Control method and charging device

Abstract

The application provides a control method and a charging device, and relates to the technical field of charging management. The control method comprises the following steps: collecting multiple types of sensing data through a multi-modal sensor; preprocessing the multiple types of sensing data to obtain multi-modal time series data; processing the multi-modal time series data based on a fusion prediction model set in the charging device to obtain a control instruction for adjusting the working parameters of the charging device; and controlling the charging device and / or a cooling device based on the control instruction. The charging device comprises a multi-modal sensor and a cooling device connection controller. The multi-modal sensor is used for collecting multiple types of sensing data. The controller is used for preprocessing the multiple types of sensing data to obtain multi-modal time series data. The multi-modal time series data is processed based on a set fusion prediction model to obtain a control instruction for adjusting the working parameters of the charging device. The charging device and / or the cooling device are controlled based on the control instruction.

Description

Technical Field

[0001] This application relates to the field of charging control technology, and more specifically, to a control method and a charging device. Background Technology

[0002] Currently, the main control methods used in the air-cooled heat dissipation systems of charging piles are as follows: Fixed speed or graded speed regulation: The fan speed is fixed at several levels based on temperature thresholds (e.g., 60°C, 70°C, 80°C), increasing the speed by one level when the temperature exceeds the threshold. Simple PID closed-loop speed regulation: The PWM duty cycle is continuously adjusted based on temperature error, but temperature is the only input, and noise is not considered. Independent noise alarm: Some high-end charging piles have built-in environmental microphones that only issue alarms or passively reduce speed when noise levels exceed the limit, lacking active predictive capabilities. Independent fault detection: Fan abnormalities are detected through vibration sensors or current monitoring, but alarms are usually only triggered after a significant fault has occurred.

[0003] However, existing charging pile air-cooling systems suffer from the following problems: A disconnect between noise, heat dissipation, and power: Heat dissipation control ignores noise levels, leading to noisy fan noises at night or in quiet areas; simply slowing down the fan to reduce noise can cause overheating and charging limitations, affecting charging efficiency. Lagging response and lack of prediction: The fan accelerates only after the temperature reaches a threshold, by which time thermal inertia has already caused temperature overshoot, forcing the fan into a high-speed, noisy state. Poor noise quality: Under conventional PWM drive, step changes in fan speed generate narrowband spike noise (such as the blade passing frequency and its harmonics), which is jarring to the human ear. Lack of health management and flexible fault tolerance: Early faults such as bearing wear cannot be detected in advance, leading to immediate shutdown once damage occurs; nor can the control strategy be proactively adjusted in the early stages of a fault to extend the operating time. Therefore, the user experience of existing charging piles is poor and fails to meet actual usage needs. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a control method and a charging device to improve the poor user experience of charging devices in the prior art.

[0005] To address the aforementioned problems, in a first aspect, embodiments of this application provide a control method, the method comprising: Multiple types of sensor data are acquired through multimodal sensors; Preprocessing of the various sensor data yields multimodal time-series data; The multimodal time-series data is processed based on the fusion prediction model set inside the charging device to obtain control commands for adjusting the operating parameters of the charging device; wherein, the operating parameters include: charging parameters and cooling parameters of the cooling devices inside the charging device; The charging device and / or the cooling device are controlled based on the control commands.

[0006] In the above implementation process, multimodal sensors can be set up to collect various sensor data characterizing different actual operating conditions of the charging device and cooling components. To improve the temporal consistency and correlation of the various sensor data, the data can be preprocessed to obtain multimodal time-series data. The charging device can be equipped with a corresponding fusion prediction model with predictive optimization capabilities. This model can use the multimodal time-series data as input, process it, and generate control commands to adaptively adjust the operating parameters of the charging device based on predicted future trends. This allows for effective control of the charging status and / or the cooling status of the cooling components within the charging device. By using the fusion prediction model to effectively control the charging device based on multiple types of sensor data, a balance can be achieved in noise control, charging efficiency, device reliability, and environmental adaptability, resulting in multi-objective optimization and significantly improving the user experience of the charging device, thus meeting the practical needs of various application scenarios.

[0007] Optionally, the preprocessing of the various sensor data to obtain multimodal time-series data includes: Time alignment processing is performed on the various sensor data to obtain synchronized data; The audio data in the synchronization data is subjected to spectrum processing to obtain spectrum data; The multimodal time series data is obtained by normalizing the spectrum data and the synchronization data.

[0008] In the above implementation process, when preprocessing multiple sensor data, time alignment can be performed first to obtain synchronized data with time sequence synchronization. Furthermore, considering that the synchronized data includes various data types, to facilitate subsequent unified processing, the audio data within the synchronized data can undergo spectral processing to obtain spectral data. This spectral data is then normalized with other synchronized data, thereby achieving multi-dimensional feature concatenation to obtain the corresponding multimodal time-series data. The ability to perform time synchronization, spectralization, and normalization processing on multiple sensor data to obtain time-series features representing multi-dimensional characteristics as corresponding multimodal time-series data effectively improves the effectiveness and comprehensiveness of the multimodal time-series data.

[0009] Optionally, the fusion prediction model includes: an input layer, a convolutional feature extraction layer, and a temporal prediction layer; The fusion prediction model based on the internal settings of the charging device processes the multimodal time-series data to obtain control commands for adjusting the operating parameters of the charging device, including: The multimodal time-series data is coupled through the input layer to obtain multidimensional data; The multi-dimensional data is reduced in dimensionality through the convolutional feature extraction layer to obtain feature vectors. The timing prediction layer dynamically predicts based on the feature vector and time relationship to obtain the control command for adjusting the operating parameters of the charging device.

[0010] In the above implementation process, the fusion prediction model can include a multi-layer neural network structure. This structure couples multimodal time-series data through the input layer to obtain multi-dimensional data. Then, it performs multi-layer convolutional processing through a convolutional feature extraction layer to extract features, achieving dimensionality reduction and obtaining a dimensionality-reduced feature vector. It should be noted that the time-series prediction layer can also learn the dependencies between features and time changes, thereby enabling dynamic prediction based on the feature vector and temporal relationships. This predicts the operating trend of the charging device over a future period, leading to control commands for adjusting the charging device's operating parameters. By using the fusion prediction model to predict the future operating trend of the charging device based on sensor data, and generating control commands to adjust the charging device's operating parameters, control can be implemented before changes in the charging device's operating state occur, effectively optimizing the charging device's performance and user experience.

[0011] Optionally, the control instructions include at least one of the following: a first instruction to control the noise upper limit threshold, a second instruction to optimize the duty cycle of the pulse width modulation of the cooling device, a third instruction to adjust the pulse width modulation mode, and a fourth instruction to adjust the charging power of the charging device.

