A low-frequency loudspeaker unit power regulation method based on voice coil temperature rise prediction

Abstract

The application discloses a low-frequency loudspeaker unit power regulation method based on voice coil temperature rise prediction, comprising: obtaining the instantaneous amplitude sequence of an audio input signal and extracting audio signal features; obtaining electrical signal features, voice coil DC resistance and voice coil temperature time series data through a voltage detection circuit and a current detection circuit; constructing a voice coil temperature prediction model to predict the voice coil temperature in the current sampling period in an audio device; obtaining the instantaneous value sequence of back electromotive force, extracting positive and negative half-cycle asymmetry indicators and waveform distortion rate indicators; constructing a voice coil local overheating risk assessment model to judge the local overheating risk grade of the voice coil; using a power regulation strategy to dynamically regulate the power of the low-frequency loudspeaker unit and performing strategy optimization. The application can significantly improve the accuracy and timeliness of voice coil overheating risk identification, effectively avoid the thermal damage problem of the low-frequency loudspeaker unit under large-stroke working conditions, and ensure the stable operation and sound quality performance of the audio device under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of thermal management technology for audio equipment, and in particular to a method for power control of low-frequency speaker units based on voice coil temperature rise prediction. Background Technology

[0002] With the widespread application of audio equipment in home theaters, car audio systems, and professional sound reinforcement systems, low-frequency speaker units have become a key component of audio systems due to their ability to provide powerful low-frequency response and an immersive listening experience. In practical use, low-frequency speaker units typically operate under high-power, low-frequency drive conditions, especially when playing low-frequency enhanced content such as movie special effects. The voice coil needs to withstand large currents and prolonged energy input, leading to a rapid rise in voice coil temperature. If the voice coil temperature continues to rise without effective control, it can easily trigger thermal runaway, causing aging of the voice coil insulation material, failure of the bonding structure, or even burnout, thus affecting the reliability and lifespan of the audio equipment. Current technologies for overheat protection of speaker units typically employ power limiting based on a fixed threshold or simple temperature estimation methods. For example, by setting a maximum power limit or simply estimating the temperature based on an electrical signal, the input signal is attenuated once a preset threshold is exceeded. However, these methods generally suffer from problems such as response lag, insufficient accuracy, and coarse control. On the one hand, fixed thresholds are difficult to adapt to the complex variations in different audio content, structural parameters of different audio devices, and different working environments. This can easily lead to premature protection resulting in decreased sound quality or delayed protection increasing the risk of damage. On the other hand, traditional methods typically only focus on overall temperature changes and fail to reflect localized overheating within the voice coil, especially under low-frequency, long-stroke conditions. Due to the voice coil's eccentric motion and nonlinear electromagnetic behavior, hot spots are more likely to form in localized areas, further exacerbating the failure risk. Furthermore, under low-frequency, high-signal driving conditions, there is a significant coupling effect between the voice coil's mechanical motion and electromagnetic characteristics. The back EMF signal not only contains temperature change information but also includes nonlinear distortion components caused by long-stroke displacement. This makes it difficult for traditional back EMF-based feature extraction methods to accurately distinguish between temperature rise and displacement effects, thus affecting the accuracy of overheating risk assessment. Simultaneously, different speaker units differ in their magnetic circuit structure, air gap size, and elastic support system. These structural parameters significantly influence the voice coil's heat dissipation capacity and thermal distribution characteristics, but current technologies rarely integrate structural parameters with operational data for analysis. Therefore, there is an urgent need for a technical solution that can accurately predict and assess the risk of voice coil temperature rise and local overheating, and on this basis, dynamically control the input power of the low-frequency speaker unit, so as to effectively reduce the risk of voice coil overheating while ensuring sound quality and improve the safety and stability of audio equipment. Summary of the Invention

[0003] To address the problems existing in the prior art, this invention provides a low-frequency speaker unit power control method based on voice coil temperature rise prediction, mainly comprising: The instantaneous amplitude sequence of the audio input signal is obtained through the analog-to-digital converter built into the audio device, and the audio signal features are extracted using the sliding time window method. By using voltage and current detection circuits and sampling rate identical to that of audio input signal, electrical signal characteristics are obtained, and pilot signal is superimposed to obtain voice coil DC resistance and voice coil temperature timing data. Based on the audio signal characteristics, electrical signal characteristics, and voice coil temperature time series data of the audio device, a voice coil temperature prediction model is constructed to predict the voice coil temperature in the audio device within the current sampling period and to determine the risk of thermal runaway of the voice coil in the current audio device. Based on the instantaneous voltage value, instantaneous current value, and voice coil DC resistance value, the instantaneous back electromotive force value sequence is obtained and filtered to extract the positive and negative half-cycle asymmetry index and waveform distortion rate index. Based on the positive and negative half-cycle asymmetry index, waveform distortion rate index, voice coil temperature data, and the structural parameters of the speaker in the corresponding audio device, a voice coil local overheating risk assessment model is constructed to determine the local overheating risk level of the voice coil in the current audio device. Based on the risk level of voice coil thermal runaway or local overheating, a power regulation strategy is used to dynamically control the power of the low-frequency speaker unit, and the strategy is optimized based on its effectiveness.

[0004] Furthermore, the step of obtaining the instantaneous amplitude sequence of the audio input signal through the analog-to-digital converter built into the audio device, and extracting audio signal features using a sliding time window method, includes: The audio input signal is continuously sampled at a preset sampling rate using the built-in analog-to-digital converter of the audio device to obtain the instantaneous amplitude sequence of the audio input signal. Based on the acquired instantaneous amplitude sequence of the audio input signal, a sliding time window method is used, with the time window length set to a preset time length and the window overlap rate set to a preset overlap ratio. The root mean square value of the audio signal is calculated within each window, and the peak factor sequence is obtained based on the ratio of the peak amplitude to the root mean square value within the same window. A fast Fourier transform is used to perform spectral analysis on the signal within the window to calculate the energy distribution of each frequency band, obtain the spectral centroid frequency value and the low-frequency energy ratio value, and calculate the signal duration value based on the cumulative calculation of the signal duration. The audio root mean square value sequence, peak factor sequence, spectral centroid frequency value, low-frequency energy ratio value, and signal duration value are standardized and combined into a feature vector to form the audio signal feature, and the audio signal feature is stored in the voice coil operation monitoring database.

