A method for diagnosing and monitoring faults of an internal inspection robot of a power transformer
By calculating the heat dissipation barrier factor and the negative sequence suppression attenuation coefficient, and adjusting the motor compensation voltage command, the problems of thermal breakdown and permanent magnet demagnetization of the power transformer internal inspection robot during stator turn short circuit faults were solved, thus achieving safe and continuous operation of the robot.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LIAONING HONGYANHE NUCLEAR POWER
- Filing Date
- 2026-05-11
- Publication Date
- 2026-06-09
Smart Images
Figure CN122178786A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motor control, and more specifically, to a method for fault diagnosis and monitoring of an internal inspection robot for power transformers. Background Technology
[0002] In the field of industrial automation, especially in the inspection of power transformers and other enclosed, oil-immersed environments, the reliability of the drive motor of the internal inspection robot is crucial. When the motor experiences early electrical faults such as stator inter-turn short circuits, existing fault-tolerant control strategies typically employ fixed-intensity negative-sequence voltage compensation to suppress torque ripple and rely on fixed temperature rise or current thresholds to trigger shutdown or derating protection. However, under the special working conditions of the transformer bottom, which is insulated and has high resistance due to high-viscosity insulating oil and deposited sludge, this traditional single-dimensional open-loop protection and control scheme is insufficient to meet the robot's need for continuous and safe escape from obstacles.
[0003] Specifically, in extremely poor heat dissipation environments, forcibly applying negative sequence voltage compensation can cause the healthy phase windings to overheat rapidly. The accumulated heat can easily trigger insulation thermal breakdown, leading to catastrophic failure. If compensation commands are simply suppressed to avoid thermal runaway, the motor's residual torque will be unable to overcome the extremely high static friction resistance of the underlying oil sludge, causing the robot to lose its mobility. At this point, attempting to restore thrust by conventionally increasing the quadrature-axis current will only lead to a vicious cycle of increased heating. On the other hand, attempting to increase thrust by injecting direct-axis current to enhance the magnetic field, under high-temperature conditions lacking dynamic monitoring, can easily cause irreversible high-temperature demagnetization of the rotor permanent magnets. Existing solutions severely lack early diagnosis of the risk of deep thermal accumulation inside the motor under complex oil-immersion conditions, as well as real-time online monitoring of the demagnetization edge state of the permanent magnets, resulting in reactive error reporting and shutdown only after damage has occurred.
[0004] Therefore, after the drive motor of the internal inspection robot fails, it is necessary to effectively compensate for the static friction thrust gap while suppressing heat generation at the source, provide online dynamic anti-demagnetization protection for the permanent magnet, and establish a real-time fault diagnosis and monitoring system covering early heat accumulation and demagnetization risks, so as to ensure that the robot can still perform its tasks safely and continuously in the fault state. Summary of the Invention
[0005] To address the technical problem in existing technologies where robots encounter stator turn-to-turn short circuits and other faults in high-viscosity insulating oil environments, traditional protection strategies, lacking multi-dimensional thermomagnetic physical countermeasures and dynamic compensation constraints, are prone to irreversible thermal breakdown of winding insulation and high-temperature demagnetization of permanent magnets, ultimately leading to the robot's loss of maneuverability, this invention provides a fault diagnosis and monitoring method for power transformer internal inspection robots. The technical solution is as follows: In a first aspect, the present invention provides a method for fault diagnosis and monitoring of an internal inspection robot for power transformers, comprising: The ambient temperature, casing temperature, three-phase stator current, input power, and mechanical power of the motor are obtained. Based on the three-phase stator current, the negative sequence intrinsic current component is extracted, and the heat dissipation barrier factor is obtained by combining the ambient temperature, the shell temperature, the input power and the mechanical power, and a negative sequence suppression attenuation coefficient is generated. The original negative sequence compensation voltage command is attenuated and adjusted using the negative sequence suppression attenuation coefficient to obtain the corrected compensation voltage command. The static friction torque gap is calculated based on the negative sequence suppression attenuation coefficient and the static viscous friction threshold. A negative direct-axis weak magnetic current component is generated based on the static friction torque gap, inductance parameters and rotor flux linkage. The extended back electromotive force is extracted by collecting the quadrature voltage and current of the motor, and the observed flux linkage amplitude is obtained by combining the electric angular velocity. The observed flux amplitude and the lower limit of the safety flux are used to generate a dynamic limiting extreme value, which is then used to clamp the negative direct-axis weak magnetic current component to obtain the final direct-axis current command. The final direct-axis current command is input to the control unit to generate a signal to drive the inverter.
[0006] Based on the negative sequence intrinsic current component, the heat dissipation barrier factor, the static friction torque gap, and the observed flux linkage amplitude, a multidimensional state assessment index is constructed, and corresponding fault diagnosis and operational health monitoring data are generated.
[0007] Preferably, the step of combining the ambient temperature, the casing temperature, the input power, and the mechanical power to obtain the heat dissipation barrier factor specifically includes: calculating the difference between the input power and the mechanical power to obtain the motor's power loss; calculating the difference between the casing temperature and the ambient temperature to obtain a temperature difference parameter; performing a time integration operation on the power loss from the fault initiation time to the current time to obtain the cumulative power loss; summing the temperature difference parameter with a preset anti-overflow constant to obtain an effective heat dissipation drive denominator; and dividing the cumulative power loss by the effective heat dissipation drive denominator, with the calculated quotient serving as the heat dissipation barrier factor.
[0008] Preferably, the generation of the negative order suppression attenuation coefficient specifically includes: when the heat dissipation barrier factor is less than or equal to a preset first safety tolerance boundary, the negative order suppression attenuation coefficient is assigned a value of 0; when the heat dissipation barrier factor is greater than the first safety tolerance boundary and less than a preset second safety tolerance boundary, the excess amount of the heat dissipation barrier factor beyond the first safety tolerance boundary is calculated, and the interval span difference between the second safety tolerance boundary and the first safety tolerance boundary is calculated, and the ratio of the excess amount to the interval span difference is used as the negative order suppression attenuation coefficient; when the heat dissipation barrier factor is greater than or equal to the second safety tolerance boundary, the negative order suppression attenuation coefficient is assigned a value of 1.