[0012] In the above implementation process, the control instructions can include a variety of instructions to control different operating parameters. For example, a first instruction to control the upper limit threshold of noise, so as to effectively suppress the high noise of the charging equipment based on the first instruction; a second instruction to optimize the duty cycle of the pulse width modulation of the cooling device; a third instruction to adjust the pulse width modulation mode, so as to dynamically adjust the cooling power of the cooling device based on the second and third instructions; and a fourth instruction to adjust the charging power of the charging equipment, so as to dynamically adjust the charging power of the charging equipment based on the fourth instruction. In this way, a balance is achieved in multiple aspects such as noise control, charging efficiency, equipment reliability and environmental adaptability, and a multi-objective optimization effect is achieved.

[0013] Optionally, controlling the charging device and / or the cooling device based on the control command includes: If the current temperature is determined to be less than or equal to the temperature threshold and the predicted noise is less than or equal to the noise upper limit threshold, the power coordination coefficient of the charging device shall be maintained or increased to the maximum value. If the current temperature is determined to be greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the predicted noise is greater than the noise upper limit threshold, the charging power of the charging device is reduced. If the current temperature is determined to be greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the charging power of the charging device reaches the minimum allowable power, the duty cycle of the pulse width modulation is increased, and the cooling power of the cooling device is maintained. If the predicted temperature is determined to be greater than or equal to the temperature threshold, the power coordination coefficient of the charging device is determined based on the difference between the predicted temperature and the temperature threshold. The pulse width modulation carrier frequency of the cooling device is spread spectrum based on the pulse width modulation mode.

[0014] In the above implementation process, to achieve multi-objective flexible optimization and safe operation, the operating parameters of the charging equipment and cooling devices can be adjusted based on the actual control commands to control the charging equipment and cooling devices accordingly. When the current temperature is less than or equal to the temperature threshold and the predicted noise is less than or equal to the upper noise limit threshold, it indicates that the cooling effect of the cooling device is normal and safe. The power coordination coefficient of the charging equipment can be maintained or increased to its maximum value to achieve faster charging. When the current temperature is greater than the warning temperature threshold corresponding to the temperature threshold and the predicted noise is greater than the upper noise limit threshold, it indicates that the current heat dissipation demand is extremely high. Even with the cooling device reaching its power limit, the heat dissipation demand cannot be met. The charging power of the charging equipment can be reduced to reduce heat generation at the source, achieving a flexible charging limitation effect. When the current temperature is greater than the warning temperature threshold corresponding to the temperature threshold and the charging power of the charging equipment has reached the minimum allowable power (i.e., the charging power of the charging equipment cannot be reduced further), it indicates that the current heat dissipation demand is extremely high. The duty cycle of the pulse width modulation can be increased, the noise reduction target can be ignored, and the cooling power of the cooling device can be maintained. Further reduction of the cooling power of the cooling device is prohibited to ensure that the cooling device has sufficient power for heat dissipation. When the predicted temperature is greater than or equal to a temperature threshold, it indicates a significant subsequent heat dissipation demand. The difference between the predicted temperature and the temperature threshold can be used to determine an appropriate power coordination factor for the charging equipment to meet this demand. Furthermore, in cases of severe overheating risk or where temperature rise cannot be suppressed through heat dissipation control, the charging equipment can be set to its minimum permissible power to reduce subsequent heat dissipation requirements. It can also spread the pulse width modulation carrier frequency of the cooling device based on pulse width modulation mode to disperse concentrated, sharp noise energy across a wide frequency band, reducing the subjective loudness of the cooling device noise. It can perform corresponding control based on various operating parameters, achieving multi-objective flexible optimization and safe operation.

[0015] Optionally, the multimodal sensor includes: a voiceprint sensor, a noise sensor, a temperature sensor, and an electrical sensor; the various sensing data include: voiceprint data, noise data, temperature data, and electrical data.

[0016] In the above implementation process, the multimodal sensor may include an acoustic fingerprint sensor for detecting near-field acoustic fingerprints of the cooling device, a noise sensor for detecting ambient noise of the charging device, a temperature sensor for detecting the temperature of the charging device, and an electrical sensor for detecting the electrical parameters of the charging device. Correspondingly, the various sensor data may include acoustic fingerprint data characterizing the near-field acoustic fingerprints of the cooling device, noise data characterizing ambient noise, temperature data characterizing the temperature of the charging device, and electrical parameters characterizing the electrical condition of the charging device. Multiple different types of sensors can be configured to collect various types of sensor data separately, so as to comprehensively reflect the actual operating status of the charging device through a combination of these various types of sensor data.

[0017] Optionally, the method further includes: A voiceprint database was established based on historical voiceprint data. Based on the fusion prediction model, the voiceprint data and the voiceprint database are processed to obtain error data; Based on the error data, the operating state of the cooling device is adjusted.

[0018] In the above implementation process, considering that acoustic fingerprint data can characterize not only the noise level of cooling devices but also their operating status, a corresponding acoustic fingerprint database can be established based on historical acoustic fingerprint data under normal operating conditions. This database is then processed using a fusion prediction model to obtain corresponding error data. Based on the actual magnitude of the error data, the operating status of the cooling devices can be adjusted in real time. This allows for proactive adjustment of the cooling devices' operating status based on the comparison between actual and historical acoustic fingerprints, achieving fault prediction and flexible fault tolerance.

[0019] Optionally, adjusting the operating state of the cooling device based on the error data includes: If the error data is greater than or equal to a preset error threshold and the duration is greater than or equal to a preset time threshold, it is determined that the cooling device has an abnormal operating speed range. The rotational speed of the cooling device is reduced, and an adjustment command is determined that does not include the specified rotational speed range.

[0020] In the above implementation process, corresponding error thresholds and time thresholds can be preset for limitation. When the error data is greater than or equal to the error threshold and the duration is greater than or equal to the time threshold, it indicates that the voiceprint data is abnormal, resulting in voiceprint drift, i.e., the cooling device is malfunctioning. The abnormal operating speed range of the cooling device can be determined, and the speed of the cooling device can be reduced. Adjustment commands that do not include this speed range can be determined to actively bypass the abnormal speed range. The operating status of the cooling device can be determined based on the actual error data, achieving a flexible operation effect under abnormal conditions and avoiding adverse situations caused by sudden shutdowns.

[0021] Secondly, embodiments of this application also provide a charging device, the charging device comprising: a multimodal sensor, a controller, and a cooling device; The multimodal sensor and the cooling device are connected to the controller; The multimodal sensor is used to collect various types of sensor data; The controller is used to preprocess various types of sensor data to obtain multimodal time-series data; process the multimodal time-series data based on a set fusion prediction model to obtain control instructions for adjusting the operating parameters of the charging device; wherein, the operating parameters include: charging parameters and cooling parameters of the cooling device; and control the charging device and / or the cooling device based on the control instructions.