[0005] Furthermore, the step of obtaining electrical signal characteristics through voltage and current detection circuits using the same sampling rate as the audio input signal, and superimposing pilot signals to obtain voice coil DC resistance and voice coil temperature timing data includes: Using the voltage and current detection circuits configured at the output of the audio equipment's power amplifier, the same sampling rate as the audio input signal is used to acquire the driving voltage and driving current signals, and the instantaneous power value at each sampling point is calculated to obtain the instantaneous voltage value sequence, the instantaneous current value sequence, and the instantaneous power value sequence. Using the same time window length as the audio characteristics, the instantaneous power value sequence is averaged using a sliding window to obtain the average power value sequence, which serves as the electrical signal feature. During the synchronous time period of acquiring the audio signal feature and the electrical signal feature, a pilot signal with a preset pilot frequency and constant amplitude is superimposed onto the audio signal. The voltage and current values ​​at the pilot signal frequency are synchronously acquired using the voltage and current detection circuits at the power amplifier output, and the voice coil DC resistance value at the pilot signal frequency is calculated. Based on the ratio of the voice coil DC resistance value to the pre-measured cold resistance value at room temperature, the temperature change is calculated using the resistance-temperature relationship to obtain the voice coil temperature time series data. The electrical signal feature and the voice coil temperature time series data are then stored in the voice coil operation monitoring database.

[0006] Furthermore, the step of constructing a voice coil temperature prediction model based on the audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data of the audio device, predicting the voice coil temperature in the audio device within the current sampling period, and determining the risk of thermal runaway of the voice coil in the current audio device includes: By using a voice coil operation monitoring database, historical audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data of audio devices are obtained. A long short-term memory network is used for model training to construct a voice coil temperature prediction model. Based on the audio signal characteristics and electrical signal characteristics acquired in real time within a preset sampling period, the voice coil temperature prediction model is used to predict the voice coil temperature in the audio device within the current sampling period and determine the temperature rise rate of the voice coil. If the current voice coil temperature in the audio device is greater than a preset temperature threshold or the temperature rise rate is greater than a preset rate threshold, it is determined that there is a risk of thermal runaway in the voice coil of the current audio device.

[0007] Furthermore, the step of obtaining the back electromotive force instantaneous value sequence based on the instantaneous voltage value, instantaneous current value, and voice coil DC resistance value, and performing filtering processing to extract the positive and negative half-cycle asymmetry index and waveform distortion rate index includes: If the voice coil in the current audio device does not pose a risk of thermal runaway, then the back electromotive force is calculated using the formula based on the instantaneous voltage value, instantaneous current value, and DC resistance of the voice coil. Obtain the instantaneous sequence of back electromotive force values, where, Let be the instantaneous value of the back electromotive force at time t. The voltage at the output terminal of the power amplifier is the instantaneous value at time t. The instantaneous value of the power amplifier output current at time t. Let be the DC resistance of the voice coil, and T be the absolute temperature of the voice coil at the current moment. Based on the pilot signal frequency, a digital notch filter is designed, with the notch center frequency set to the pilot signal frequency and the notch bandwidth set to a preset value. The instantaneous back EMF value sequence is input into the notch filter for filtering to obtain the de-piloted back EMF signal. Based on the de-piloted back EMF signal, a zero-crossing detection method is used to identify the boundary points between the positive and negative half-cycles of the waveform. The instantaneous values ​​within each positive half-cycle are integrated to obtain the waveform area value for the positive half-cycle. The area of ​​the waveform during the negative half-cycle is obtained by integrating the absolute values ​​of the instantaneous values ​​within each negative half-cycle. According to the formula for the asymmetry index of positive and negative half cycles The asymmetry index A of the positive and negative half-cycles is obtained; the fundamental amplitude is obtained by performing a fast Fourier transform on the instantaneous back electromotive force value sequence after windowing using a window function. With the amplitude of the third harmonic The waveform distortion rate index formula is adopted. The waveform distortion rate index D is obtained. Based on the instantaneous value sequence of back electromotive force, the instantaneous value sequence of current, and the DC resistance value of the voice coil, it is determined whether the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency speaker unit. The positive and negative half-cycle asymmetry index and the waveform distortion rate index are corrected. The corrected positive and negative half-cycle asymmetry index and waveform distortion rate index are stored in the voice coil operation monitoring database.

[0008] It also includes determining whether the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency speaker unit based on the instantaneous back electromotive force value sequence, the instantaneous current value sequence, and the voice coil DC resistance value; and performing feature correction processing on the positive and negative half-cycle asymmetry index and waveform distortion rate index, specifically including: Based on the instantaneous sequence of back electromotive force, the instantaneous sequence of current, and the DC resistance value of the voice coil, the electromagnetic coupling consistency index formula is used. Calculate the electromagnetic coupling consistency index K; if the electromagnetic coupling consistency index is greater than the preset index threshold, it is determined that the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency horn unit, and the correction coefficient is calculated based on the formula. Determine the correction factor ,in, Reference asymmetry index collected under small displacement conditions Compared with the benchmark distortion rate index The calculated K value under the baseline working condition, To preset the lower limit of the correction factor, The preset saturation parameters are used; based on the obtained correction coefficients, the asymmetry index of the positive and negative half cycles and the waveform distortion rate index are subjected to feature correction processing.

