Method for measuring slip ratio of three-phase asynchronous motor

Abstract

The invention relates to the technical field of electrical mechanical parameter measurement, in particular to a slip ratio measurement method of a three-phase asynchronous motor. The method comprises the following steps: carrying out temperature gradient analysis and copper loss temperature rise ratio calculation on the three-phase asynchronous motor to obtain a thermal feature vector; carrying out air gap area capacitance measurement on the motor stator to obtain a fusion capacitance signal; performing rotor period feature extraction on the fused capacitance signal to obtain a rotor slot frequency feature spectrum; performing thermal field-capacitance field characteristic correlation on the thermal characteristic vector and the rotor slot frequency characteristic spectrum to obtain a characteristic correlation matrix; performing double-field reliability weight dynamic allocation under different working conditions according to the feature correlation matrix to obtain a fusion credibility index; and calculating an initial slip rate estimated value based on the thermal feature vector and the rotor slot frequency feature spectrum to obtain a dual-source slip rate value. According to the invention, the precision and accuracy of slip ratio measurement in a variable-temperature environment are significantly improved through an electrical mechanical parameter measurement technology.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical machinery parameter measurement, and particularly to a method for measuring the slip rate of a three-phase asynchronous motor. Background Art

[0002] In the prior art, direct speed measurement methods (such as encoders and tachogenerators) need to be physically installed on the motor shaft, belonging to invasive or contact measurement. This is difficult to implement under harsh working conditions such as explosive, high temperature, high humidity, and strong corrosion, requiring shutdown for modification or even unable to be installed. While the method based on electrical signal harmonic analysis is a non-contact electrical measurement, the signal sources (current, voltage) it depends on are easily affected by various factors such as power grid quality fluctuations, inverter harmonics, and internal electromagnetic interference of the motor, and it is not a truly non-contact and stable speed information source in the physical sense.

[0003] Whether directly measuring the speed or analyzing electrical harmonics, both rely on a single measurement information. This single source greatly reduces the accuracy and stability of the measurement results and has poor robustness when the environmental interference is severe (such as strong electromagnetic interference, vibration), the working conditions are complex (such as frequent start-stop, variable load, non-rated voltage / frequency operation), or the sensor / signal source itself is abnormal, making it difficult to achieve continuous and reliable monitoring.

[0004] During the operation of the motor, the temperatures of components such as windings, iron cores, and bearings will change significantly. These temperature changes will directly affect the equivalent circuit parameters of the motor (such as winding resistance, magnetization reactance, iron loss resistance) and mechanical losses. Existing methods often fail to fully consider or accurately compensate the influence of these temperatures on the slip rate calculation model, resulting in systematic deviations in the measurement results under different temperature environments, especially when the temperature rise is high, affecting the accuracy.

[0005] In summary, there are problems in the existing technology such as the measurement method being limited by working conditions, the low reliability of a single information source, and insufficient compensation for temperature influence, which need to be solved urgently. Summary of the Invention

[0006] Based on this, it is necessary to provide a method for measuring the slip rate of a three-phase asynchronous motor to solve at least one of the above technical problems.

[0007] To achieve the above object, a method for measuring the slip rate of a three-phase asynchronous motor includes the following steps:

[0008] Step S1: Analyze the temperature gradient of different regions of the three-phase asynchronous motor to obtain regional hot spot data and a temperature gradient matrix; calculate the copper loss temperature rise ratio according to the regional hot spot data and the temperature gradient matrix to obtain a thermal feature vector;

[0009] Step S2: Measure the capacitance of the air-gap region of the motor stator to obtain a fused capacitance signal; extract the rotor period characteristics from the fused capacitance signal to obtain a rotor slot frequency characteristic spectrum;

[0010] Step S3: Correlate the thermal characteristic vector and the rotor slot frequency characteristic spectrum to obtain a characteristic correlation matrix; perform dynamic allocation of the dual-field reliability weights under different working conditions based on the characteristic correlation matrix to obtain a fused credibility index;

[0011] Step S4: Calculate the preliminary slip rate estimation values based on the thermal characteristic vector and the rotor slot frequency characteristic spectrum respectively to obtain a dual-source slip rate value; analyze the conductor temperature characteristics based on the thermal characteristic vector to obtain a temperature correction coefficient; perform weight mapping and dual-field slip rate weighted fusion on the dual-source slip rate value according to the fused credibility index to obtain a preliminary fused slip rate; perform temperature characteristic compensation on the preliminary fused slip rate according to the temperature correction coefficient to obtain an optimized slip rate.

[0012] The present invention uses a high-precision infrared thermal imager to perform all-round, high-frequency surface thermal imaging, and then performs sophisticated denoising, correction, and regional segmentation on the original thermal image. This allows accurate capture of subtle temperature differences and hotspot distribution on the motor surface, overcoming the limitations of traditional single-point temperature measurement. By extracting the relative temperature rise, temperature rise ratio, and temperature gradient information of each functional area, particularly focusing on the regional temperature rise characteristics related to the rotor copper loss heat dissipation path, a method is established to indirectly perceive the rotor loss state without directly contacting the rotor or measuring current. The constructed multidimensional thermal feature vector comprehensively reflects the distribution intensity and heat transfer pattern of the internal heat source of the motor, laying a solid foundation for subsequent slip estimation based on thermal information. In particular, under steady-state heavy-load conditions, the heat generated by the rotor copper loss is an important indicator of slip, providing a reliable thermal basis. An array of microcapacitive sensors installed on the inner surface of the stator enables non-invasive, real-time monitoring of dynamic physical property changes in the air gap region. These changes are directly caused by the passage of the rotor slot tooth structure, providing a more direct source of rotor mechanical speed information that is less susceptible to grid interference than current and voltage harmonic analysis. A high-frequency oscillator circuit drives the sensor and performs high-speed data acquisition, ensuring accurate capture of high-frequency slot-passing signals. Bandpass filtering, wavelet noise reduction, and envelope detection are applied to the raw capacitance signal, effectively improving the signal-to-noise ratio and highlighting the periodic characteristics of rotor slot passage. Spatial capacitance field fusion, combining information from multiple sensors, further enhances the signal's interference resistance and stability. Time-frequency analysis, such as FFT and STFT, on the fused signal precisely locates the rotor slot-passing frequency, its harmonics, and modulation components. These modulation features (frequency modulation depth and phase modulation) are closely correlated with the rotor current distribution and load conditions, providing rich rotor dynamic information. The constructed rotor slot frequency signature directly reflects the rotor's high-precision mechanical speed, providing a critical, physical displacement-based measurement basis for slip calculation and exhibiting strong robustness to power quality and electromagnetic interference. Precise time alignment of thermal eigenvectors and rotor slot frequency signatures, acquired at different frequencies, ensures temporal synchronization between the two physical characteristic data, facilitating subsequent joint analysis. By calculating the basic correlation coefficient, lagged correlation coefficient, and nonlinear correlation metric between thermal and slot-pass characteristics, and performing significance tests, we can deeply explore and quantify the inherent correlation patterns between the two different physical phenomena, identifying which thermal characteristics have reliable correspondences with which air gap capacitance characteristics. Based on the feature correlation matrix and current data, the motor operating conditions are identified, enabling subsequent signal quality assessment and weight assignment to be targeted. By performing separate signal quality assessments on the thermal and slot-pass characteristics, we can determine in real time which measurement method provides the most reliable data under the current operating conditions.Dynamically allocate the fusion weights of the two data sources according to signal quality, operating condition type, and feature correlation to achieve adaptive weighted fusion. When the quality of a certain signal deteriorates or its reliability is low under specific operating conditions, its weight is reduced, and the weight of the other signal is increased, effectively overcoming the limitations of a single measurement method and enhancing the robustness of the overall measurement. Calculate the fusion credibility index, which provides a quantitative reliability indicator for the final slip ratio estimation result. It is possible to independently calculate the preliminary slip ratio estimation values based on the thermal feature vector and the rotor slot frequency feature spectrum respectively, providing two independent verification paths for subsequent fusion. Estimate the temperature of the key components of the motor through the thermal feature vector and map it to the stator winding, rotor conductor, etc., enabling the acquisition of real-time temperature information of the internal conductors of the motor, which is a prerequisite for accurate temperature compensation. Calculate the temperature correction coefficients of parameters such as resistance, mechanical loss, and iron loss according to the conductor temperature and the temperatures of other components (such as bearings), providing a basis for the temperature correction of the equivalent circuit model parameters. Dynamically adjust the weights of the two preliminary slip ratio estimation values according to the fusion credibility index for weighted fusion, which can make full use of the advantages of the two data sources and obtain a more stable and closer-to-real preliminary fusion slip ratio under different operating conditions. Calculate the comprehensive compensation factor based on the temperature-corrected motor parameters and multiple detailed temperature effects (main effect of rotor resistance, interaction effect of stator impedance, influence of mechanical loss, magnetic characteristic correction), and use the iterative optimization method to perform temperature compensation on the preliminary fusion slip ratio, which can accurately correct the influence of temperature changes on the equivalent parameters and operating characteristics of the motor, significantly improving the accuracy and precision of slip ratio measurement in a variable-temperature environment. The finally obtained optimized slip ratio is a high-precision result after multi-source information fusion, adaptive weighting, and comprehensive temperature compensation, providing reliable data support for motor condition monitoring, efficiency evaluation, and fault diagnosis.

[0013] Therefore, the present invention provides a slip ratio measurement method for a three-phase asynchronous motor. Through the infrared thermal imaging technology, non-contact acquisition of the surface temperature distribution of the motor is realized. Combining with the detection of the air gap dynamic characteristics by the micro-capacitor sensor array on the inner surface of the stator, a dual-physical characteristic measurement system of the thermal field-capacitance field is established; the adaptive weighted fusion algorithm is used to dynamically allocate the weights of the two physical characteristic data, and the optimal fusion of the measurement results is realized according to the signal reliability under different operating conditions; at the same time, a complete temperature characteristic compensation mechanism is established, considering the influence of temperature on the parameters of each component of the motor, and high-precision slip ratio calculation is realized through iterative optimization. This method does not require shutdown for installation, is applicable to harsh industrial environments, maintains high-precision measurement within the full operating range, and provides reliable technical support for the state monitoring and control of three-phase asynchronous motors. Description of the Drawings

[0014] Figure 1 It is a schematic diagram of the step flow of a slip ratio measurement method for a three-phase asynchronous motor.

[0015] The realization of the object, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0016] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0017] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0018] It should be understood that although the terms "first", "second", etc. may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0019] In the embodiments of the present invention, refer to Figure 1 As shown, it is a schematic diagram of the step flow of the slip ratio measurement method for a three-phase asynchronous motor of the present invention. In this example, the slip ratio measurement method for the three-phase asynchronous motor includes the following steps:

[0020] Step S1: Analyze the temperature gradient of different regions of the three-phase asynchronous motor to obtain regional hot spot data and a temperature gradient matrix; calculate the copper loss temperature rise ratio based on the regional hot spot data and the temperature gradient matrix to obtain a thermal feature vector;

[0021] In an embodiment of the present invention, an original thermal image sequence of the motor surface is collected by a high-resolution infrared thermal imager at a frequency of 8 Hz, and a corrected thermal image is obtained through geometric correction, Gaussian filtering, and background temperature compensation; based on the motor structure, the corrected thermal image is divided into six functional regions, and in each region, the coordinates, area, and peak temperature of the three characteristic hot spots with the highest temperature are identified and extracted by using an adaptive threshold method to form regional hot spot data; the temperature gradient is calculated along 8 radial directions with the characteristic hot spot as the center, and the temperature transition between adjacent functional regions is calculated to construct a temperature gradient matrix; a regional cold-state temperature table is established as a reference in the cold state of the motor, and the relative temperature rise of each region and hot spot is calculated in real time to form regional temperature rise data, and the temperature rise rate and steady-state temperature rise are analyzed to obtain the dynamic characteristics of the temperature rise; the temperature rise ratio between different regions is calculated to form a set of regional temperature rise ratios, especially the temperature rise ratio between the top of the motor housing and the end cover; key parameters sensitive to the slip rate are extracted from the set of regional temperature rise ratios and the dynamic characteristics of the temperature rise to form slip rate thermal characteristic parameters, and they are integrated with the set of regional temperature rise ratios to obtain temperature rise characteristic data; finally, the regional hot spot data, the temperature gradient matrix, and the temperature rise characteristic data are normalized and integrated to construct a multi-dimensional thermal characteristic vector describing the distribution state of internal heat sources in the motor.

