A heat dissipation performance optimization method and system for a Bluetooth speaker

Abstract

The present invention discloses a method and system for optimizing the heat dissipation performance of a Bluetooth speaker, which relates to the technical field of data analysis. The method includes the following steps: Step 1, set the data acquisition frequency, and collect the speaker status data and environmental factor data; Step 2, synchronize and preprocess the collected data, and perform feature extraction based on the preprocessed data; Step 3, combine thermodynamics and statistical modeling, construct a regression model based on the speaker status, environmental factors and historical temperature, and calculate the internal temperature of the speaker; Step 4, quantify the residual heat of the speaker, integrate the feature vector and the real temperature label to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation. The present invention can effectively improve the situation in the prior art that the applicability is poor due to the single active heat dissipation condition of the Bluetooth speaker.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and particularly to a method and system for optimizing the heat dissipation performance of a Bluetooth speaker. Background Art

[0002] As a portable device integrating audio playback and wireless connection functions, Bluetooth speakers are widely used in scenarios such as smart home and outdoor entertainment. During its operation, components such as the power amplifier module and the main control chip continuously generate heat, and the enclosed box structure and complex and changeable usage environment (such as high temperature and high humidity) are prone to heat accumulation. Excessive temperature inside the speaker will not only degrade the audio output quality and affect the sound quality performance, but also accelerate the aging of electronic components, shorten the service life of the device, and even cause failures in severe cases.

[0003] Currently, the heat dissipation design of most Bluetooth speakers mainly relies on passive heat dissipation methods, such as increasing heat dissipation holes and using heat-conducting materials. This method has low heat dissipation efficiency and is difficult to meet the heat dissipation requirements in high-power and long-time usage scenarios. For some Bluetooth speakers with active heat dissipation functions, the setting of their heat dissipation trigger conditions is often based on a single temperature threshold, without fully considering the working state of the speaker (such as the input voltage of the power amplifier and the duty cycle of the audio signal), environmental factors (such as the air flow rate and humidity), and the impact of historical heat accumulation on the thermal state, resulting in a lack of accuracy and dynamic adaptability in heat dissipation control. In addition, in terms of temperature calculation and prediction in the prior art, simple empirical formulas or fixed models are mostly used, which cannot effectively cope with complex and changeable actual usage scenarios, and are prone to misjudgment or delayed heat dissipation, reducing the heat dissipation performance and usage experience of the speaker. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for optimizing the heat dissipation performance of a Bluetooth speaker to solve the problems raised in the prior art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A method for optimizing the heat dissipation performance of a Bluetooth speaker, the method comprising the following steps:

[0006] Step 1, set the data acquisition frequency and acquire the speaker state data and environmental factor data;

[0007] Step 2, synchronize and preprocess the acquired data, and perform feature extraction based on the preprocessed data;

[0008] Step 3, combine thermodynamics and statistical modeling, construct a regression model based on the speaker state, environmental factors and historical temperature, and calculate the internal temperature of the speaker;

[0009] Step 4: Quantify the residual heat of the speaker, integrate the feature vectors and true temperature labels to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation.

[0010] In Step 1, set the data acquisition frequency according to the working characteristics and application scenarios of the speaker;

[0011] The specific method for collecting the speaker status data and environmental factor data is as follows:

[0012] Arrange sensors inside the speaker and on the main control board to collect the speaker status data, expressed as: [D1(t), D2(t), …, D m (t)]; where m is a positive integer representing the number of speaker statuses; t represents the data acquisition time, and D1(t), D2(t), …, D m (t) respectively represent the 1st, 2nd, …, mth speaker status data collected at time t;

[0013] The speaker status signals include but are not limited to the power amplifier input voltage, current, audio signal duty cycle, sound pressure level, etc.;

[0014] Obtain the environmental factor data corresponding to the location through the meteorological API call, expressed as: [E1(t), E2(t), …, E n (t)]; where n is a positive integer representing the number of environmental factors; E1(t), E2(t), …, E n (t) respectively represent the 1st, 2nd, …, nth environmental factor data obtained at time t;

[0015] The environmental factor signals include but are not limited to environmental temperature, air flow rate, humidity, etc.

[0016] In Step 2, the specific method for synchronizing and preprocessing the collected data is as follows:

[0017] Set the sliding window length w and the step size s;

[0018] Apply median filtering with a window length of w to the collected speaker status data and environmental factor data to eliminate short-term spikes and jitters;

[0019] Perform linear interpolation on the speaker status data and environmental factor data based on the same time reference to synchronize the data;

[0020] The specific method for feature extraction based on the preprocessed data is as follows: within a sliding window with a length of w and a step size of s, calculate the short-term average D j of the speaker status data D j mean and the short-term peak D j max, as well as the environmental factor data Ek Short - term mean value E k mean; where j is a positive integer, j ∈ {1, 2, …, m}, representing the speaker status quantity sequence, D j represents the j - th speaker status data; k is a positive integer, k ∈ {1, 2, …, n}, representing the environmental factor quantity sequence, E k represents the k - th environmental factor data;

[0021] Normalize the above - mentioned features according to the maximum reference value to form a feature vector: X(t) = [D1(t)mean, D1(t)max, …, D m (t)mean, D m (t)max, E1(t)mean, …, E n (t)mean];

[0022] The average value reflects the "continuous power consumption" level of the power amplifier, battery or other components during this period, and has the greatest impact on long - term heat accumulation (cumulative temperature); the maximum value captures the short - term "power peak" or "transient impact", which is the key factor causing the instantaneous temperature jump; environmental temperature, air flow rate, humidity, etc. usually change slowly within several seconds or even several minutes, and the extreme value change range within a short time is limited, having a small impact on the instantaneous temperature shock.

