A multi-dimensional information fusion perception wireless charging foreign matter detection method and device

Abstract

The application provides a wireless charging foreign matter detection method and device based on multi-dimensional information fusion perception. The method comprises the following steps: acquiring power data of a wireless charging transmitting end and an original infrared thermal image of a wireless charging area; processing the power data and the original infrared thermal image to obtain a thermal anomaly probability and a deviation probability; the deviation probability represents a deviation degree of a ratio between average radiation brightness of the power data and the infrared thermal image relative to an expected ratio; the thermal anomaly probability represents a deviation degree of a space temperature of the wireless charging area relative to a space temperature of the wireless charging area when there is no foreign matter; calculating a foreign matter detection probability based on the thermal anomaly probability and the deviation probability; adjusting a wireless charging mode according to the foreign matter detection probability; the wireless charging mode comprises at least one of the following: a safety mode, an alert mode and a block mode. The above scheme can realize spatial and power consumption dual-dimensional perception and fusion during foreign matter detection, and improve the accuracy of wireless charging foreign matter detection.

Description

Technical Field

[0001] This application relates to the field of foreign object detection technology, specifically to a wireless charging foreign object detection method and device based on multi-dimensional information fusion sensing. Background Technology

[0002] With the development of wireless charging technology in consumer electronics, smart homes and electric vehicles, a reliable and fast foreign object detection mechanism is key to ensuring the safety of wireless charging.

[0003] In existing technologies, foreign object detection methods for wireless charging mainly include electrical methods based on coil parameter detection and thermal methods based on temperature sensing. Electrical methods based on coil parameter detection rely on electrical parameters (such as coil impedance and quality factor). This method only focuses on overall power consumption changes, is insensitive to small-sized metallic foreign objects, and is easily affected by coil offset and load changes. Thermal methods based on temperature sensing use infrared thermal imaging with temperature sensors. This method identifies localized overheating areas by detecting hot spots in space. While it can detect small-sized metallic foreign objects, it suffers from problems such as delayed heat diffusion, susceptibility to ambient temperature interference, and the potential for the infrared thermal image to be confused with normal charging heat generation. Summary of the Invention

[0004] This application aims to provide a method and device for detecting foreign objects in wireless charging using multi-dimensional information fusion sensing, which can perform spatial and power consumption sensing and fusion in foreign object detection, thereby improving the accuracy of foreign object detection in wireless charging.

[0005] The technical solution of this application is implemented as follows: In a first aspect, this application provides a method for detecting foreign objects in wireless charging based on multi-dimensional information fusion sensing. The method includes: acquiring power data from a wireless charging transmitter and an original infrared thermal image of the wireless charging area; processing the power data and the original infrared thermal image to obtain a thermal anomaly probability and a shift probability of the infrared thermal image; wherein the shift probability characterizes the degree of deviation between the ratio of the power data and the average radiance of the infrared thermal image relative to a desired ratio; the desired ratio is determined based on desired power data and desired radiance, where the desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects; the thermal anomaly probability characterizes the degree of deviation between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects; calculating a foreign object detection probability based on the thermal anomaly probability and the shift probability; and adjusting the wireless charging mode according to the foreign object detection probability; wherein the wireless charging mode includes at least one of the following: a safety mode, a warning mode, and a blocking mode.

[0006] According to the method of the first aspect of this application, power data and raw infrared thermal images are processed to obtain thermal anomaly probability and offset probability, including: converting the raw infrared thermal image to obtain a thermal energy distribution matrix; obtaining thermal anomaly probability based on the thermal energy distribution matrix; obtaining average radiance based on the thermal energy distribution matrix; and obtaining offset probability based on power data and average radiance.

[0007] According to the method of the first aspect of this application, the original infrared thermal image is converted to obtain a thermal energy distribution matrix, including: preprocessing the original infrared thermal image to obtain an infrared thermal image; and converting the pixel values ​​in the infrared thermal image into temperature values ​​to obtain a thermal energy distribution matrix.

[0008] According to the method of the first aspect of this application, obtaining the thermal anomaly probability based on the thermal energy distribution matrix includes: obtaining spatial discontinuity features of the infrared thermal image based on the thermal energy distribution matrix; wherein the spatial discontinuity features include local hotspot features and temperature gradient abrupt change features; and obtaining the thermal anomaly probability based on the spatial discontinuity features. Obtaining the average radiance based on the thermal energy distribution matrix includes: calculating the statistical average of all elements in the thermal energy distribution matrix, and using the statistical average as the average radiance.