[0009] Furthermore, the step of constructing a voice coil local overheating risk assessment model based on the positive and negative half-cycle asymmetry index, waveform distortion rate index, voice coil temperature data, and the structural parameters of the speaker in the corresponding audio device, to determine the local overheating risk level of the voice coil in the current audio device, includes: By using a voice coil operation monitoring database, historical positive and negative half-cycle asymmetry indices, waveform distortion rate indices, voice coil temperature data, structural parameters of the corresponding speaker in the audio equipment, and voice coil local overheating risk level labels are obtained. A gradient boosting decision tree is used to train the model and construct a voice coil local overheating risk assessment model. The structural parameters include magnetic circuit structure dimensions, voice coil height, air gap width, and spider linear range values. The risk levels are high, medium, and low. By using product design drawings or product manuals, the magnetic circuit structure parameters of the speaker in the audio equipment are obtained. Combined with the real-time positive and negative half-cycle asymmetry indices, waveform distortion rate indices, and voice coil temperature data, the voice coil local overheating risk assessment model is used to determine the current local overheating risk level of the voice coil in the audio equipment.

[0010] Furthermore, the step of dynamically regulating the power of the low-frequency speaker unit using a power adjustment strategy based on the risk level of voice coil thermal runaway or local overheating, and optimizing the strategy based on its effectiveness, includes: If the voice coil in the current audio device has a high risk of thermal runaway or local overheating, the power of the low-frequency speaker unit is dynamically adjusted according to a preset power adjustment strategy. The power adjustment strategy includes amplitude compression processing of the low-frequency band in the audio signal, adjusting the input gain coefficient of the power amplifier driving the low-frequency speaker unit, and limiting the low-frequency dynamic range. After implementing the power adjustment strategy, the voice coil temperature change data, temperature rise rate, positive and negative half-cycle asymmetry index, and waveform distortion rate index are obtained to determine the level of thermal runaway and local overheating risk of the voice coil in the audio device after implementing the power adjustment strategy, and to evaluate the effectiveness of the power adjustment strategy. If the effectiveness does not meet expectations, the power adjustment strategy is optimized until the level of thermal runaway and local overheating risk of the voice coil in the audio device returns to a safe range. The optimization process of the power adjustment strategy includes adjusting the low-frequency power threshold parameter and improving the low-frequency audio amplitude compression function.

[0011] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention provides a low-frequency speaker unit power control method based on voice coil temperature rise prediction. By jointly analyzing audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data, this invention achieves early identification of voice coil temperature change trends and thermal runaway risks. Furthermore, by combining back electromotive force feature extraction and a large-stroke displacement influence correction mechanism, the acquired asymmetry and distortion rate indices can more accurately reflect the nonlinear characteristics caused by temperature rise, thereby improving the accuracy of local overheating risk assessment. Based on this, by introducing speaker structural parameters for comprehensive judgment, and dynamically controlling and adaptively optimizing the low-frequency speaker unit power according to the level of thermal runaway risk or local overheating risk, this method ensures safe voice coil operation while reducing the impact of excessive limiting on sound quality. Compared to traditional protection methods based on fixed thresholds, this method achieves more refined, reliable, and stable thermal management and power control effects. The method provided by this invention can accurately sense and predict the temperature and local overheating state of the voice coil of a low-frequency speaker unit. Based on the comprehensive analysis of audio signal characteristics, electrical signal characteristics and nonlinear distortion characteristics, it can achieve adaptive dynamic control of low-frequency power, significantly improve the accuracy and timeliness of voice coil overheating risk identification, effectively avoid thermal damage to the low-frequency speaker unit under large stroke conditions, and thus ensure the stable operation and sound quality performance of audio equipment under complex conditions. Attached Figure Description

[0012] Figure 1 This is a flowchart of a low-frequency speaker unit power control method based on voice coil temperature rise prediction according to the present invention. Figure 2 This is a schematic diagram of a low-frequency speaker unit power control method based on voice coil temperature rise prediction according to the present invention. Figure 3 This is another schematic diagram of a low-frequency speaker unit power control method based on voice coil temperature rise prediction according to the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0014] like Figure 1-3 This embodiment of a low-frequency speaker unit power control method based on voice coil temperature rise prediction may specifically include: Step S101: Obtain the instantaneous amplitude sequence of the audio input signal through the analog-to-digital converter built into the audio device, and extract the audio signal features using the sliding time window method.

[0015] The audio input signal is continuously sampled at a preset sampling rate using the analog-to-digital converter built into the audio device to obtain the instantaneous amplitude sequence of the audio input signal. Based on the acquired instantaneous amplitude sequence of the audio input signal, a sliding time window method is used, with the time window length set to a preset time length and the window overlap rate set to a preset overlap ratio. The root mean square (RMS) value of the audio signal is calculated within each window, and the peak factor sequence is obtained based on the ratio of the peak amplitude to the RMS value within the same window. A fast Fourier transform is used to perform spectral analysis on the signal within the window, calculating the energy distribution of each frequency band, obtaining the spectral centroid frequency value and the low-frequency energy proportion value, and accumulating the signal duration value to obtain the signal duration value. The audio RMS value sequence, peak factor sequence, spectral centroid frequency value, low-frequency energy proportion value, and signal duration value are standardized and combined into a feature vector to constitute the audio signal features, which are then stored in the voice coil operation monitoring database.

[0016] For example, during the operation of a low-frequency speaker device, the input audio signal is continuously sampled at a sampling rate of 48kHz using its built-in analog-to-digital converter. A sliding time window of 200ms with a window overlap rate of 50% is captured at a certain moment. The audio signal within this time period is analyzed. The root mean square value of the audio signal within this window is calculated to be 0.35, and the peak amplitude within this window is detected to be 0.70, corresponding to a peak factor of 2.0, indicating that the signal has certain dynamic impact characteristics. Fast Fourier Transform analysis of the signal within this window yields a spectral centroid frequency of 85Hz, indicating that the signal energy is mainly concentrated in the low-frequency range. Furthermore, the energy proportion below 100Hz is calculated to be 78%, indicating that this segment of the signal is a typical low-frequency dominant signal. Furthermore, by accumulating the duration of this type of signal, it was found that the current low-frequency high-energy signal has lasted continuously for 3.5 seconds, indicating the presence of significant low-frequency continuous excitation. Subsequently, the above-mentioned root mean square value of 0.35, peak factor of 2.0, spectral centroid of 85Hz, low-frequency energy ratio of 78%, and duration of 3.5 seconds were normalized according to preset standardization ranges, such as normalizing the amplitude to the 0 to 1 range and the frequency to the 0 to 500Hz range. Finally, a five-dimensional feature vector (0.35, 2.0, 0.17, 0.78, 0.35) was formed, where the spectral centroid of 85Hz corresponds to a normalized value of 0.17. Assuming the maximum reference time is 10 seconds, the duration of 3.5 seconds corresponds to a normalized value of 0.35. This feature vector was then stored as the audio signal feature at the current moment in the voice coil operation monitoring database.