[0022] In the above implementation process, multiple sensor data representing various actual operating conditions of the charging equipment and cooling devices are collected through multimodal sensors. To improve the temporal consistency and correlation of the multiple sensor data, the controller preprocesses the multiple sensor data to obtain multimodal time-series data. The controller can be equipped with a corresponding fusion prediction model with predictive optimization function. It can take the multimodal time-series data as input data, process it into the fusion prediction model, and obtain control commands to adaptively adjust the operating parameters of the charging equipment based on the predicted future trends. Thus, based on the control commands, the charging status of the charging equipment and / or the cooling status of the cooling devices in the charging equipment can be effectively controlled.

[0023] Optionally, the multimodal sensor includes: an acoustic signature sensor, a noise sensor, a temperature sensor, and an electrical sensor; the various sensing data include: acoustic signature data, noise data, temperature data, and electrical data; The acoustic signature sensor is located at the detection position of the cooling device; The noise sensor is mounted on the housing of the charging device; The temperature sensor is located inside the charging device; The electrical sensor is installed in the charging circuit of the charging device.

[0024] In the above implementation process, the multimodal sensor may include an acoustic fingerprint sensor disposed at the detection location of the cooling device to detect the near-field acoustic fingerprint of the cooling device; a noise sensor disposed on the housing of the charging device to detect the ambient noise of the charging device; a temperature sensor disposed inside the charging device to detect the temperature of the charging device; and an electrical sensor disposed in the charging circuit of the charging device to detect the electrical parameters of the charging device. Correspondingly, the various sensor data may include acoustic fingerprint data characterizing the near-field acoustic fingerprint of the cooling device, noise data characterizing the ambient noise, temperature data characterizing the temperature of the charging device, and electrical parameters characterizing the electrical condition of the charging device. Multiple different types of sensors can be configured to collect various different types of sensor data separately, so as to comprehensively reflect the actual working condition of the charging device through the comprehensive collection of various types of sensor data.

[0025] In summary, the embodiments of this application provide a control method and charging device that can effectively control the charging device based on multiple types of sensor data by fusing predictive models, so as to achieve a balance in terms of noise control, charging efficiency, device reliability and environmental adaptability, realize multi-objective optimization effect, effectively improve the user experience of the charging device, and thus meet the actual use needs in various application scenarios. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart illustrating a control method provided in an embodiment of this application; Figure 2 A detailed flowchart of step S200 provided for an embodiment of this application; Figure 3 A detailed flowchart of step S300 provided for an embodiment of this application; Figure 4 A detailed flowchart of step S400 provided for an embodiment of this application; Figure 5 A flowchart illustrating another control method provided in an embodiment of this application; Figure 6A detailed flowchart of step S530 provided for an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a charging device provided in an embodiment of this application.

[0028] Icons: 610 - Multimodal sensor; 620 - Controller; 630 - Cooling device. Detailed Implementation

[0029] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of the embodiments of this application.

[0030] Existing air-cooled heat dissipation systems for charging piles suffer from the following problems: A disconnect between noise, heat dissipation, and power: Heat dissipation control ignores noise levels, leading to noisy fan noises at night or in quiet areas; simply slowing down the fan to reduce noise can cause overheating and charging limitations, affecting charging efficiency. Lagging response and lack of prediction: The fan accelerates only after the temperature reaches a threshold, by which time thermal inertia has already caused temperature overshoot, forcing the fan into a high-speed, noisy state. Poor noise quality: Under conventional PWM drive, abrupt changes in fan speed generate narrowband spike noise (such as the blade passing frequency and its harmonics), which is jarring to the human ear. Lack of health management and flexible fault tolerance: Early faults such as bearing wear cannot be detected in advance, leading to immediate shutdown upon failure; nor can control strategies be proactively adjusted in the early stages of a fault to extend the operating time. Therefore, existing charging piles offer a poor user experience and fail to meet actual usage needs.

[0031] To address the aforementioned issues, this application provides a control method and charging device that can effectively control the charging device based on multiple types of sensor data by fusing predictive models. This achieves a balance between noise control, charging efficiency, device reliability, and environmental adaptability, resulting in multi-objective optimization and significantly improving the user experience of the charging device. Consequently, it meets the practical needs of various application scenarios.

[0032] This application provides a control method for use in charging equipment, which may include various devices such as charging piles and charging stations with multiple charging piles.

[0033] Please see Figure 1 , Figure 1 This is a flowchart illustrating a control method provided in an embodiment of this application. The method may include steps S100-S400.

[0034] In step S100, various sensing data are acquired through a multimodal sensor.

[0035] Among them, multimodal sensors can be set up to collect various sensor data that characterize the charging equipment and cooling devices under various actual working conditions.

[0036] Optionally, the sensing data may include various data related to temperature, power, noise, etc.

[0037] Step S200: Preprocess the various sensor data to obtain multimodal time series data.

[0038] In order to improve the temporal consistency and correlation of multiple sensor data, multiple sensor data can be preprocessed to obtain multimodal time series data.

[0039] Optionally, multimodal time series data may include time series feature maps containing multiple features.

[0040] Step S300: Based on the fusion prediction model set inside the charging device, the multimodal time series data is processed to obtain control commands for adjusting the operating parameters of the charging device.

[0041] The charging equipment can be equipped with a corresponding fusion prediction model with prediction and optimization functions. It can take multimodal time series data as input data, process it into the fusion prediction model, and obtain control commands to adaptively adjust the operating parameters of the charging equipment based on the predicted future trends.

[0042] Optionally, the operating parameters may include: charging parameters and cooling parameters of the cooling devices inside the charging device. The charging parameters may include parameters such as the charging power of the charging device, and the cooling parameters may include parameters such as the cooling power of the cooling devices.

[0043] Step S400: Control the charging device and / or cooling device based on control commands.

[0044] Among them, the charging status of the charging equipment and / or the cooling status of the cooling devices in the charging equipment can be effectively controlled based on control commands.

[0045] It should be noted that since control commands can adjust the operating parameters of the charging equipment, the charging equipment and / or cooling devices can be controlled accordingly based on the actual situation of the charging equipment and the control commands.

[0046] exist Figure 1In the illustrated embodiment, the charging device can be effectively controlled based on multiple types of sensor data by fusing predictive models to achieve a balance in noise control, charging efficiency, device reliability, and environmental adaptability, thereby achieving multi-objective optimization and effectively improving the user experience of the charging device, thus meeting the actual usage needs in various application scenarios.

[0047] Optionally, please refer to Figure 2 , Figure 2 The following is a detailed flowchart of step S200 provided in an embodiment of this application. Step S200 may include steps S210-S230.