[0023] In step 3, the specific method of combining thermodynamics and statistical modeling is as follows:

[0024] Combine thermodynamics and statistics for modeling to calculate the internal temperature T int (t):

[0025] ;

[0026] where T int (t) represents the internal temperature of the speaker corresponding to time t, directly obtained by the sensor; a j represents the weight coefficient corresponding to the speaker status data D j , reflecting the unit sensitivity of this status to temperature; b k represents the weight coefficient corresponding to the environmental factor data E k , reflecting the thermal coupling intensity of the environmental factor; c represents the weight coefficient of the historical temperature cumulative term, determining the contribution degree of exponential smoothing in the current temperature calculation; g k () represents the environmental factor mapping function, which maps the environmental factor data E k to the model space, defined by the staff in combination with the sensor characteristics;

[0027] The internal temperature of the speaker is determined by the dynamic balance between heat generation (power consumption) and heat dissipation (environmental conditions). The linear superposition in the formula directly reflects the physical contributions of the heat generation term (such as the Joule heat of the amplifier voltage and current) and the heat dissipation term (such as the convective heat dissipation of wind speed and humidity).

[0028] Simplified modeling of statistical regression: The actual heat transfer process involves complex partial differential equations (such as the heat conduction equation), but the main factors can be extracted through a linear regression model, and the parameters can be fitted using a data-driven approach to avoid complex calculations;

[0029] ;

[0030] Among them, EMA[T int (t) represents the exponentially weighted moving average, which is used to characterize the residual influence of past temperatures on the current thermal state; α represents the smoothing factor, and by adjusting α, the EMA curve can achieve a balance between "tracking" and "smoothing". Usually, cross-validation is used to evaluate the MSE to determine the optimal α; Δt represents the time interval between two data acquisitions;

[0031] Taking D j (t), g k (E k (t)) and EMA[T int (t) as independent variables and T int (t) as the dependent variable to construct a regression model, and the least squares method is used to solve for a j , b k and c.

[0032] In step 4, the specific method for quantifying the residual heat of the speaker is as follows:

[0033] Quantify the residual heat of the speaker:

[0034] ;

[0035] Among them, H res (t) represents the residual heat index, indicating the heat that has not dissipated yet due to historical heat generation inside the speaker up to time t; H res (t - Δt) represents the residual heat value at the previous sampling moment; the initial residual heat H res (0) is calibrated manually; α represents the smoothing factor; T std represents the preset calibrated ambient temperature value;

[0036] The recursive form of the residual heat (depending on the value at the previous moment) simulates the physical law of the exponential decay of heat over time (similar to the discharge of an RC circuit). Heat capacity and heat decay: The heat capacity characteristics of the speaker material and structure result in a lag in temperature change. The smoothing factor in the formula is related to the thermal time constant. The larger it is, the slower the residual heat decays, corresponding to an environment with low heat dissipation efficiency (such as an enclosed space).

[0037] By taking the difference from the calibrated ambient temperature (such as 25 °C), the baseline effect of the ambient temperature on the residual heat is eliminated, focusing on the heat accumulation generated by the speaker itself;

[0038] The specific way to train the LSTM to predict temperature is as follows: Integrate the feature vector sequence X(t i ) and the corresponding true temperature label T int (t i ); where i is a positive integer representing the data acquisition time sequence; X(t i ) represents the preprocessed speaker state and environmental factor features corresponding to time t i ; T int (t i ) represents the true temperature corresponding to time t i ;

[0039] Construct sample-label pairs: Each sample consists of a feature subsequence of length τ [X(t - τ + 1),..., X(t)]; the label is the temperature T int (T + Δ);

[0040] Select the long short-term memory network LSTM. By training on the sample-label pairs, output the predicted temperature T pre (t + Δ);

[0041] The calculation of the comprehensive heat load index and the dynamic temperature threshold includes:

[0042] Calculate the comprehensive heat load index S(t):

[0043] ;

[0044] where D j,max represents the reference maximum value for normalization of the j-th speaker state; T1 represents the safety critical temperature reference value calibrated by the staff; w 1,j represents the weight corresponding to the j-th speaker state, reflecting the relative contribution of this state to the heat load; w2 represents the weight of the predicted temperature term, measuring the influence degree of the future temperature on the current heat load judgment; w3 represents the weight of the residual heat term, measuring the influence degree of the historical heat accumulation on the current heat load judgment; w 4,k represents the weight corresponding to the k-th environmental factor mapping term, reflecting the influence intensity of this environmental factor on the heat dissipation ability;

[0045] Normalize physical quantities with different dimensions (voltage, temperature, residual heat) and then perform weighted summation to convert them into a dimensionless index S(t), which is convenient for setting a unified threshold.