[0022] Step S2: Measure the capacitance of the air gap region of the motor stator to obtain a fused capacitance signal; extract the rotor period characteristics from the fused capacitance signal to obtain a rotor slot frequency characteristic spectrum;

[0023] In the embodiment of the present invention, nine 4mm×4mm×0.8mm micro-capacitive sensors are installed on the inner surface of the stator along the positions of phases A, B, and C according to the stator structure parameters of the motor to form a sensor position map; each sensor is driven by a high-frequency LC oscillation circuit operating at 2MHz, and a high-speed acquisition system with a sampling rate of 800kHz is used to synchronously acquire the output signals of the sensors, capture the air-gap capacitance changes caused by the passing of the rotor slot teeth, and obtain the original capacitance signals; the original capacitance signals are subjected to band-pass filtering with a frequency range of 100Hz - 10kHz and wavelet noise reduction, and then the modulation characteristics are extracted through envelope detection to obtain the noise-reduced capacitance signals; the noise-reduced capacitance signals are subjected to time-phase correction and weighted average fusion based on signal quality according to the sensor position map to obtain the fused capacitance signals with a high signal-to-noise ratio; the FFT is performed on the fused capacitance signals to calculate the spectrum, the passing frequency of the rotor slot is accurately located through peak detection to obtain the accurate slot frequency, and the amplitude ratios and phase differences of its first few harmonics are calculated to construct harmonic relationship parameters; the frequency modulation depth is calculated by analyzing the sideband amplitudes near the accurate slot frequency to obtain the frequency modulation parameters; the fused capacitance signals are synchronously demodulated and the instantaneous phase fluctuations are analyzed to calculate the phase modulation index to obtain the phase modulation characteristics; the harmonic relationship parameters, frequency modulation parameters, and phase modulation characteristics are input into a pre-trained model for load feature correlation analysis to obtain load correlation features; the rotor mechanical speed is calculated based on the accurate slot frequency and the number of rotor slots to obtain the speed characteristic data; finally, the speed characteristic data and the load correlation features are integrated to construct a rotor slot frequency feature spectrum including the rotor speed and dynamic modulation information.

[0024] Step S3: Perform thermal field-capacitance field characteristic correlation on the thermal feature vector and the rotor slot frequency feature spectrum to obtain a feature correlation matrix; perform dynamic allocation of the double-field reliability weights under different working conditions according to the feature correlation matrix to obtain a fusion credibility index;

[0025] In the embodiments of the present invention, the thermal feature vector (5Hz) and the rotor slot frequency feature spectrum (100Hz) are aligned in time series according to timestamps. For example, by performing time downsampling on the rotor slot frequency feature spectrum or interpolating the thermal feature vector, a series of synchronized data pairs with consistent timestamps are formed; key parameter features such as the housing temperature rise ratio, rotor speed, and frequency modulation depth are extracted from the synchronized data pairs; the Pearson correlation coefficients between the key parameter features are calculated pairwise to form the original correlation matrix; the cross-correlation coefficients at different time lags are calculated to find the maximum lag correlation and its lag time to form lag correlation data; methods such as mutual information or MIC are used to mine the non-linear association strength between the key parameters to form non-linear association data; a statistical significance test is performed on the association metric values in the original correlation matrix, lag correlation data, and non-linear association data, and the statistically reliable association relationships are marked to form the significance test results; the most important key correlation feature pairs are selected according to the significance test results and association strength; the selected key correlation feature pairs and their association metric values are organized into a feature correlation matrix, which reflects the most significant associations between the thermal field and capacitance field data. According to the feature correlation matrix and the parameter features in the current synchronized data pair, the current motor operating condition type is identified through a classification model to obtain an operating condition type identifier; based on the operating condition type identifier and the feature correlation matrix, signal quality indicators such as temperature stability and frequency stability of the thermal feature vector and the rotor slot frequency feature spectrum under the current operating condition are evaluated to obtain a quality score table; according to the quality score table, operating condition type identifier, and feature correlation matrix, a weighted sum model or a fuzzy logic system is used to dynamically calculate the fusion weight coefficients of the thermal feature and the slot passing feature in the slip rate estimation, so that the data sources with high quality, high correlation, and more suitable for the current operating condition obtain greater weights; finally, based on the fusion weight coefficients and the quality score table, and considering the consistency of the preliminary slip rate estimation values of the two methods, the overall fusion credibility index of the current slip rate estimation result is calculated.

[0026] Step S4: Based on the thermal feature vector and the rotor slot frequency feature spectrum respectively, calculate the preliminary slip rate estimation values to obtain the dual-source slip rate values; analyze the conductor temperature characteristics according to the thermal feature vector to obtain the temperature correction coefficient; perform weight mapping and dual-field slip rate weighted fusion on the dual-source slip rate values according to the fusion credibility index to obtain the preliminary fusion slip rate; perform temperature characteristic compensation on the preliminary fusion slip rate according to the temperature correction coefficient to obtain the optimized slip rate.

[0027] In the embodiments of the present invention, the mapping relationship between the temperature rise ratio and the slip ratio in the thermal feature vector and the difference between the rotor speed and the synchronous speed in the rotor slot frequency characteristic spectrum are respectively utilized to calculate the preliminary slip ratio based on thermal information and the preliminary slip ratio based on speed information, so as to obtain a dual-source slip ratio value; the highest temperature of the machine shell, the end cover temperature, the bearing seat temperature, etc. are extracted from the thermal feature vector, and the average temperatures of the stator winding, the rotor conductor and the bearing are estimated through a thermal model or a mapping relationship to form a key component temperature set; the stator and rotor resistance correction coefficients are calculated according to the estimated conductor temperature and the copper resistance temperature characteristic to form a resistance temperature correction value; the lubricating viscosity change rate is estimated according to the bearing temperature; the mechanical loss correction factor is calculated by using the viscosity change rate and the bearing temperature through a mechanical loss model; the iron loss temperature coefficient is estimated according to the core estimated temperature; the resistance temperature correction value, the mechanical loss correction factor and the iron loss temperature coefficient are integrated to form a temperature correction coefficient; according to the relative reliability information included in the fusion credibility index, the weights of the preliminary slip ratios of the thermal feature and the slot passing feature are dynamically adjusted, weighted average is performed and smoothing processing is carried out to obtain a preliminary fusion slip ratio; the temperature correction coefficient is used to update the motor equivalent circuit parameters to construct temperature-corrected motor parameters; the main increment of the slip ratio caused by the change of the rotor resistance temperature is calculated as the rotor main effect coefficient; based on the temperature-corrected motor parameters and the rotor main effect coefficient, the indirect influence of the changes of the stator resistance and leakage reactance on the slip ratio is analyzed through an equivalent circuit to obtain a stator-rotor interaction coefficient; the influence amount of the mechanical loss on the slip ratio is calculated according to the preliminary fusion slip ratio and the mechanical loss correction factor to obtain a mechanical influence coefficient; the influence amount of the magnetic characteristic temperature change on the slip ratio is evaluated according to the temperature correction coefficient and the temperature-corrected motor parameters to obtain a magnetic characteristic coefficient; the rotor main effect coefficient, the stator-rotor interaction coefficient, the mechanical influence coefficient and the magnetic characteristic coefficient are added together to generate a comprehensive compensation factor; finally, using the preliminary fusion slip ratio as a starting point, the comprehensive compensation factor is applied for iterative correction until the slip ratio converges, and an optimized slip ratio after comprehensive compensation of temperature characteristics is obtained.

[0028] Preferably, the temperature gradient analysis in step S1 is specifically as follows:

[0029] The three-phase asynchronous motor in operation is scanned and imaged in all directions by an infrared thermal imager at a frequency above 5 Hz to obtain an original thermal map sequence;

[0030] The thermal field noise of the original thermal map sequence is filtered to obtain a corrected thermal map;

[0031] The functional area of the corrected thermal map is divided, and then the characteristics of the hot spot area are extracted to obtain area hot spot data;

[0032] According to the area hot spot data, the heat flux density gradient of the corrected thermal map is extracted to obtain a temperature gradient matrix.

[0033] In an embodiment of the present invention, a high-precision infrared thermal imager, for example, a device with a resolution of 640×480 pixels and a temperature resolution of 0.03°C, is fixed at a position 1.5 meters away from the outer surface of the measured three-phase asynchronous motor. The perspective of the thermal imager is adjusted to ensure that the main heat dissipation surfaces such as the motor housing, front and rear end covers, and bearing seats are all completely within the field of view. The acquisition frequency of the thermal imager is set to 8 Hz, and the running motor is continuously scanned and imaged. The thermal imager captures the infrared radiation signals emitted from the motor surface in real time and converts them into raw digital images containing the temperature information of each pixel point. These images are arranged in chronological order to form a raw thermal map data stream, that is, a raw thermal map sequence.

[0034] Each frame of the acquired raw thermal map sequence is processed. First, geometric correction is performed. By identifying four non-collinear marker points (for example, small black circular stickers with high emissivity) pre-pasted on the motor surface, the homography matrix between the current image and the standard reference image is calculated, and this matrix is applied to perform a perspective transformation on the image to correct the geometric distortion caused by slight changes in the position or angle of the thermal imager. Then, noise filtering is carried out. A Gaussian filter with a size of 5×5 is applied to the corrected image for two-dimensional convolution operation to smooth the image and remove the random high-frequency temperature noise caused by environmental heat source fluctuations, dust, sensor self-noise, etc. Finally, background temperature compensation is performed. Using the environmental temperature value measured by an independent Pt100 environmental temperature sensor, combined with the fixed emissivity (for example, 0.95) parameter of the motor surface material, each pixel temperature value in the thermal image is corrected according to the principle of radiative heat transfer to deduct the influence of background environmental radiation on the measurement result. After the above processing, a series of clear and standardized corrected thermal maps are obtained.

[0035] On the obtained corrected thermal maps, according to the standard structural dimensions and layout of the target three-phase asynchronous motor, six main functional regions are predefined and automatically identified: the front end cover region, the rear end cover region, the top region of the housing, the side region of the housing, the front bearing seat region, and the rear bearing seat region. In each functional region, an adaptive threshold segmentation algorithm is used to identify local hot spots. The specific method is to calculate the average temperature μ and standard deviation σ of the current region, and set the threshold T = μ + k·σ, where k is a coefficient that can be adjusted according to the motor type and load (for example, k = 1.5). The regions where the pixel temperature is higher than T and are connected to each other are identified as a potential hot spot. For each identified hot spot, calculate its pixel center coordinates ( , ), the number of pixels it contains (representing the area ), the highest temperature value and its two-dimensional boundary pixel set. In each functional region, according to Sort all the identified hotspots in descending order of value, select the three hottest hotspots as the characteristic hotspots of this area, and store their ( , ), , and the boundary data to form the area hotspot data.

[0036] For each characteristic hotspot identified in the previous step, using its central coordinates ( , ) as the starting point, extend outward a fixed distance L (for example, L = 3 cm) along 8 equally spaced directions (0°, 45°, 90°,..., 315°) in the radial direction. Along these 8 radial directions, collect the temperature value every fixed step Δr (for example, Δr = 0.5 cm). Calculate the temperature change rate, that is, the temperature gradient component, within each Δr interval in each direction. For example, along the 0° direction, collect the temperatures T( and ) at distances and respectively. Then the temperature gradient of this section is approximately (T( ) - T( )) / ( ). Combine the gradient components in these 8 directions to form the temperature gradient vector around this hotspot. At the same time, for each pair of adjacent functional areas (for example, the top area of the housing and the front cover area), define one or more sampling line segments perpendicular to the boundary at or near the boundary. Collect the temperature profile data along these sampling line segments, and calculate the average temperature difference and the spatial temperature change rate (i.e., the temperature gradient between regions) across the region boundary. Organize the temperature gradient vectors around all characteristic hotspots and the temperature gradient values between all adjacent functional areas into a multi-dimensional data structure, that is, the temperature gradient matrix. This matrix records the temperature change rate distribution information between the key hotspot areas on the motor surface and different functional areas.