[0009] Preferably, the step of adjusting the original negative-sequence compensation voltage command by using the negative-sequence suppression attenuation coefficient to obtain the corrected compensation voltage command specifically includes: subtracting the value 1 from the negative-sequence suppression attenuation coefficient to obtain a retention ratio factor; multiplying the original negative-sequence compensation voltage command calculated by the symmetrical component method with the retention ratio factor, and using the resulting product as the corrected compensation voltage command.
[0010] Preferably, the step of calculating the static friction torque gap based on the negative-sequence suppression attenuation coefficient and the static viscous friction threshold specifically includes: multiplying a preset torque attenuation mapping coefficient by the negative-sequence suppression attenuation coefficient to obtain an attenuation multiplier; calculating a power function value with the natural constant as the base and the negative of the attenuation multiplier as the exponent; subtracting the power function value from the value 1 to obtain a torque drop ratio factor; multiplying the static viscous friction threshold obtained from the table by the torque drop ratio factor, and using the resulting product as the static friction torque gap.
[0011] Preferably, the step of generating a negative direct-axis field-weakening current component based on the static friction torque gap, inductance parameters, and rotor flux linkage specifically includes: obtaining the number of pole pairs, direct-axis inductance parameters, quadrature-axis inductance parameters, and current quadrature-axis current of the motor; calculating the difference between the direct-axis inductance parameters and the quadrature-axis inductance parameters; multiplying the constant 3, the number of pole pairs, the difference, and the current quadrature-axis current together to obtain a torque conversion reference number; and dividing the product of the constant 2 and the static friction torque gap by the torque conversion reference number, with the resulting quotient being the negative direct-axis field-weakening current component.
[0012] Preferably, the step of obtaining the observed flux linkage amplitude by combining electric angular velocity specifically includes: performing adaptive filtering on the extracted extended back electromotive force to obtain a back electromotive force distribution vector and extracting the amplitude of the back electromotive force distribution vector; obtaining the larger value between the electric angular velocity and a preset minimum rotational speed clamping threshold as the denominator of the effective angular velocity; dividing the amplitude of the back electromotive force distribution vector by the denominator of the effective angular velocity, and calculating the quotient as the observed flux linkage amplitude characterizing the magnetic field strength of the permanent magnet.
[0013] Preferably, the step of generating a dynamic limiting extreme value using the observed flux linkage amplitude and the lower limit of the safety flux linkage specifically includes: calculating the difference between the observed flux linkage amplitude and the lower limit of the safety flux linkage; multiplying the difference by a preset flux linkage decay sensitivity factor to obtain a decay independent variable parameter; calculating a decay power function value with the natural constant as the base and the negative of the decay independent variable parameter as the exponent; summing the value 1 with the decay power function value to obtain a smoothed tightening denominator term; dividing the preset allowed physical upper limit of the direct-axis current of the motor hardware by the smoothed tightening denominator term, and using the quotient value as the dynamic limiting extreme value.
[0014] Preferably, the step of clamping the negative direct-axis field-weakening current component to obtain the final direct-axis current command specifically includes: taking the opposite of the physical upper limit value of the direct-axis current preset by the motor hardware, dividing it by the smoothing and tightening denominator term to obtain the negative dynamic limiting extreme value; comparing the negative direct-axis field-weakening current component with the negative dynamic limiting extreme value, and selecting the larger of the two values as the final direct-axis current command after rigid clamping.
[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By integrating physical quantities such as temperature and power in real time, the thermal runaway barrier factor, which characterizes the antagonistic relationship between heat generation and dissipation, is quantitatively calculated. When heat dissipation conditions deteriorate or heat continues to accumulate, a negative-sequence suppression attenuation coefficient can be dynamically generated to proactively and smoothly adjust the original compensation voltage command. This mechanism cuts off the positive feedback loop of thermal runaway by reducing the excitation at the source before the thermophysical boundary is broken down, significantly reducing the risk of motor burnout under extremely poor heat dissipation conditions.
[0016] 2. By using the negative-sequence suppression attenuation coefficient and the static viscous friction threshold, the static friction torque gap required for the robot to escape high-viscosity sludge was precisely quantified. Utilizing the salient-pole characteristics of the permanent magnet synchronous motor, a negative direct-axis weak magnetic current component was calculated in reverse and injected into the stator side. The change in the relative angle between the stator current phasor and the rotor flux linkage was used to excite the reluctance torque effect to compensate for thrust loss. This strategy logically isolates the connection between thrust recovery and the main heat source (quadrature-axis current), successfully breaking the traditional control dilemma of "heat preservation leads to power loss, and power increase leads to heat increase," thus ensuring the robot's ability to escape obstacles.
[0017] 3. By extracting the extended back electromotive force from the port electrical signal, the transient observed flux linkage amplitude, characterizing the magnetic field strength of the permanent magnet, is calculated online. Based on how close this observed value is to the lower limit of the safe flux linkage, a limiting extremum is dynamically generated, and the negative direct-axis weak magnetic current component is rigidly clamped. This mechanism achieves closed-loop protection of the magnetic field state, precisely limiting the amplitude of the direct-axis injected current before irreversible decay of the permanent magnet occurs. It physically constructs the ultimate safety barrier against high-temperature demagnetization, ensuring long-term sustainable operation under fault conditions. Attached Figure Description
[0018] Figure 1 This is a flowchart illustrating the implementation of a fault diagnosis and monitoring method for an internal inspection robot of a power transformer according to an embodiment of the present invention. Detailed Implementation
[0019] S1: Obtain the ambient temperature, casing temperature, three-phase stator current, input power, and mechanical power of the motor.
[0020] Due to the extremely high viscosity of the insulating oil and the harsh environment of sludge buildup inside power transformers, the Joule heat generated inside the robot drive motor during early faults such as stator turn-to-turn short circuits can easily lead to thermal runaway risks due to obstructed heat transfer paths.
[0021] Single-dimensional protection strategies typically rely solely on the drive motor current threshold or a single temperature monitoring point. This makes it difficult to accurately perceive the real-time dynamic relationship between heat accumulation and the external environment's heat dissipation capacity under complex oil-immersion heat exchange conditions. Consequently, protection actions are often delayed and cannot effectively prevent irreversible thermal breakdown damage to the motor winding insulation layer caused by instantaneous local high temperatures.