[0046] Significance of weight assignment: w 1,j : Reflect the contribution differences of different speaker states (such as power amplifier voltage, sound pressure level) to heat generation, which can be determined by experimental calibration or principal component analysis (PCA). w2, w3: Balance the importance of the current state, future prediction, and historical residue (for example, if the weight of the predicted temperature is high, it is more inclined to preventive heat dissipation).

[0047] Negative sign environmental term: Directly reflect the enhancement effect of environmental factors on heat dissipation (for example, the higher the wind speed, the stronger the heat dissipation ability, equivalently reducing the heat load);

[0048] Design a dynamic threshold T th (t):

[0049] ;

[0050] Among them, l k represents the sensitivity coefficient corresponding to the environmental factor E k ; E k,std represents the preset calibration value corresponding to the environmental factor E k ;

[0051] In a harsh environment (such as high temperature, high humidity), the heat dissipation ability of the speaker decreases, and the safety threshold needs to be reduced to avoid the critical state. For example: when the environmental temperature E k (t) > E k,std (calibration value), the threshold decreases and heat dissipation is triggered in advance.

[0052] The linear sensitivity coefficient measures the influence gradient of environmental factors on the heat dissipation efficiency through experiments.

[0053] The specific way to define the condition for triggering active heat dissipation is as follows:

[0054] Define the triggering condition for active heat dissipation:

[0055] 1. S(t) ≥ S th ; It means that the comprehensive heat load index reaches or exceeds the preset triggering threshold S th ;

[0056] 2. T int (t) ≥ T th (t); It means that the actual internal temperature of the current speaker reaches or exceeds the dynamic temperature threshold Tth(t);

[0057] 3. T pre (t + Δ) ≥ T th(t); indicates that the predicted temperature will reach or exceed the temperature threshold T within the next time step th (t);

[0058] When any of the conditions is met, active heat dissipation is triggered.

[0059] A heat dissipation performance optimization system for a Bluetooth speaker, the system includes a data acquisition module, a preprocessing module, a temperature model module, and a heat dissipation prediction module;

[0060] The data acquisition module is used to set the data acquisition frequency and collect speaker status data and environmental factor data;

[0061] The preprocessing module is used to synchronize and preprocess the collected data, and perform feature extraction based on the preprocessed data;

[0062] The temperature model module is used to combine thermodynamics and statistics for modeling, construct a regression model based on the speaker status, environmental factors, and historical temperature, and calculate the internal temperature of the speaker;

[0063] The heat dissipation prediction module is used to quantify the residual heat of the speaker, integrate the feature vector and the true temperature label to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation.

[0064] The data acquisition module includes a frequency setting unit, a status acquisition unit, and an environment acquisition unit;

[0065] The frequency setting unit is used to set the data acquisition frequencies of the sensors and the API;

[0066] The status acquisition unit is used to collect speaker status data through sensors;

[0067] The environment acquisition unit is used to call the meteorological API to obtain environmental factor data.

[0068] The preprocessing module includes a filtering processing unit, a data synchronization unit, and a feature extraction unit;

[0069] The filtering processing unit is used to apply median filtering to eliminate data jitter;

[0070] The data synchronization unit is used to perform time alignment and linear interpolation on multi-source data;

[0071] The feature extraction unit is used to calculate the short-term mean and peak values of the data and generate a normalized feature vector.

[0072] The temperature model module includes a model construction unit and a parameter solution unit;

[0073] The model building unit is used to establish a thermodynamic and statistical hybrid model equation including EMA;

[0074] The parameter solving unit is used to calculate the weight coefficient by the least square method.

[0075] The heat dissipation prediction module includes a residual heat unit, a temperature prediction unit, an index calculation unit and a threshold trigger unit;

[0076] The residual heat unit is used to quantify the residual heat of the speaker;

[0077] The temperature prediction unit is used to train an LSTM model to predict the future temperature;

[0078] The index calculation unit is used to calculate the heat load index by synthesizing the state, prediction and residual heat;

[0079] The threshold trigger unit is used to dynamically calculate the temperature threshold and define the active heat dissipation trigger condition.

[0080] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes a method for quantifying the residual heat of the speaker, constructs a comprehensive heat load index, normalizes physical quantities with different dimensions and then sums them up with weights, and also designs a dynamic threshold to reasonably trigger active heat dissipation, improve the efficiency and accuracy of the heat dissipation system, avoid wasting energy by starting heat dissipation when it is unnecessary, and at the same time ensure that the speaker can obtain timely and effective heat dissipation protection under various environmental conditions; The present invention combines thermodynamics and statistics for modeling, constructs a regression model to calculate the internal temperature of the speaker through data collection, preprocessing and feature extraction, and uses LSTM for temperature prediction, which can more comprehensively capture various factors affecting the temperature, thereby providing a more accurate basis for heat dissipation control, and helping to avoid problems such as performance degradation, shortened life and even failure of the speaker due to overheating. Description of the Drawings

[0081] Figure 1 It is a step schematic diagram of a method for optimizing the heat dissipation performance of a Bluetooth speaker according to the present invention;

[0082] Figure 2 It is a process schematic diagram of a system for optimizing the heat dissipation performance of a Bluetooth speaker according to the present invention. Detailed Embodiments

[0083] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0084] Embodiment: AsFigure 1 - Figure 2 As shown in Figure 1 - Figure 2 , the present invention provides a technical solution, an optimization method for the heat dissipation performance of a Bluetooth speaker, and the method includes the following steps:

[0085] Step 1: Set the data acquisition frequency, and acquire the speaker status data and environmental factor data;

[0086] Step 2: Synchronize and preprocess the acquired data, and perform feature extraction based on the preprocessed data;

[0087] Step 3: Combine thermodynamics and statistical modeling, construct a regression model based on the speaker status, environmental factors, and historical temperature, and calculate the internal temperature of the speaker;

[0088] Step 4: Quantify the residual heat of the speaker, integrate the feature vectors and real temperature labels to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation.