[0009] According to the method of the first aspect of this application, the method further includes: inputting power data and thermal energy distribution matrix into a composite neural network to perform foreign object detection and obtain a foreign object detection probability; wherein, the composite neural network includes a thermal anomaly analysis subnetwork, a displacement analysis subnetwork, and a fusion decision subnetwork; the thermal anomaly analysis subnetwork is used to take the thermal energy distribution matrix as input and output the thermal anomaly probability; the displacement analysis subnetwork is used to take the power data and the average radiance of the infrared thermal image as input and output the displacement probability; the fusion decision subnetwork is used to take the thermal anomaly probability and the displacement probability as input and output the foreign object detection probability.

[0010] According to the method of the first aspect of this application, the foreign object detection probability is calculated based on the thermal anomaly probability and the offset probability, including: performing nonlinear fusion of the thermal anomaly probability and the offset probability to obtain the foreign object detection probability.

[0011] According to the method of the first aspect of this application, adjusting the wireless charging mode based on the foreign object detection probability includes: controlling the wireless charging system to enter a safe mode when the foreign object detection probability is less than a first threshold; controlling the wireless charging system to enter a warning mode when the foreign object detection probability is greater than or equal to the first threshold and less than a second threshold; and controlling the wireless charging system to enter a blocking mode when the foreign object detection probability is greater than or equal to the second threshold; wherein the first threshold is less than the second threshold.

[0012] Secondly, this application provides a wireless charging foreign object detection device with multi-dimensional information fusion sensing, including: an acquisition module, a processing module and an adjustment module; The acquisition module is used to acquire power data from the wireless charging transmitter and the original infrared thermal image of the wireless charging area. The processing module processes the power data and raw infrared thermal image acquired by the acquisition module to obtain the thermal anomaly probability and offset probability of the infrared thermal image. The offset probability characterizes the degree of deviation between the ratio of the power data and the average radiance of the infrared thermal image relative to a desired ratio. The desired ratio is determined based on desired power data and desired radiance. The desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects. The thermal anomaly probability characterizes the degree of deviation between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects. The foreign object detection probability is calculated based on the thermal anomaly probability and offset probability obtained by the processing module. An adjustment module is used to adjust the wireless charging mode based on the foreign object detection probability calculated by the processing module; wherein the wireless charging mode includes at least one of the following: safety mode, alert mode, and blocking mode.

[0013] Thirdly, this application provides a wireless charging foreign object detection device based on multi-dimensional information fusion sensing, comprising: a processor and a memory, wherein, Memory, used to store computer programs; A processor is used to retrieve and run computer programs from memory to perform methods such as those described in the first aspect.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing executable instructions for causing a processor to perform the method as described in the first aspect.

[0015] This application provides a method, apparatus, and medium for detecting foreign objects in wireless charging based on multi-dimensional information fusion sensing. The method includes: acquiring power data of the wireless charging transmitter and an original infrared thermal image of the wireless charging area; processing the power data and the original infrared thermal image to obtain a thermal anomaly probability and a shift probability; wherein, the shift probability is used to characterize the degree of shift between the ratio of the power data and the average radiance of the infrared thermal image and a desired ratio; the desired ratio is determined based on desired power data and desired radiance, wherein the desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects; the thermal anomaly probability is used to characterize the degree of shift between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects; calculating the foreign object detection probability based on the thermal anomaly probability and the shift probability; adjusting the wireless charging mode according to the foreign object detection probability; wherein, the wireless charging mode includes at least one of the following: a safety mode, a warning mode, and a blocking mode. In the above scheme, since the thermal anomaly probability and offset probability are first obtained based on power data and the original infrared thermal image, and then the foreign object detection probability obtained by fusing the thermal anomaly probability and offset probability is used to adjust the wireless charging mode, it is possible to perceive the impact of foreign objects on the power consumption of wireless charging through power data, and to perceive the local overheating phenomenon of wireless charging caused by foreign objects through the original infrared thermal image. Thus, it is possible to perform spatial and power consumption dual-dimensional perception and fusion during foreign object detection, thereby improving the accuracy of foreign object detection in wireless charging. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application. Obviously, the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0017] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0018] Figure 1 This is an optional flowchart illustrating a multi-dimensional information fusion sensing method for detecting foreign objects in wireless charging, provided in an embodiment of this application. Figure 2 A schematic diagram of the structure of a wireless charging foreign object detection device with multi-dimensional information fusion sensing provided in an embodiment of this application; Figure 3This is a schematic diagram of the structure of a wireless charging foreign object detection device based on multi-dimensional information fusion sensing, provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of this application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0020] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this application is for the purpose of describing embodiments of this application only and is not intended to be limiting of this application.

