A method for overheat protection of a direct current brush motor

Abstract

The application discloses a kind of direct current brush motor overheat protection methods, it is related to motor protection field, including the following steps: electromagnetic wave receiver arranged in the outside of motor real-time acquisition electromagnetic radiation signal generated by brush and commutator when motor operates, the electromagnetic radiation signal collected is filtered with multiple frequency band to extract the characteristic frequency component related to overheat, the signal after filtering is processed using dynamic feature enhancement algorithm to highlight the waveform mode related to overheat and suppress irrelevant noise, the waveform features after processing are input into a pre-trained classification model, the overheat risk level of motor is determined by analyzing the waveform features, and the corresponding motor control operation is executed when the risk level reaches the preset threshold according to the overheat risk level to trigger load reduction operation or shutdown protection measures.The application solves the problem of strong invasiveness and insufficient accuracy of traditional overheat detection methods.

Description

Technical Field

[0001] This invention relates to the field of motor protection, and more specifically, to a method for overheat protection of a DC brushed motor. Background Technology

[0002] DC brushed motors are widely used in power tools, electric vehicles, industrial automation, and other fields due to their simple structure and convenient control. Their working principle involves switching the direction of current through the contact between the brushes and the commutator, thereby driving the rotor to rotate continuously. However, under high load, prolonged operation, or harsh environmental conditions, the friction between the brushes and the commutator inside the motor generates a large amount of heat, leading to motor overheating. Overheating not only reduces the motor's performance and lifespan but can also cause safety hazards such as insulation damage, short circuits, and even fires. Therefore, timely and accurate detection and prevention of motor overheating are crucial to ensuring the safe operation of equipment.

[0003] In existing technologies, motor overheat protection typically relies on methods such as temperature sensors or current monitoring. Temperature sensors, such as thermistors, are often installed inside the motor to directly measure temperature changes. However, this method has significant limitations. In space-constrained or high-temperature environments, sensor installation not only increases assembly difficulty but may also lead to failure due to mechanical vibration or high temperatures. Another common method, current monitoring, indirectly infers heating by detecting the motor's input current. However, this method cannot accurately reflect the motor's actual thermal state and is prone to misjudgment due to load fluctuations or electrical noise. Although these methods alleviate overheating problems to some extent, their shortcomings remain significant, especially the invasive installation and low detection accuracy, limiting their application in compact or complex environments. Therefore, a non-contact solution is urgently needed to improve the reliability of overheat protection. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for overheat protection of a DC brushed motor, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for overheat protection of a DC brushed motor includes the following steps:

[0007] An electromagnetic wave receiver placed outside the motor collects electromagnetic radiation signals generated by the brushes and commutator during motor operation in real time. The collected electromagnetic radiation signals are processed by multi-frequency band filtering to extract characteristic frequency components related to overheating.

[0008] The filtered signal is processed using a dynamic feature enhancement algorithm to highlight overheat-related waveform patterns and suppress irrelevant noise. The processed waveform features are then input into a pre-trained classification model.

[0009] The overheating risk level of the motor is determined by analyzing waveform characteristics. Based on the overheating risk level, load reduction or shutdown protection measures are triggered. When the risk level reaches the preset threshold, the corresponding motor control operation is executed.

[0010] In an optional embodiment, the electromagnetic wave receiver is a small broadband antenna installed outside the motor housing, which wirelessly captures the electromagnetic radiation signals generated when the motor is running.

[0011] The antenna operates in a frequency range of 5kHz to 150kHz. The acquired signals are converted into digital signals by an analog-to-digital converter and transmitted to the signal processing unit.

[0012] In an optional embodiment, the multi-frequency band filtering process uses a set of parallel bandpass filters to decompose the acquired electromagnetic radiation signal. The frequency range of the bandpass filters is set to 10kHz to 100kHz, decomposing the signal into three sub-bands: low frequency, mid frequency, and high frequency.

[0013] The low-frequency sub-band covers 10kHz to 30kHz, the mid-frequency sub-band covers 30kHz to 70kHz, and the high-frequency sub-band covers 70kHz to 100kHz.