[0089] In Step 1, set the data acquisition frequency according to the working characteristics and application scenarios of the speaker;

[0090] The specific method for acquiring the speaker status data and environmental factor data is as follows:

[0091] Arrange sensors inside the speaker and on the main control board to acquire the speaker status data, expressed as: [D1(t), D2(t), …, D m (t)]; where m is a positive integer representing the number of speaker statuses; t represents the data acquisition time, and D1(t), D2(t), …, D m (t) respectively represent the 1st, 2nd, …, mth speaker status data acquired at time t;

[0092] The speaker status signals include but are not limited to the power amplifier input voltage, current, audio signal duty cycle, sound pressure level, etc.;

[0093] Obtain the environmental factor data corresponding to the location through the weather API call, expressed as: [E1(t), E2(t), …, E n (t)]; where n is a positive integer representing the number of environmental factors; E1(t), E2(t), …, E n (t) respectively represent the 1st, 2nd, …, nth environmental factor data acquired at time t;

[0094] The environmental factor signals include but are not limited to environmental temperature, air flow rate, humidity, etc.

[0095] In Step 2, the specific method for synchronizing and preprocessing the acquired data is as follows:

[0096] Set the sliding window length w and the step size s;

[0097] Apply median filtering with window length w to the collected speaker status data and environmental factor data to eliminate short-term spikes and jitters;

[0098] Perform linear interpolation on the speaker status data and environmental factor data based on the same time reference to synchronize the data;

[0099] The specific method for feature extraction based on the preprocessed data is as follows: within a sliding window of length w and step size s, calculate the short-term average D j of the speaker status data D j mean and the short-term peak D j max, as well as the short-term mean E k mean of the environmental factor data E k ; where j is a positive integer, j ∈ {1, 2,..., m}, representing the speaker status number sequence, D j represents the j-th speaker status data; k is a positive integer, k ∈ {1, 2,..., n}, representing the environmental factor number sequence, E k represents the k-th environmental factor data;

[0100] Normalize the above features according to the maximum reference value to form a feature vector: X(t) = [D1(t)mean, D1(t)max,..., D m (t)mean, D m (t)max, E1(t)mean,..., E n (t)mean];

[0101] The average value reflects the "continuous power consumption" level of the power amplifier, battery or other components during this period, and has the greatest impact on long-term heat accumulation (cumulative temperature); the maximum value captures the short-term "power peak" or "transient impact", which is the key factor causing the instantaneous temperature jump; environmental temperature, air flow rate, humidity, etc. usually change slowly within seconds or even minutes, and the extreme value change range within a short time is limited, having a small impact on the instantaneous temperature shock.

[0102] In step 3, the specific method of combining thermodynamic and statistical modeling is as follows:

[0103] Combine thermodynamics and statistics for modeling and calculate the internal temperature T int (t) of the speaker:

[0104] ;

[0105] where T int (t) represents the internal temperature of the speaker corresponding to time t, directly obtained by the sensor; a j represents the speaker status data D jThe corresponding weight coefficient reflects the unit sensitivity of this state to temperature; b k represents the environmental factor data E k The corresponding weight coefficient reflects the thermal coupling intensity of the environmental factors; c represents the weight coefficient of the historical temperature accumulation term, which determines the contribution of exponential smoothing in the current temperature calculation; g k () represents the environmental factor mapping function, which maps the environmental factor data E k to the model space and is defined by the staff in combination with the sensor characteristics;

[0106] The internal temperature of the speaker is determined by the dynamic balance between heat generation (power consumption) and heat dissipation (environmental conditions). The linear superposition in the formula directly reflects the physical contributions of the heat generation term (such as the Joule heat of the amplifier voltage and current) and the heat dissipation term (such as the convective heat dissipation of wind speed and humidity).

[0107] Simplified modeling of statistical regression: The actual heat transfer process involves complex partial differential equations (such as the heat conduction equation), but the main factors can be extracted through a linear regression model, and the parameters can be fitted using a data-driven method to avoid complex calculations;

[0108] ;

[0109] Among them, EMA[T int (t) represents the exponentially weighted moving average, which is used to characterize the residual influence of past temperatures on the current thermal state; α represents the smoothing factor, and by adjusting α, the EMA curve can achieve a balance between "tracking" and "smoothing". Usually, cross-validation is used to evaluate the MSE to determine the optimal α; Δt represents the time interval between two data acquisitions;

[0110] Using D j (t), g k (E k (t)) and EMA[T int (t) as independent variables and T int (t) as the dependent variable to construct a regression model, and using the least squares method to solve for a j 、b k and c.