[0022] To obtain complete data on the energy flow and dissipation within the motor, ambient reference oil temperature is collected in real time. Transient temperature of motor casing and three-phase stator current Combined with input electrical power With mechanical power The real-time difference is used to calculate the power loss converted into heat energy. This method of obtaining parameters based on the law of conservation of energy aims to provide accurate basic data support for the subsequent calculation of the heat dissipation barrier factor.
[0023] The ambient reference oil temperature of the insulating oil inside the transformer is continuously collected by a temperature sensor. Meanwhile, a platinum resistance sensor mounted in the middle of the motor housing is used to monitor the motor housing temperature in real time. .
[0024] These two temperature parameters can be used to evaluate the motor's heat dissipation driving force. To ensure data validity, the sensor sampling period must be synchronized with the control period, and the calibration latching period of the ambient reference oil temperature must be specified. The empirical value of 500ms is set to effectively filter out high-frequency thermal disturbance signals. The calibrated latch period value can be adjusted by the implementer according to the specific implementation scenario.
[0025] When acquiring power parameter data, transient three-phase stator current is extracted using a high-bandwidth Hall current sensor at a sampling frequency of no less than 10 kHz. The sampling frequency can be adjusted by the implementer according to the specific implementation scenario.
[0026] The direct-axis voltage command is calculated by the current loop proportional-integral regulator inside the motor control unit. With cross-axis voltage command And combined with the direct-axis current fed back after coordinate transformation With cross-axis current Real-time calculation of the stator input power of the drive motor Simultaneously, the real-time mechanical speed feedback from the photoelectric encoder is utilized. Electromagnetic torque estimated by the internal state observer Calculate the actual mechanical power output of the motor .
[0027] The total internal power loss of the motor is calculated in real time based on the parameters collected above. The specific calculation formula is as follows: In the formula, It represents the total power loss of the motor in converting energy into heat, and is used to indicate the heat source intensity per unit time. The total electrical power input to the motor in real time; The effective mechanical power output by the motor; and These are the direct-axis and quadrature-axis voltage commands under constant amplitude coordinate transformation, respectively. and These are the direct-axis and quadrature-axis currents, respectively.
[0028] Through calculation and The dynamic difference can bypass the complex internal structural parameters of the motor and directly find the source of heat intensity from the perspective of energy conservation.
[0029] S2: Based on the three-phase stator current, the negative sequence intrinsic current component is extracted, and the heat dissipation barrier factor is obtained by combining the ambient temperature, shell temperature, input power and mechanical power, and the negative sequence suppression attenuation coefficient is generated.
[0030] After a stator-to-turn short circuit fault occurred in the motor of the internal inspection robot, the three-phase stator current... It will exhibit obvious asymmetry, thereby inducing a negative sequence intrinsic current component with frequency doubling characteristics. The additional Joule heat generated by this sequence intrinsic current component in the stator winding will directly lead to a sharp increase in local temperature rise.
[0031] Due to the high viscosity of the insulating oil inside the transformer, the heat exchange path is obstructed, and the fixed compensation strategy is very likely to cause thermal breakdown of the winding insulation layer under extreme operating conditions.
[0032] Therefore, it is necessary to be able to dynamically assess the balance between the energy conflict between heat generation and heat dissipation during a fault.
[0033] By introducing the real-time acquired input power respectively With mechanical power The dynamic difference is taken as the heat source intensity, and the time integral of the heat source intensity is calculated to represent the total heat entering the motor.
[0034] At the same time, the temperature of the motor casing is introduced. With ambient temperature The temperature difference acts as a barrier to heat dissipation, thus constructing a feature that can predict the risk of thermal breakdown a priori, namely the heat dissipation barrier factor. .
[0035] In the specific implementation process, the first step is to collect the three-phase stator current. Perform Discrete Fourier Transform or Sliding Window Symmetric Component Decomposition.
[0036] To obtain the negative-sequence intrinsic current component reflecting the actual operating state of the internal inspection robot The three-phase stator current is decoupled using a symmetrical component method combined with digital low-pass filtering to extract the pure negative-sequence intrinsic current component. It is independently latched, which can provide accurate raw input data for subsequent acquisition and correction of compensation instructions.
[0037] Subsequently, the heat dissipation barrier factor The specific construction process is as follows: Obtain the real-time input power With mechanical power And its difference, i.e., the transient power loss at the onset of the fault. up to the current moment Integral operations are performed within the interval.
[0038] To avoid energy integral divergence and overflow due to long-term operation, an exponential forgetting factor is introduced to construct a sliding dissipation integral model.
[0039] Divide the integral energy value by and The effective heat dissipation driving denominator is used to calculate the heat dissipation barrier factor, which characterizes the current heat dissipation bottleneck. The specific calculation process is as follows: In the formula, It is the heat dissipation barrier factor; and These are the start and end times of the integration process, respectively. For integral infinitesimal variables; To suppress the forgetting factor that causes the integral to diverge infinitely; Real-time temperature of the motor housing; The reference ambient temperature for transformer oil; This is a preset overflow prevention constant, typically set to a very small positive number; an empirical value is [value missing]. It can be adjusted by the implementer according to the specific implementation scenario.
[0040] The poor heat dissipation environment formed by the high-viscosity insulating oil and sludge deposits inside the transformer has significant thermal inertia and thermal hysteresis characteristics. Relying solely on the transient temperature threshold for judgment cannot truly reflect the cumulative thermal stress borne by the insulating material over a long period of time. This can easily lead to insulation breakdown when the temperature has not reached the threshold but heat has already accumulated significantly.
[0041] To this end, a heat dissipation barrier factor is constructed and used as a key intermediate variable connecting the thermal field and the electrical control dimension. By integrating the dynamic temperature difference in the time domain with attenuation, it represents the dynamic balance process of heat generation and dissipation of the motor under harsh operating conditions inside the transformer.
[0042] Specifically, in the formula for the heat dissipation barrier factor, the difference between the real-time temperature of the motor casing and the reference ambient temperature of the transformer oil is used to represent the natural driving force and transient heat generation intensity of the current outward heat dissipation. The introduction of a minimal positive constant as an anti-overflow constant in the denominator is to safely map the temperature difference signal into a dimensionless or dimensionless control factor when performing discretization calculations, so as to avoid the internal inspection robot motor control crashing due to division by zero overflow or calculation truncation.