[0014] Frequency feature vectors are generated by extracting amplitude and phase information from each sub-band. The formula for calculating the amplitude of the feature vector is as follows:

[0015] ;

[0016] in, The amplitude at frequency f. The real part of the signal, This represents the imaginary part of the signal.

[0017] In an optional embodiment, the dynamic feature enhancement algorithm employs wavelet transform technology to perform time-frequency analysis on the filtered signal. It decomposes the signal into multiple scales by selecting discrete wavelet basis functions, retaining low-order and high-order wavelet coefficients related to overheating. The algorithm then generates enhanced waveform feature signals through reconstruction processing and applies an amplitude threshold to the reconstructed signal to filter out signal components below a preset amplitude threshold. The wavelet transform decomposition formula is as follows: ;

[0018] in, These are wavelet coefficients. For input signal, Let be the conjugate wavelet basis function, s be the scaling parameter, t be the time shift parameter, and τ be the integral variable in the formula, which represents the time variable of the input signal x(τ).

[0019] In an optional embodiment, the pre-trained classification model is a one-dimensional convolutional neural network, which is deployed and runs in an edge computing module;

[0020] The network structure consists of an input layer, multiple convolutional layers, pooling layers, and a fully connected layer connected sequentially. The input layer receives the enhanced waveform feature signal. The convolutional layers extract the time-series features of the waveform using one-dimensional convolutional kernels. The pooling layers perform dimensionality reduction on the convolutional results. The fully connected layer outputs the classification result of the motor overheating risk, which includes three levels: normal, medium risk, and high risk. The formula for calculating the output feature of the convolutional layer is as follows: ;

[0021] in, For the i-th output feature, For input signal, is the kernel weight, b is the bias, and k is the kernel length.

[0022] In an optional embodiment, the protection measures include load reduction operation and shutdown protection. When the overheating risk level output by the classification model is medium risk, the control unit sends a command to the motor drive circuit to reduce the input current or speed of the motor. When the overheating risk level output by the classification model is high risk, the control unit immediately cuts off the motor power supply to stop the motor from running.

[0023] In an optional embodiment, the method further includes an online learning mechanism, which incrementally updates the pre-trained classification model using real-time collected electromagnetic radiation waveform data. During motor operation, the newly collected waveform feature signals are compared with the output results of the classification model. If a classification deviation is detected, the new data is added to the training set, and the model weights are updated to adapt to changes in the motor's operating state. The update process is performed once every fixed time interval, such as 30 seconds.

[0024] In an optional embodiment, the online learning mechanism employs transfer learning technology, using the pre-trained classification model weights as initial parameters and fine-tuning the model using real-time collected waveform data. During the fine-tuning process, the stochastic gradient descent algorithm is used to optimize the loss function, which is based on the cross-entropy between the classification model output and the actual overheating state. For example, the optimization step size can be set to 0.001.

[0025] In an optional embodiment, the method further includes a multi-source data fusion step, in which ambient temperature and motor vibration data are collected by an auxiliary sensor integrated into the motor system. The temperature sensor is installed on the surface of the motor housing, and the vibration sensor is installed at the bearing position of the motor. The collected temperature data and vibration data are synchronized with the electromagnetic radiation waveform characteristic signal in time. The three sets of data are input into the data fusion unit, and a comprehensive overheating risk index is calculated by a weighted average algorithm. For example, the weights can be 0.5 for electromagnetic radiation waveform characteristics, 0.3 for temperature data, and 0.2 for vibration data.

[0026] In an optional embodiment, the multi-source data fusion adopts a Bayesian network model, using electromagnetic radiation waveform features, temperature data, and vibration data as input nodes. The posterior probability of overheating is calculated through a pre-constructed conditional probability table. The output of the Bayesian network is a comprehensive overheating risk probability value. A multi-level early warning mechanism is triggered based on the probability value. For example, a low-level early warning is issued when the probability value exceeds 50%, and a motor shutdown protection is triggered when the probability value exceeds 80%. The early warning signal is transmitted to the motor management system through the control unit.

