temperature value according to the third temperature data includes the following steps: The third temperature data includes multiple temperature values, which are data measured at multiple positions of other units inside the power module and outside the heat dissipation channel; The unit temperature value is the average of the multiple temperature values.

4. The method for detecting a power module based on temperature measurement according to claim 1 or 2, characterized in that The method for calculating the unit temperature value according to the third temperature data includes the following steps: The third temperature data includes multiple temperature values, which are data measured at multiple positions of other units inside the power module and outside the heat dissipation channel; Calculate the unit temperature value by using the weighted average method according to the temperature data of each temperature measurement point and its corresponding weight.

5. The method for detecting a power module based on temperature measurement according to claim 1 or 2, characterized in that, The method for calculating the channel temperature value according to the second temperature data includes the following steps: The second temperature data includes multiple temperature values, which are data measured at multiple positions inside the power module and within the heat dissipation channel; The channel temperature value is the average of the multiple temperature values.

6. The temperature measurement-based power module detection method according to claim 1 or 2, characterized in that The method for calculating the channel temperature value according to the second temperature data includes the following steps: The second temperature data includes multiple temperature values, which are data measured at multiple positions inside the power module and within the heat dissipation channel; Calculate the channel temperature value by using the weighted average method according to the temperature data of each temperature measurement point and its corresponding weight.

7. The temperature-based power module detection method according to claim 1, wherein The method for calculating the curve matching value between the first change curve and the second change curve further includes the following steps: Wherein, the first change curve and the second change curve are temperature change curves of corresponding heat difference temperature values on the time axis; Calculate the curvature change rate, inflection point position and extreme point distribution of the two curves, and calculate the shape similarity according to the curvature change rate, inflection point position and extreme point distribution; Calculate the first derivative and second derivative of the two curves at multiple time points, and calculate the trend similarity according to the first derivative and second derivative; Analyze the phase relationship between the two curves by using the cross-correlation function, and calculate the phase similarity; Calculate the matching value according to the shape similarity, trend similarity and phase similarity, and the matching value = α×shape similarity + β×trend similarity + γ×phase similarity; wherein, α, β and γ are weight coefficients; Normalize the matching value to the interval [0, 1] as the curve matching value. The closer the curve matching value is to 1, the more similar the curves are, and the closer it is to 0, the greater the difference is.

8. The temperature-based power module detection method according to claim 7, wherein The method further includes the following steps: Obtain the data fluctuation amplitude of the temperature sensor at the set temperature measurement point; Anticorrelate and adjust the magnitude of α according to the data fluctuation amplitude. The greater the data fluctuation amplitude, the smaller α; the smaller the data fluctuation amplitude, the greater α; Obtain the load rate of the power module; Correlate and adjust the value of β according to the load rate. The greater the load rate, the greater β; the smaller the load rate, the smaller β; Obtain the electromagnetic interference intensity in the environment of the power module; Anticorrelate and adjust the value of γ according to the electromagnetic interference intensity. The stronger the electromagnetic interference intensity, the smaller γ; the weaker the electromagnetic interference intensity, the greater γ.

9. The method for detecting a power module based on temperature measurement according to claim 6, wherein A thermally deformable channel made of a thermally deformable material is arranged in the heat dissipation channel. The higher the temperature in the channel, the narrower the thermally deformable channel after the thermally deformable material is bent; the lower the temperature in the channel, the wider the thermally deformable channel after the thermally deformable material is bent; The method further includes the following steps: Obtain the width data of the channel at the temperature measurement point in the thermally deformable channel; Anticorrelate and set the distribution proportion of the corresponding weights according to the distribution proportion of the width data. The greater the channel width at the temperature measurement point, the smaller the weight proportion.

10. A power module detection system based on temperature measurement, characterized in that, It includes a processor, and the steps of the power module detection method based on temperature measurement according to any one of claims 1-9 are executed in the processor.

Images