[0111] In step 4, the specific method for quantifying the residual heat of the speaker is as follows:

[0112] Quantify the residual heat of the speaker:

[0113] ;

[0114] Among them, H res (t) represents the residual heat index, indicating the heat that has not dissipated due to historical heat generation inside the speaker up to time t; H res(t - Δt) represents the residual heat value at the previous sampling moment; the initial residual heat H res (0) is calibrated manually; α represents the smoothing factor; T std represents the preset calibrated ambient temperature value;

[0115] The recursive form of the residual heat (where (t) depends on the previous moment value) simulates the physical law of the exponential decay of heat over time (similar to the discharge of an RC circuit). Heat capacity and heat decay: The heat capacity characteristics of the speaker material and structure result in a lag in temperature change. The smoothing factor in the formula is related to the thermal time constant. The larger it is, the slower the residual heat decays, corresponding to an environment with low heat dissipation efficiency (such as an enclosed space).

[0116] By taking the difference from the calibrated ambient temperature (such as 25 °C), the baseline influence of the ambient temperature on the residual heat is eliminated, focusing on the heat accumulation generated by the speaker itself;

[0117] The specific way to train the LSTM to predict temperature is as follows: Integrate the feature vector sequence X(t i ) and the corresponding true temperature label T int (t i ); where i is a positive integer representing the data acquisition moment sequence; X(t i ) represents the preprocessed speaker state and environmental factor features corresponding to the moment t i ; T int (t i ) represents the true temperature corresponding to the moment t i ;

[0118] Construct sample - label pairs: Each sample consists of a feature subsequence of length τ [X(t - τ + 1), …, X(t)]; the label is the temperature T int (t + Δ);

[0119] Select the long - short - term memory network LSTM. By training the sample - label pairs, output the predicted temperature T pre (t + Δ);

[0120] The calculation of the comprehensive heat load index and the dynamic temperature threshold includes:

[0121] Calculate the comprehensive heat load index S(t):

[0122] ;

[0123] where D j,max represents the reference maximum value for normalization of the j - th speaker state; T1 represents the safety critical temperature reference value, calibrated by the staff; w 1,j$w_1$ represents the weight corresponding to the $j$-th speaker state, reflecting the relative contribution of this state to the heat load; $w_2$ represents the weight of the predicted temperature term, measuring the influence degree of future temperature on the current heat load judgment; $w_3$ represents the weight of the residual heat term, measuring the influence degree of historical heat accumulation on the current heat load judgment; $w$ 4,k $w_4$ represents the weight corresponding to the $k$-th environmental factor mapping term, reflecting the influence intensity of this environmental factor on the heat dissipation capacity;

[0124] The physical quantities with different dimensions (voltage, temperature, residual heat) are normalized and then weighted and summed to be transformed into a dimensionless index $S(t)$, which is convenient for setting a unified threshold;

[0125] Significance of weight allocation: $w$ 1,j : Reflect the contribution differences of different speaker states (such as power amplifier voltage, sound pressure level) to heat generation, which can be determined by experimental calibration or principal component analysis (PCA). $w_2, w_3$: Balance the importance of the current state with future prediction and historical residuals (such as a high weight of predicted temperature, then more inclined to preventive heat dissipation).

[0126] Negative sign environmental term: Directly reflects the enhancement effect of environmental factors on heat dissipation (such as the higher the wind speed, the stronger the heat dissipation capacity, equivalently reducing the heat load);

[0127] Design a dynamic threshold $T$ th (t):

[0128] ;

[0129] where, $l$ k represents the sensitivity coefficient corresponding to the environmental factor $E$ k ; $E$ k,std represents the preset calibration value corresponding to the environmental factor $E$ k ;

[0130] In a harsh environment (such as high temperature, high humidity), the heat dissipation capacity of the speaker decreases, and the safety threshold needs to be reduced to avoid the critical state. For example: when the environmental temperature $E$ k (t) > $E$ k,std (calibration value), the threshold decreases and the heat dissipation is triggered in advance.

[0131] The linear sensitivity coefficient measures the influence gradient of environmental factors on the heat dissipation efficiency through experiments.

[0132] The specific way to define the condition for triggering active heat dissipation is as follows:

[0133] Define the trigger condition for active heat dissipation:

[0134] 1. $S(t) \geq S$ th ; It means that the comprehensive heat load index reaches or exceeds the preset trigger threshold $S$ th ;

[0135] 2.T int (t) ≥ T th (t); indicates that the actual internal temperature of the current speaker reaches or exceeds the dynamic temperature threshold Tth(t);

[0136] 3.T pre (t + Δ) ≥ T th (t); indicates that the predicted temperature will reach or exceed the temperature threshold T within the next time step th (t);

[0137] When any of the conditions is met, active heat dissipation is triggered.

[0138] A heat dissipation performance optimization system for a Bluetooth speaker, which includes a data acquisition module, a preprocessing module, a temperature model module, and a heat dissipation prediction module;

[0139] The data acquisition module is used to set the data acquisition frequency and collect speaker status data and environmental factor data;

[0140] The preprocessing module is used to synchronize and preprocess the collected data, and perform feature extraction based on the preprocessed data;

[0141] The temperature model module is used to combine thermodynamics and statistics for modeling, construct a regression model based on the speaker status, environmental factors, and historical temperature, and calculate the internal temperature of the speaker;

[0142] The heat dissipation prediction module is used to quantify the residual heat of the speaker, integrate the feature vector and the true temperature label to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation.