[0017] In step S102, the voltage detection circuit and the current detection circuit use the same sampling rate as the audio input signal to obtain the electrical signal characteristics, and then superimpose the pilot signal to obtain the timing data of the voice coil DC resistance and voice coil temperature.

[0018] Using voltage and current detection circuits configured at the output of the audio equipment's power amplifier, and employing the same sampling rate as the audio input signal, the driving voltage and current signals are acquired. The instantaneous power value at each sampling point is calculated, resulting in a sequence of instantaneous voltage, current, and power values. A sliding window averaging process is applied to the instantaneous power value sequence using the same time window length as the audio characteristics, yielding an average power value sequence as the electrical signal feature. During the synchronous acquisition of the audio and electrical signal features, a pilot signal with a preset pilot frequency and constant amplitude is superimposed onto the audio signal. The voltage and current detection circuits at the power amplifier output synchronously acquire the voltage and current values ​​at the pilot signal frequency and calculate the voice coil DC resistance at that frequency. Based on the ratio of the voice coil DC resistance to a pre-measured cold-state resistance at room temperature, the temperature change is calculated using a resistance-temperature formula, obtaining the voice coil temperature time-series data. The electrical signal features and voice coil temperature time-series data are then stored in the voice coil operation monitoring database.

[0019] For example, the voltage and current detection circuits configured at the output of the audio device's power amplifier synchronously acquire the drive signal at the same 48kHz sampling rate as the audio input. Within a 200ms time window, the instantaneous voltage value changes within ±12V, and the instantaneous current value changes within ±2A. At a certain sampling point, when the voltage is 10V and the current is 1.5A, the corresponding instantaneous power is 15W. By statistically averaging the instantaneous power of all sampling points within this 200ms window, the average power of this window is obtained as 12W, which serves as the electrical signal characteristic at the current moment. Simultaneously, within this time window, a pilot signal with a frequency of 1kHz and a constant amplitude of 0.1 is superimposed onto the audio signal. The voltage and current components corresponding to this pilot frequency are extracted using the same voltage and current detection circuit. The effective voltage value at the pilot frequency is measured to be 2V, and the effective current value is measured to be 0.4A. Therefore, the equivalent DC resistance of the voice coil at this moment is calculated to be 5Ω. Based on the pre-measured cold-state resistance of the voice coil at room temperature (25℃) of 4Ω, and utilizing the linear relationship between resistance and temperature (e.g., temperature change is proportional to resistance change), and considering that the temperature coefficient of copper is approximately 0.004 / ℃, it can be calculated that a 25% increase in resistance corresponds to a rise in the voice coil temperature to approximately 85℃. This provides the voice coil temperature data for that moment. Finally, the average power of 12W, voltage and current variation characteristics, and the corresponding voice coil temperature of 85℃ obtained within this time window are compiled into a set of time-series data and stored in the voice coil operation monitoring database.

[0020] Step S103: Based on the audio signal characteristics, electrical signal characteristics and voice coil temperature time series data of the audio device, construct a voice coil temperature prediction model to predict the voice coil temperature in the audio device within the current sampling period and determine the risk of thermal runaway of the voice coil in the current audio device.

[0021] By utilizing a voice coil operation monitoring database, historical audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data of audio devices are acquired. A long short-term memory network is used for model training to construct a voice coil temperature prediction model. Based on the real-time acquired audio and electrical signal characteristics within a preset sampling period, the voice coil temperature prediction model is used to predict the voice coil temperature in the audio device within the current sampling period and determine the temperature rise rate. If the current voice coil temperature in the audio device exceeds a preset temperature threshold or the temperature rise rate exceeds a preset rate threshold, it is determined that there is a risk of thermal runaway in the voice coil of the current audio device.

[0022] For example, during the long-term operation of audio equipment, approximately 10,000 sets of historical data samples were accumulated and stored in the voice coil operation monitoring database. Each set of samples contains audio signal features within a corresponding 200ms time window, such as a root mean square value of 0.30, a peak factor of 1.8, and a low-frequency energy ratio of 75%, as well as electrical signal features, such as an average power of 10W, and the corresponding voice coil temperature, such as 70℃. During the model training phase, 80% of the data was selected as the training set and 20% as the validation set to construct a long short-term memory network model with two hidden layers. Each layer contains 64 neurons. The time series input length was set to 10 consecutive time windows, i.e., 2 seconds of historical data. The learning rate was set to 0.001, and the mean squared error was used as the loss function. A voice coil temperature prediction model was constructed. Through approximately 50 iterations of training, the model was able to learn the temporal relationship between audio features, electrical signal features, and voice coil temperature. After training, the voice coil temperature prediction model is deployed on an audio device for real-time prediction. If the audio features from the last two seconds are obtained (RMS 0.40, low-frequency energy percentage 80%), and the electrical signal features (average power 15W), and input into the voice coil temperature prediction model, the predicted voice coil temperature is 86℃. Based on the prediction results of adjacent time windows, the temperature rise rate is calculated to be 3℃ per second. When the preset temperature threshold is 120℃ and the preset temperature rise rate threshold is 2℃ per second, since the current temperature rise rate has exceeded the threshold, even though the temperature has not yet reached its limit, it can still be determined that the voice coil is in a rapid heating phase, posing a risk of thermal runaway.

[0023] Step S104: Based on the instantaneous voltage value, instantaneous current value, and voice coil DC resistance value, obtain the instantaneous back electromotive force value sequence and perform filtering processing to extract the positive and negative half-cycle asymmetry index and waveform distortion rate index.