[0021] In the following description, references to "some embodiments," "this embodiment," "this application embodiment," and examples, etc., describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subset of all possible embodiments and may be combined with each other without conflict.

[0022] If the application documents contain similar descriptions such as "first / second", the following explanation shall be added: In the following description, the terms "first / second / third" are used only to distinguish similar objects and do not represent a specific order of objects. It is understood that "first / second / third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0023] This application provides a method for detecting foreign objects in wireless charging using multi-dimensional information fusion sensing. Figure 1 This is an optional flowchart illustrating a multi-dimensional information fusion sensing-based foreign object detection method for wireless charging, provided in an embodiment of this application. Figure 1 The steps shown are explained.

[0024] S101. Acquire power data from the wireless charging transmitter and the original infrared thermal image of the wireless charging area.

[0025] In some embodiments of this application, the power data is first collected at the wireless charging transmitter via a power sampling circuit before the power data is acquired.

[0026] A wireless charging transmitter is the transmitting end of a wireless charging device that charges other devices. In a wireless charging system, the transmitter is the hardware component responsible for converting electrical energy into an alternating electromagnetic field and transmitting it outwards. It typically includes a power inverter circuit, a transmitting coil, and a power sampling circuit. It is the starting point for energy output.

[0027] Before acquiring the raw infrared thermal image, data is first collected from the wireless charging area using an infrared thermal imaging sensor to obtain the raw infrared thermal image, which represents the full-field data of the wireless charging area. The wireless charging area refers to the spatial range above or around the wireless charging transmitter where an effective electromagnetic field can be formed, allowing for the placement of receiving devices (such as mobile phones or electric vehicle receivers) for energy transfer. The wireless charging area is also the location where foreign objects (such as metal objects) may be heated and generate hot spots. The wireless charging transmitter generates an alternating electromagnetic field that covers the charging area. If there are metal foreign objects within the wireless charging area, they will generate heat due to the eddy current effect. Therefore, the wireless charging transmitter is the source of the thermal effect, and the wireless charging area is the location where the thermal effect occurs.

[0028] In some embodiments of this application, the infrared thermal image is a single-channel infrared thermal image.

[0029] In some embodiments of this application, an infrared thermal imaging sensor is mounted on a device within the wireless charging area, and its imaging area completely covers the area where the wireless charging device is located. The wireless charging area refers to the spatial range above or near the wireless charging transmitter where power can be effectively transmitted; devices within the wireless charging area are the devices being charged. In other words, devices within the wireless charging area are the receiving devices.

[0030] In some embodiments of this application, the sampling period of the power data may be the same as or different from the sampling period of the original infrared thermal image. For example, the sampling period of the power data may be a multiple of the sampling period of the original infrared thermal image.

[0031] In some embodiments of this application, the power data and the infrared image frame sequence of the original infrared thermal image can be time-synchronized using the same clock triggering mechanism to align the power data with the original infrared thermal image on the time scale.

[0032] In some embodiments of this application, after real-time acquisition of power data and raw infrared thermal images, the composite neural network can be periodically triggered based on the acquired power data and raw infrared thermal images to perform calculations for foreign object detection in wireless charging.

[0033] In some embodiments of this application, the sampling frequency of the power data can be 200Hz–1000Hz. Of course, the sampling frequency can also be other frequencies, and this application does not limit this.

[0034] In some embodiments of this application, the sampling frequency of the original infrared thermal image can be 20Hz–50Hz. Of course, the sampling frequency can also be other frequencies, and this application does not limit this.

[0035] In some embodiments of this application, the calling frequency used to trigger the operation of the composite neural network can be 20Hz–50Hz. Of course, the calling frequency can be other frequencies, and this application does not limit this.

[0036] In some embodiments of this application, by acquiring power data and raw infrared thermal images, and subsequently fusing these two types of data to perform foreign object detection from both spatial and power consumption dimensions, this approach can provide complementary perception of evidence of the presence of foreign objects compared to traditional methods that rely on information from a single dimension for foreign object detection.

[0037] S102. Process the power data and the original infrared thermal image to obtain the thermal anomaly probability and offset probability.

[0038] In some embodiments of this application, thermal anomaly probability is used to characterize spatial hot spot features in infrared thermal images.

[0039] In some embodiments of this application, the higher the probability of thermal anomalies, the higher the proportion of heat from the spatial hotspot in the infrared thermal image to the total heat in the infrared thermal image.

[0040] In some embodiments of this application, the offset probability is used to characterize the degree of offset between the ratio of power data and the average radiance of the infrared thermal image relative to the ratio between the power data and the average radiance of the infrared thermal image of the wireless charging transmitter when there are no foreign objects.