[0048] Step S210: Time alignment processing is performed on multiple sensor data to obtain synchronized data.

[0049] In the process of preprocessing multiple sensor data, time alignment can be performed on the multiple sensor data to obtain synchronized data with time synchronization.

[0050] Optionally, hardware-level timestamp alignment can be performed on various sensor data to ensure that multiple data types, such as audio frames and temperature changes, are synchronized at the millisecond level to obtain corresponding synchronized data.

[0051] Step S220: Perform spectrum processing on the audio data in the synchronization data to obtain spectrum data.

[0052] Considering that the synchronization data includes various types of data, in order to facilitate unified processing later, the audio data in the synchronization data can be further processed to obtain spectrum data.

[0053] Optionally, the audio data may include near-field acoustic signature data of the cooling device and far-field noise data of the environment where the charging device is located. The near-field / far-field audio may be processed by frame segmentation (e.g., frame length 1024 points, hop length 512) and windowed (Hamming window). The power spectrum is calculated by FFT (Fast Fourier Transform) to generate a Mel spectrum as the corresponding spectrum data.

[0054] Step S230: Normalize the spectrum data and synchronization data to obtain multimodal time series data.

[0055] In this process, the spectral data is further normalized with other synchronization data to achieve multi-dimensional feature splicing and obtain the corresponding multimodal time series data.

[0056] Optionally, for other non-audio data such as power and temperature present in the synchronization data, the temperature and power data can be combined with audio features in a multi-dimensional feature concatenation to form a time-series feature map, which can be used as the corresponding multimodal time-series data.

[0057] exist Figure 2 In the illustrated embodiment, time synchronization, spectralization, and normalization processing can be performed on various sensor data to obtain time-series features that characterize multidimensional features as corresponding multimodal time-series data, effectively improving the effectiveness and comprehensiveness of multimodal time-series data.

[0058] It should be noted that the hardware platform for the fusion prediction model can be an embedded processor equipped with an NPU (Neural Processing Unit) to meet the real-time inference requirements of the fusion prediction model. The fusion prediction model includes a multi-layer neural network structure, including an input layer, a convolutional feature extraction layer, and a temporal prediction layer, and the output control commands can be multi-dimensional.

[0059] Optionally, the fusion prediction model may include a CNN-LSTM (Convolutional Neural Network-Long Short-Term Memory Network) hybrid network structure, where CNN is responsible for spatial feature extraction and LSTM is responsible for temporal feature modeling, taking into account both spatial information and temporal dependencies. This model can be implemented within a neural network processor or a lightweight neural network chip. The fusion prediction model may also include other predictive models such as simplified rule models.

[0060] Optionally, please refer to Figure 3 , Figure 3 This is a detailed flowchart of step S300 provided in an embodiment of the present application. Step S300 may include steps S310-S330.

[0061] Step S310: Through the input layer, the multimodal time series data is coupled to obtain multidimensional data.

[0062] Among them, multimodal time series data can be coupled through the input layer to obtain multidimensional data.

[0063] Optionally, the input dimensions of the input layer may include: [batch_size, time_steps, features]. Here, time_steps can take timing data from the past 5-10 seconds. Since the cooling device can be controlled based on PWM (Pulse Width Modulation), features can include near-field spectral features (128 dimensions), far-field spectral features (128 dimensions), temperature value (2 dimensions), charging power (1 dimension), and current PWM duty cycle (1 dimension), for a total of approximately 260 feature dimensions.

[0064] Step S320: The multi-dimensional data is reduced in dimensionality by using a convolutional feature extraction layer to obtain feature vectors.

[0065] In this process, multiple convolutional layers can be used to extract features, thereby achieving dimensionality reduction and obtaining the dimensionality-reduced feature vector.

[0066] Optionally, the convolutional feature extraction layer can be a feature extraction layer of a CNN, which can treat the audio spectrum as a two-dimensional image and extract local time-frequency patterns. The structure of the convolutional feature extraction layer can include: two two-dimensional convolutional layers + pooling layers. The convolutional kernel size is (3, 3), and the stride is 1. The first convolutional layer extracts narrowband peak features (such as the passing frequency of the blades of a cooling device's fan). The second convolutional layer extracts abnormal sound pattern patterns such as whistling and friction.

[0067] Step S330: Through the time-series prediction layer, dynamic prediction is performed based on feature vectors and time relationships to obtain control commands for adjusting the operating parameters of the charging device.

[0068] It should be noted that the time-series prediction layer can also learn the dependencies of features over time, thereby making dynamic predictions based on feature vectors and time relationships to predict the working trend of charging equipment in the future, and thus obtain control commands to adjust the working parameters of the charging equipment.

[0069] Optionally, the temporal prediction layer can be an LSTM temporal prediction layer, capable of learning the dependencies of features over time and predicting operating conditions such as temperature and noise values ​​over a future period (e.g., 5-10 seconds, which can be selected based on actual conditions and needs). The structure of the temporal prediction layer can include two LSTM layers with 128 hidden units, capable of capturing the dynamic coupling relationship between charging device power increase → temperature increase → cooling device power increase → noise spectrum change.

[0070] It should be noted that the fusion prediction model may also include an output layer or a decision layer, which can output multi-dimensional control commands.

[0071] exist Figure 3 In the illustrated embodiment, the future operating trend of the charging device can be predicted based on sensor data by fusion prediction model, thereby generating control commands to adjust the operating parameters of the charging device. This allows for control before the operating state of the charging device changes, effectively optimizing the working effect and user experience of the charging device.

[0072] It should be noted that the multi-dimensional control instructions may include at least one of the following: a first instruction to control the noise upper limit threshold, a second instruction to optimize the duty cycle of the pulse width modulation of the cooling device, a third instruction to adjust the pulse width modulation mode, and a fourth instruction to adjust the charging power of the charging device. The control instructions may include multiple instructions for controlling different operating parameters. For example, the first instruction to control the noise upper limit threshold effectively suppresses high noise levels in the charging device; the second instruction to optimize the duty cycle of the pulse width modulation of the cooling device; the third instruction to adjust the pulse width modulation mode dynamically adjusts the cooling power of the cooling device based on the second and third instructions; and the fourth instruction to adjust the charging power of the charging device dynamically adjusts the charging power based on the fourth instruction. This achieves a balance in multiple aspects such as noise control, charging efficiency, equipment reliability, and environmental adaptability, realizing a multi-objective optimization effect.

[0073] For example, the noise upper limit threshold in the first instruction can be pre-set by the upper-level rules of the fusion prediction model based on factors such as time period (day / night), ambient noise, and surrounding sensitivity level. The fusion prediction model can learn how to adjust subsequent instructions to approximate this target. The duty cycle of the pulse width modulation in the second instruction ranges from 0-100%. The pulse width modulation mode in the third instruction can include two types: conventional PWM mode and spread spectrum modulation mode. When a significant narrowband spike is detected in the acoustic waveform of the cooling device, the spread spectrum modulation mode can be triggered. In the fourth instruction, the charging power of the charging device can be adjusted by a power coordination coefficient (used to request a reduction in charging power from the charging module) ranging from 0.5 to 1.0, based on a factor adjustment method.