[0043] Furthermore, by accumulating the historical states within the calculation cycle using integral infinitesimals, the total amount of heat accumulated inside the motor can be calculated, thus accurately reflecting the risk of continuous fatigue damage to the insulation layer caused by thermal inertia.
[0044] Meanwhile, to prevent the time integral term from exhibiting infinite numerical divergence during long-term steady-state operation of the motor, a forgetting factor based on natural exponential decay is introduced into the integrator kernel.
[0045] This forgetting factor gives the calculation results the characteristic of memory decay. That is, the longer the heat accumulation is, the lower its contribution to the current thermal breakdown risk because it has been gradually dissipated to the surrounding transformer oil. On the other hand, the heat source surge that is closer to the current moment is given a higher weight.
[0046] The calculated Mapped to the attenuation coefficient used to actively suppress the original negative sequence compensation voltage command. .
[0047] The analytical expression of its piecewise linear mapping model is as follows: In the formula, The negative order suppression attenuation coefficient; the first safety tolerance boundary. Set to 100 J / K, second safety tolerance boundary Set to 500J / K.
[0048] When faults such as stator inter-turn short circuits occur, forced full compensation can easily lead to irreversible thermal breakdown of the insulation layer. On the other hand, simply cutting off the compensation command will cause the residual torque of the robot motor to be unable to overcome the extremely high static friction resistance, thus resulting in loss of maneuverability and introducing a heat dissipation barrier factor. It serves as an independent variable to dynamically characterize the physical antagonistic relationship between heat generation and heat dissipation.
[0049] Specifically, the mapping relationship and control logic of the piecewise linear mapping model are divided into three stages based on the degree of heat accumulation.
[0050] The first stage is the safe operation stage, when the heat dissipation barrier factor... When the value is less than or equal to the first safety tolerance boundary, it indicates that the current environment's heat dissipation capacity is sufficient to handle the redundant heat generated inside the motor. In this case, the model will suppress the attenuation coefficient with a negative order. Assigning a value of zero ensures that the motor control of the internal inspection robot fully retains the original negative sequence compensation voltage command, providing the maximum torque required for the robot to stably escape from obstacles.
[0051] The second stage is the adaptive smooth decay stage, when the heat dissipation barrier factor... When the temperature exceeds the first safety tolerance boundary but is less than the second safety tolerance boundary, it indicates that heat has begun to accumulate inside the motor, and the heat dissipation conditions are gradually deteriorating.
[0052] At this point, through calculation The ratio of the excess amount beyond the first boundary to the difference in span between the two boundary intervals generates a linear ratio that varies continuously between 0 and 1. This enables the internal inspection robot to actively and smoothly reduce the amplitude of the original negative sequence compensation voltage command during motor control, cutting off the positive feedback loop of thermal runaway at the source, while avoiding torque pulses caused by compensation abrupt changes.
[0053] The third stage is the rigid thermal protection clamping stage, when the heat dissipation barrier factor... When the temperature is greater than or equal to the second safety tolerance boundary, it indicates that the heat accumulation has approached the insulation material's tolerance limit, and the model forces the temperature to be lower than or equal to the second safety tolerance boundary. The value is set to 1, thus completely suppressing instructions that could induce dangerous temperature rises.
[0054] S3: The original negative sequence compensation voltage command is attenuated and adjusted using the negative sequence suppression attenuation coefficient to obtain the corrected compensation voltage command.
[0055] When a stator-turn short-circuit fault occurs in the drive motor of the transformer inspection robot, the conventional fault-tolerant control strategy usually generates an original negative-sequence compensation voltage command based on the symmetrical component method to counteract the electromagnetic torque pulsation caused by the asymmetrical current.
[0056] However, in a closed environment with high insulating oil viscosity and extremely poor heat dissipation, the fixed strength compensation command will force the healthy phase winding to carry a current far exceeding the rated amplitude for a long time. The resulting surge in additional Joule heat can easily trigger secondary thermal breakdown of the insulating material.
[0057] Therefore, in the event of thermal runaway, extreme measures such as power outages that result in loss of mobility should not be taken. Instead, feedforward intervention should be implemented by introducing pre-generated real-time thermal state characterization variables. A retention factor can be constructed to proactively reduce the redundant compensation magnitude in the original voltage command that may induce dangerous temperature rises.
[0058] In the specific implementation process, the conventional voltage command calculation process is executed first.
[0059] The extracted negative-sequence intrinsic current component is input into a preset resonant regulator, and the original negative-sequence compensation voltage command is calculated using the symmetrical component method. .
[0060] Subsequently, the negative-order suppression attenuation coefficient generated in real time based on the heat dissipation barrier factor is invoked. The retention ratio factor, which characterizes the safety compensation margin, is obtained by subtracting the value of 1 from the negative order suppression attenuation coefficient. .
[0061] Then, an adaptive command amplitude scaling operation is performed, and the calculated product is used as a correction compensation voltage command that can balance torque smoothness and thermal safety. .
[0062] The specific calculation formula for this adaptive attenuation adjustment process is as follows: In the formula, This is the corrected compensation voltage command after thermal constraint modulation. The negative order suppression attenuation coefficient; The calculated original negative sequence compensation voltage command; The retention scaling factor generated for intermediate calculations.
[0063] Under the complex working conditions inside a power transformer, directly controlling the start and stop of the motor based on the temperature threshold can cause the robot to experience severe mechanical shock or become trapped in the sludge accumulation area due to loss of thrust.
[0064] Therefore, by constructing a retention ratio factor This allows a dynamic buffer layer to be built between the original control command and the safety boundary, enabling feedforward control of the heat generation intensity of the faulty motor by finely clipping the peak of the negative sequence voltage amplitude.
[0065] Specifically, retain the scaling factor As the negative order suppression attenuation coefficient The complement of represents the percentage of control energy that can be retained under the current thermal environment.
[0066] When the heat dissipation barrier factor is within the safe range If it is zero, then The output correction compensation voltage command is 1. Equivalent to the original negative sequence compensation voltage command This ensures that the robot has rated self-healing and obstacle-avoidance thrust.
[0067] With the increasing risk of heat accumulation, linear increase leads to The corresponding reduction and the corrected voltage command achieve a smooth reduction in amplitude through multiplication and dissociation.
[0068] The advantage of this proportional adjustment logic is that it does not change the phase or frequency characteristics of the negative sequence voltage, but only modulates its voltage value, thereby effectively reducing the thermal load of the healthy phase winding while still maintaining a certain degree of torque ripple suppression effect.