[0024] If the voice coil in the current audio device does not pose a risk of thermal runaway, then the back electromotive force is calculated using the formula based on the instantaneous voltage value, instantaneous current value, and DC resistance of the voice coil. Obtain the instantaneous sequence of back electromotive force values, where, Let be the instantaneous value of the back electromotive force at time t. The voltage at the output terminal of the power amplifier is the instantaneous value at time t. The instantaneous value of the power amplifier output current at time t. Let be the DC resistance of the voice coil, and T be the absolute temperature of the voice coil at the current moment. Based on the pilot signal frequency, a digital notch filter is designed. The notch center frequency is set to the pilot signal frequency, and the notch bandwidth is set to a preset bandwidth value. The instantaneous back EMF value sequence is input into the notch filter for filtering to obtain the de-pilot back EMF signal. Based on the de-pilot back EMF signal, a zero-crossing detection method is used to identify the boundary points between the positive and negative half-cycles of the waveform. The instantaneous values ​​within each positive half-cycle are integrated to obtain the waveform area value for the positive half-cycle. The area of ​​the waveform during the negative half-cycle is obtained by integrating the absolute values ​​of the instantaneous values ​​within each negative half-cycle. According to the formula for the asymmetry index of positive and negative half cycles. The asymmetry index A of the positive and negative half-cycles is obtained. The fundamental amplitude is obtained by windowing the instantaneous back electromotive force sequence using a window function and then performing a Fast Fourier Transform. With the amplitude of the third harmonic The waveform distortion rate index formula is adopted. The waveform distortion rate index D is obtained. Based on the instantaneous values ​​of the back electromotive force, the instantaneous values ​​of the current, and the DC resistance value of the voice coil, it is determined whether the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency speaker unit. Feature correction processing is performed on the positive and negative half-cycle asymmetry index and the waveform distortion rate index, and the corrected positive and negative half-cycle asymmetry index and waveform distortion rate index are stored in the voice coil operation monitoring database.

[0025] For example, assuming an audio device is operating normally and no risk of voice coil thermal runaway is detected, at a certain sampling time t, the instantaneous voltage at the power amplifier output is 8V and the instantaneous current is 1.6A. Simultaneously, based on pilot measurements, the current DC resistance of the voice coil is 5Ω, corresponding to a voice coil temperature of approximately 80℃. Then, according to the back electromotive force calculation formula... We can obtain that the back electromotive force at that moment is 0V, where, Let be the instantaneous value of the back electromotive force at time t. The voltage at the output terminal of the power amplifier is the instantaneous value at time t. The instantaneous value of the power amplifier output current at time t. Let be the DC resistance of the voice coil, and T be the absolute temperature of the voice coil at the current moment. In subsequent continuous sampling, if the voltage is 9V and the current is 1.5A at the next moment, the back electromotive force (EMF) will be 1.5V, thus forming a complete instantaneous sequence of back EMF values. Since a 1kHz pilot signal is superimposed on the audio signal, a digital notch filter with a center frequency of 1kHz and a bandwidth of 50Hz is designed to filter this back EMF sequence to remove the pilot component, obtaining a de-piloted back EMF signal containing only actual motion information. A complete period is selected from this signal; for example, the area obtained by integrating the positive half-cycle within the period is... The area is 0.012. The area is obtained by integrating the absolute value of the negative half-cycle. If the value is 0.010, then according to the asymmetry formula... The calculated asymmetry index A for the positive and negative half-cycles is 0.167, indicating a certain degree of asymmetry in the waveform. Further spectral analysis of this periodic signal reveals that if the fundamental amplitude... The third harmonic amplitude is 2.0. If it is 0.4, then according to the distortion rate formula... The calculated waveform distortion rate index D is 0.2, indicating a certain degree of nonlinear distortion. Then, considering the current changes, it is determined whether the current back electromotive force (EMF) characteristics are affected by the large stroke displacement of the low-frequency speaker unit. If the peak current reaches 2A and the back EMF amplitude fluctuates significantly over time, it is determined that the current state is a low-frequency, large stroke state. Therefore, the positive and negative half-cycle asymmetry index and waveform distortion rate index are corrected, resulting in corrected asymmetry and distortion rate indices. These corrected features, along with the corresponding timestamps, are stored in the voice coil operation monitoring database.

[0026] Specifically, based on the instantaneous back EMF value sequence, the instantaneous current value sequence, and the voice coil DC resistance value, it is determined whether the current back EMF characteristics are affected by the large stroke displacement factor of the low-frequency speaker unit, and feature correction processing is performed on the positive and negative half-cycle asymmetry index and waveform distortion rate index.

[0027] Based on the instantaneous sequence of back electromotive force, the instantaneous sequence of current, and the DC resistance value of the voice coil, the electromagnetic coupling consistency index formula is used. Calculate the electromagnetic coupling consistency index K. If the electromagnetic coupling consistency index is greater than the preset index threshold, it is determined that the current back electromotive force characteristics are affected by the large-stroke displacement factor of the low-frequency horn unit, and the calculation is based on the correction coefficient formula. Determine the correction factor ,in, Reference asymmetry index collected under small displacement conditions Compared with the benchmark distortion rate index The calculated K value under the baseline working condition, To preset the lower limit of the correction factor, The preset saturation parameters are used. Based on the obtained correction coefficients, characteristic correction processing is performed on the positive and negative half-cycle asymmetry index and the waveform distortion rate index.