[0074] Optionally, please refer to Figure 4 , Figure 4 The following is a detailed flowchart of step S400 provided in an embodiment of this application. Step S400 may include steps S410-S450.

[0075] Step S410: If the current temperature is determined to be less than or equal to the temperature threshold and the predicted noise is less than or equal to the noise upper limit threshold, the power coordination coefficient of the charging device is maintained or increased to the maximum value.

[0076] Among them, if the current temperature is less than or equal to the temperature threshold and the predicted noise is less than or equal to the noise upper limit threshold, it indicates that the cooling effect of the current cooling device is normal and safe, and the power coordination coefficient of the charging device can be maintained or increased to the maximum value. For example, the power coordination coefficient of the charging device can be maintained at 1.0, or the power coordination coefficient of the charging device can be increased to 1.0 to achieve a faster charging effect.

[0077] Step S420: If the current temperature is determined to be greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the predicted noise is greater than the upper limit threshold of the noise, the charging power of the charging device is reduced.

[0078] In this case, if the current temperature is higher than the warning temperature threshold corresponding to the temperature threshold, and the predicted noise is higher than the upper limit of the noise threshold, it indicates that the current heat dissipation demand is extremely high, and the cooling device still cannot meet the heat dissipation demand even when it reaches the power limit. In this case, the charging power of the charging device can be reduced to reduce heat generation from the source and achieve a flexible charging limit effect.

[0079] Alternatively, the charging power of the charging device can be reduced by adjusting the power coordination factor, for example, by making the power coordination factor less than 1.0 (such as setting it to 0.8 or 0.9).

[0080] It should be noted that the temperature threshold can be set based on the actual operating conditions of the charging device. When the temperature is greater than or equal to the temperature threshold, it indicates that the charging device has reached an overheating state. The temperature threshold is T1, and the corresponding warning temperature threshold is T2, where T2 = T1 - T2. , It can be set to 2℃ or other configurable temperature values ​​based on actual needs. When the temperature reaches the warning temperature threshold, it indicates that the charging device is approaching overheating.

[0081] Step S430: If it is determined that the current temperature is greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the charging power of the charging device reaches the minimum allowable power, the duty cycle of the pulse width modulation is increased, and the cooling power of the cooling device is maintained.

[0082] Among them, when the current temperature is greater than the warning temperature threshold corresponding to the temperature threshold, and the charging power of the charging device has reached the minimum allowable power, that is, the charging power of the charging device can no longer be reduced, it indicates that the current heat dissipation demand is extremely high. In this case, the duty cycle of the pulse width modulation can be increased, the noise reduction target can be ignored, and the cooling power of the cooling device can be maintained. Further reduction of the cooling power of the cooling device is prohibited so that the cooling device has a sufficiently high power for heat dissipation and cooling treatment.

[0083] Optionally, the duty cycle of the pulse width modulation can be forced to 100%, while ignoring noise reduction requirements and prohibiting further reduction of the cooling power of the cooling device.

[0084] Step S440: If the predicted temperature is greater than or equal to the temperature threshold, the power coordination coefficient of the charging device is determined based on the difference between the predicted temperature and the temperature threshold.

[0085] Specifically, if the predicted temperature is greater than or equal to the temperature threshold, it indicates a significant subsequent heat dissipation requirement. The appropriate power coordination factor for the charging device can be determined based on the difference between the predicted temperature and the temperature threshold to meet this requirement. Furthermore, in cases of severe overheating risk or where temperature rise cannot be suppressed through heat dissipation control, the charging device can be set to its minimum permissible power to reduce subsequent heat dissipation demands.

[0086] Optionally, the minimum allowable power of the charging device can be set based on actual needs and circumstances, for example, it can be set to 0.8 times the rated power.

[0087] Step S450: Spread the pulse width modulation carrier frequency of the cooling device based on the pulse width modulation mode.

[0088] Furthermore, it can spread the pulse width modulation carrier frequency of the cooling device based on the pulse width modulation mode to disperse the concentrated sharp noise energy into a wide frequency band and reduce the subjective loudness of the noise of the cooling device.

[0089] Optionally, the cooling device may include a fan or other device that achieves air cooling. In conventional PWM mode, the PWM duty cycle of the cooling device can be adjusted in advance based on the temperature trend predicted by the fusion prediction model to achieve linear speed change and avoid abrupt changes. Spread spectrum modulation may include randomly and slightly fluctuating the PWM carrier frequency between 20kHz and 25kHz to disperse concentrated sharp noise energy across a wide frequency band, thereby reducing subjective loudness.

[0090] exist Figure 4 In the illustrated embodiment, corresponding control can be performed based on a variety of different operating parameters to achieve the effects of multi-objective flexible optimization and safe operation.

[0091] It should be noted that multimodal sensors can include acoustic fingerprint sensors for detecting near-field acoustic signatures of cooling devices, noise sensors for detecting ambient noise of charging devices, temperature sensors for detecting the temperature of charging devices, and electrical sensors for detecting the electrical parameters of charging devices. Correspondingly, the various sensor data can include acoustic fingerprint data characterizing the near-field acoustic signatures of cooling devices, noise data characterizing ambient noise, temperature data characterizing the temperature of charging devices, and electrical parameters characterizing the electrical properties of charging devices. Multiple types of sensors can be configured to collect various types of sensor data separately, so as to comprehensively reflect the actual operating status of the charging device through a combination of these diverse sensor data.

[0092] Optionally, please refer to Figure 5 , Figure 5 This is a flowchart illustrating another control method provided in an embodiment of this application. The method may further include steps S510-S530.

[0093] Step S510: Establish a voiceprint database based on historical voiceprint data.

[0094] Considering that voiceprint data can characterize not only the noise level of cooling devices but also their operating status, a corresponding voiceprint database can be established based on historical voiceprint data under normal operating conditions.

[0095] Optionally, taking a fan as an example of a cooling device, near-field acoustic signature spectra at different speeds can be collected under the normal operating conditions of the fan to establish a multi-dimensional baseline database as an acoustic signature database. The acoustic signature database includes the mean μ and the standard deviation σ.

[0096] Optionally, the historical voiceprint data can be data collected before the charging device leaves the factory, or data collected during the historical normal operation phase after the charging device leaves the factory.

[0097] Step S520: Based on the fusion prediction model, the voiceprint data and voiceprint database are processed to obtain error data.