[0069] When the extreme thermal protection logic is triggered The value of the compensation voltage command is reset to zero, thus cutting off the electrical source that could lead to thermal runaway.
[0070] S4: Calculate the static friction torque gap based on the negative sequence suppression attenuation coefficient and the static viscous friction threshold, and generate a negative direct-axis weak magnetic current component based on the static friction torque gap, inductance parameters and rotor flux linkage.
[0071] After implementing negative sequence voltage compensation suppression on the drive motor of the internal inspection robot to avoid the risk of thermal breakdown, the electromagnetic torque output will inevitably experience a drop in overall amplitude. The high-viscosity insulating sludge deposited at the bottom of the transformer constitutes a huge static viscous frictional resistance to the robot's movement.
[0072] If the residual output torque of the motor is insufficient to overcome the static friction resistance in this environment, the robot will become stuck and lose its ability to escape.
[0073] If the quadrature current is directly increased to restore torque, the stator copper loss will increase sharply, thus triggering thermal runaway again.
[0074] To achieve thrust compensation without thermal additives, the static friction torque gap is first calculated based on the negative-sequence suppression attenuation coefficient generated previously and the real-time ambient temperature. The calculation formula is shown below: In the formula, The calculated static friction torque gap; The static viscous friction threshold is obtained by looking up a table based on the ambient temperature; and These are the preset first torque attenuation mapping coefficient and the second torque attenuation mapping coefficient, respectively.
[0075] The static viscous friction threshold obtained by looking up a table based on ambient temperature is derived from prior experimental calibration based on the oil sludge environment inside the transformer.
[0076] Specifically, in the initialization phase, a simulated test environment containing standard transformer insulating oil and typical deposited sludge is constructed. The temperature of this environment is discretized within a preset typical operating temperature range of the transformer using a constant temperature control box.
[0077] Specifically, at each discrete temperature node, the output torque of the internal inspection robot's drive motor is continuously increased by a high-precision dynamometer or torque sensor. The critical torque value at the moment when the robot just overcomes the static friction resistance and begins to slide relative to the ground is recorded. By mapping and storing the critical torque values obtained from the test at different ambient temperature nodes, a one-dimensional static viscous friction threshold lookup table can be constructed.
[0078] First preset torque attenuation mapping coefficient The primary characteristic is the trigger sensitivity of the attenuation adjustment. Its value is inversely correlated with the motor thermal time constant and the stator core heat capacity. An empirical value of 1.2 is used, indicating that under the maximum allowable compensation state, it provides a torque margin of 20% more than the basic static friction resistance, thereby ensuring that the robot can effectively break free from the sludge. This value can be adjusted by the implementer according to the specific implementation scenario. The second preset torque attenuation mapping coefficient... It mainly characterizes the slope or intensity of the attenuation command decline. An empirical value of 0.5 is used, which means that a moderate curve decline is introduced in the middle and late stages of the compensation command rise to prevent secondary thermal breakdown or mechanical resonance caused by excessive compensation. It can be adjusted by the implementer according to the specific implementation scenario.
[0079] After obtaining the torque gap, thrust compensation is performed using the reluctance torque characteristics of the salient-pole motor, and the number of pole pairs of the motor is collected in real time. Direct-axis inductor quadrature axis inductance and the current quadrature axis feedback current .
[0080] Because internal inspection robots generally use built-in permanent magnet synchronous motors for drive, the quadrature axis inductance is often greater than the direct axis inductance, i.e. .
[0081] To compensate for the torque gap required for getting out of trouble By adjusting the direct-axis current, reluctance torque can be utilized. Based on the torque equation of a permanent magnet synchronous motor, given the current quadrature-axis feedback current... Under these conditions, the required negative direct-axis magnetic weakening current component to be injected The following formula is obtained through reverse calculation: In the formula, This represents the number of pole pairs in the motor. This formula is derived based on the complete torque equation of a salient-pole motor, because... And the torque to be compensated Calculation results It must be a negative value. This negative direct-axis weak magnetic current component is injected into the stator as a bottom-level incremental command, accurately extracting the positive reluctance torque to compensate for the escape thrust without significantly increasing the main heat source.
[0082] In extreme conditions where there is a short circuit between stator turns and the transformer sludge severely hinders heat dissipation, if the traditional strategy of directly increasing the quadrature axis current to increase thrust is adopted, the copper loss of the stator winding will increase exponentially, which will easily trigger thermal runaway again and cause secondary breakdown of the insulation layer.
[0083] Therefore, the conventional approach of adding heat sources was abandoned, and instead the salient pole effect of the built-in permanent magnet synchronous motor was utilized. In the standard torque equation, the total electromagnetic torque is composed of the superposition of the permanent magnet excitation torque and the reluctance torque generated by the magnetic reluctance asymmetry.
[0084] Because the quadrature-axis inductance of the built-in motor is always greater than the direct-axis inductance, that is... Under the premise of maintaining the current quadrature-axis current for positive traction, that is In order to obtain a positive torque gap compensation value, i.e., T_{gap}>0, the direct-axis current compensation command calculated in reverse by the above formula must be negative.
[0085] When this negative direct-axis weak magnetic current component is injected into the stator side as a bottom-level incremental command, the resulting armature reaction magnetic field can not only interact with the rotor magnetic field to generate considerable reluctance torque to accurately compensate for the escape thrust, but also weaken the main magnetic field to a certain extent, reduce the eddy current loss and hysteresis loss of the iron core, and achieve the robot's thermal additional power compensation without significantly increasing the main heat source.
[0086] S5: Collect the quadrature voltage and current of the motor to extract the extended back electromotive force, and combine it with the electric angular velocity to obtain the observed flux linkage amplitude.
[0087] In the process of injecting a negative direct-axis weak magnetic current component into the drive motor of the internal inspection robot to compensate for the escape thrust, the strong demagnetizing armature reaction will inevitably significantly interfere with the magnetic field state on the stator side.
[0088] However, the harsh high-temperature insulating oil environment inside the transformer is very likely to weaken the magnetic properties of the rotor permanent magnet. At this time, the strong armature reaction can easily induce irreversible high-temperature demagnetization of the permanent magnet.