[0037] Preferably, the calculation of the copper loss temperature rise ratio in step S1 is specifically as follows:

[0038] Establish the cold-state reference temperature for the area hotspot data to obtain the area cold-state temperature table;

[0039] Calculate the relative temperature rise according to the area cold-state temperature table and the area hotspot data to obtain the area temperature rise data;

[0040] Analyze the temperature rise time characteristics of the area temperature rise data to obtain the temperature rise dynamic characteristics;

[0041] Construct the area temperature rise ratio based on the area temperature rise data and the temperature gradient matrix to obtain the area temperature rise ratio set;

[0042] Extract the slip rate thermal-related parameters based on the regional temperature rise ratio set and the temperature rise dynamic characteristics to obtain the slip rate thermal characteristic parameters;

[0043] Integrate the slip rate thermal characteristic parameters and the regional temperature rise ratio set to obtain the temperature rise characteristic data;

[0044] Conduct multi-dimensional thermal characteristic construction on the regional hot spot data, temperature gradient matrix, and temperature rise characteristic data to obtain the thermal characteristic vector.

[0045] In the embodiment of the present invention, before the three-phase asynchronous motor starts running, when the motor is in a long-term shutdown state and the surface temperature is basically the same as the ambient temperature, use an infrared thermal imager to collect the average temperature of each functional area (front end cover, rear end cover, top of the machine shell, side of the machine shell, front bearing seat, rear bearing seat) of the motor at this moment and the temperature of each characteristic hot spot. Record these temperature values as the reference temperature of the motor in the cold state. For example, record the average temperature of the front end cover area as 25.1 °C, the average temperature of the top of the machine shell area as 24.9 °C, the temperature of the characteristic hot spot of the front bearing seat as 25.3 °C, etc. Store these cold state temperature values as a data structure to form a regional cold state temperature table. This table provides a zero reference for calculating the relative temperature rise during operation.

[0046] During the operation of the motor, obtain the average temperature of each functional area and the temperature of each characteristic hot spot in real time (these data are included in the regional hot spot data). For each area or hot spot, subtract the corresponding cold state reference temperature value from its current temperature value to obtain the temperature rise value (ΔT) relative to the cold state of this area or hot spot. For example, if the current average temperature of the top of the machine shell area is 58.7 °C and the cold state reference temperature is 24.9 °C, then its relative temperature rise is 58.7 °C - 24.9 °C = 33.8 °C. Calculate and record the average temperature rise of all functional areas and the temperature rise of all characteristic hot spots to form the regional temperature rise data. These temperature rise values directly reflect the heat accumulation and distribution generated during the operation of the motor.

[0047] Continuously monitor the change of the relative temperature rise values of each functional area and characteristic hot spot during the operation of the motor over time. Record the time series data of the temperature rise value of each area. Analyze these time series data and calculate key parameters: for example, calculate the average rate of temperature rise (ΔT / Δt) of each area from the start to the current moment; identify the inflection point in the temperature rise curve to judge whether the motor enters the thermal steady state; after the motor reaches the thermal steady state, record the steady state temperature rise value of each area. These dynamic parameters describe the characteristics of the motor's thermal process and form the temperature rise dynamic characteristic data.

[0048] Calculate the relative temperature rise ratio between different functional regions. Pay special attention to the regions directly related to the heat generation of rotor copper loss. For example, calculate the ratio of the average temperature rise of the top region of the motor housing to the average temperature rise of the front end cover region (ΔT_top of motor housing / ΔT_front end cover), or the ratio of the temperature rise of the characteristic hot spot in the top region of the motor housing to the temperature rise of the characteristic hot spot in the rear end cover region. These ratios reflect the differences in the conduction and convection efficiencies of the heat generated by different heat sources (such as stator copper loss, rotor copper loss, iron loss, mechanical loss) inside the motor through different paths to the surface. Since the heat generated by rotor copper loss is mainly transferred through the air gap and the shaft, it has a greater impact on the temperature rise of the top region of the motor housing and the bearing housing region, while stator copper loss has a greater impact on the overall temperature rise of the stator core and the motor housing. Therefore, the temperature rise ratio of different regions is related to the proportion of rotor copper loss in the total loss, and thus related to the slip ratio. Combining the temperature transition characteristics between regions described in the temperature gradient matrix, the calculation of the regional temperature rise ratio can be further refined to construct a set of regional temperature rise ratios.

[0049] Extract the key ratios sensitive to the change of slip ratio from the set of regional temperature rise ratios, such as the ratio of the temperature rise of the top of the motor housing to the temperature rise of the end cover, or the ratio of the temperature rise of a specific characteristic hot spot. These ratios are directly used as candidates for the thermal-related parameters of the slip ratio. At the same time, extract the parameters reflecting the speed or steady state degree of the thermal process from the dynamic characteristics of the temperature rise, such as the temperature rise rate, the time to reach the steady state, etc. These parameters are also indirectly related to the motor load and the slip ratio. Combine these extracted key ratios and dynamic parameters to form the thermal characteristic parameters of the slip ratio. For example, the thermal characteristic parameters of the slip ratio include: the ratio of the temperature rise of the top of the motor housing to the temperature rise of the end cover at steady state, the temperature rise rate of the top of the motor housing at the initial stage of startup, the time required to reach 90% of the steady state temperature rise, etc. Among them, the rotor copper loss P_cu2 is proportional to the square of the slip ratio s and the square of the rotor current: P_cu2 = ;

[0050] Where is the square value of the rotor current, representing the square of the rotor current, which is used to calculate the rotor copper loss; is the rotor resistance (referred to the stator side), representing the resistance value of the rotor winding. When the air gap voltage , the rotor leakage reactance changes little, and near the rated slip ratio, P_cu2 is approximately proportional to and also proportional to Is directly proportional. The heat generated by the rotor copper loss is mainly conducted to the surface of the motor housing through the rotor core, air gap, and stator core. Among them, the heat flux components conducted to the top and side parts of the motor housing through the air gap are highly correlated with the rotor copper loss. Therefore, there is an approximate non-linear monotonic relationship between the temperature rise ΔT_top in the top area of the motor housing and the rotor copper loss P_cu2, such as ΔT_top≈ f(P_cu2), where f is an increasing function. The ratio of the temperature rise in the top of the motor housing to the temperature rise in the end cover (ΔT_top / ΔT_endcap) can partially offset the influence of changes in the ambient temperature and the overall ventilation and heat dissipation conditions, and focus more on the proportion of the rotor copper loss in the total loss, where ΔT_endcap is the temperature rise value of the end cover, indicating the temperature increase value of the motor end cover area relative to the cold-state reference temperature. When constructing the thermal-related parameters of the slip ratio, the value of (ΔT_top / ΔT_endcap) under steady-state conditions can be extracted, for example, and a mapping model between it and the slip ratio s can be established, such as fitting a polynomial through experimental data , where a, b, and c are all polynomial coefficients, representing the parameters of the mapping relationship between the temperature rise ratio and the slip ratio, or using a look-up table. The thermal characteristic parameters of the slip ratio include these strongly correlated temperature rise ratios and temperature rise dynamic parameters.

[0051] Integrate the thermal characteristic parameters of the slip ratio (refined key ratios and dynamic parameters) extracted in the previous step with a more comprehensive set of temperature rise ratios for different regions (including all calculated temperature rise ratios between regions). Form a comprehensive data set containing multi-faceted temperature rise information, that is, temperature rise characteristic data. This data set contains both the temperature rise ratios and dynamic parameters that are directly highly correlated with the slip ratio, as well as the temperature rise information of other regions, providing comprehensive thermal information for subsequent multi-dimensional feature construction.

[0052] Integrate and normalize the regional hot spot data (including hot spot location, area, temperature, etc.), temperature gradient matrix (describing the spatial rate of change of temperature), and temperature rise characteristic data (including relative temperature rise, temperature rise ratio, temperature rise dynamic parameters) obtained above. Construct a unified multi-dimensional thermal feature vector. The dimension of this vector is relatively high and contains rich information about the thermal distribution on the surface of the motor. For example, the thermal feature vector can include: the average temperature rise of each functional area, the temperature rise of the top three characteristic hot spots in each area, the normalized coordinates of the hot spot location, the average temperature gradient along the radial direction in the top area of the motor housing, the average temperature gradient between the motor housing and the end cover, the steady-state ratio of the temperature rise in the top of the motor housing to the temperature rise in the front end cover, the temperature rise rate in the top of the motor housing at the initial stage of startup, etc. Normalize all numerical characteristic parameters, for example, using the Z-score normalization method( =(x - μ_x) / σ_x), where, is the standardized value, x is the original value, μ_x is the mean of x, and σ_x is the standard deviation of x. By eliminating the influence of different physical dimensions and numerical ranges, the components in the eigenvector become comparable. This multi-dimensional thermal eigenvector comprehensively reflects the internal heat source distribution state and heat transfer characteristics of the motor, providing a key heat-information-based input for subsequent slip ratio estimation.

[0053] Preferably, the capacitance measurement in the air gap region in step S2 is specifically as follows:

[0054] Obtain the structural parameters of the inner surface of the motor stator; install capacitance sensors at the positions of phases A, B, and C of the three-phase windings of the motor stator according to the structural parameters of the inner surface of the motor stator. The size of each capacitance sensor does not exceed 5mm×5mm×1mm to form a sensor position map;

[0055] According to the sensor position map, drive each capacitance sensor through a high-frequency oscillation circuit to capture the change in air gap characteristics caused by three-phase imbalance, and obtain the original capacitance signal;

[0056] Perform noise reduction on the original capacitance signal to obtain a noise-reduced capacitance signal;

[0057] Perform spatial capacitance field fusion on the noise-reduced capacitance signal according to the sensor position map to obtain a fused capacitance signal.

[0058] In the embodiment of the present invention, first, obtain the detailed design parameters of the measured three-phase asynchronous motor, including the inner diameter of the stator, the number of stator slots, the number of rotor slots, the air gap length, and the distribution information of the stator windings (the starting and ending angles of the three-phase windings A, B, and C on the circumference). According to these structural parameters, determine the positions for installing micro-capacitance sensors along the circumferential direction on the inner surface of the stator. Metal film capacitance sensors with a size of 4mm×4mm×0.8mm are selected as the sensors. To capture the electromagnetic field distribution during the operation of the three-phase windings and the medium change caused by the passing of the rotor slots and teeth, 9 capacitance sensors (3 sensors per phase) are installed at equal intervals along the circumferential direction on the inner surface of the stator, corresponding to the approximate center positions of the three-phase windings A, B, and C, forming an array containing 9 sensors in total. The sensors are directly pasted or installed on the surface of the stator teeth near the air gap through non-conductive fixtures, with the electrodes facing the rotor, forming a capacitance coupling with the outer surface of the rotor. During installation, ensure that the surface of the sensor fits as closely as possible to the inner arc surface of the stator, and the leads of the sensors are insulated and properly fixed without affecting the rotation of the rotor. Record the precise angular positions of each sensor on the stator circumference (relative to a certain reference point on the stator, such as the center line of the starting side of phase A) to form a sensor position map.

[0059] Connect each installed micro-capacitive sensor to an independent high-frequency LC oscillation circuit. The operating frequency of this oscillation circuit is set at 2 MHz. The capacitive sensor serves as the capacitive element in the LC circuit. When the rotor rotates, the slots (with a larger air gap and a lower equivalent dielectric constant) and teeth (with a smaller air gap and a higher equivalent dielectric constant) on the rotor surface periodically pass under the sensor. This periodic structural change causes a slight change in the equivalent capacitance C of the air gap between the sensor and the rotor. According to the principle of the LC oscillation circuit, the oscillation frequency has the relationship with the capacitance C as = 1 / (2π ), where π is the ratio of the circumference of a circle to its diameter, is the inductance value, representing the inductance of the inductor in the oscillation circuit. Therefore, the slight change in the air-gap capacitance causes a slight modulation (frequency modulation) of the oscillation circuit frequency. A high-speed data acquisition system, such as a data acquisition card with 9 synchronous acquisition channels and a sampling rate set at 800 kHz, is connected to the output of each oscillation circuit. The data acquisition system synchronously records the frequency modulation signals output by each sensor, forming a 9-channel original capacitance signal data stream containing information on the change in the air-gap capacitance. These signals reflect the periodic change in the air-gap dielectric characteristics caused by the rotation of the rotor and also contain additional changes caused by the magnetic field distortion due to unbalanced three-phase currents.