[0027] This invention is also applicable to motors with more pole pairs and slots, such as four-pole, six-slot brushed motors. Four poles indicate four magnetic poles, resulting in more frequent magnetic field changes compared to common two-pole motors. Six slots mean that each pole has multiple slot positions for winding distribution, thus making the contact changes of the brushes at different slot positions more complex. The commutator must handle more current switching events, resulting in more information in the electromagnetic radiation signal during commutation, leading to more diverse electromagnetic radiation signals with higher spectral characteristics. Therefore, this invention, through multi-frequency band filtering and dynamic feature enhancement processing of the electromagnetic radiation signal, can more accurately extract characteristic frequency components related to motor overheating, ensuring accurate assessment of overheating risks in motors such as four-pole, six-slot brushed motors.

[0028] In practical applications, four-pole, six-slot brushed motors are typically controlled by drive circuits composed of MOSFETs or IGBTs. However, the power semiconductor devices (such as MOSFETs, IGBT chips, and modules) in the drive circuit also generate heat during high-frequency switching, affecting system stability. Therefore, in optional embodiments, the technical principles of this invention can be further extended to power electronic devices in motor drive systems, such as MOSFET and IGBT components. These devices not only generate characteristic electromagnetic radiation signals during high-frequency switching but may also cause abnormal electromagnetic fluctuations due to switching losses, parasitic parameter effects, and changes in the operating environment. These abnormal signals are often early signs of increased heat loss, increased breakdown risk, or device aging. If not detected in time, they may lead to system failure or even damage. The method of this invention, through further extension, can also be used to analyze the electromagnetic radiation characteristics of these devices, thereby indirectly assessing the health status of the drive circuit without adding additional sensors, providing a more comprehensive system monitoring capability.

[0029] Therefore, the advantages of this invention over existing technologies lie in its ability to accurately determine the overheating state of a motor by collecting and analyzing the electromagnetic radiation signals generated during motor operation. This core innovation effectively solves the problems of insufficient detection accuracy and the need for invasive installation in traditional methods. Its advantages include eliminating the need for any components inside the motor, avoiding damage to the equipment from high temperatures or mechanical stress, and directly reflecting the contact state between the brushes and commutator through waveform analysis, significantly improving detection accuracy and response speed. This non-contact method also has strong environmental adaptability and can be widely applied to different types of motors. Furthermore, this invention further enhances the protection effect through several auxiliary technologies. Multi-frequency band filtering and dynamic feature enhancement technologies improve signal quality, ensuring accurate extraction of overheating characteristics even under noise interference. An online learning mechanism enables the system to dynamically adjust according to the specific operating characteristics of the motor, reducing false alarms and improving long-term reliability. In addition, the application of multi-source data fusion and Bayesian networks integrates multiple information sources, optimizing the risk assessment process. These improvements collectively construct an intelligent protection system that not only prevents damage caused by overheating in a timely manner but also extends motor life, reduces maintenance costs, and provides significant technological advancements for related fields. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the overall method of the present invention;

[0031] Figure 2 This is a schematic diagram of the multi-frequency band filtering process of the present invention;

[0032] Figure 3 This is a schematic diagram of the dynamic feature enhancement algorithm of the present invention;

[0033] Figure 4 This is a schematic diagram illustrating the online learning process of this invention;

[0034] Figure 5 This is a schematic diagram of the multi-source data fusion mechanism of the present invention. Detailed Implementation

[0035] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0036] This invention provides an overheat protection method for a four-pole, six-slot DC brushed motor. Its core is to collect the electromagnetic radiation waveform generated during motor operation, combine it with the algorithm to analyze the waveform characteristics, determine whether the motor is in an overheating state, and trigger corresponding protection measures based on the judgment result.

[0037] The method proposed in this invention stems from the shortcomings of traditional overheat protection technologies in certain scenarios. For example, relying on internal temperature sensors or current monitoring may fail or be difficult to implement in space-constrained, high-temperature environments, or situations requiring high reliability. Therefore, this invention employs a non-contact detection method, indirectly determining overheating conditions using electromagnetic radiation signals from outside the motor. This not only avoids direct intervention in the motor's internal structure but also improves the system's applicability and durability. This method, combining external signal acquisition with intelligent analysis, provides an innovative solution for motor overheat protection.