[0089] Magnetic flux monitoring relies on built-in Hall sensors, but such hardware is easily damaged and fails in high-temperature and high-pressure oil immersion environments, resulting in the inability to accurately sense the health status of the magnetic field, and thus facing the risk of getting out of control.
[0090] According to the core electromagnetic theory of permanent magnet synchronous motors, the back electromotive force generated by the rotor rotating and cutting the stator windings has a strictly proportional relationship with the effective flux linkage in the air gap and the rotational angular velocity of the motor.
[0091] Therefore, by extracting the electrical parameters at the motor's ports to calculate the extended back electromotive force and removing the speed variable, the observed flux linkage amplitude, which characterizes the magnetic field strength of the permanent magnet, can be indirectly obtained.
[0092] In the specific implementation process, a two-phase stationary orthogonal coordinate system is first obtained by acquiring and mapping through a coordinate transformation matrix. Voltage components in coordinate system and and the corresponding current components and .
[0093] Subsequently, combined with the pre-calibrated stator resistance Based on the stator voltage equations in a two-phase stationary coordinate system, and considering inductance parameters: An extended back EMF observer is constructed to perform forward state calculations. Adaptive filtering is then applied to obtain the back EMF distribution vector components. and .
[0094] After obtaining the back electromotive force distribution vector, the amplitude is extracted using a numerical algorithm that takes the square root of the sum of squares. Simultaneously, the current transient electric angular velocity is obtained. .
[0095] Finally, the amplitude of the observed magnetic flux was calculated. The calculation formula is as follows: In the formula, The observed flux linkage amplitude represents the magnetic field strength of the permanent magnet. and The back electromotive force distribution vector components in an orthogonal coordinate system; It is the transient electric angular velocity; To prevent division overflow anomalies at extremely low speeds, a preset minimum speed clamping threshold is set at 5% of the motor's rated electrical angular velocity, which can be adjusted by the implementer according to the specific implementation scenario.
[0096] The magnitude of the back electromotive force distribution vector after adaptive filtering is extracted by solving the square root of the sum of squares of the back electromotive force components in the two-phase orthogonal coordinate system.
[0097] Since the amplitude of the back electromotive force distribution vector is generated by the coupling effect of magnetic flux and rotational speed, in order to obtain characteristic quantities that can represent the health state and residual magnetic properties of permanent magnets, the electric angular velocity is decoupled and stripped by a division operator.
[0098] The denominator introduces a maximum value comparison function as a protection barrier for the underlying operation. When the robot is stuck in the mud and is preparing to perform an escape action in the initial stage, the motor speed is slowly built up from zero. At this time, the transient electric angular velocity is extremely small or even close to zero. Furthermore, if it is directly used as a divisor, it will not only cause the operation to overflow due to division by zero, leading to program crash, but also the back electromotive force extracted in the extremely low speed region has a very low signal-to-noise ratio. Direct division will cause the observed magnetic flux to oscillate and become distorted.
[0099] By introducing a preset minimum speed clamping threshold to construct a lower limit for the divisor, a stable and conservative state estimation can be forcibly performed according to the safety set value in the extremely low speed start-up zone. Once the speed exceeds this threshold, it can smoothly switch to a precise division calculation mode.
[0100] S6: By using the observed flux amplitude and the lower limit of the safety flux to generate a dynamic limiting extreme value, the negative direct-axis weak magnetic current component is clamped to obtain the final direct-axis current command.
[0101] The negative direct-axis weak magnetic current component calculated for the pre-compensation escape thrust generates a demagnetizing armature reaction magnetic field that, under the high-temperature oil immersion environment of the transformer, can easily cause the rotor permanent magnet operating point to cross the demagnetization curve inflection point.
[0102] The difference between the currently observed flux linkage amplitude and the preset lower limit of the safety flux linkage is calculated to construct a negative dynamic limiting extremum to dynamically evaluate the residual capacity of the permanent magnet to withstand demagnetizing magnetomotive force. The formula for calculating the negative dynamic limiting extremum is as follows: In the formula, This represents the negative dynamic limiting extreme value generated; This represents the physical upper limit of the current allowed by the motor hardware. This is the flux linkage decay sensitivity factor; To observe the flux linkage amplitude; The lower limit of the safety flux is set as 0.85 of the rated nominal flux amplitude at room temperature. The flux decay sensitivity factor mainly represents the sensitivity of the dynamic limiting curve near the demagnetization critical point, with an empirical value of 200. Both the lower limit of the safety flux and the flux decay sensitivity factor can be adjusted by the implementer according to the specific implementation scenario.
[0103] Because the poor heat dissipation conditions inside the transformer can easily lead to a sharp increase in the temperature of the motor rotor, and the magnetic force of the permanent magnet material decreases significantly with increasing temperature, if a static constant current limiting strategy is adopted, the operating point of the permanent magnet can easily be pushed below the inflection point of the demagnetization curve under high temperature conditions, thereby causing an irreversible high-temperature demagnetization disaster.
[0104] Using the real-time difference between the current observed flux amplitude and the lower limit of the safe flux as an independent variable, the physical upper limit amplitude of the allowed injected negative direct-axis current is dynamically modulated through an exponential function.
[0105] When the observed flux linkage amplitude is much greater than the lower limit of the safe flux linkage, it indicates that the current permanent magnet has good magnetic properties and the ambient temperature has not yet posed a substantial threat to it. At this time, the exponent term approaches 0 and the denominator approaches 1, allowing the full output of the negative physical upper limit of the current amplitude to maximize the escape thrust.
[0106] As the thermal load intensifies, causing the observed flux linkage amplitude to gradually decline and approach the safety lower limit, the exponential term increases rapidly, leading to a sharp increase in the denominator. The negative dynamic limiting extremum then shrinks rapidly and smoothly towards the zero axis. This nonlinear modulation mechanism allows for a proactive and gentle approach to the output boundary of the demagnetizing current before the permanent magnet faces the demagnetization critical point, avoiding the torque abrupt change and mechanical shock caused by hard limiting switching.
[0107] After obtaining the dynamic negative state limiting extreme value, obtain the final direct axis current command and use the generated negative dynamic limiting extreme value as the demagnetization safety boundary.