[0060] Perform digital signal processing on the 9-channel original capacitance signals acquired at high speed. First, apply a digital band-pass filter with a passband of 100 Hz to 10 kHz to filter out power frequency (50 / 60 Hz) and its low-order harmonic interference, as well as high-frequency random noise. Then, apply wavelet decomposition technology to the filtered signals, for example, perform three-layer decomposition using Daubechies wavelet basis functions, identify and remove the wavelet coefficients corresponding to interference sources such as motor vibration and power transients, and retain the signal components related to the rotor slot and tooth passing frequency. Next, perform envelope detection on the processed signals. Since the change in the air-gap capacitance causes frequency modulation, convert the frequency change into a voltage or amplitude change signal through a frequency discriminator circuit or software demodulation technology and extract its envelope. The periodic change in this envelope signal more prominently shows the characteristic frequency caused by the passing of the rotor slots and teeth. After the above filtering, wavelet noise reduction, and envelope detection processing, 9-channel noise-reduced capacitance signals with a higher signal-to-noise ratio are obtained.

[0061] Using the precise angular information of each sensor recorded in the sensor position map, perform temporal phase correction on the nine-channel noise-reduced capacitance signals. Since the sensors are arranged at equal intervals along the circumference, there is a fixed time delay when the rotor slot teeth pass through different sensors. Calculate these delays based on the approximate rotational speed of the rotor and the angular interval between sensors, and perform corresponding time-shift operations on the signals to align all sensor signals in time as if they were measured at a single point. Then, perform weighted average fusion on the nine-channel phase-corrected signals. The weighting coefficients can be dynamically determined according to the signal quality of each sensor (e.g., signal amplitude, signal-to-noise ratio), and sensors with higher signal quality are given larger weights. This spatial fusion method can effectively suppress the random noise and local anomalies of individual sensors, improving the robustness and accuracy of the overall signal. For example, simple averaging or signal-to-noise ratio weighted averaging is used. The fused signal is a comprehensive capacitance change signal, that is, the fused capacitance signal, which more stably reflects the overall dynamic characteristics of the air-gap medium.

[0062] Preferably, the rotor period feature extraction in step S2 is specifically as follows:

[0063] Perform slot tooth modulation characteristic analysis on the fused capacitance signal to obtain spectral data;

[0064] Perform precise slot frequency positioning on the spectral data to obtain the precise slot frequency;

[0065] Construct harmonic structure features of the spectral data based on the precise slot frequency to obtain harmonic relationship parameters;

[0066] Analyze the frequency modulation depth of the spectral data based on the precise slot frequency to obtain frequency modulation parameters;

[0067] Extract the phase modulation characteristics of the fused capacitance signal based on the precise slot frequency to obtain phase modulation features;

[0068] Perform load feature correlation analysis on the harmonic relationship parameters, frequency modulation parameters, and phase modulation features to obtain load correlation features;

[0069] Extract rotational speed information from the precise slot frequency and spectral data to obtain rotational speed characteristic data;

[0070] Construct a rotor slot frequency feature spectrum based on the rotational speed characteristic data and load correlation features.

[0071] In the embodiments of the present invention, a fast Fourier transform (FFT) is performed on the fused capacitance signal. The window length of the FFT is set to 16384 sampling points, the overlap rate is 50%, and a Hanning window function is used. The power spectral density or amplitude spectrum of the signal is calculated. In the obtained spectrum, the components related to the rotor slot passing frequency are identified. The rotor slot passing frequency f_slot = N_r × f_r, where f_slot is the rotor slot passing frequency, representing the frequency at which the rotor slot teeth structure passes through a fixed point of the stator, N_r is the number of rotor slots, and f_r is the rotor mechanical speed (unit: revolutions per second). On the spectrogram, the rotor slot passing frequency and its harmonics usually appear as prominent spectral lines. The frequency values, amplitudes, and phase information of these spectral lines are accurately measured. For example, for a motor with 28 rotor slots and 2 pole pairs, if the power frequency is 50 Hz and the slip rate is 0.03, the synchronous speed is 25 revolutions per second, the rotor speed is approximately 24.25 revolutions per second, and the slot passing frequency is approximately 28 × 24.25 = 679 Hz. Search for and accurately locate the spectral line near 679 Hz in the spectrum. At the same time, apply the short-time Fourier transform (STFT) or wavelet time-frequency analysis method to analyze the joint distribution of the fused capacitance signal in time and frequency, and generate a time-frequency diagram. Observe the change of the rotor slot passing frequency over time through the time-frequency diagram, which reflects the small fluctuations of the rotor speed. Combine the frequency, amplitude, and phase information obtained from the FFT analysis and the time-frequency diagram data obtained from the STFT to form spectral data.

[0072] In the obtained spectral data, identify the spectral line with the strongest energy and whose frequency is approximately within the expected rotor slot passing frequency range. The expected rotor slot passing frequency range can be estimated based on the rated synchronous speed of the motor and the number of rotor slots. For example, for a motor with a rated synchronous speed of 1500 revolutions per minute (25 revolutions per second) and 28 rotor slots, the slot passing frequency should be around 25 × 28 = 700 Hz. Considering the slip rate, the actual slot frequency will be slightly lower than 700 Hz. Use a peak detection algorithm to search for local maxima in the spectrogram, and combine the frequency threshold and amplitude threshold to accurately locate the rotor slot passing frequency f_slot. For example, search for peaks in the range of 650 Hz to 700 Hz. If the peak amplitude is higher than the background noise level by 10 dB, then the frequency corresponding to this peak is determined as the accurate slot frequency. Record this accurate frequency value to obtain the accurate slot frequency.

[0073] In the spectrum data, in addition to the exact slot frequency f_slot, its harmonics such as 2f_slot, 3f_slot,... etc. will also appear. The amplitude distribution of these harmonics and the phase relationship between them are affected by the rotor current distribution, air-gap non-uniformity, and even load fluctuations. Taking the amplitude A_slot of the exact slot frequency f_slot as a reference, calculate the amplitudes A_2slot, A_3slot, A_4slot of its first few main harmonics (e.g., 2f_slot, 3f_slot, 4f_slot). Calculate the ratio of the amplitude of each harmonic to the amplitude of the fundamental frequency, such as A_2slot / A_slot, A_3slot / A_slot, etc. At the same time, extract the phase information φ_slot, φ_2slot, φ_3slot, etc. of the exact slot frequency and its first few main harmonics, and calculate their phase differences relative to the fundamental frequency, such as φ_2slot - 2φ_slot, φ_3slot - 3φ_slot. These amplitude ratios and phase differences constitute the harmonic structure characteristics and form the harmonic relationship parameters. For example, the harmonic relationship parameter can be a vector [A_2slot / A_slot, A_3slot / A_slot, A_4slot / A_slot, (φ_2slot - 2φ_slot), (φ_3slot - 3φ_slot), (φ_4slot - 4φ_slot)].

[0074] In addition to the dominant slot passing frequency, the frequency of the air-gap capacitance signal is also modulated by the rotor current frequency (i.e., the slip frequency f_s = s×f_1, where s is the slip ratio and f_1 is the power supply frequency). This modulation is manifested as sidebands appearing near the main frequency f_slot, i.e., f_slot ± k·f_s, where k is an integer. In the spectrum data, identify the sideband spectral lines on both sides of the exact slot frequency f_slot. For example, look for spectral lines with frequencies of f_slot ± f_s and f_slot ± 2f_s. Measure the amplitudes of these sideband spectral lines. The frequency modulation depth is usually related to the amplitudes of the sidebands and the amplitude of the main frequency. For example, calculate the ratio of the average amplitude of the first-order sidebands (f_slot ± f_s) to the amplitude of the main frequency A_slot as a measure of the frequency modulation depth. This frequency modulation depth reflects the non-uniformity of the rotor current distribution and is thus related to the slip ratio. These parameters constitute the frequency modulation parameters. For example, the frequency modulation parameter includes [(A_(f_slot + f_s) + A_(f_slot - f_s)) / (2A_slot)].

[0075] The rotor current frequency \(f_s\) also causes phase modulation of the air-gap capacitance signal. Synchronous demodulation of the fused capacitance signal based on the precise slot frequency \(f_{slot}\) is performed to extract the instantaneous phase of the signal. Analyze the variation of the instantaneous phase over time. If there is phase modulation, the instantaneous phase will fluctuate periodically at the slip frequency \(f_s\) around an average value. Calculate the amplitude of this phase fluctuation, i.e., the phase modulation index. For example, obtain the analytic signal of the signal through the Hilbert transform, calculate its instantaneous phase, and filter the instantaneous phase to remove high-frequency noise and DC offset, and then find the maximum amplitude of the phase fluctuation as the phase modulation index. The phase modulation index also reflects the rotor electromagnetic state and is related to the slip ratio. These parameters constitute the phase modulation characteristics.

[0076] Take the previously obtained harmonic relationship parameters, frequency modulation parameters, and phase modulation characteristics as inputs, and analyze their correlations with the motor load. Under different load conditions (e.g., no-load, half-load, full-load), collect these characteristic parameters and record the corresponding motor output power or torque data. Through machine learning algorithms, such as support vector regression or neural networks, establish a mapping model between these characteristic parameters and the motor load. This model can learn how the harmonic structure, frequency modulation, and phase modulation change under different loads. These change rules indirectly reflect the changes in the rotor current and magnetic field distribution with the load, and these changes are closely related to the slip ratio. Through this correlation model, these modulation characteristics can be interpreted as characteristics reflecting the current load state, forming load-related characteristics. For example, the load-related characteristics can be a vector, whose value is obtained by inputting the harmonic relationship parameters, frequency modulation parameters, and phase modulation characteristics into a pre-trained correlation model, representing the "load fingerprint" in the current state.

[0077] Based on the precise slot frequency \(f_{slot}\) and the known number of rotor slots \(N_r\), directly calculate the rotor mechanical speed \(f_r = f_{slot} / N_r\). For example, if the precise slot frequency is 679 Hz and the number of rotor slots is 28, then the rotor speed \(f_r=679\ Hz / 28 = 24.25\) revolutions per second. Convert the rotor speed to a commonly used unit, such as revolutions per minute (RPM): \(N = f_r×60 = 24.25×60 = 1455\) RPM. In addition, analyze the amplitude information of the power frequency \(f_1\) in the spectrum data, as well as the relative amplitude relationship between \(f_{slot}\) and \(f_1\), which can also be used as auxiliary speed information or signal quality indicators. These calculated rotor speed values and related spectrum amplitude information constitute the speed characteristic data. For example, the speed characteristic data contains \([f_r, A_{slot} / A_{f1}]\), where \(A_{f1}\) is the amplitude of the power frequency in the spectrum.

[0078] Integrate the previously obtained rotational speed characteristic data (the core is the accurately calculated rotor speed) and the load-related characteristics (the modulation parameters reflecting the load state and the rotor electromagnetic state). Construct a comprehensive feature vector or data structure that includes the rotor mechanical speed and reflects its dynamic characteristics and load-related modulation characteristics, namely the rotor slot frequency characteristic spectrum. For example, the rotor slot frequency characteristic spectrum can be a vector [f_r, frequency modulation parameter, phase modulation characteristic, harmonic relationship parameter]. This characteristic spectrum directly provides high-precision rotor mechanical speed information and indirectly reflects the rotor current, magnetic field distribution, and load state through the modulation characteristics, providing a key air-gap capacitance signal-based input for subsequent slip ratio calculation.