[0038] More specifically, this invention is based on the fact that during motor operation, the contact between the brushes and the commutator generates friction and electrical sparks, the intensity and frequency of which change with motor temperature. As temperature rises, the contact between the brushes and the commutator may deteriorate due to thermal expansion or lubricant failure, leading to an increase in the frequency and intensity of spark discharges. These spark discharges generate specific electromagnetic radiation signals, the frequency and amplitude of which are closely related to temperature conditions. Therefore, by collecting these signals, the temperature changes of the motor can be indirectly inferred. The paper "Electromagnetic emissions with DC motors – maxon Support" mentions that the electric arc in the brush system is a major source of electromagnetic interference, and the arc intensity may change with increasing temperature. This provides a theoretical basis for using electromagnetic signals to monitor temperature.

[0039] To ensure that the collected electromagnetic radiation signals accurately reflect the motor temperature, advanced signal processing techniques can be employed. For example, multi-frequency band filtering can be used to extract characteristic frequency components related to spark discharge and temperature changes. Then, dynamic feature enhancement algorithms (such as wavelet transform) can be used to highlight temperature-related waveform patterns while suppressing irrelevant noise (such as environmental noise). Finally, a pre-trained classification model (such as a one-dimensional convolutional neural network) can be combined to analyze the signal features and accurately determine the motor's temperature status. These techniques can significantly improve the accuracy of temperature monitoring.

[0040] Furthermore, by conducting experiments on the motor under different temperature conditions, recording the changes in electromagnetic radiation signals, and analyzing their correlation with temperature, a mapping model between signal characteristics and temperature can be established. In practical applications, the system can accurately infer the motor temperature based on the real-time acquired signals and this model.

[0041] More specifically, as shown in Figure 1, in practice, an electromagnetic wave receiver placed outside the motor is used to collect the electromagnetic radiation signals generated by the interaction between the brushes and the commutator during motor operation in real time. The electromagnetic wave receiver uses a small broadband antenna, installed outside the motor housing, to wirelessly capture the electromagnetic radiation. The operating frequency range of this antenna is set from 5kHz to 150kHz. This range is selected based on extensive experimental analysis of the motor's operating characteristics, covering the typical electromagnetic radiation frequency band generated when the brushes contact the commutator.

[0042] The acquired analog signals are converted into digital signals by an analog-to-digital converter and then transmitted to the signal processing unit for further processing. The design of the small, wideband antenna ensures that it can still capture signals under non-contact conditions without interfering with the normal operation of the motor. In practical applications, the antenna is typically placed near the brushes and commutator area to ensure signal strength, while the electromagnetic shielding effect of the housing material must be considered to avoid signal attenuation.

[0043] The advantage of broadband antennas lies in their ability to capture a variety of electromagnetic components, ranging from low-frequency mechanical vibrations to high-frequency spark discharges, providing comprehensive data support for subsequent feature extraction. For example, low-frequency signals may dominate when the motor is running under light load, while high-frequency spark signals will increase significantly under overload or overheating conditions. This diversity lays the foundation for overheat detection.

[0044] Next, the collected electromagnetic radiation signals need to undergo multi-frequency band filtering to extract characteristic frequency components closely related to overheating.

[0045] As shown in Figure 2, this processing employs a set of parallel bandpass filters to decompose the signal into multiple sub-bands, with a frequency range set from 10kHz to 100kHz. Specifically, this is divided into a low-frequency sub-band (10kHz to 30kHz), a mid-frequency sub-band (30kHz to 70kHz), and a high-frequency sub-band (70kHz to 100kHz). By analyzing the amplitude and phase information in each sub-band, a frequency feature vector is generated, and its amplitude is calculated using the following formula:

[0046] ;

[0047] in, The amplitude at frequency f. The real part of the signal, This represents the imaginary part of the signal.

[0048] The motivation behind this decomposition method is that when a motor overheats, the contact state between the brushes and the commutator changes, resulting in increased sparking or noise, which is particularly noticeable within a specific frequency range. By splitting the signal into different sub-bands, these characteristics can be captured more precisely.