[0098] Step S530: Adjust the operating status of the cooling device based on the error data.

[0099] Among them, based on the fusion prediction model, combined with voiceprint data and voiceprint database, the corresponding error data can be obtained, so as to adjust the working state of the cooling device in real time according to the actual size of the error data.

[0100] Optionally, the fusion prediction model can simultaneously output the reconstruction error between the spectrum corresponding to the current voiceprint data and the baseline between the voiceprint database as error data.

[0101] exist Figure 5 In the illustrated embodiment, the operating state of the cooling device can be adjusted in advance based on the comparison between the actual voiceprint and the historical voiceprint, so as to achieve the effects of fault prediction and flexible fault tolerance.

[0102] Optionally, please refer to Figure 6 , Figure 6 The following is a detailed flowchart of step S530 provided in an embodiment of this application. Step S530 may include steps S531-S532.

[0103] Step S531: If the error data is greater than or equal to a preset error threshold and the duration is greater than or equal to a preset time threshold, determine that the cooling device has an abnormal operating speed range.

[0104] Among them, corresponding error thresholds and time thresholds can be preset for limitation. If the error data is greater than or equal to the error threshold and the duration is greater than or equal to the time threshold, it indicates that there is an abnormality in the voiceprint data and a voiceprint drift phenomenon, that is, there is an abnormality in the cooling device. The speed range in which the cooling device is malfunctioning can be determined.

[0105] For example, the error threshold and time threshold can be selected based on the actual situation and needs to reduce the adverse situation of misjudgment. For example, the error threshold can be set to μ+3σ and the time threshold can be set to 72 hours. When the error data is greater than or equal to μ+3σ for 72 hours, it is judged as voiceprint drift, that is, there is an abnormality in the cooling device. An early warning can be issued to the cloud operation and maintenance platform to notify the staff to carry out maintenance and other processing.

[0106] Optionally, the speed range in which abnormal resonance occurs can be identified by analyzing the distribution of error data, and this range can be used as the speed range in which the operation is abnormal.

[0107] Step S532: Reduce the rotation speed of the cooling device and determine an adjustment command that does not include the rotation speed range.

[0108] Among these measures, the rotational speed of the cooling device can be reduced to identify adjustment commands that do not include the specified speed range, thereby proactively bypassing the abnormal speed range.

[0109] Optionally, the adjustment instructions may include various control methods such as limiting the maximum speed of the fan to 80% of the normal speed limit, and prioritizing the use of spread spectrum modulation mode to mask abnormal frequency components.

[0110] exist Figure 6 In the illustrated embodiment, the operating status of the cooling device can be determined based on the actual situation of the error data, achieving a flexible operation effect under abnormal conditions and avoiding adverse situations caused by sudden shutdown.

[0111] Please see Figure 7 , Figure 7 This is a schematic diagram of the structure of a charging device provided in an embodiment of this application. The charging device may include: a multimodal sensor 610, a controller 620, and a cooling device 630. The multimodal sensor 610 and the cooling device 630 are respectively connected to the controller 620; The multimodal sensor 610 is used to acquire various types of sensor data; The controller 620 is used to preprocess various sensor data to obtain multimodal time-series data; based on the set fusion prediction model, it processes the multimodal time-series data to obtain control commands to adjust the operating parameters of the charging device; wherein, the operating parameters include: charging parameters and cooling parameters of the cooling device 630; and controls the charging device and / or cooling device 630 based on the control commands.

[0112] Optionally, the controller 620 can be an electronic device with logic computing capabilities, or an integrated circuit chip with signal processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a digital signal processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0113] Optionally, the multimodal sensor 610 includes: an acoustic sensor, a noise sensor, a temperature sensor, and an electrical sensor; and various sensing data include: acoustic data, noise data, temperature data, and electrical data. The acoustic signature sensor is located at the detection position of the cooling device 630; The noise sensor is mounted on the casing of the charging device; The temperature sensor is located inside the charging device; Electrical sensors are installed in the charging circuit of the charging equipment.

[0114] For example, the acoustic sensor may include a MEMS digital microphone (such as INMP441) with a sampling rate of ≥51.2kHz. The detection position of the cooling device 630 may be 10-15cm away from the wind turbine blades to achieve wind protection. The acoustic data can capture high-frequency features such as wind turbine bearing friction, blade dynamic balance, and airflow whistling.

[0115] For example, a noise sensor may include a waterproof and dustproof microphone array (multiple microphones, such as 2-4), which may be located at the air inlet and outlet on the housing of the charging device, as well as on the outside of the housing. The noise data can characterize the noise contribution of the environment and the background ambient noise.

[0116] For example, the temperature sensor may include an NTC thermistor or a PT1000, which can be installed inside the charging device's air inlet, air outlet, and inside the power module. The temperature data can reflect the actual situation of heat load and heat dissipation efficiency in real time.

[0117] For example, an electrical sensor may include a current / voltage sampling chip, and the electrical parameters may characterize actual conditions such as charging power, current, and voltage.

[0118] Optionally, the control method of this application can be applied not only to the application scenarios of charging equipment, but also to various other application scenarios such as air-cooled noise reduction of data center servers, thermal management systems of electric vehicles, and active noise reduction of outdoor units of household air conditioners.

[0119] Optionally, taking a fan as an example, the fan noise mainly originates from narrowband spikes (blade passing frequency and its harmonics) generated by blade rotation. In conventional PWM modulation mode, a step change in fan speed leads to a sudden change in noise. This application achieves noise reduction by adjusting the speed in advance (avoiding speed jumps) and using spread spectrum modulation mode (dispersing narrowband energy). The noise reduction calculation process may include: first determining the noise reduction contribution of smooth speed adjustment; the relationship between fan aerodynamic noise and speed typically follows an empirical formula: ; in, Sound pressure level (dB) For rotational speed, The reference speed is 55, which is derived from the 5th to 6th power law of fan noise (approximately 55 after taking the logarithm).

[0120] In the existing solution, when the power suddenly increases, the fan speed is reduced within 1 second. =3000 rpm step to =6000 rpm, noise increase is: ; This application predicts the temperature rise 5 seconds in advance using a fusion prediction model, linearly increasing the rotation speed within those 5 seconds. While the peak speed remains at 6000 rpm, the human ear's integration time is approximately 200 ms, making the short-term peak speed imperceptible. More importantly, due to the advance heat dissipation, the actual required maximum rotation speed can be reduced by approximately 10% (6000 → 5400 rpm). Therefore, this noise reduction is: .