[0028] For example, during a certain sampling period, the instantaneous voltage at the power amplifier output is detected to be 11V and the instantaneous current is 2A. Simultaneously, pilot measurement shows the current voice coil DC resistance to be 4Ω, corresponding to a voice coil temperature of approximately 72℃. Therefore, according to the back electromotive force (EMF) calculation formula, the back EMF at that moment can be calculated to be 3V. Further calculation yields a back EMF / current ratio of 1.5Ω, which is then compared with the current voice coil DC resistance of 4Ω. Based on the electromagnetic coupling consistency index formula... The calculated value of K is 2.5. If the benchmark asymmetry index collected under small displacement conditions... Compared with the benchmark distortion rate index Reference under the reference working condition measured below If the value of K is 0.5 and the preset exponent threshold is 1.5, then the current K value is significantly greater than the threshold. Therefore, it is determined that the back electromotive force characteristic has been affected by the large-stroke displacement of the low-frequency horn unit. Thus, a lower limit for the correction coefficient is set. The saturation parameter is 0.8. If the value is 3, then the correction factor is calculated using the formula. We can obtain the correction coefficient. The value is approximately 0.67. Since this value is less than the lower limit of the correction coefficient of 0.8, the final correction coefficient is set to 0.8. Subsequently, this correction coefficient is applied to the positive and negative half-cycle asymmetry index of 0.20 and the waveform distortion rate index of 0.25 calculated in the current cycle, and corrected to 0.16 and 0.20 respectively. This effectively reduces the spurious nonlinear components caused by low-frequency large-stroke displacement while ensuring that the correction magnitude is not too large, making the corrected characteristics more stable and reliable. The corrected characteristic results are then stored in the voice coil operation monitoring database.

[0029] Step S105: Based on the positive and negative half-cycle asymmetry index, waveform distortion rate index, voice coil temperature data, and the structural parameters of the speaker in the corresponding audio device, construct a voice coil local overheating risk assessment model to determine the local overheating risk level of the voice coil in the current audio device.

[0030] By utilizing a voice coil operation monitoring database, historical positive and negative half-cycle asymmetry indices, waveform distortion rate indices, voice coil temperature data, corresponding speaker structural parameters in audio equipment, and voice coil local overheating risk level labels are obtained. A gradient boosting decision tree is used for model training to construct a voice coil local overheating risk assessment model. Structural parameters include magnetic circuit structure dimensions, voice coil height, air gap width, and spider linear range values. Risk levels are categorized as high, medium, and low. Using product design drawings or product manuals, the magnetic circuit structure parameters of the speaker in the audio equipment are obtained. Combined with real-time acquired positive and negative half-cycle asymmetry indices, waveform distortion rate indices, and voice coil temperature data, the voice coil local overheating risk assessment model is used to determine the current local overheating risk level of the voice coil in the audio equipment.

[0031] For example, approximately 10,000 sets of historical sample data have been accumulated through the voice coil operation monitoring database. Each set of data includes the positive and negative half-cycle asymmetry index extracted within a 200ms time window, with typical values ​​between 0.05 and 0.30; waveform distortion rate index, such as between 0.08 and 0.35; corresponding voice coil temperature, such as 60℃ to 130℃; and structural parameter information of the speaker, such as the effective magnetic gap height of the magnetic circuit structure being 6mm, the voice coil height being 8mm, the air gap width being 0.5mm, and the spider linearity range being ±4mm. In addition, each set of data also corresponds to manually marked local overheating risks. The risk level is categorized as follows: a temperature exceeding 110℃ and a distortion rate greater than 0.25 indicates a high risk, signifying that the voice coil is in a state of significant nonlinearity and heat accumulation, posing a high risk of burnout. A voice coil temperature between 90℃ and 110℃ and a waveform distortion rate between 0.15 and 0.25 indicates a medium risk, signifying that the voice coil has entered a stage of high temperature rise and gradually increasing nonlinearity, requiring attention but not yet reaching a dangerous state. A voice coil temperature below 90℃ and a waveform distortion rate less than 0.15 indicates a low risk, signifying that the voice coil is within its normal operating range, with stable thermal conditions and low nonlinearity. During model training, 70% of the data is used as the training set and 30% as the validation set. A gradient boosting decision tree model is used for training, with 120 trees, a maximum tree depth of 5, and a learning rate of 0.05. Classification error is used as the optimization objective to construct a voice coil local overheating risk assessment model. Through iterative training, the model learns the correspondence between asymmetry, distortion rate, temperature, and structural parameters with the risk level. After the voice coil local overheating risk assessment model is deployed, when the positive and negative half-cycle asymmetry index is detected to be 0.22, the waveform distortion rate is 0.28, and the voice coil temperature is 105℃ at a certain moment, and combined with the structural parameters of the device, such as the air gap width of 0.5mm leading to weak heat dissipation, the data is input into the trained voice coil local overheating risk assessment model for local overheating risk assessment. The model outputs that the current local overheating risk level is high, thus determining that the voice coil of the audio device is in a potential local overheating state, and further power regulation measures need to be taken to avoid damage.

[0032] Step S106: Based on the risk level of voice coil thermal runaway or local overheating, a power adjustment strategy is used to dynamically regulate the power of the low-frequency speaker unit, and the strategy is optimized based on its effectiveness.

[0033] If the voice coil in the current audio device has a high risk of thermal runaway or local overheating, the power of the low-frequency speaker unit is dynamically adjusted according to a preset power adjustment strategy. This strategy includes amplitude compression of the low-frequency band in the audio signal, adjustment of the input gain coefficient of the power amplifier driving the low-frequency speaker unit, and limitation of the low-frequency dynamic range. Data on voice coil temperature changes, temperature rise rate, positive and negative half-cycle asymmetry, and waveform distortion rate are acquired after implementing the power adjustment strategy to determine the risk levels of thermal runaway and local overheating in the audio device, and to evaluate the effectiveness of the power adjustment strategy. If the effectiveness is not as expected, the power adjustment strategy is optimized until the risk levels of thermal runaway and local overheating in the audio device return to a safe range. The optimization process includes adjusting the low-frequency power threshold parameter and improving the low-frequency audio amplitude compression function.