[0079] Preferably, the association between the thermal field-capacitance field characteristics in step S3 is specifically as follows:

[0080] Perform thermal field-capacitance field time series alignment on the thermal feature vector and the rotor slot frequency characteristic spectrum to obtain synchronous data pairs;

[0081] Extract the key parameter characteristics of the synchronous data pairs;

[0082] Perform basic correlation coefficient calculation on the key parameter characteristics to obtain the original correlation matrix;

[0083] Perform lag correlation analysis on the key parameter characteristics to obtain lag correlation data;

[0084] Perform non-linear correlation mining on the key parameter characteristics to obtain non-linear correlation data;

[0085] Perform significance test on the correlation of the original correlation matrix, lag correlation data, and non-linear correlation data to obtain the significance test result;

[0086] Select key relevant features according to the significance test result to obtain key relevant feature pairs;

[0087] Construct a feature correlation matrix based on the key relevant feature pairs.

[0088] In the embodiments of the present invention, the acquisition frequency of the thermal feature vectors is relatively low (e.g., 5 Hz), while the calculation of the rotor slot frequency characteristic spectrum is based on the capacitively sensed signals acquired at high speed, and its effective update rate can be relatively high (e.g., 100 Hz). To perform joint analysis, it is necessary to align these two types of data in time. The precise timestamps recorded by the data acquisition system are used as the basis. Since the thermal features are updated slowly, the rotor slot frequency characteristic spectrum data is downsampled or aggregated according to the timestamps of the thermal features. For example, for the timestamp \(t_i\) of each thermal feature vector, all the rotor slot frequency characteristic spectrum data within the time period \([t_i - \Delta t / 2, t_i + \Delta t / 2]\) (where \(\Delta t\) is the thermal feature acquisition period, e.g., 0.2 seconds) is collected, and their average value or median is calculated as the synchronized rotor slot frequency characteristic spectrum corresponding to this thermal feature vector. Another method is to perform linear interpolation or spline interpolation on the thermal feature sequence to increase its time resolution to a level close to that of the rotor slot frequency characteristic spectrum. For example, if the thermal feature is one point every 0.2 seconds and the rotor slot frequency characteristic spectrum is one point every 0.01 seconds, then the thermal feature sequence is interpolated 20 times. Regardless of which method is adopted, the goal is to construct a series of data pairs with consistent timestamps, where each data pair contains a thermal feature vector and a corresponding rotor slot frequency characteristic spectrum, forming synchronized data pairs. For example, the synchronized data pairs can be represented as \(\{(thermal feature vector 1, rotor slot frequency characteristic spectrum 1, ), (thermal feature vector 2, rotor slot frequency characteristic spectrum 2, ),...\}.

[0089] From each thermal feature vector and rotor slot frequency characteristic spectrum in the synchronized data pairs, key parameters for correlation analysis are selected. For example, the relative temperature rise in the top region of the machine housing, the temperature rise ratio between the top of the machine housing and the end cover, the temperature rise value of the key hot spot, etc. are extracted from the thermal feature vector. The rotor mechanical speed (calculated from the precise slot frequency), the first sideband amplitude ratio (frequency modulation parameter), the phase modulation index (phase modulation feature), etc. are extracted from the rotor slot frequency characteristic spectrum. These selected parameters constitute the key parameter feature set for correlation analysis. For example, the key parameter feature can be a combined vector \([\Delta T_{top of housing}, \Delta T_{top of housing} / \Delta T_{end cover}, f_r, frequency modulation depth, phase modulation index]\).

[0090] For the extracted key parameter feature set, the Pearson correlation coefficient between any two parameters is calculated. The Pearson correlation coefficient \(r_{xy}\) is used to measure the linear correlation degree between two variables \(X\) and \(Y\), and its calculation formula is: \(r_{xy} = where and are the parameter values of the \(i\)-th sample in the synchronized data pair, and is the average value of parameters X and Y. Calculate the correlation coefficients between all pairs of selected key parameters to form a symmetric correlation coefficient matrix, i.e., the original correlation matrix. The element r_ij in the matrix represents the correlation coefficient between the i-th parameter and the j-th parameter. For example, the matrix includes the correlation coefficient between the temperature rise at the top of the motor housing and the rotor speed, the correlation coefficient between the temperature rise ratio and the frequency modulation depth, etc.

[0091] Considering that the heat conduction process has a certain inertia, the temperature change on the motor surface often lags behind the changes in rotor speed and load. Therefore, it is necessary to analyze the lag correlation between thermal characteristics and slot passing characteristics. For each pair of key parameters (e.g., the temperature rise ΔT_top at the top of the motor housing and the rotor speed f_r), calculate the correlation coefficient ρ_xy(τ) between the ΔT_top sequence and the f_r sequence at different time lags τ = , where E[...] is the expectation operator, representing the statistical average of the expression in the brackets; is the value of the X time series at time t; is the average value of the X time series; is the value of the Y time series at time t + τ; μ y is the average value of the Y time series; is the standard deviation of the X time series; is the standard deviation of the Y time series; τ is the time lag value, representing the time shift of the Y sequence relative to the X sequence. τ can take positive values (indicating that the thermal characteristic lags behind the capacitance characteristic) or negative values (indicating that the capacitance characteristic lags behind the thermal characteristic, which is usually not significant). Calculate the correlation coefficients at a series of different lag values, and find the lag time τ_max that maximizes the correlation coefficient. For example, it is found that the temperature rise at the top of the motor housing has the highest correlation when lagging the rotor speed by about 10 seconds. Record the lag time that maximizes the correlation coefficient and the corresponding maximum correlation coefficient value to form lag correlation data. These data reveal the time delay and the strongest correlation between different physical processes.

[0092] In addition to linear correlation, there is a non-linear relationship between thermal characteristics and capacitance characteristics. For example, the rotor copper loss is proportional to the square of the slip ratio, and the temperature rise is not a simple linear relationship with the copper loss, but is also affected by heat dissipation conditions, etc. Use non-linear correlation measurement methods, such as Mutual Information or Maximal Information Coefficient (MIC), to quantify the strength of the non-linear association between key parameters. The mutual information I(X;Y) = p(x,y) log2[p(x,y) / (p(x) p(y))] is used to quantify the mutual dependence between two random variables X and Y, where p(x,y) is the joint probability distribution, and p(x) and p(y) are the marginal probability distributions, measuring the degree of reduction in the uncertainty of one variable after knowing the information content of the other variable. The MIC value ranges from 0 to 1 and can capture a wide range of functional relationships. Calculate the mutual information or MIC value between every pair of the selected key parameters to form non-linear correlation data. These data complement the results of linear correlation analysis and reveal more complex correlation patterns.

[0093] Conduct statistical significance tests on the calculated original correlation coefficients, lag correlation coefficients, and non-linear correlation metric values (such as MIC). For example, for the Pearson correlation coefficient, a t-test can be used to determine whether it is significantly different from zero statistically. For mutual information and MIC, non-parametric methods such as permutation tests can be used to evaluate their significance. Set a significance level (e.g., α = 0.05). Only the correlation relationships that pass the significance test are considered reliable and non-random associations. For example, if the Pearson correlation coefficient between the temperature rise at the top of the motor housing and the rotor speed is 0.85 and the p-value of the t-test is less than 0.05, it is considered that there is a significant linear correlation between them. The results of the significance test mark which associations are statistically reliable.

[0094] According to the results of the significance test, screen out the feature pairs with high statistical significance and strong correlation strength from all the calculated correlation relationships. For example, screen out the linear correlation pairs with the absolute value of the Pearson correlation coefficient greater than 0.7 and passing the significance test; screen out the lag correlation pairs with the maximum value of the lag correlation coefficient greater than 0.6 and the corresponding lag time within a reasonable range; screen out the non-linear correlation pairs with the MIC value greater than 0.5 and passing the significance test. These screened feature pairs are considered the most important associations between the thermal features and the slot passing features, and they can effectively reflect the operating state of the motor. These screened parameters constitute the key correlation feature pairs. For example, the key correlation feature pairs include {(ΔT_top of motor housing, f_r), (ΔT_top of motor housing / ΔT_end cover, frequency modulation depth)}.

[0095] Organize the key relevant feature pairs and their corresponding association metric values (e.g., Pearson correlation coefficient, maximum lag correlation coefficient, MIC value) screened out in the previous step into a concise feature correlation matrix. This matrix only contains the relationships between those key features considered to have significant associations. For example, if the linear correlation between ΔT_top of the housing and f_r and the non-linear association between ΔT_top of the housing / ΔT_end cover and the frequency modulation depth are screened out, the matrix only shows these two associated values. For each pair of key relevant features, record their association type (linear, lag, non-linear) and the association strength value. This feature correlation matrix condenses the most valuable association information in the thermal field and capacitance field data, providing a basic basis for dynamically adjusting the data fusion weights according to different working conditions. For example, the element M_corr[i][j] of the matrix M_corr can represent the association strength (e.g., the absolute value of the Pearson correlation coefficient or the MIC value) between the i-th thermal feature parameter and the j-th capacitance feature parameter, and is 0 if the association is not significant.

[0096] Preferably, the dynamic allocation of the dual-field reliability weights in step S3 is specifically as follows:

[0097] Identify the motor operating conditions according to the feature correlation matrix to obtain the condition type identifier;

[0098] Conduct a dual-field signal quality assessment on the condition type identifier and the feature correlation matrix to obtain a quality score table;

[0099] Perform dynamic weight allocation according to the quality score table, the condition type identifier, and the feature correlation matrix to obtain the fusion weight coefficient;

[0100] Calculate the overall credibility index of the slip rate under the current working condition based on the fusion weight coefficient and the quality score table to obtain the fusion credibility index.

[0101] In the embodiments of the present invention, the current operating condition of the motor is identified by using the feature correlation matrix and the current synchronization data pair (thermal feature vector and rotor slot frequency feature spectrum). Several typical motor operating condition types are predefined, for example: 1) steady-state light load (load less than 30% of the rated load, stable temperature rise, small speed fluctuation); 2) steady-state heavy load (load greater than 70% of the rated load, relatively high and stable temperature rise, small speed fluctuation); 3) starting process (rapid change of speed and temperature rise); 4) variable load process (frequent load fluctuation, dynamic change of speed and temperature rise); 5) no-load; 6) fault condition (for example, abnormal increase in bearing housing temperature caused by bearing fault, local overheating caused by stator winding fault, etc.). Using a rule-based method or a trained classification model (such as a decision tree or a support vector machine), according to the key parameter features in the current synchronization data pair (such as the average value and standard deviation of the rotor speed, the temperature rise rate at the top of the motor housing, the difference between the temperature rise and the ambient temperature, the stability index of the slot passing frequency, etc.) and the correlation strength between features reflected in the feature correlation matrix, the current motor state is classified into the most suitable operating condition type. For example, if the rotor speed is close to the synchronous speed and the temperature rise rate is close to zero, it is identified as steady-state light load; if the speed is much lower than the synchronous speed and the temperature rise rate is large, it is identified as the starting process. The identified operating condition type and the duration of this condition are output as the operating condition type identifier.

[0102] After identifying the current operating condition of the motor, the signal quality and reliability of the thermal feature vector and the rotor slot frequency feature spectrum under this type of operating condition are evaluated. For the thermal feature vector, its quality score is calculated. For example, in the steady-state condition, temperature stability is an important indicator. Calculate the ratio of the standard deviation of the temperature rise at the top of the motor housing in the most recent minute to the average temperature rise. The smaller the ratio, the higher the quality score. At the same time, the consistency of hot spot identification is evaluated. If the position and number of feature hot spots change frequently in a short period of time, the thermal feature quality is considered to be low. For the rotor slot frequency feature spectrum, its quality score is calculated. For example, calculate the ratio of the standard deviation of the exact slot frequency in the most recent second to the average value, which reflects the stability of speed measurement. The smaller the ratio, the higher the quality score. At the same time, evaluate the ratio of the main frequency to the harmonic energy in the spectrum. If the harmonic energy is abnormally high or the spectrum is diffuse, it indicates that the signal is strongly interfered and the quality score is reduced. The quality assessment also refers to the typical quality performance of each signal under different operating conditions in the historical data. The thermal feature quality score and the slot passing feature quality score are organized into a quality score table. For example, the quality score table contains [thermal feature quality score (range 0 - 1), slot passing feature quality score (range 0 - ))].