[0049] For example, the low-frequency subband may reflect the cumulative effect of mechanical vibration, the mid-frequency subband may correlate with changes in electrical characteristics as heat slowly increases, while the high-frequency subband is more likely to capture rapid discharge phenomena caused by overheating. In practice, the bandwidth and frequency range of the filter can be fine-tuned according to the motor model, load conditions, or operating environment to improve the specificity and accuracy of the detection. This flexibility ensures the adaptability of the method in different application scenarios.

[0050] Subsequently, the filtered signal is further processed by a dynamic feature enhancement algorithm to highlight waveform patterns related to overheating while suppressing irrelevant background noise.

[0051] As shown in Figure 3, this algorithm employs wavelet transform technology to perform time-frequency analysis on the signal. It achieves multi-scale decomposition through discrete wavelet basis functions, preserving low-order and high-order wavelet coefficients related to overheating. The formula is:

[0052] ;

[0053] in, These are wavelet coefficients. For input signal, is the conjugate wavelet basis function, s is the scaling parameter, and t is the time shift parameter.

[0054] After decomposition, an enhanced waveform feature signal is generated through reconstruction. An amplitude threshold is then applied to the reconstructed signal to filter out components below a preset threshold. The advantage of wavelet transform lies in its ability to simultaneously analyze the temporal and frequency characteristics of a signal, making it particularly suitable for processing non-stationary signals such as electromagnetic radiation from motors. For example, when a motor generates a sudden spark discharge due to overheating, high-frequency abrupt changes occur in the signal; wavelet transform can effectively capture these features. Using the Daubechies wavelet as the basis function allows for better extraction of such abrupt changes or high-frequency components. Simultaneously, thresholding removes irrelevant noise, improving the signal-to-noise ratio. This processing not only improves the accuracy of feature extraction but also lays a solid foundation for subsequent classification.

[0055] The processed waveform feature signal is input into a pre-trained classification model to analyze and determine the overheating risk level of the motor. This model employs a one-dimensional convolutional neural network, deployed in an edge computing module. Its network structure includes an input layer, multiple convolutional layers, pooling layers, and a fully connected layer connected sequentially. The input layer receives the enhanced waveform feature signal; the convolutional layers extract time-series features using one-dimensional convolutional kernels; the pooling layers reduce the dimensionality of the convolutional results; and the fully connected layer finally outputs the overheating risk classification result, categorized into three levels: normal, medium risk, and high risk. The formula for calculating the output features of the convolutional layers is as follows:

[0056] ;

[0057] in, For the i-th output feature, For input signal, is the kernel weight, b is the bias, and k is the kernel length.

[0058] The initial intention behind this design was to leverage deep learning to automatically learn the complex relationship between waveform features and overheating conditions, offering higher accuracy and robustness compared to traditional threshold-based judgments. Deploying the model on an edge computing module enables real-time analysis and rapid response, avoiding the latency associated with data transmission to the cloud. For example, when a motor operates under high load for extended periods, the model can identify gradually increasing risk trends through training on historical waveform samples. Model training requires waveform data covering a variety of operating conditions to ensure its generalization ability across different scenarios.

[0059] Based on the overheating risk level output by the classification model, the system triggers corresponding protective measures. When the classification model determines that the motor is at a medium risk, the system sends a command to the motor drive circuit through the control unit to reduce the motor's input current or speed. This measure relies on the adjustability of the motor drive circuit, for example, by adjusting the input voltage using pulse width modulation (PWM) technology to reduce current input. Reducing the current or speed effectively reduces heat generation in the motor, allowing it to cool down gradually without overheating.

[0060] For example, in industrial production lines, motors may need to run for extended periods. In such cases, reducing the current by 20% or the speed by 15% can alleviate overheating without completely halting production. The specific reduction range can be preset based on the motor's rated power and the current load conditions, and optimized through experiments in practical applications.

[0061] When the risk level rises to high, the system immediately cuts off the motor power, triggering a shutdown protection mechanism. This is achieved through a hardware interrupt mechanism, where the control unit directly sends a shutdown signal to the power relay, ensuring a response time within milliseconds, such as less than 10 milliseconds, to prevent insulation damage or fire risks caused by overheating. The motivation behind this tiered protection strategy is to balance operational continuity with safety, avoid unnecessary downtime, and take decisive action in high-risk situations. In electric vehicle applications, this mechanism can protect critical components and extend their service life when the motor is overloaded.