[0034] For example, if it is determined that the voice coil in an audio device has a high risk of thermal runaway or local overheating, the power of the low-frequency speaker unit is dynamically adjusted according to a preset power adjustment strategy. This includes compressing the amplitude of low-frequency signals, such as those in the 20Hz to 120Hz range, to 70% of their original value, reducing the input gain of the amplifier driving the low-frequency speaker unit from 1.0 to 0.8, and limiting the low-frequency dynamic range. This smooths out the low-frequency impact signals that originally had large peak values, thereby reducing the instantaneous power input. If, during the operation of an audio device, the current voice coil temperature is detected to be 108℃, the temperature rise rate is 1.2℃ per second, the waveform distortion rate is 0.28, and the positive and negative half-cycle asymmetry index is 0.23, although the current voice coil temperature has not exceeded the preset thermal runaway temperature threshold and the temperature rise rate is also lower than the preset rate threshold of 2℃, according to the voice coil local overheating risk assessment model, the local overheating risk level of the voice coil is high. After implementing this power adjustment strategy for about 2 seconds, the voice coil operating data is collected again, and the voice coil temperature drops to 102℃, the temperature rise rate drops to 0.8℃ per second, the waveform distortion rate drops to 0.22, and the asymmetry index drops to 0.19. At this time, the local overheating risk level drops from high risk to medium risk, indicating that the control strategy is initially effective, but it has not yet entered the safe range. Since the temperature was still close to the high-risk threshold, the power adjustment strategy was further optimized. For example, the low-frequency amplitude compression ratio was adjusted from 70% to 60%, and the compression function was improved to make the compression process smoother and avoid sudden changes in sound quality. At the same time, the low-frequency power threshold was lowered by 10%. After further adjustments, the voice coil temperature further decreased to 92℃, the temperature rise rate decreased to 0.3℃ per second, the distortion rate decreased to 0.19, and the asymmetry index decreased to 0.17. Finally, the risk level was restored to the low-risk range, thus achieving effective control of the overheating state of the low-frequency speaker unit's voice coil.

[0035] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the concept of this application. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for power control of a low-frequency speaker unit based on voice coil temperature rise prediction, characterized in that, The method includes: The instantaneous amplitude sequence of the audio input signal is obtained through the analog-to-digital converter built into the audio device, and the audio signal features are extracted using the sliding time window method. By using voltage and current detection circuits and sampling rate identical to that of audio input signal, electrical signal characteristics are obtained, and pilot signal is superimposed to obtain voice coil DC resistance and voice coil temperature timing data. Based on the audio signal characteristics, electrical signal characteristics, and voice coil temperature time series data of the audio device, a voice coil temperature prediction model is constructed to predict the voice coil temperature in the audio device within the current sampling period and to determine the risk of thermal runaway of the voice coil in the current audio device. Based on the instantaneous voltage value, instantaneous current value, and voice coil DC resistance value, the instantaneous back electromotive force value sequence is obtained and filtered to extract the positive and negative half-cycle asymmetry index and waveform distortion rate index. Based on the positive and negative half-cycle asymmetry index, waveform distortion rate index, voice coil temperature data, and the structural parameters of the speaker in the corresponding audio device, a voice coil local overheating risk assessment model is constructed to determine the local overheating risk level of the voice coil in the current audio device. Based on the risk level of voice coil thermal runaway or local overheating, a power regulation strategy is used to dynamically control the power of the low-frequency speaker unit, and the strategy is optimized based on its effectiveness.

2. The method according to claim 1, wherein, The process involves acquiring the instantaneous amplitude sequence of the audio input signal through the analog-to-digital converter built into the audio device, and extracting audio signal features using a sliding time window method, including: The audio input signal is continuously sampled at a preset sampling rate using the built-in analog-to-digital converter of the audio device to obtain the instantaneous amplitude sequence of the audio input signal. Based on the acquired instantaneous amplitude sequence of the audio input signal, a sliding time window method is used, with the time window length set to a preset time length and the window overlap rate set to a preset overlap ratio. The root mean square value of the audio signal is calculated within each window, and the peak factor sequence is obtained based on the ratio of the peak amplitude to the root mean square value within the same window. A fast Fourier transform is used to perform spectral analysis on the signal within the window to calculate the energy distribution of each frequency band, obtain the spectral centroid frequency value and the low-frequency energy ratio value, and calculate the signal duration value based on the cumulative calculation of the signal duration. The audio root mean square value sequence, peak factor sequence, spectral centroid frequency value, low-frequency energy ratio value, and signal duration value are standardized and combined into a feature vector to form the audio signal feature, and the audio signal feature is stored in the voice coil operation monitoring database.

3. The method according to claim 1, wherein, The process involves using a voltage detection circuit and a current detection circuit, employing the same sampling rate as the audio input signal, to acquire electrical signal characteristics. A pilot signal is then superimposed to obtain timing data of the voice coil DC resistance and voice coil temperature. This includes: Using the voltage and current detection circuits configured at the output of the audio device's power amplifier, the same sampling rate as the audio input signal is used to acquire the driving voltage and driving current signals, and the instantaneous power value at each sampling point is calculated to obtain the instantaneous voltage value sequence, the instantaneous current value sequence, and the instantaneous power value sequence. Using the same time window length as the audio characteristics, the instantaneous power value sequence is averaged using a sliding window to obtain the average power value sequence, which serves as the electrical signal characteristic. During the synchronous time period of acquiring audio signal characteristics and electrical signal characteristics, a pilot signal with a preset pilot frequency and constant amplitude is superimposed on the audio signal. The voltage and current values ​​at the pilot signal frequency are synchronously acquired using the voltage and current detection circuits at the power amplifier output, and the DC resistance value of the voice coil at the pilot signal frequency is calculated. Based on the ratio of the DC resistance value of the voice coil to the pre-measured cold resistance value at room temperature, the temperature change is calculated using the resistance-temperature relationship to obtain the voice coil temperature time series data. The electrical signal characteristics and the voice coil temperature time series data are then stored in the voice coil operation monitoring database.

4. The method according to claim 1, wherein, The step involves constructing a voice coil temperature prediction model based on the audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data of the audio device. This model predicts the voice coil temperature in the audio device within the current sampling period and assesses the risk of thermal runaway in the voice coil of the current audio device. This includes: By using a voice coil operation monitoring database, historical audio signal characteristics, electrical signal characteristics, and voice coil temperature time-series data of audio devices are obtained. A long short-term memory network is used for model training to construct a voice coil temperature prediction model. Based on the audio signal characteristics and electrical signal characteristics acquired in real time within a preset sampling period, the voice coil temperature prediction model is used to predict the voice coil temperature in the audio device within the current sampling period and determine the temperature rise rate of the voice coil. If the current voice coil temperature in the audio device is greater than a preset temperature threshold or the temperature rise rate is greater than a preset rate threshold, it is determined that there is a risk of thermal runaway in the voice coil of the current audio device.