[0108] Subsequently, the previously generated negative direct-axis magnetizing current component is compared with the dynamic limiting extreme value of this negative direction. The larger of the two values is selected, i.e., tightened towards the zero axis to limit the demagnetizing intensity, as the final direct-axis current command. The specific calculation formula is as follows: In the formula, This is the final direct-axis current command; This refers to the negative direct-axis magnetic weakening current component calculated previously; This generates a negative safety boundary limit. This instruction ultimately balances dynamic compensation and demagnetization protection.
[0109] Demagnetization defense is implemented by taking the maximum value function. Since the direct-axis currents that weaken and demagnetize are both negative in the synchronous rotating coordinate system of the motor, the negative direct-axis field-weakening current component calculated in the previous step and the dynamically generated negative safety boundary limit are both negative in mathematical terms.
[0110] Taking the maximum value between two negative numbers means selecting the value with the smaller absolute value, that is, the value closer to the zero point of the coordinate system.
[0111] When the absolute value of the negative current required for power compensation is greater than the absolute value of the safety boundary limit, the large value function will forcibly eliminate the dangerous over-limit part and rigidly clamp the output command on the negative safety boundary; otherwise, if the compensation requirement is within the safety margin, it will be released as is.
[0112] This ensures that no matter how extreme the external demand for escape thrust is, the output demagnetizing current intensity will be absolutely controlled within the limits that the permanent magnet can withstand under its current thermal state.
[0113] Furthermore, comparing the minimum value function with zero constitutes a mechanism to prevent false triggering of the control direction. When a digital microcontroller performs complex floating-point operations, high-frequency noise from the sensor or transient discretization truncation errors are highly likely to cause the command to exceed the positive limit near the zero axis. Since injecting positive direct-axis current not only fails to generate positive reluctance torque to help the system escape, it also increases the reactive copper losses of the stator windings, thereby further exacerbating the risk of thermal runaway.
[0114] By using this minimum value function, any possible command output greater than zero is forcibly cut off, ensuring that the direct-axis current is always strictly limited to the weak magnetic operating region in the second or third quadrant.
[0115] S7: Input the final direct-axis current command into the control unit to generate a signal to drive the inverter.
[0116] Due to the asymmetry of motor parameters and severe cross-coupling effect caused by stator inter-turn short circuit faults, directly applying the final direct-axis current command in the underlying execution circuit can easily lead to overshoot and high-frequency oscillation of the current response.
[0117] A closed-loop current tracking mechanism based on feedforward decoupling and space vector modulation is constructed. It receives the final direct-axis current command after boundary verification and coordinates with the quadrature-axis current command to maintain basic operation.
[0118] By introducing a proportional-integral regulator with cross-coupling compensation terms, the static current error under dynamic operating conditions can be effectively eliminated.
[0119] Subsequently, the continuous voltage expectation value is discretized into a high-frequency duty cycle sequence using a space vector pulse width modulation algorithm to achieve physical mapping.
[0120] In the specific implementation process, the final direct-axis current command will be... With real-time sampled direct-axis feedback current The difference is calculated and input to the direct-axis proportional-integral controller.
[0121] To eliminate dynamic coupling interference between the two axes, an electric angular velocity is introduced. This constitutes the cross-feedforward compensation term. Its decoupling control analytical model is as follows: In the formula, and Reference voltages for the direct axis and quadrature axis; and These are the proportional and integral gains of the direct-axis and quadrature-axis adjusters, respectively. Quadrature axis current command to maintain basic operation; and This is a precise cross-voltage decoupling compensation term.
[0122] Subsequently, the direct-axis and quadrature-axis reference voltages are mapped to the two-phase stationary coordinate system through the inverse Park transformation matrix to obtain the positive-sequence fundamental voltage vector.
[0123] At this point, the previously acquired correction and compensation voltage command will be... The fundamental voltage vector is then superimposed with the vector feedforward to complete the final voltage command modulation.
[0124] Finally, space vector pulse width modulation is performed, and the combined voltage vector is synthesized based on the DC bus voltage to calculate the duty cycle sequence of the six power switches of the three-phase inverter. After dead-zone compensation, the inverter is driven, where space vector pulse width modulation is a well-known technique and will not be described in detail.
[0125] S8: Construct a multi-dimensional fault diagnosis and operational health monitoring system based on parameters and control commands, and upload the diagnosis and monitoring data to the host computer interactive terminal in real time.
[0126] While the underlying control unit executes the aforementioned fault-tolerant instructions and anti-demagnetization measures, it performs state aggregation and data fusion on the key parameters generated in the preceding steps for diagnosis and monitoring.
[0127] The specific implementation process is divided into two parts: fault diagnosis and operation monitoring. In the fault diagnosis section: the effective value of the extracted negative sequence intrinsic current component is compared with a preset short-circuit characteristic threshold. If the threshold is exceeded for several consecutive electrical cycles, a diagnostic flag for a stator inter-turn short-circuit fault is generated. Simultaneously, the calculated heat dissipation barrier factor is used in conjunction with this flag. To classify and diagnose the risk of thermal runaway caused by a fault, for example, when The initial diagnosis was mild thermal contamination; the area is under safety monitoring. The diagnosis was then extremely high risk of thermal breakdown, and rigid thermal protection had been implemented.
[0128] In the operation monitoring section: the calculated static friction torque gap will be... As a monitoring value for the mechanical obstruction state of the robot chassis, the flux linkage amplitude will be observed. As a monitoring value for the health and demagnetization edge state of the rotor permanent magnet.
[0129] Finally, the above-mentioned graded diagnostic results, monitoring values, correction compensation voltage commands, and final direct-axis current commands are encapsulated in a protocol to generate a comprehensive telemetry data frame with a timestamp, which is then reported in real time to the host computer interactive terminal outside the transformer via an industrial CAN bus or Ethernet.
[0130] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the counting principle of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.