[0062] As shown in Figure 4, an online learning mechanism can be further employed to update the classification model using real-time collected electromagnetic radiation waveform data, adapting to changes in specific motors and operating environments. The initial classification model is pre-trained on a general dataset, but due to individual differences in motors (such as manufacturing tolerances or aging) and changes in external conditions (such as ambient temperature or load fluctuations), the model may gradually deviate from the actual state. The online learning process involves periodically collecting electromagnetic radiation signals during motor operation and comparing them with the actual overheating state.

[0063] For example, the system collects data every 30 seconds, inputs waveform features into the classification model, and calculates the bias based on the detection results (whether it is truly overheating). If a prediction error is found, new data is added to the training set, and the model parameters are adjusted. This process is accomplished through incremental learning, avoiding retraining on all data and reducing the computational burden. In embedded systems, the update frequency and data volume need to be adjusted according to processor performance; for example, on a medium-performance microcontroller, the number of updates per day can be limited to less than 50. The advantage of this mechanism is that it allows the system to dynamically adapt to the individual characteristics of the motor. For example, a motor that has been used for many years may exhibit different electromagnetic radiation patterns due to wear. Online learning can capture these changes and optimize the model, thereby improving detection accuracy and reducing false alarms or false negatives.

[0064] In a further embodiment, online learning employs transfer learning and stochastic gradient descent (SGD) algorithms. The core idea of ​​transfer learning is to use the general knowledge of a pre-trained model as a starting point, then fine-tune it using real-time data. The pre-trained model may be trained based on a large amount of electromagnetic radiation data from standard motors, containing general features related to overheating. In practical applications, the system inputs real-time acquired waveform features into the model, calculates the loss between the predicted result and the actual state (such as cross-entropy loss), and updates the model weights using SGD. The advantage of SGD lies in its incremental update characteristic, using only a small amount of new data (e.g., a batch containing 10 samples) to calculate the gradient each time, making it suitable for scenarios with high real-time requirements. The learning rate is set to 0.001; this small value ensures smooth weight adjustment and avoids model non-convergence due to excessively large step sizes.

[0065] For example, on a newly installed motor, the system can quickly adapt to its characteristics through transfer learning within the first few hours of operation, significantly reducing misjudgments in the initial stage. Compared to training from scratch, transfer learning requires less data, typically only a few hundred samples to achieve high accuracy. This method is particularly suitable for scenarios involving motor replacement or changes in the operating environment, such as mobile devices operating under different climatic conditions, enabling rapid model adjustments to maintain reliability.

[0066] As shown in Figure 5, in a further embodiment, a multi-source data fusion mechanism can be employed. Specifically, this involves integrating electromagnetic radiation waveform characteristics with temperature and vibration data. A temperature sensor is mounted on the motor housing to monitor the surface temperature in real time; a vibration sensor is placed at the bearing to detect mechanical vibrations during operation. These two types of data are synchronized in time with the electromagnetic radiation signal, forming a multimodal dataset.

[0067] Data fusion can employ a weighted average algorithm, with the weight of each data source determined based on its correlation with overheating. For example, the weight of electromagnetic radiation waveform features is set to 0.5, temperature data to 0.3, and vibration data to 0.2. The data fusion process involves standardizing the signals from each sensor (e.g., normalizing to the 0-1 range) and then calculating a weighted sum as a comprehensive risk indicator. In practical applications, if the temperature suddenly rises but the electromagnetic radiation signal remains normal, the system may determine that it is due to external environmental influences rather than motor overheating, thus avoiding misjudgment. The motivation for multi-source fusion is that a single signal may fail due to noise or anomalies, while multimodal data can mutually verify each other, improving robustness. For example, during heavy-load operation, vibration data may indicate bearing problems, while temperature and electromagnetic radiation signals together confirm the overheating risk, thereby triggering protective measures. The specific values ​​of the weights can be optimized through historical data analysis or machine learning to ensure the optimal fusion result.