[0143] The data acquisition module includes a frequency setting unit, a status acquisition unit, and an environment acquisition unit;

[0144] The frequency setting unit is used to set the data acquisition frequencies of the sensors and the API;

[0145] The status acquisition unit is used to collect speaker status data through sensors;

[0146] The environment acquisition unit is used to call the meteorological API to obtain environmental factor data.

[0147] The preprocessing module includes a filtering processing unit, a data synchronization unit, and a feature extraction unit;

[0148] The filtering processing unit is used to eliminate data jitter by applying median filtering;

[0149] The data synchronization unit is used to perform time alignment and linear interpolation on multi-source data;

[0150] The feature extraction unit is used to calculate the short-time mean and peak value of data and generate a normalized feature vector.

[0151] The temperature model module includes a model construction unit and a parameter solving unit;

[0152] The model construction unit is used to establish a thermodynamic and statistical hybrid model equation including EMA;

[0153] The parameter solving unit is used to calculate the weight coefficient by the least square method.

[0154] The heat dissipation prediction module includes a residual heat unit, a temperature prediction unit, an index calculation unit, and a threshold trigger unit;

[0155] The residual heat unit is used to quantify the residual heat of the speaker;

[0156] The temperature prediction unit is used to train an LSTM model to predict the future temperature;

[0157] The index calculation unit is used to calculate the heat load index by integrating the state, prediction, and residual heat;

[0158] The threshold trigger unit is used to dynamically calculate the temperature threshold and define the active heat dissipation trigger condition.

[0159] In this embodiment, there is a certain portable Bluetooth speaker. The speaker is powered by a built-in lithium battery and is equipped with a digital power amplifier module. When playing music at a high volume for a long time in an outdoor high-temperature and high-humidity environment, it is prone to overheating problems. To optimize its heat dissipation performance, it is implemented according to the above method, and the specific device configuration is as follows:

[0160] Sensor arrangement: Install a voltage sensor, a current sensor, a temperature sensor, etc. on the main control board of the speaker; arrange a sound pressure level sensor inside the speaker cavity; at the same time, call the meteorological API through the device GPS positioning information to obtain environmental data.

[0161] Collect the power amplifier input voltage D1(t), the power amplifier input current D2(t), and the frequency output sound pressure level D3(t), and record them in the form of an array, such as [D1(t), D2(t), D3(t)].

[0162] At t = 10, the collected power amplifier input voltage is 3.8V, the current is 0.5A, and the sound pressure level is 85dB, D1(10) = 3.8, D2(10) = 0.5, D3(10) = 85.

[0163] Obtain the environmental temperature E1(t), the air flow rate E2(t), and the humidity E3(t) at the corresponding location through the meteorological API, expressed as [E1(t), E2(t), E3(t)].

[0164] At time t = 10, the ambient temperature is obtained as 30, the air flow rate is 1.2, the humidity is 65, E1(10) = 30, E2(10) = 1.2, and E3(10) = 65.

[0165] The length of the sliding window w = 60 (corresponding to 6 seconds of data, considering the change period of audio signals and environmental factors), and the step size s = 10 (i.e., the feature calculation is updated every 1 second).

[0166] For the collected speaker status data and environmental factor data, median filtering with a window length of 60 is applied to eliminate abnormal data such as short - time voltage spikes; linear interpolation is performed based on the same time reference.

[0167] Within a sliding window of length 60 and step size 10, calculate the short - time average D j of the speaker status data D j mean and the short - time peak D j max, as well as the short - time mean E k mean of the environmental factor data E k mean.

[0168] For the power amplifier input voltage D1, within the window from t = 10 to t = 70, calculate the average Dmean and the maximum D1max of the voltage data within this window; for the ambient temperature E1, calculate the average E1mean of the temperature data within this window.

[0169] Normalize the above - mentioned features by the maximum reference value to form the feature vector X(t)=[D1(t)mean,D1(t)max,…,D3(t)mean,D3(t)max,E1(t)mean,…,E3(t)mean].

[0170] Combined with thermodynamics and statistics for modeling, calculate the internal temperature T int (t); a j 、b k 、c The initial values can be set as empirical values first, such as a1 = 0.2, a2 = 0.3, a3 = 0.1, b1 = 0.1, b2 = 0.05, b3 = 0.02, c = 0.4; g k () function is defined according to the sensor characteristics. For example, for the ambient temperature E1, g1(E1(t)) = E1(t), and for the air flow rate E2, g2(E2(t)) = E2(t)^2 (considering that heat dissipation is approximately proportional to the square of the wind speed).

[0171] Calculate the exponentially weighted moving average EMA[T int (t); α = 0.3, Δt = 0.1 second (data acquisition interval), the initial EMA[T int(0) is set as the internal temperature value of the speaker for the first collection.

[0172] Using D j (t), g k (E k (t)) and EMA[T int (t) as independent variables, and T int (t) as the dependent variable to construct a regression model. By collecting data for a certain period of time (such as 1 hour) and using the least squares method to solve, more optimal values of a j , b k and c are obtained.

[0173] Quantify the residual heat of the speaker: For example, at t = 10, T int (10) = 37.5, H res (9) = 2 (the residual heat value at the previous moment), then H res (10) = 0.2×2+(1 - 0.2)×(37.5 - 25) = 10.4.