[0103] Based on the quality score table, the current operating condition type identifier, and the correlation strength between features reflected in the feature correlation matrix, dynamically calculate the fusion weight coefficients of the thermal feature vector and the rotor slot frequency feature spectrum in the subsequent slip ratio calculation. The weight allocation adopts an adaptive algorithm. For example, the following logic can be used: Initialize the thermal feature weight w_heat and the slot passing feature weight w_slot, satisfying w_heat + w_slot = 1. According to the quality score table, if the quality score of the thermal feature is higher than that of the slot passing feature, it is initially inclined to increase w_heat; otherwise, it is inclined to increase w_slot. At the same time, consider the operating condition type: In the startup and high-speed variable load conditions, the rotational speed changes rapidly, and the rotor slot frequency feature spectrum can more realistically reflect the rotational speed, giving w_slot a higher basic weight (e.g., 0.7); in the steady-state heavy load condition, the temperature rise is more stably correlated with the copper loss (square of the slip ratio), giving w_heat a higher basic weight (e.g., 0.6). Combine the feature correlation matrix: If under the current operating condition, a certain parameter in the thermal feature shows a very strong correlation with a certain parameter in the slot passing feature (e.g., the correlation coefficient is close to 1), this indicates that the two are highly consistent in the current state, and more direct weighting can be performed according to the quality score. If the correlation is weak, it indicates that there is a problem with one of the signals or the correlation is not strong under the current operating condition, and the weight needs to be allocated carefully. The weight dynamic adjustment formula can be designed as: w_heat = f_heat(quality score_heat, quality score_slot, operating condition type, feature correlation matrix), w_slot = 1 - w_heat. The function f_heat is a mapping function that dynamically calculates the weight of the thermal feature according to the input parameters. For example, simple linear weighting: w_heat = · quality score_heat + · operating condition basic weight_heat + · average correlation strength_heat, where , , are adjustment coefficients, and the operating condition basic weight and the average correlation strength are determined according to the operating condition type and the feature correlation matrix. More complex non-linear mapping or a fuzzy logic-based inference system can also be used for weight calculation. Finally, obtain the thermal feature fusion weight w_heat and the slot passing feature fusion weight w_slot at the current moment to form the fusion weight coefficient.

[0104] Calculate the overall confidence index of the current slip ratio estimation result according to the dynamically determined fusion weight coefficient and the quality scoring table. The confidence index reflects the reliability of the current measurement result. The calculation method comprehensively considers the intrinsic quality of the data source and the consistency after the two are fused. For example, the confidence index Confidence_Index can be calculated as: Confidence_Index = w_heat · Quality_Score_heat + w_slot · Quality_Score_slot + Consistency_Score. Consistency_Score is a score that measures the consistency between the slip ratio results independently estimated by the thermal feature and the slot passing feature. If the slip ratio estimated values obtained by the two methods are very close in the preliminary slip ratio estimation step, then Consistency_Score is high, otherwise it is low. Consistency_Score can be calculated as 1 - |s_heat_estimate - s_slot_estimate| / max(|s_heat_estimate|, |s_slot_estimate|, ε), where s_heat_estimate and s_slot_estimate are the slip ratio estimated values independently obtained by the two methods, and ε is a very small positive number to prevent division by zero. Synthesize the weighted average quality score and the consistency score according to a preset ratio (for example, 60% quality score, 40% consistency score) to obtain the final fusion confidence index, and the range is usually normalized to 0 - 100 or 0 - 1. For example, Confidence_Index = 0.6×(w_heat · Quality_Score_heat + w_slot · Quality_Score_slot) + 0.4×Consistency_Score. This fusion confidence index provides a reliability reference for the subsequent use of the slip ratio result.

[0105] Preferably, the analysis of the conductor temperature characteristics in step S4 is specifically as follows:

[0106] Extract the temperatures of key regions from the thermal feature vector to obtain the set of key component temperatures;

[0107] Map the conductor temperature distribution for the set of key component temperatures to obtain the conductor temperature distribution map;

[0108] Calculate the copper resistance temperature coefficient according to the conductor temperature distribution map to obtain the resistance temperature correction value;

[0109] Analyze the bearing lubrication characteristics based on the set of key component temperatures to obtain the lubrication viscosity change rate;

[0110] Perform mechanical loss temperature mapping according to the lubrication viscosity change rate and the set of key component temperatures to obtain the mechanical loss correction factor;

[0111] Estimate the iron loss temperature characteristics based on the key component temperature set to obtain the iron loss temperature coefficient;

[0112] Integrate the resistance temperature correction value, mechanical loss correction factor, and iron loss temperature coefficient according to the double-source slip ratio value to obtain the temperature correction coefficient.

[0113] In the embodiment of the present invention, surface temperature information highly correlated with the temperature of key components inside the motor is extracted from the thermal feature vector. These key components include the stator winding, rotor winding (or squirrel cage bars), stator core, rotor core, and front and rear bearings. Although the infrared thermal imager measures the surface temperature of the motor, the approximate temperature of internal components can be inferred from the surface temperature through a pre-established thermal model or empirical relationship. For example, the average temperature rise in the top area of the motor housing is related to the temperature rise of the stator core and stator winding; the temperature rise ratio in the area of the motor housing facing the rotor is related to the rotor copper loss, indirectly reflecting the rotor temperature; the temperature in the end cover and bearing housing area is related to the bearing temperature. Extract values such as the highest temperature at the top of the motor housing, the average temperature of the end cover, and the characteristic hot spot temperature of the bearing housing from the thermal feature vector. Combine these extracted temperature values to form the key component temperature set. For example, the key component temperature set can be a vector [T_housing_top_max, T_end_cover_avg, T_bearing_housing_front, T_bearing_housing_rear].

[0114] Using the key component temperature set, estimate the average operating temperatures of the stator winding and rotor conductors through a pre-established motor thermal model or data-driven mapping model. The motor thermal model can be a lumped parameter thermal network model, which simplifies the various components of the motor into a thermal capacitance and thermal resistance network, and calculates the internal temperature with the surface temperature as the boundary condition. The data-driven model can establish a regression relationship between the surface temperature and the internal conductor temperature based on a large amount of experimental data. For example, estimate the stator winding temperature T_stator≈α×T_housing_top_max + β×T_end_cover_avg + γ×ambient temperature + δ, where α, β, γ, δ are coefficients determined through experiments or simulations. For a squirrel-cage induction motor, the rotor conductor temperature T_rotor can be indirectly estimated through the relationship between the temperature rise ratio in a specific area of the motor housing and the slip ratio, or a more detailed thermal model can be used. Obtain the estimated average temperatures of the stator winding and rotor conductors, which constitute the conductor temperature distribution map (although it is a simplified average temperature). For example, the conductor temperature distribution map contains [T_stator, T_rotor].

[0115] According to the estimated average temperature T_stator of the stator winding and the average temperature T_rotor of the rotor conductor, calculate the correction coefficients of the resistance values of the stator and rotor copper conductors relative to the standard temperature (usually 25°C or 75°C). The resistivity of copper (or other conductor materials) varies approximately linearly with temperature: , where is the reference temperature of the resistivity, and is the temperature coefficient of resistance (for copper,

[0116] near 20 °C). . Calculate the stator resistance correction factor . Similarly, calculate the rotor resistance correction factor . Here and are the nominal values of the stator and rotor resistances at the reference temperature . Obtain the resistance temperature correction value, for example .

[0117] Using the bearing housing temperatures (T_bearing_housing_front, T_bearing_housing_rear) where the temperatures of key components are concentrated, estimate the operating temperature of the bearings. The bearing temperature directly affects the viscosity of the grease or lubricating oil. The viscosity of most lubricants decreases as the temperature increases. According to the viscosity-temperature curve or empirical formula of the lubricant, calculate the change rate of the lubricant viscosity at the bearing operating temperature relative to the viscosity at the standard operating temperature. For example, the viscosity change rate η_ratio = viscosity(T_bearing) / viscosity(T_standard). The higher the bearing temperature, the lower the viscosity, and the friction force usually decreases as the viscosity decreases (within a certain range). Obtain the lubricant viscosity change rate.

[0118] The mechanical losses of the motor mainly include bearing friction losses and windage losses. The bearing friction losses are related to the bearing type, load, speed, and lubrication state (viscosity). The windage losses are related to the speed and the duct structure, and are also affected by the air density (which changes with temperature and altitude). Using the estimated bearing temperature and the lubricant viscosity change rate, through a mechanical loss model, calculate the change of the bearing friction losses with temperature. For example, the frictional torque M_friction ≈ C × η_ratio × , where C is a constant and β is usually close to ¹. The windage losses also change slightly as the air density is affected by the increase in the internal temperature of the motor. Considering these factors comprehensively, calculate the correction factor k_mech_loss = mechanical_loss(T_bearing, T_internal_air) / mechanical_loss(T_standard) of the total mechanical losses at the current temperature relative to the mechanical losses at the standard temperature. Obtain the mechanical loss correction factor.

[0119] The iron loss of the motor includes hysteresis loss and eddy current loss, which mainly occur in the stator and rotor cores. The iron loss is affected by the magnetic flux density, frequency, and the temperature of the core material. The hysteresis loss decreases slightly with the increase in temperature, while the eddy current loss decreases as the increase in temperature leads to an increase in resistivity. Estimate the temperature of the stator core (which can be inferred from the temperature at the top of the machine shell) and the rotor core where the temperature of the key components is concentrated. Through empirical formulas or by looking up tables, estimate the change in iron loss at the current temperature relative to the iron loss at the standard temperature. For example, calculate the iron loss temperature coefficient k_iron_loss = iron loss(T_core) / iron loss(T_standard). Obtain the iron loss temperature coefficient.

[0120] Integrate the resistance temperature correction values (k_R_stator, k_R_rotor), the mechanical loss correction factor k_mech_loss, and the iron loss temperature coefficient k_iron_loss to form a set of temperature correction coefficients for subsequent slip ratio temperature compensation. These correction coefficients will be used to adjust the resistance, mechanical loss, and iron loss parameters in the motor equivalent circuit model to reflect the current actual temperature state. When integrating, it is also necessary to refer to the dual-source slip ratio value because the slip ratio itself affects the rotor current and frequency, and thus indirectly affects the iron loss and part of the loss distribution. For example, the temperature correction coefficient can be a data structure containing all the correction factors: {k_R_stator, k_R_rotor, k_mech_loss, k_iron_loss}. These coefficients are the key inputs for temperature compensation in the subsequent slip ratio calculation model to ensure that the calculation results are not affected by the change in the motor temperature.

[0121] Preferably, the temperature characteristic compensation in step S4 is specifically:

[0122] Construct the motor equivalent parameters according to the temperature correction coefficients and the preliminarily fused slip ratio to obtain the temperature-corrected motor parameters;

[0123] Calculate the main effect of the rotor resistance on the preliminarily fused slip ratio according to the temperature correction coefficients to obtain the rotor main effect coefficient;

[0124] Analyze the interaction effect of the stator impedance according to the temperature correction coefficients, the temperature-corrected motor parameters, and the rotor main effect coefficient to obtain the stator-rotor interaction coefficient;

[0125] Evaluate the influence of mechanical loss on the preliminarily fused slip ratio to obtain the mechanical influence coefficient;

[0126] Perform magnetic characteristic temperature correction according to the temperature correction coefficients and the temperature-corrected motor parameters to obtain the magnetic characteristic coefficient;

[0127] Generate a comprehensive compensation factor according to the rotor main effect coefficient, the stator-rotor interaction coefficient, the mechanical influence coefficient, and the magnetic characteristic coefficient;

[0128] Iteratively optimize the slip compensation based on the preliminary fusion slip ratio, the comprehensive compensation factor, and the temperature-corrected motor parameters to obtain the optimized slip ratio.

[0129] In the embodiments of the present invention, the equivalent circuit parameters of the motor at the current temperature and slip ratio are constructed by using the temperature correction coefficients (k_R_stator, k_R_rotor, k_mech_loss, k_iron_loss) and the preliminary fusion slip ratio s_preliminary. A commonly used single-phase equivalent circuit model of a squirrel-cage induction motor includes the stator resistance , the stator leakage reactance , the magnetizing reactance X_m, the core loss resistance R_c, the rotor resistance (referred to the stator side), the rotor leakage reactance (referred to the stator side), and the resistance representing the mechanical load R_mech = (1 - s) / s. According to the temperature correction coefficients, update the temperature-related parameters: the current stator resistance _temp = k_R_stator × _standard; the current rotor resistance _temp = k_R_rotor × _standard; the current core loss resistance R_c_temp ≈ R_c_standard / k_iron_loss (the core loss increases, and the equivalent resistance decreases). The leakage reactance and the magnetizing reactance are less affected by temperature and are usually regarded as constants or slightly corrected. The mechanical load equivalent resistance R_mech_temp _temp × (1 - s_preliminary) / s_preliminary. Combine these corrected parameters to form the temperature-corrected motor parameters. For example, the temperature-corrected motor parameters can be a set containing all the corrected equivalent circuit parameters { _temp, , X_m, R_c_temp, _temp, , R_mech_temp}.