[0068] In a further embodiment, data fusion can employ Bayesian networks and be combined with a multi-level early warning mechanism for risk management. A Bayesian network is a probabilistic graphical model containing nodes such as electromagnetic radiation characteristics, temperature, vibration, and overheating state. The conditional dependencies between nodes are defined by a preset conditional probability table (CPT). For example, a CPT might indicate that when the electromagnetic radiation signal is abnormal and the temperature is above 80°C, the probability of overheating is 90%. The system inputs real-time data into the network and calculates the posterior probability of the overheating state using a probability propagation algorithm. This probability value drives a multi-level early warning mechanism: if the probability exceeds 50%, a low-level warning is triggered, such as alerting the operator through flashing LEDs or a buzzer; if the probability exceeds 80%, a high-level warning is triggered and the system is immediately shut down.

[0069] Threshold settings are based on actual risk management needs, providing 50% early intervention opportunities and ensuring timely protection in high-risk situations (80%). In practical systems, the inference process of Bayesian networks can be completed within milliseconds, meeting real-time requirements. The construction of conditional probability tables can combine historical data and expert knowledge, for example, determining the correlation of variables by analyzing 1000 hours of operational data. Compared to simple threshold methods, Bayesian networks can quantify uncertainty and provide more accurate risk assessments. For example, in scenarios with high ambient temperatures, the network can avoid false alarms triggered by a single temperature increase through comprehensive analysis. The advantages of this method lie in its flexibility and reliability, making it suitable for motor protection under complex operating conditions.

[0070] This invention is also applicable to motors with more pole pairs and slots, such as four-pole, six-slot brushed motors. Four poles indicate four magnetic poles, resulting in more frequent magnetic field changes compared to common two-pole motors. Six slots mean that each pole has multiple slot positions for winding distribution, thus making the contact changes of the brushes at different slot positions more complex. The commutator must handle more current switching events, resulting in more information in the electromagnetic radiation signal during commutation, leading to more diverse electromagnetic radiation signals with higher spectral characteristics. Therefore, this invention, through multi-frequency band filtering and dynamic feature enhancement processing of the electromagnetic radiation signal, can more accurately extract characteristic frequency components related to motor overheating, ensuring accurate assessment of overheating risks in motors such as four-pole, six-slot brushed motors.

[0071] In summary, this invention achieves accurate monitoring and timely protection of overheating conditions in a four-pole, six-slot brushed DC motor through non-contact electromagnetic radiation signal acquisition, combined with multi-frequency band filtering, wavelet transform enhancement, and convolutional neural network classification. This method not only overcomes the limitations of traditional technologies but also provides innovative ideas for the field of motor protection, demonstrating significant application potential. Further implementation of graded protection measures, dynamic adaptation through online learning, comprehensive analysis through multi-source data fusion, and probabilistic decision-making using Bayesian networks collectively construct a flexible and reliable overheating protection system. Through load reduction and shutdown protection, the system can effectively cope with different risk levels; online learning and transfer learning ensure the model's adaptability to individual motor and environmental changes; multi-source fusion and Bayesian networks significantly improve the accuracy and robustness of detection. In practical applications, such as industrial motors or electric vehicle motors, this system can extend equipment life, reduce maintenance costs, and ensure operational safety, demonstrating significant practical value and widespread application potential.

[0072] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for overheat protection of a DC brushed motor, characterized in that, Includes the following steps: An electromagnetic wave receiver placed outside the motor collects electromagnetic radiation signals generated by the brushes and commutator during motor operation in real time. The collected electromagnetic radiation signals are processed by multi-frequency band filtering to extract characteristic frequency components related to overheating. The filtered signal is processed using a dynamic feature enhancement algorithm to highlight overheat-related waveform patterns and suppress irrelevant noise. The processed waveform features are then input into a pre-trained classification model. The overheating risk level of the motor is determined by analyzing waveform characteristics. Based on the overheating risk level, load reduction or shutdown protection measures are triggered. When the risk level reaches the preset threshold, the corresponding motor control operation is executed. The electromagnetic wave receiver is a small broadband antenna installed outside the motor housing, which wirelessly captures the electromagnetic radiation signals generated by the motor during operation. The antenna operates in a frequency range of 5kHz to 150kHz. The acquired signals are converted into digital signals by an analog-to-digital converter and transmitted to the signal processing unit.

2. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that: The multi-frequency band filtering process employs a set of parallel bandpass filters to decompose the acquired electromagnetic radiation signal. The frequency range of the bandpass filters is set to 10kHz to 100kHz, decomposing the signal into three sub-bands: low frequency, mid frequency, and high frequency. The low-frequency sub-band covers 10kHz to 30kHz, the mid-frequency sub-band covers 30kHz to 70kHz, and the high-frequency sub-band covers 70kHz to 100kHz. Frequency feature vectors are generated by extracting amplitude and phase information from each sub-band. The formula for calculating the amplitude of the feature vector is as follows: ； in, The amplitude at frequency f. The real part of the signal, This represents the imaginary part of the signal.

3. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that, The dynamic feature enhancement algorithm employs wavelet transform technology to perform time-frequency analysis on the filtered signal. It decomposes the signal into multiple scales by selecting discrete wavelet basis functions, retaining low-order and high-order wavelet coefficients related to overheating. The algorithm then reconstructs the enhanced waveform feature signal and applies an amplitude threshold to the reconstructed signal to filter out signal components below a preset amplitude threshold. The wavelet transform decomposition formula is as follows: ； in, These are wavelet coefficients. For input signal, Let be the conjugate wavelet basis function, s be the scaling parameter, t be the time shift parameter, and τ be the integral variable in the formula, which represents the time variable of the input signal x(τ).

4. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that, The pre-trained classification model is a one-dimensional convolutional neural network, which is deployed and runs in the edge computing module; The network structure consists of an input layer, multiple convolutional layers, pooling layers, and a fully connected layer connected sequentially. The input layer receives the enhanced waveform feature signal. The convolutional layers extract the time-series features of the waveform using one-dimensional convolutional kernels. The pooling layers perform dimensionality reduction on the convolutional results. The fully connected layer outputs the classification result of the motor overheating risk, which includes three levels: normal, medium risk, and high risk. The formula for calculating the output feature of the convolutional layer is as follows: ; in, For the i-th output feature, For input signal, is the kernel weight, b is the bias, and k is the kernel length.

5. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that, The protection measures include load reduction operation and shutdown protection. When the overheating risk level output by the classification model is medium risk, the control unit sends a command to the motor drive circuit to reduce the input current or speed of the motor. When the overheating risk level output by the classification model is high risk, the control unit immediately cuts off the motor power supply to stop the motor operation.

6. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that, The method also includes an online learning mechanism, which incrementally updates the pre-trained classification model using real-time collected electromagnetic radiation waveform data. During motor operation, the newly collected waveform feature signals are compared with the output of the classification model. If a classification deviation is detected, the new data is added to the training set, and the model weights are updated to adapt to the changes in the motor's operating state. The update process is performed once every fixed time interval.

7. The method for overheat protection of a DC brushed motor according to claim 6, characterized in that, The online learning mechanism employs transfer learning technology, using the pre-trained classification model weights as initial parameters and fine-tuning the model using real-time collected waveform data. During the fine-tuning process, the stochastic gradient descent algorithm is used to optimize the loss function, which is based on the cross-entropy between the classification model output and the actual overheating state.

8. The method for overheat protection of a DC brushed motor according to claim 1, characterized in that, The method also includes a multi-source data fusion step, in which ambient temperature and motor vibration data are collected by auxiliary sensors integrated into the motor system. The temperature sensor is installed on the surface of the motor housing, and the vibration sensor is installed at the bearing position of the motor. The collected temperature data and vibration data are synchronized with the electromagnetic radiation waveform characteristic signal in time. The three sets of data are input into the data fusion unit, and a comprehensive overheating risk index is calculated by a weighted average algorithm.

9. The method for overheat protection of a DC brushed motor according to claim 8, characterized in that, The multi-source data fusion adopts a Bayesian network model, taking electromagnetic radiation waveform characteristics, temperature data, and vibration data as input nodes. The posterior probability of overheating is calculated through a pre-constructed conditional probability table. The output of the Bayesian network is a comprehensive overheating risk probability value. Based on the probability value, a multi-level early warning mechanism is triggered, and the early warning signal is transmitted to the motor management system through the control unit.

Images