Claims
1. A method for fault diagnosis and monitoring of an internal inspection robot for power transformers, comprising: The ambient temperature, casing temperature, three-phase stator current, input power, and mechanical power of the motor are obtained. Based on the three-phase stator current, the negative sequence intrinsic current component is extracted, and the heat dissipation barrier factor is obtained by combining the ambient temperature, the shell temperature, the input power and the mechanical power, and a negative sequence suppression attenuation coefficient is generated. The original negative sequence compensation voltage command is attenuated and adjusted using the negative sequence suppression attenuation coefficient to obtain the corrected compensation voltage command. The static friction torque gap is calculated based on the negative sequence suppression attenuation coefficient and the static viscous friction threshold. A negative direct-axis weak magnetic current component is generated based on the static friction torque gap, inductance parameters and rotor flux linkage. The extended back electromotive force is extracted by collecting the quadrature voltage and current of the motor, and the observed flux linkage amplitude is obtained by combining the electric angular velocity. The observed flux amplitude and the lower limit of the safety flux are used to generate a dynamic limiting extreme value, which is then used to clamp the negative direct-axis weak magnetic current component to obtain the final direct-axis current command. The final direct-axis current command is input to the control unit to generate a signal to drive the inverter; Based on the negative sequence intrinsic current component, the heat dissipation barrier factor, the static friction torque gap, and the observed flux linkage amplitude, a multidimensional state assessment index is constructed, and corresponding fault diagnosis and operational health monitoring data are generated.
2. A method for diagnosing and monitoring the fault of a robot for internal inspection of a power transformer according to claim 1, characterized in that, The method of combining the ambient temperature, the casing temperature, the input power, and the mechanical power to obtain the heat dissipation barrier factor specifically includes: Calculate the difference between the input power and the mechanical power to obtain the motor's power loss; calculate the difference between the casing temperature and the ambient temperature to obtain the temperature difference parameter; The power loss from the fault initiation time to the current time is integrated over time to obtain the cumulative power loss. The effective heat dissipation driving denominator is obtained by summing the temperature difference parameter with the preset anti-overflow constant; the quotient obtained by dividing the accumulated energy loss by the effective heat dissipation driving denominator is used as the heat dissipation barrier factor.
3. A method for diagnosing and monitoring the fault of a robot for inspecting inside a power transformer according to claim 2, characterized by, The generation of the negative-order suppression attenuation coefficient specifically includes: When the heat dissipation barrier factor is less than or equal to the preset first safety tolerance boundary, the negative order suppression attenuation coefficient is assigned a value of 0; When the heat dissipation barrier factor is greater than the first safety tolerance boundary and less than the preset second safety tolerance boundary, the amount by which the heat dissipation barrier factor exceeds the first safety tolerance boundary is calculated, and the interval span difference between the second safety tolerance boundary and the first safety tolerance boundary is calculated. The ratio of the amount of excess to the interval span difference is used as the negative order suppression attenuation coefficient. When the heat dissipation barrier factor is greater than or equal to the second safety tolerance boundary, the negative order suppression attenuation coefficient is assigned a value of 1.
4. A method for diagnosing and monitoring the fault of a robot for inspecting inside a power transformer according to claim 3, characterized by, The step of adjusting the original negative-sequence compensation voltage command by utilizing the negative-sequence suppression attenuation coefficient to obtain the corrected compensation voltage command specifically includes: The retention ratio factor is obtained by subtracting the value 1 from the negative order suppression attenuation coefficient. The original negative-sequence compensation voltage command calculated by the symmetrical component method is multiplied by the retention scaling factor, and the resulting product is used as the corrected compensation voltage command.
5. The method for fault diagnosis and monitoring of an internal inspection robot for power transformers according to claim 1, characterized in that, The calculation of the static friction torque gap based on the negative sequence suppression attenuation coefficient and the static viscous friction threshold specifically includes: The attenuation multiplier is obtained by multiplying the preset torque attenuation mapping coefficient by the negative sequence suppression attenuation coefficient. Calculate the value of the power function with the natural constant as the base and the negative of the decay multiplier as the exponent; Subtracting the power function value from the numerical value 1 yields the torque drop ratio factor. The static viscous friction threshold obtained from the table is multiplied by the torque drop ratio factor, and the resulting product is used as the static friction torque gap.
6. The method for fault diagnosis and monitoring of an internal inspection robot for power transformers according to claim 5, characterized in that, The generation of a negative direct-axis weak magnetic current component based on the static friction torque gap, inductance parameters, and rotor flux linkage specifically includes: Obtain the number of pole pairs, direct-axis inductance parameters, quadrature-axis inductance parameters, and current quadrature-axis current of the motor; calculate the difference between the direct-axis inductance parameters and the quadrature-axis inductance parameters; Multiply the constant 3, the number of pole pairs, the difference, and the current quadrature axis current together to obtain the torque conversion reference number; The product of constant 2 and the static friction torque gap is divided by the torque conversion reference number, and the resulting quotient is taken as the negative direct-axis weak magnetic current component.
7. The method for fault diagnosis and monitoring of an internal inspection robot for power transformers according to claim 1, characterized in that, The method of obtaining the observed flux linkage amplitude by combining electric angular velocity specifically includes: The extracted extended back EMF is subjected to adaptive filtering to obtain the back EMF distribution vector and the amplitude of the back EMF distribution vector is extracted. The larger of the electric angular velocity and the preset minimum rotational speed clamping threshold is used as the denominator of the effective angular velocity. The quotient obtained by dividing the magnitude of the back electromotive force distribution vector by the denominator of the effective angular velocity is used as the observed flux linkage magnitude characterizing the magnetic field strength of the permanent magnet.
8. The method for fault diagnosis and monitoring of an internal inspection robot for power transformers according to claim 7, characterized in that, The process of generating dynamic limiting extreme values using the observed flux amplitude and the lower limit of the safety flux specifically includes: Calculate the difference between the observed flux amplitude and the lower limit of the safe flux; multiply the difference by the preset flux decay sensitivity factor to obtain the decay independent variable parameter; Calculate the value of the decay power function with the natural constant as the base and the negative of the decay independent variable parameter as the exponent; The sum of the value 1 and the value of the attenuation power function is used to obtain a smoothed tightening denominator term; the physical upper limit of the direct axis current preset by the motor hardware is divided by the smoothed tightening denominator term, and the quotient obtained is used as the dynamic limiting extreme value.
9. The method for fault diagnosis and monitoring of an internal inspection robot for power transformers according to claim 8, characterized in that, The clamping of the negative direct-axis magnetic weakening current component to obtain the final direct-axis current command specifically includes: The negative value of the physical upper limit of the direct-axis current preset by the motor hardware is divided by the smoothing and tightening denominator to obtain the negative dynamic limiting extreme value. The negative direct-axis magnetic weakening current component is compared with the negative dynamic limiting extreme value, and the larger of the two values is selected as the final direct-axis current command after being rigidly clamped.