[0174] Train the LSTM to predict the temperature: Integrate the feature vector sequence X(t i ) and the corresponding true temperature label T int (t i ) to construct sample - label pairs. Each sample consists of a feature subsequence of length τ = 30, [X(t - 30 + 1), …, X(t)], and the label is the temperature T int (t + Δ), where Δ = 10 time steps (i.e., 1 second later).

[0175] Use the Keras library in Python to build an LSTM network, set appropriate network parameters (such as the number of neurons in the hidden layer is 64, the number of training epochs is 50, the batch size is 32, etc.). By training on the sample - label pairs, output the predicted temperature T pre (t + Δ).

[0176] Calculate the comprehensive heat load index and the dynamic temperature threshold:

[0177] Calculate the comprehensive heat load index S(t): w 1,1 =0.3, w 1,2 =0.2, w 1,3 =0.1, w2 = 0.2, w3 = 0.1, w 4,1 =0.05, w 4,2 =0.03, w 4,3 =0.02; D 1,max =5, D 2,max =1, D 3,max =100, T1 = 60 (the safety critical temperature reference value), H max= 50 (maximum reference value of residual heat).

[0178] At t = 10, D1(10) = 3.8, D2(10) = 0.5, D3(10) = 85, T pre (20) = 40 (predicted temperature 1 second later), H res (10) = 10.4, E1(10) = 30, E2(10) = 1.2, E3(10) = 65, substitute into the formula to calculate the value of S(10).

[0179] Design the dynamic threshold T th (t); At t = 10, substitute the data to calculate T th (10) = 60 - 0.5×(30 - 25) - 0.3×(1.2 - 2) - 0.2×(65 - 50) = 54.74;

[0180] Define the triggering condition for active heat dissipation: When S(t) ≥ S th (assuming S th = 0.8 is the preset triggering threshold), trigger active heat dissipation.

[0181] When T int (t) ≥ T th (t), trigger active heat dissipation. At t = 10, T int (10) = 37.5, T th (10) = 54.74, at this time this condition is not satisfied.

[0182] When T pre (t + Δ) ≥ T th (t), trigger active heat dissipation. At t = 10, T pre (20) = 40, T th (10) = 54.74, at this time this condition is not satisfied.

[0183] When any of the above conditions is satisfied, start the active heat dissipation device of the speaker, such as turning on the built-in fan, adjusting the power amplifier power, etc., to reduce the internal temperature of the speaker and optimize the heat dissipation performance.

[0184] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be encompassed within the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.

Claims

1. A method for optimizing the heat dissipation performance of a Bluetooth speaker, characterized in that: The method includes the following steps: Step 1: Set the data acquisition frequency and collect speaker status data and environmental factor data; Step 2: Synchronize and preprocess the collected data, and perform feature extraction based on the preprocessed data; Step 3: Combine thermodynamic and statistical modeling to construct a regression model based on the speaker status, environmental factors, and historical temperature, and calculate the internal temperature of the speaker; Step 4: Quantify the residual heat of the speaker, integrate the feature vectors and true temperature labels to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation; In Step 3, the specific method of combining thermodynamic and statistical modeling is as follows: Modeling is carried out by combining thermodynamics and statistics to calculate the internal temperature T int (t): ； Among them, T int (t) represents the internal temperature of the speaker corresponding to the moment t, which is directly obtained by the sensor; j is a positive integer, j ∈ {1, 2, …, m}, representing the speaker status quantity sequence, D j represents the j-th speaker status data; m is a positive integer, representing the number of speaker statuses; a j represents the weight coefficient corresponding to the speaker status data D j , reflecting the unit sensitivity of this status to temperature; n is a positive integer, representing the number of environmental factors; b k represents the weight coefficient corresponding to the environmental factor data E k , reflecting the thermal coupling intensity of the environmental factor; k is a positive integer, k ∈ {1, 2, …, n}, representing the environmental factor quantity sequence, E k represents the k-th environmental factor data; c represents the weight coefficient of the historical temperature cumulative term, determining the contribution of exponential smoothing in the current temperature calculation; g k () represents the environmental factor mapping function, which converts the environmental factor data E k to the model space and is defined by the staff in combination with the sensor characteristics; ； Among them, EMA[T int (t) represents the exponentially weighted moving average, which is used to characterize the residual influence of past temperatures on the current thermal state; α represents the smoothing factor; Δt represents the time interval between two data acquisitions; With D j (t), g k (E k (t)) and EMA[T int (t) as independent variables, and T int (t) as the dependent variable to construct a regression model, and using the least squares method to solve for a j , b k and c; In Step 4, the specific method of quantifying the residual heat of the speaker is as follows: The residual heat index H res (t) is calculated using a heat accumulation model, which is based on historical temperature data and heat dissipation characteristic parameters; The specific method for training the LSTM to predict temperature is as follows: integrating the preprocessed speaker state, environmental factor features, and the corresponding true temperature label; selecting the long short-term memory network (LSTM) for training and outputting the predicted temperature T pre (t + Δ); The calculation of the comprehensive heat load index and the dynamic temperature threshold includes: Calculate the comprehensive heat load index S(t): Use a comprehensive evaluation function to calculate S(t), and the function includes a speaker status index item, a temperature prediction item, a residual heat index item, and an environmental factor mapping item, and each item is equipped with a corresponding weight coefficient; Design a dynamic threshold T th (t): Dynamically adjust the temperature threshold based on environmental factors and preset parameters, where the preset parameters include the environmental factor sensitivity coefficient, the preset calibration value, and the safety critical temperature reference value.