[0121] Next, we calculate the noise reduction contribution of the spread spectrum modulation mode. With a fixed carrier frequency of 20kHz in conventional PWM, a 20kHz fundamental frequency and its harmonics are generated, concentrating energy within an extremely narrow frequency band (one FFT bin width). Since the human ear is sensitive to narrowband pure tones, according to equal loudness curves, narrowband noise is louder than broadband noise at the same sound pressure level. The spread spectrum modulation mode randomly varies the carrier frequency between 20–25kHz, resulting in a uniform distribution of spectral energy. =5000 Hz bandwidth.

[0122] Assuming the original narrowband peak energy is Then the energy density (energy per unit frequency) is The energy density after spread spectrum is The energy density ratio is: ; This means the peak energy density is reduced to 1 / 213 of its original value. While human loudness perception is approximately proportional to the logarithm of energy density, in psychoacoustics, a decrease in sharpness (SS) more directly reflects a reduction in subjective discomfort. According to the Zwicker sharpness model: ; in, Characteristic loudness (related to energy density). This is a weighting function. A significant decrease in energy density leads to a substantial drop in the characteristic loudness peak, resulting in a sharp reduction in the integral result. Simulations show that spread spectrum modulation can reduce sharpness by 40%–60%. For every 10% reduction in sharpness, the perceived subjective loudness decreases by approximately 2–3 dB(A). Taking a 50% reduction in sharpness, the perceived subjective loudness decreases by approximately 0.5 × 20 = 10 dB (taking the median value). More directly, according to the ISO 532 standard, broadband noise has a perceived loudness approximately 5–10 dB lower than narrowband noise at the same sound pressure level. In this example, the noise reduction contribution of spread spectrum modulation is taken as 8 dB. The total noise reduction can be calculated in the following ways: ; The total noise reduction ranges from 8 to 12 dB, which meets the noise reduction requirements.

[0123] The derivation process for a sharpness reduction of more than 30% can include: in sharpness calculations, energy dispersion affects the characteristic loudness. The peak value drops significantly. Let the characteristic loudness corresponding to the original narrowband peak value be... The characteristic loudness reduction factor of the same frequency band after spread spectrum is =1 / 213, but considering the human ear's frequency band grouping (Bark scale), the energy dispersion may span multiple Bark bands, and the integration effect reduces sharpness. A simplified model can be used: the original narrow band is located within a certain Bark band, and the sharpness mainly comes from that band. After spreading, the energy is evenly distributed across multiple Bark bands, and the characteristic loudness of each band is reduced to 1 / K of the original (K is the number of Bark bands covered; here, approximately 5kHz covers about 10 Bark bands, K≈10). Therefore, the integral value is reduced to approximately 1 / K of the original, and the sharpness is reduced by approximately 1 / K. 1 / K = 90%, but in reality, due to the weighting of g(z), it is conservatively estimated to be reduced by 30% to 50%. Therefore, ≥30% is reasonable.

[0124] The derivation process of the optimization effects of heat dissipation and charging efficiency in this application may include: The derivation for increasing charging time by less than 3 minutes can include: taking a typical 60kW DC fast charging station as an example, with a battery capacity of 60kWh, the required charge from 20% to 80% SOC is Q = 60 × 0.6 = 36 kWh. Ideal full-power charging time: ; Under the worst-case scenario, with an ambient temperature of 40°C and an initial charging power of 60kW, the predicted temperature determined by the fusion prediction model will approach the 85°C threshold, enabling flexible charging limitation with a power coordination coefficient α = 0.85 (i.e., power reduced to 51kW). Assuming that charging limitation lasts throughout the entire charging process (in reality, it may only occur for a portion of the time), the charging time is as follows: ; Extended by 42.35 36 = 6.35 minutes. However, this assumption is too conservative because: charging limitation only activates when the temperature approaches the threshold, and power may recover as the temperature drops. Since the rate of temperature rise is slowed by adjusting the speed in advance, the actual charging limitation time is usually <30%. A more accurate estimate can be achieved based on a thermal model: power module thermal capacity CC, thermal resistance RR, and temperature rise. When the temperature reaches 80°C, charging is limited, reducing the power to 51kW. The temperature then stabilizes at 82°C and no longer rises. Assuming the charging curve is in a constant power range, the average power is 60 × (1... 0.15 × percentage of limited charging time). According to simulation, the percentage of limited charging time is about 20%, so the average power is 60 × (1 0.15 × 0.2) = 60 × 0.97 = 58.2 kW, charging time 36 / 58.2 = 0.618 h = 37.1 minutes, only an extension of 1.1 minutes. Taking the average of multiple operating conditions, the extension is less than 3 minutes.

[0125] The derivation of extending wind turbine life by 20%-30% can be summarized as follows: Wind turbine life is mainly affected by bearing wear and motor winding fatigue, and is related to speed and the number of start-stop cycles. According to the L10 bearing life formula (rated life): ; Where P is the equivalent dynamic load, which is approximately positively correlated with the square of the rotational speed. Life index (ball bearing) =3). Therefore, lifespan is inversely proportional to the 6th power of the rotational speed: ; In existing technologies, frequent step speed jumps to the maximum speed of 6000 rpm are common. This application, through predictive speed regulation, reduces the average speed by 10% (from 4000 rpm to 3600 rpm), resulting in a lifespan extension factor: ; This translates to an 87% lifespan extension, far exceeding the 20%-30% limit. However, actual lifespan is also affected by start-stop shocks; step speed regulation generates impact loads, while smooth speed regulation reduces these impacts. Considering both factors, a conservative estimate of 20%-30% is taken.

[0126] The technical effects of improving equipment reliability: The derivation process for fault early warning 7-15 days in advance with an accuracy rate >90% is as follows: Early warning principle: By continuously monitoring the fan's acoustic signature through a near-field microphone, characteristic frequencies of bearing faults (such as the outer race fault frequency) are extracted. An alert is issued when the characteristic frequency amplitude exceeds the baseline threshold. Characteristic frequency calculation: ; Taking a certain type of fan as an example: n=3000 rpm, Nb=8, d=7 mm, D=50 mm, =0, then ; Baseline establishment: Under healthy conditions, data is collected continuously for 30 days, and the mean μ and standard deviation σ of the amplitude at this frequency are calculated daily. A baseline threshold of μ + 3σ (99.7% upper confidence limit) is taken. If the amplitude exceeds this threshold on a certain day and continues for 72 hours, an alert is triggered.

[0127] Lead time derivation: In the early stages of bearing wear, the characteristic amplitude increases slowly. According to accelerated life test data, the time for the amplitude to rise from the baseline to the failure threshold (typically twice the baseline) is approximately 10–20 days. Taking the failure threshold as 2μ, the amplitude growth curve can be fitted as an exponential type: .

[0128] when hour, .generally Therefore, 1 + 0.15 = 1.15. .like To achieve If the time is 15 days, then , Substituting... However, early warning systems are actually more sensitive and employ [methods]. It can predict wear patterns much earlier. Furthermore, bearing wear characteristics are not only reflected in amplitude but also in increased sideband intensity; these combined characteristics can predict wear patterns 7-15 days in advance. Accuracy is verified through historical fault data, with an accuracy rate >90%. .