[0130] Analyze the most significant effect of temperature rise on the slip ratio, that is, the increase in the rotor resistance . According to the torque equation or power balance relationship of the induction motor, when the output torque is constant, the increase in the rotor resistance will cause the slip ratio s to increase. This main effect can be approximated by a simple proportional relationship or equivalent circuit analysis. For example, when ignoring the stator impedance drop and the change in magnetic circuit saturation, the torque . When and the reactance change is small, and , the torque is approximately proportional to , so at the same torque, . That is to say, the slip rate and the rotor resistance are approximately linearly related. Calculate the main increment of the slip rate caused by the increase in the rotor resistance . For example, . This increment represents the direct and main influence of the temperature change of the rotor resistance on the initially fused slip rate, and the rotor main effect coefficient is obtained. For example, rotor main effect coefficient = .

[0131] Consider the stator resistance and leakage reactance temperature changes (although relatively small) and their interaction with rotor parameters on the slip rate. The increase in the stator resistance will increase the stator voltage drop and reduce the air-gap voltage, which will affect the magnetic flux and torque, and thus affect the slip rate. This effect is relatively complex and the entire equivalent circuit needs to be considered. Using the temperature-corrected motor parameters and the initially fused slip rate s_preliminary, calculate parameters such as the input impedance, input current, air-gap voltage, and air-gap power at the current temperature through the motor equivalent circuit model. For example, the input impedance . Calculate the air-gap voltage . The torque is related to the air-gap power. Analyze when changes from _standard to _temp, or when undergoes a small change, how it affects the air-gap voltage and torque, and thus affects the change in the slip rate required to achieve the same torque. This change amount Δs_stator_interaction is defined as the slip rate correction amount caused by the stator-rotor interaction effect. Obtain the stator-rotor interaction coefficient. For example, stator-rotor interaction coefficient = Δs_stator_interaction. This coefficient reflects the indirect influence of the temperature change of the stator parameters on the slip rate through the equivalent circuit coupling.

[0132] Mechanical losses (bearing friction and wind resistance) generate a mechanical drag torque that is dual to the rotational speed. This drag torque requires the motor to output an electromagnetic torque to overcome, thereby consuming a part of the air-gap power and affecting the slip ratio. Using the preliminary fusion of the slip ratio s_preliminary and the mechanical loss correction factor k_mech_loss, the influence of mechanical losses on the slip ratio at the current temperature is evaluated. The mechanical loss power P_mech_loss = P_mech_loss_standard × k_mech_loss. The mechanical loss torque T_mech_loss = P_mech_loss / ω_r, where ω_r is the rotor angular velocity. In the total electromagnetic torque T_em, a part is used to overcome T_mech_loss, and the remaining part is provided to the load torque T_load. T_em = T_load + T_mech_loss. Since the torque is related to the slip ratio, a change in mechanical losses will cause a small change in the slip ratio Δs_mech. For example, the greater the mechanical losses, the greater the electromagnetic torque required to achieve the same load torque, resulting in a slightly increased slip ratio. The mechanical influence coefficient is obtained, for example, mechanical influence coefficient = Δs_mech.

[0133] The magnetization characteristics (magnetization reactance X_m) and iron losses (R_c) of the motor are also affected by temperature. An increase in temperature usually causes a slight decrease in the magnetic permeability of the core material, thereby affecting the magnetization reactance X_m. The iron loss resistance R_c is related to the eddy current loss, and the eddy current loss is affected by the resistivity of the core. The resistivity increases with the increase in temperature, resulting in a decrease in the eddy current loss and an increase in the equivalent iron loss resistance R_c. Using the temperature correction factor k_iron_loss and the temperature-corrected motor parameters {..., X_m, R_c_temp,...}, the influence of the temperature change of the magnetic characteristics (mainly X_m and R_c) on the slip ratio is evaluated. These influences are relatively small, but still need to be considered to improve the accuracy. For example, calculate the slip ratio correction amount Δs_magnetic caused by the small change in the magnetization reactance ΔX_m and the change in the iron loss resistance ΔR_c_temp. The magnetic characteristic coefficient is obtained, for example, magnetic characteristic coefficient = Δs_magnetic.

[0134] Integrate the correction amounts of the slip ratio changes caused by various temperatures (Δs_R2 caused by the main rotor effect, the stator-rotor interaction coefficient Δs_stator_interaction caused by the stator-rotor interaction, Δs_mech caused by mechanical losses, and Δs_magnetic caused by magnetic characteristics). These correction amounts represent the compensation values for the temperature change to the preliminarily fused slip ratio. The comprehensive compensation factor Δs_compensation = Δs_R2 + Δs_stator_interaction + Δs_mech + Δs_magnetic. This comprehensive compensation factor converts the influence of temperature on each part of the motor into the overall correction amount for the slip ratio.

[0135] Using the preliminary slip ratio s_preliminary as the initial value and applying the comprehensive compensation factor for correction to obtain the estimated slip ratio after the first temperature compensation = s_preliminary + Δs_compensation. However, the compensation factor Δs_compensation itself is calculated based on s_preliminary, and R_mech_temp in the temperature correction coefficient also depends on the slip ratio. Therefore, an iterative optimization method needs to be adopted to improve the accuracy of compensation. Using Update R_mech_temp in the temperature-corrected motor parameters, recalculate the stator-rotor interaction coefficient, mechanical influence coefficient, and magnetic characteristic coefficient, and generate a new comprehensive compensation factor . Calculate the slip ratio after the second compensation . Repeat this process to calculate , ,... until the difference in the slip ratios calculated in two adjacent iterations is less than a preset convergence threshold (e.g., ). The finally converged slip ratio value is the optimized slip ratio after temperature characteristic compensation. This iterative process ensures that the temperature compensation is self-consistent with the current estimated slip ratio. Finally, output this converged optimized slip ratio

[0136] Preferably, the analysis of the stator impedance interaction effect is specifically as follows:

[0137] Conduct a current distribution balance analysis according to the temperature correction coefficient to obtain the current distribution change rate;

[0138] Conduct an air-gap magnetic field intensity correction according to the current distribution change rate to obtain the magnetic field intensity correction coefficient;

[0139] Conduct a rotor induced electromotive force calculation according to the magnetic field intensity correction coefficient to obtain the induced electromotive force coefficient;

[0140] Conduct a torque balance characteristic analysis according to the induced electromotive force coefficient to obtain the torque balance coefficient;

[0141] Conduct a stator leakage reactance temperature effect evaluation according to the torque balance coefficient to obtain the leakage reactance effect coefficient;

[0142] Determine the stator-rotor coupling factor for the current distribution change rate, magnetic field intensity correction coefficient, induced electromotive force coefficient, torque balance coefficient, and leakage reactance effect coefficient to obtain the stator-rotor coupling factor;

[0143] Conduct a comprehensive calculation of the interaction coefficient for the stator-rotor coupling factor, temperature-corrected motor parameters, and rotor main effect coefficient to obtain the stator-rotor interaction coefficient.

[0144] In the embodiments of the present invention, the temperature correction coefficient, especially the stator resistance correction coefficient k_R_stator, is used to analyze the influence of the increase in the stator winding resistance on the balance of the three-phase current distribution. Although the power supply is a three-phase symmetrical power supply, the increase in the temperature of the winding resistance (even if it is uniformly increased) will change the impedance of the equivalent circuit. Especially when considering the internal resistance of the power supply or the connection impedance, this impedance change causes a slight imbalance in the amplitudes or phases of the three-phase currents. Using the _temp, ,X_m,R_c_temp, _temp, and the preliminary fusion slip rate s_preliminary, a complete three-phase equivalent circuit model including the internal resistance R_s and reactance X_s of the power supply is established. Calculate the currents I_a, I_b, and I_c flowing through the three-phase stator windings when a symmetrical three-phase voltage is applied under this model. Analyze whether there are deviations in the amplitudes and phases of these three-phase currents from the ideal symmetrical situation. Calculate the current unbalance degree. For example, the unbalance degree V_unbalance defined by the IEC standard is used, where V_unbalance = (the maximum value of the difference between the highest phase current and the average current) / the average current. This unbalance degree reflects the change in the current distribution and forms the current distribution change rate. For example, the current distribution change rate can be a numerical value representing the unbalance degree.

[0145] The imbalance in the current distribution will cause the rotating magnetic field in the air gap of the motor to no longer be an ideal circle, but an ellipse or contain harmonic components. This magnetic field distortion will affect the fundamental wave amplitude of the air gap magnetic field. Using the current distribution change rate calculated in the previous step, through the motor magnetic circuit model or empirical formula, calculate the correction of the fundamental wave amplitude of the air gap magnetic field relative to the ideal symmetrical current situation. For example, if the current unbalance degree is ε_I, the air gap magnetic field intensity correction coefficient k_B_gap ≈ 1 - c × ε_ , where c is a coefficient related to the motor design. This correction coefficient k_B_gap reflects the influence of the current imbalance on the air gap magnetic field intensity and forms the magnetic field intensity correction coefficient.

[0146] The fundamental wave amplitude of the air gap magnetic field directly determines the amplitude of the induced electromotive force in the rotor winding. The rotor induced electromotive force ≈4.44× × _eff×Φ_gap, where is the rotor current frequency ( =s× ), _eff is the equivalent number of turns of the rotor, and Φ_gap is the air-gap magnetic flux. The air-gap magnetic flux Φ_gap is proportional to the air-gap magnetic field intensity B_gap. Using the magnetic field intensity correction coefficient k_B_gap, correct the air-gap magnetic flux or directly correct the rotor induced electromotive force. For example, the corrected rotor induced electromotive force _temp ≈ _standard × k_B_gap. This corrected electromotive force value reflects the influence of the air-gap magnetic field change on the rotor induced electromotive force, forming an induced electromotive force coefficient. For example, the induced electromotive force coefficient = k_B_gap.

[0147] The change in the rotor induced electromotive force directly affects the rotor current and electromagnetic torque. The electromagnetic torque T_em is approximately proportional to the product of the air-gap magnetic flux and the rotor current, and is related to the rotor power factor. The rotor current . Using the induced electromotive force coefficient k_B_gap, update the rotor induced electromotive force to _temp, combined with the _temp and in the temperature-corrected motor parameters, calculate the corrected rotor current I2'_temp. Analyze the change in the slip rate required to achieve the same load torque T_load under the new rotor current and air-gap magnetic field conditions. For example, under a constant load torque, T_load = T_em - T_mech_loss. Calculate the electromagnetic torque T_em(s, temperature-corrected parameters) through the corrected equivalent circuit parameters and slip rate s. Solve the equation T_em(s, temperature-corrected parameters) = T_load + T_mech_loss to obtain the new slip rate that satisfies the torque balance . The torque balance coefficient can be expressed as the difference between the new slip rate and the preliminary fusion slip rate Δs_torque = - s_preliminary. Obtain the torque balance coefficient.

[0148] Stator leakage reactance is mainly generated by the end leakage flux and slot leakage flux of the stator winding, and its value is affected by the winding size, shape, and the slight influence of temperature on the conductor resistivity and the dielectric constant of the insulating material. Although it is generally considered that the leakage reactance is little affected by temperature, it still needs to be evaluated in high-precision calculations. The torque balance coefficient reflects the change in the slip rate required to achieve torque balance after considering factors such as resistance and magnetic field. This change amount Δs_torque includes the influence brought by the slight change in the stator leakage reactance. For example, through sensitivity analysis, evaluate the slight change in the stator leakage reactance slight change Effect on the torque-slip curve, thus estimating the change in slip Δs_X1 required to achieve the same torque. This Δs_X1 can be separated from Δs_torque or calculated through an independent model as a measure of the temperature effect of the stator leakage reactance. Obtain the leakage reactance effect coefficient, for example, leakage reactance effect coefficient = Δs_X1.