2. The heat dissipation performance optimization method of a Bluetooth speaker according to claim 1, wherein: In Step 1, the specific method of collecting speaker status data and environmental factor data is as follows: Sensors are arranged inside the speaker and on the main control board to collect speaker status data, expressed as: [D1(t),D2(t),…,D m (t)]; where t represents the data collection time, D1(t), D2(t),…, D m (t) represents the 1st, 2nd, ..., mth speaker status data collected at time t respectively; Through the meteorological API call, obtain the environmental factor data of the corresponding location, expressed as: [E1(t), E2(t), …, E n (t)]; where E1(t), E2(t), …, E n (t) respectively represent the 1st, 2nd, …, nth environmental factor data obtained at time t.

3. The heat dissipation performance optimization method of a Bluetooth speaker according to claim 2, characterized in that: In Step 2, the specific method of synchronizing and preprocessing the collected data is as follows: Set the sliding window length w and the step size s; Apply median filtering with a window length of w to the collected speaker status data and environmental factor data to eliminate short-term spikes and jitters; Perform linear interpolation on the speaker status data and environmental factor data based on the same time reference to synchronize the data; The specific method for feature extraction based on the preprocessed data is as follows: within a sliding window of length w and step size s, calculate the short-time average value D j of the speaker status data D j mean and the short-time peak value D j max, as well as the short-time average value E k of the environmental factor data E k mean; Normalize the above features according to the maximum reference value to form a feature vector: X(t) = [D1(t)mean, D1(t)max, …, D m (t)mean, D m (t)max, E1(t)mean, …, E n (t)mean].

4. The heat dissipation performance optimization method of a Bluetooth speaker according to claim 3, characterized in that: In Step 4, the specific method of defining the conditions for triggering active heat dissipation is as follows: Define the triggering conditions for active heat dissipation: Condition 1: S(t) ≥ S th ; indicating that the comprehensive heat load index reaches or exceeds the preset trigger threshold S th ; Condition 2: T int (t) ≥ T th (t); indicates that the actual internal temperature of the current speaker reaches or exceeds the dynamic temperature threshold Tth(t); Condition Three: T pre (t + Δ) ≥ T th (t); indicating that the predicted temperature will reach or exceed the temperature threshold T within the next time step th (t); When any condition is met, trigger active heat dissipation.

5. A heat dissipation performance optimization system for a Bluetooth speaker, applied to the heat dissipation performance optimization method for a Bluetooth speaker according to any one of claims 1-4, characterized in that: The system includes a data acquisition module, a preprocessing module, a temperature model module, and a heat dissipation prediction module; The data acquisition module is used to set the data acquisition frequency and collect speaker status data and environmental factor data; The preprocessing module is used to synchronize and preprocess the collected data, and perform feature extraction based on the preprocessed data; the temperature model module is used to combine thermodynamic and statistical modeling to construct a regression model based on the speaker status, environmental factors, and historical temperature, and calculate the internal temperature of the speaker; the heat dissipation prediction module is used to quantify the residual heat of the speaker, integrate the feature vectors and true temperature labels to train the LSTM to predict the temperature, calculate the comprehensive heat load index and the dynamic temperature threshold, and define the conditions for triggering active heat dissipation.

6. The heat dissipation performance optimization system of a Bluetooth speaker according to claim 5, characterized in that: The data acquisition module includes a frequency setting unit, a status acquisition unit, and an environment acquisition unit; The frequency setting unit is used to set the data acquisition frequencies of the sensors and the API; the status acquisition unit is used to collect speaker status data through sensors; the environment acquisition unit is used to call the meteorological API to obtain environmental factor data.

7. The heat dissipation performance optimization system of a Bluetooth speaker according to claim 6, characterized in that: The preprocessing module includes a filtering processing unit, a data synchronization unit, and a feature extraction unit; The filtering processing unit is used to eliminate data jitters by applying median filtering; the data synchronization unit is used to perform time alignment and linear interpolation on multi-source data; the feature extraction unit is used to calculate the short-term mean and peak values of the data and generate a normalized feature vector.

8. The heat dissipation performance optimization system of a Bluetooth speaker according to claim 7, characterized in that: The temperature model module includes a model construction unit and a parameter solving unit; The model construction unit is used to establish a thermodynamic and statistical hybrid model equation including EMA; the parameter solving unit is used to calculate the weight coefficient by the least squares method.

9. The heat dissipation performance optimization system of a Bluetooth speaker according to claim 8, characterized in that: The heat dissipation prediction module includes a residual heat unit, a temperature prediction unit, an index calculation unit, and a threshold trigger unit; The residual heat unit is used to quantify the residual heat of the speaker; the temperature prediction unit is used to train the LSTM model to predict the future temperature; the index calculation unit is used to calculate the heat load index by integrating the state, prediction, and residual heat; the threshold trigger unit is used to dynamically calculate the temperature threshold and define the active heat dissipation trigger condition.

Images