[0129] The derivation process of improving availability to 99.5%: Availability is defined as: ; Where MTBF is the Mean Time Between Failures and MTTR is the Mean Time To Repair. The original system experienced downtime due to wind turbine failure, resulting in a low MTBF. Assuming the original system had an MTBF of 2000 hours and an MTTR of 24 hours, its availability was 2000 / (2000+24) = 0.988 = 98.8%. This proposed solution, through predictive maintenance, replaces the wind turbines before failures occur, transforming turbine failures into planned maintenance. This increases the MTBF to 4000 hours (doubling the lifespan), while maintaining the MTTR at 24 hours (but shortening it by allowing for earlier replacement of parts). Therefore, the availability is 4000 / (4000+24) = 0.994 = 99.4%, close to 99.5%.

[0130] The technical effects of environmental adaptability: The derivation process of setting the dynamic target noise threshold: The system determines the target noise threshold based on the ambient noise. Dynamically adjust target noise based on time period (day / night) Let the maximum contribution be allowed during the day. =15 dB, nighttime t=5 dB. According to the sound pressure level superposition formula: ; To ensure that the total noise does not exceed the ambient background noise plus the allowable increment, the following requirements are specified. The upper limit of pile noise is obtained: For example, during the day =50dB, =15, then Difference Taking the logarithm, we get This means the noise level from the pile body can reach as high as 65 dB. At night... =30dB, =5, then The difference is 2162. Therefore, the noise level of the pile body at night needs to be strictly controlled to around 33dB, and the system is set accordingly. This is achieved by adjusting the fan. This dynamic threshold ensures that the contribution of the charging pile noise to the total noise is kept within an acceptable range under different environments, while maximizing charging efficiency (higher noise is allowed during the day).

[0131] In the several embodiments provided in this application, it should be understood that the disclosed device can also be implemented in other ways. The device embodiments described above are merely illustrative; for example, the block diagrams in the accompanying drawings illustrate the possible architecture, functions, and operations of the device according to various embodiments of this application. In this regard, each block in the block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram, and combinations of block diagrams, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0132] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0133] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0134] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

[0136] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A control method, characterized in that, The method includes: Multiple types of sensor data are acquired through multimodal sensors; Preprocessing of the various sensor data yields multimodal time-series data; The multimodal time-series data is processed based on the fusion prediction model set inside the charging device to obtain control commands for adjusting the operating parameters of the charging device; wherein, the operating parameters include: charging parameters and cooling parameters of the cooling devices inside the charging device; The charging device and / or the cooling device are controlled based on the control commands.

2. The method according to claim 1, characterized in that, The preprocessing of various sensor data to obtain multimodal time-series data includes: Time alignment processing is performed on the various sensor data to obtain synchronized data; The audio data in the synchronization data is subjected to spectrum processing to obtain spectrum data; The multimodal time series data is obtained by normalizing the spectrum data and the synchronization data.

3. The method according to claim 1, characterized in that, in, The fusion prediction model includes: an input layer, a convolutional feature extraction layer, and a temporal prediction layer; The fusion prediction model based on the internal settings of the charging device processes the multimodal time-series data to obtain control commands for adjusting the operating parameters of the charging device, including: The multimodal time-series data is coupled through the input layer to obtain multidimensional data; The multi-dimensional data is reduced in dimensionality through the convolutional feature extraction layer to obtain feature vectors. The timing prediction layer dynamically predicts based on the feature vector and time relationship to obtain the control command for adjusting the operating parameters of the charging device.

4. The method according to any one of claims 1-3, characterized in that, in, The control commands include at least one of the following: a first command to control the noise upper limit threshold, a second command to optimize the duty cycle of the pulse width modulation of the cooling device, a third command to adjust the pulse width modulation mode, and a fourth command to adjust the charging power of the charging device.

5. The method according to claim 4, characterized in that, The control of the charging device and / or the cooling device based on the control command includes: If the current temperature is determined to be less than or equal to the temperature threshold and the predicted noise is less than or equal to the noise upper limit threshold, the power coordination coefficient of the charging device shall be maintained or increased to the maximum value. If the current temperature is determined to be greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the predicted noise is greater than the noise upper limit threshold, the charging power of the charging device is reduced. If the current temperature is determined to be greater than or equal to the warning temperature threshold corresponding to the temperature threshold, and the charging power of the charging device reaches the minimum allowable power, the duty cycle of the pulse width modulation is increased, and the cooling power of the cooling device is maintained. If the predicted temperature is determined to be greater than or equal to the temperature threshold, the power coordination coefficient of the charging device is determined based on the difference between the predicted temperature and the temperature threshold. The pulse width modulation carrier frequency of the cooling device is spread spectrum based on the pulse width modulation mode.

6. The method according to any one of claims 1-3, characterized in that, in, The multimodal sensor includes: a voiceprint sensor, a noise sensor, a temperature sensor, and an electrical sensor; the various sensing data include: voiceprint data, noise data, temperature data, and electrical data.

7. The method according to claim 6, characterized in that, The method further includes: A voiceprint database was established based on historical voiceprint data; Based on the fusion prediction model, the voiceprint data and the voiceprint database are processed to obtain error data; Based on the error data, the operating state of the cooling device is adjusted.

8. The method according to claim 7, characterized in that, Adjusting the operating state of the cooling device based on the error data includes: If the error data is greater than or equal to a preset error threshold and the duration is greater than or equal to a preset time threshold, it is determined that the cooling device has an abnormal operating speed range. The rotational speed of the cooling device is reduced, and an adjustment command is determined that does not include the specified rotational speed range.

9. A charging device, characterized in that, The charging device includes: a multimodal sensor, a controller, and a cooling device; The multimodal sensor and the cooling device are respectively connected to the controller; The multimodal sensor is used to collect various types of sensor data; The controller is used to preprocess various types of sensor data to obtain multimodal time-series data; process the multimodal time-series data based on a set fusion prediction model to obtain control instructions for adjusting the operating parameters of the charging device; wherein, the operating parameters include: charging parameters and cooling parameters of the cooling device; and control the charging device and / or the cooling device based on the control instructions.

10. The charging device according to claim 9, characterized in that, in, The multimodal sensor includes: an acoustic signature sensor, a noise sensor, a temperature sensor, and an electrical sensor; the various sensing data include: acoustic signature data, noise data, temperature data, and electrical data; The acoustic signature sensor is located at the detection position of the cooling device; The noise sensor is mounted on the housing of the charging device; The temperature sensor is located inside the charging device; The electrical sensor is installed in the charging circuit of the charging device.

Images