5. The method according to claim 1, wherein, The process involves obtaining a sequence of instantaneous back electromotive force values ​​based on instantaneous voltage, instantaneous current, and voice coil DC resistance, followed by filtering to extract positive and negative half-cycle asymmetry indices and waveform distortion rate indices, including: If the voice coil in the current audio device does not pose a risk of thermal runaway, then the back electromotive force is calculated using the formula based on the instantaneous voltage value, instantaneous current value, and DC resistance of the voice coil. Obtain the instantaneous sequence of back electromotive force values, where, Let be the instantaneous value of the back electromotive force at time t. The voltage at the output terminal of the power amplifier is the instantaneous value at time t. The instantaneous value of the power amplifier output current at time t. Let be the DC resistance of the voice coil, and T be the absolute temperature of the voice coil at the current moment. Based on the pilot signal frequency, a digital notch filter is designed, with the notch center frequency set to the pilot signal frequency and the notch bandwidth set to a preset value. The instantaneous back EMF value sequence is input into the notch filter for filtering to obtain the de-piloted back EMF signal. Based on the de-piloted back EMF signal, a zero-crossing detection method is used to identify the boundary points between the positive and negative half-cycles of the waveform. The instantaneous values ​​within each positive half-cycle are integrated to obtain the waveform area value for the positive half-cycle. The area of ​​the waveform during the negative half-cycle is obtained by integrating the absolute values ​​of the instantaneous values ​​within each negative half-cycle. According to the formula for the asymmetry index of positive and negative half cycles The asymmetry index A of the positive and negative half-cycles is obtained; the fundamental amplitude is obtained by performing a fast Fourier transform on the instantaneous back electromotive force value sequence after windowing using a window function. With the amplitude of the third harmonic The waveform distortion rate index formula is adopted. The waveform distortion rate index D is obtained. Based on the instantaneous value sequence of back electromotive force, the instantaneous value sequence of current, and the DC resistance value of the voice coil, it is determined whether the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency speaker unit. The positive and negative half-cycle asymmetry index and the waveform distortion rate index are corrected. The corrected positive and negative half-cycle asymmetry index and waveform distortion rate index are stored in the voice coil operation monitoring database.

6. The method according to claim 5, wherein, The process involves determining whether the current back electromotive force (EMF) characteristics are affected by the large stroke displacement of the low-frequency speaker unit based on the instantaneous back EMF sequence, instantaneous current sequence, and voice coil DC resistance value. Feature correction processing is then applied to the positive and negative half-cycle asymmetry indicators and waveform distortion rate indicators, including: Based on the instantaneous sequence of back electromotive force, the instantaneous sequence of current, and the DC resistance value of the voice coil, the electromagnetic coupling consistency index formula is used. Calculate the electromagnetic coupling consistency index K; if the electromagnetic coupling consistency index is greater than the preset index threshold, it is determined that the current back electromotive force characteristics are affected by the large stroke displacement factor of the low-frequency horn unit, and the correction coefficient is calculated based on the formula. Determine the correction factor ,in, Reference asymmetry index collected under small displacement conditions Compared with the benchmark distortion rate index The calculated K value under the baseline working condition, To preset the lower limit of the correction factor, The preset saturation parameters are used; based on the obtained correction coefficients, the asymmetry index of the positive and negative half cycles and the waveform distortion rate index are subjected to feature correction processing.

7. The method according to claim 1, wherein, The method involves constructing a voice coil local overheating risk assessment model based on positive and negative half-cycle asymmetry indices, waveform distortion rate indices, voice coil temperature data, and the structural parameters of the speaker in the corresponding audio device. This model determines the local overheating risk level of the voice coil in the current audio device, including: By using a voice coil operation monitoring database, historical positive and negative half-cycle asymmetry indices, waveform distortion rate indices, voice coil temperature data, structural parameters of the corresponding speaker in the audio equipment, and voice coil local overheating risk level labels are obtained. A gradient boosting decision tree is used to train the model and construct a voice coil local overheating risk assessment model. The structural parameters include magnetic circuit structure dimensions, voice coil height, air gap width, and spider linear range values. The risk levels are high, medium, and low. By using product design drawings or product manuals, the magnetic circuit structure parameters of the speaker in the audio equipment are obtained. Combined with the real-time positive and negative half-cycle asymmetry indices, waveform distortion rate indices, and voice coil temperature data, the voice coil local overheating risk assessment model is used to determine the current local overheating risk level of the voice coil in the audio equipment.

8. The method according to claim 1, wherein, The process involves dynamically adjusting the power of the low-frequency speaker unit using a power regulation strategy based on the risk level of voice coil thermal runaway or local overheating, and optimizing the strategy based on its effectiveness. This includes: If the voice coil in the current audio device has a high risk of thermal runaway or local overheating, the power of the low-frequency speaker unit is dynamically adjusted according to a preset power adjustment strategy. The power adjustment strategy includes amplitude compression processing of the low-frequency band in the audio signal, adjusting the input gain coefficient of the power amplifier driving the low-frequency speaker unit, and limiting the low-frequency dynamic range. After implementing the power adjustment strategy, the voice coil temperature change data, temperature rise rate, positive and negative half-cycle asymmetry index, and waveform distortion rate index are obtained to determine the risk level of thermal runaway and local overheating of the voice coil in the audio device after implementing the power adjustment strategy, and to evaluate the effectiveness of the power adjustment strategy. If the effectiveness does not meet expectations, the power regulation strategy will be optimized until the risk of thermal runaway and local overheating of the voice coil in the audio device is restored to a safe range. The optimization process of the power regulation strategy includes adjusting the low-frequency power threshold parameter and improving the low-frequency audio amplitude compression function.

Images