[0149] Integrate the intermediate quantities such as the current distribution change rate, magnetic field strength correction coefficient, induced electromotive force coefficient, torque balance coefficient, and leakage reactance effect coefficient, which reflect how the temperature change on the stator side affects the air gap and rotor side parameters, and ultimately affects torque and slip. These intermediate quantities together describe the "interaction" or "coupling" degree between the temperature change of the stator parameters and the rotor behavior. For example, current imbalance causes magnetic field distortion (through k_B_gap), magnetic field distortion affects the rotor electromotive force (through the induced electromotive force coefficient), and further affects the rotor current and torque (through the torque balance coefficient). The change in the stator leakage reactance also directly affects the stator impedance and air gap voltage, and indirectly affects the torque (through the leakage reactance effect coefficient). The combined effect of these factors can be quantified as one or more stator-rotor coupling factors. For example, the stator-rotor coupling factor can be a vector [current distribution change rate, k_B_gap, induced electromotive force coefficient, torque balance coefficient, leakage reactance effect coefficient].

[0150] Utilize the various interaction effect information contained in the stator-rotor coupling factor, combined with the temperature-corrected motor parameters { , ,...} and the rotor main effect coefficient (k_R_rotor - 1), to calculate the final stator-rotor interaction coefficient Δs_stator_interaction. The stator-rotor interaction coefficient is the total additional effect of the temperature change on the stator side on the slip through complex electromagnetic coupling. This effect quantity Δs_stator_interaction is not simply linear and depends on the current operating state (reflected by the temperature-corrected motor parameters and the preliminary slip). For example, a function f_interaction based on an equivalent circuit model can be used, whose input is the stator-rotor coupling factor, temperature-corrected motor parameters, and rotor main effect coefficient, and the output is Δs_stator_interaction = f_interaction(stator-rotor coupling factor, temperature-corrected motor parameters, rotor main effect coefficient). This function calculates the effect on the slip by simulating the chain reaction caused by the changes in parameters such as the stator resistance and leakage reactance in the equivalent circuit. For example, by calculating at = _temp, = _temp (assuming a small change) under the temperature-corrected parameters, the slip required to achieve the same load torque as at the standard temperature , then Δs_stator_interaction≈ -s_preliminary - Δs_R2_simple, where Δs_R2_simple is the approximate increment of slip rate caused by a simple rotor resistance change. This calculation process requires iterative solution of the equivalent circuit equation. Finally, the accurate stator-rotor interaction coefficient Δs_stator_interaction is obtained.

[0151] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Thus, all changes falling within the meaning and scope of the equivalent elements of the application document are intended to be embraced by the present invention.

[0152] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A method for measuring the slip rate of a three-phase asynchronous motor, characterized in that, It includes the following steps: Step S1: Analyze the temperature gradient of different regions of a three-phase asynchronous motor to obtain regional hot spot data and a temperature gradient matrix; calculate the copper loss temperature rise ratio based on the regional hot spot data and the temperature gradient matrix to obtain a thermal feature vector; Step S2: Measure the air gap region capacitance of the motor stator to obtain a fused capacitance signal; extract the rotor periodic characteristics of the fused capacitance signal to obtain a rotor slot frequency characteristic spectrum; Step S3: Correlate the thermal field-capacitance field characteristics of the thermal feature vector and the rotor slot frequency characteristic spectrum to obtain a characteristic correlation matrix; Perform dynamic allocation of the dual-field reliability weights under different working conditions according to the characteristic correlation matrix to obtain a fused credibility index; Step S4: Calculate the preliminary slip rate estimation values based on the thermal feature vector and the rotor slot frequency characteristic spectrum respectively to obtain a dual-source slip rate value; analyze the conductor temperature characteristics based on the thermal feature vector to obtain a temperature correction coefficient; Perform weight mapping on the dual-source slip rate value according to the fused credibility index and perform weighted fusion of the dual-field slip rate to obtain a preliminary fused slip rate; perform temperature characteristic compensation on the preliminary fused slip rate according to the temperature correction coefficient to obtain an optimized slip rate.

2. The slip rate measurement method of the three-phase asynchronous motor according to claim 1, characterized in that, The temperature gradient analysis in Step S1 is specifically as follows: Use an infrared thermal imager to perform an all-round scanning imaging of the operating three-phase asynchronous motor at a frequency above 5 Hz to obtain an original thermal image sequence; Filter the thermal field noise of the original thermal image sequence to obtain a corrected thermal image; Divide the functional regions of the corrected thermal image, and then extract the characteristics of the hot spot regions to obtain regional hot spot data; Extract the heat flux density gradient of the corrected thermal image according to the regional hot spot data to obtain a temperature gradient matrix.

3. The slip ratio measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The calculation of the copper loss temperature rise ratio in Step S1 is specifically as follows: Establish a cold-state reference temperature for the regional hot spot data to obtain a regional cold-state temperature table; Calculate the relative temperature rise according to the regional cold-state temperature table and the regional hot spot data to obtain regional temperature rise data; Analyze the time characteristics of the temperature rise of the regional temperature rise data to obtain the dynamic characteristics of the temperature rise; Construct a regional temperature rise ratio for the regional temperature rise data and the temperature gradient matrix to obtain a set of regional temperature rise ratios; Extract the slip rate thermal-related parameters according to the set of regional temperature rise ratios and the dynamic characteristics of the temperature rise to obtain the slip rate thermal characteristic parameters; Integrate the slip rate thermal characteristic parameters and the set of regional temperature rise ratios to obtain temperature rise characteristic data; Construct multi-dimensional thermal characteristics for the regional hot spot data, the temperature gradient matrix, and the temperature rise characteristic data to obtain a thermal feature vector.

4. The slip ratio measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The measurement of the air gap region capacitance in Step S2 is specifically as follows: Obtain the structural parameters of the inner surface of the motor stator; install capacitance sensors along the A, B, and C phase positions of the three-phase windings of the motor stator according to the structural parameters of the inner surface of the motor stator. The size of each capacitance sensor does not exceed 5 mm × 5 mm × 1 mm to form a sensor position map; According to the sensor position map, drive each capacitance sensor through a high-frequency oscillation circuit to capture the changes in the air gap characteristics caused by the three-phase imbalance to obtain an original capacitance signal; Reduce the noise of the response signal of the original capacitance signal to obtain a noise-reduced capacitance signal; Perform spatial capacitance field fusion on the noise-reduced capacitance signal according to the sensor position map to obtain a fused capacitance signal.

5. The slip ratio measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The extraction of the rotor periodic characteristics in Step S2 is specifically as follows: Analyze the slot-tooth modulation characteristics of the fused capacitance signal to obtain spectral data; Precisely locate the slot frequency of the spectral data to obtain the accurate slot frequency; Construct the harmonic structure characteristics of the spectral data based on the accurate slot frequency to obtain harmonic relationship parameters; Analyze the frequency modulation depth of the spectral data based on the accurate slot frequency to obtain frequency modulation parameters; Extract the phase modulation characteristics of the fused capacitance signal based on the accurate slot frequency to obtain phase modulation features; Conduct a load characteristic correlation analysis on the harmonic relationship parameters, frequency modulation parameters, and phase modulation features to obtain load correlation features; Extract the rotational speed information from the accurate slot frequency and spectral data to obtain rotational speed characteristic data; Construct a rotor slot frequency characteristic spectrum based on the rotational speed characteristic data and load correlation features.

6. The slip rate measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The specific correlation between the thermal field and capacitance field characteristics in step S3 is as follows: Perform thermal field-capacitance field time series alignment on the thermal feature vector and the rotor slot frequency characteristic spectrum to obtain synchronous data pairs; Extract the key parameter features of the synchronous data pairs; Calculate the basic correlation coefficients of the key parameter features to obtain the original correlation matrix; Conduct a lag correlation analysis on the key parameter features to obtain lag correlation data; Mine the non-linear correlation of the key parameter features to obtain non-linear correlation data; Conduct a significance test on the correlation of the original correlation matrix, lag correlation data, and non-linear correlation data to obtain the significance test results; Screen the key correlation features according to the significance test results to obtain key correlation feature pairs; Construct a feature correlation matrix based on the key correlation feature pairs.

7. The slip rate measurement method of the three-phase asynchronous motor according to claim 1, characterized in that, The specific dynamic allocation of the reliability weights of the two fields in step S3 is as follows: Identify the operating conditions of the motor based on the feature correlation matrix to obtain the operating condition type identifier; Evaluate the signal quality of the two fields for the operating condition type identifier and the feature correlation matrix to obtain a quality score table; Perform dynamic weight allocation based on the quality score table, operating condition type identifier, and feature correlation matrix to obtain the fused weight coefficient; Calculate the overall credibility index of the slip ratio under the current operating condition based on the fused weight coefficient and the quality score table to obtain the fused credibility index.

8. The slip rate measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The specific analysis of the conductor temperature characteristics in step S4 is as follows: Extract the temperature of the key area from the thermal feature vector to obtain the set of key component temperatures; Map the conductor temperature distribution for the set of key component temperatures to obtain a conductor temperature distribution map; Calculate the copper resistance temperature coefficient based on the conductor temperature distribution map to obtain the resistance temperature correction value; Analyze the bearing lubrication characteristics based on the set of key component temperatures to obtain the lubrication viscosity change rate; Map the mechanical loss temperature based on the lubrication viscosity change rate and the set of key component temperatures to obtain the mechanical loss correction factor; Estimate the iron loss temperature characteristics based on the set of key component temperatures to obtain the iron loss temperature coefficient; Integrate the resistance temperature correction value, mechanical loss correction factor, and iron loss temperature coefficient according to the dual-source slip ratio value to obtain the temperature correction coefficient.

9. The slip ratio measurement method of the three-phase asynchronous motor according to claim 1, characterized in that The specific temperature characteristic compensation in step S4 is as follows: Construct the equivalent parameters of the motor based on the temperature correction coefficient and the preliminarily fused slip ratio to obtain the temperature-corrected motor parameters; Calculate the main effect coefficient of the rotor resistance for the preliminarily fused slip ratio based on the temperature correction coefficient; Analyze the interaction effect of stator impedance according to the temperature correction coefficient, temperature-corrected motor parameters, and rotor main effect coefficient to obtain the stator-rotor interaction coefficient; Evaluate the influence of mechanical loss on the preliminary fusion slip rate to obtain the mechanical influence coefficient; Perform magnetic characteristic temperature correction according to the temperature correction coefficient and temperature-corrected motor parameters to obtain the magnetic characteristic coefficient; Generate a comprehensive compensation factor based on the rotor main effect coefficient, stator-rotor interaction coefficient, mechanical influence coefficient, and magnetic characteristic coefficient; Perform iterative optimization of slip compensation according to the preliminary fusion slip rate, comprehensive compensation factor, and temperature-corrected motor parameters to obtain the optimized slip rate.

10. The slip rate measurement method of the three-phase asynchronous motor according to claim 9, characterized in that, The analysis of the interaction effect of stator impedance is specifically as follows: Analyze the balance of current distribution according to the temperature correction coefficient to obtain the current distribution change rate; Correct the air-gap magnetic field intensity according to the current distribution change rate to obtain the magnetic field intensity correction coefficient; Calculate the rotor induced electromotive force according to the magnetic field intensity correction coefficient to obtain the induced electromotive force coefficient; Analyze the torque balance characteristics according to the induced electromotive force coefficient to obtain the torque balance coefficient; Evaluate the temperature effect of stator leakage reactance according to the torque balance coefficient to obtain the leakage reactance effect coefficient; Determine the stator-rotor coupling factor for the current distribution change rate, magnetic field intensity correction coefficient, induced electromotive force coefficient, torque balance coefficient, and leakage reactance effect coefficient to obtain the stator-rotor coupling factor; Perform comprehensive calculation of the interaction coefficient for the stator-rotor coupling factor, temperature-corrected motor parameters, and rotor main effect coefficient to obtain the stator-rotor interaction coefficient.

Images