[0041] Understandably, small metallic foreign objects, such as metal scraps and aluminum foil, do not show significant temperature increases or exhibit weak thermal radiation in the early stages. Therefore, foreign object detection methods based on spatial hot spots are not sensitive to them. However, the power consumption caused by these small metallic foreign objects in wireless charging actually deviates from the normal value. Therefore, by detecting the deviation between the ratio of power data and the average radiance of the infrared thermal imaging image relative to the ratio when there are no foreign objects, the presence of small metallic foreign objects can be detected in a timely manner, improving the real-time performance of foreign object detection.

[0042] In some embodiments of this application, step S102 can be specifically implemented by the following steps S102a, S102b and S102c: S102a: Convert the original infrared thermal image to obtain the thermal energy distribution matrix; S102b: Obtain the probability of thermal anomalies based on the thermal energy distribution matrix; S102c: The average radiance is obtained based on the thermal energy distribution matrix, and the offset probability is obtained based on the power data and the average radiance.

[0043] In some embodiments of this application, the original infrared thermal image is converted, including converting the pixel values ​​of the original infrared thermal image into temperature values. Specifically, a one-to-one correspondence between pixel grayscale values ​​(e.g., 0–255) and temperature values ​​is established based on a pre-calibrated temperature lookup table. During conversion, the grayscale level of each pixel is read, and its temperature value is directly obtained by looking up the table; if the grayscale level is not directly listed in the table, the corresponding temperature is calculated using linear interpolation.

[0044] In some embodiments of this application, the size of the thermal distribution matrix corresponds to the image size of the original infrared thermal image, and the number of elements in the thermal distribution matrix is ​​equal to the number of pixels in one frame of the original infrared thermal image.

[0045] In some embodiments of this application, average radiance is used to characterize the overall thermal radiation level of the wireless charging area.

[0046] In some embodiments of this application, after obtaining the average radiance, the power data, thermal energy distribution matrix and average radiance can be fused and encapsulated, and the encapsulated data can be used as input samples for a composite neural network.

[0047] For example, assuming the input power is P_in, the thermal energy distribution matrix is ​​T_energy, and the average radiance is L_avg, the input sample of the composite neural network can be {P_in, T_energy, L_avg}, where P_in is a real scalar, T_energy is a thermal energy distribution matrix of size H×W, and L_avg is a real scalar.

[0048] In some embodiments of this application, step S102a can be specifically implemented by the following steps A1 and A2: A1: Preprocess the original infrared thermal image to obtain an infrared thermal image; A2: Convert the pixel values ​​in the infrared thermal image data into temperature values ​​to obtain the thermal energy distribution matrix.

[0049] In some embodiments of this application, preprocessing may include at least one of the following: radiometric calibration and denoising; or other image preprocessing methods, which are not limited in this application.

[0050] Radiometric calibration is used to convert the raw pixel values ​​of an original infrared thermal image into radiance or temperature values ​​with practical physical meaning. Specifically, through a pre-defined calibration formula, the value of each pixel in the image is linearly stretched and corrected to eliminate uneven brightness in the image caused by differences in the hardware manufacturing of the infrared sensor (i.e., non-uniformity correction). Denoising is used to eliminate random noise and abnormal extrema in infrared images. Specifically, blind pixels are first detected and removed from the image, and then interpolated using the average value of their surrounding pixels. Subsequently, a bilateral filtering algorithm is used for spatial smoothing, which filters out high-frequency noise while fully preserving the temperature gradient abrupt changes at the edges of foreign objects, thus avoiding image edge blurring.

[0051] In some embodiments of this application, step S102b can be specifically implemented through steps B1 and B2. Based on this, obtaining the average radiance based on the thermal energy distribution matrix can be achieved through step B3: B1: Spatial discontinuity features are obtained based on the thermal energy distribution matrix. Spatial discontinuity features include local hot spot features and temperature gradient abrupt changes. B2: The probability of thermal anomalies is obtained based on spatial discontinuity characteristics; B3: The statistical average of all elements in the statistical thermal energy distribution matrix, which is used as the average radiance.

[0052] In some embodiments of this application, the thermal energy distribution matrix is ​​input into a thermal anomaly analysis subnetwork of a composite neural network. The subnetwork learns local hotspot features and temperature gradient abrupt change features in the infrared thermal image through multiple convolutional blocks, and outputs a thermal anomaly probability. Thus, the thermal anomaly analysis subnetwork can perceive spatial hot spots in the infrared thermal image of the wireless charging area, thereby outputting a thermal anomaly probability to characterize the spatial hot spot features.

[0053] S103. Calculate the foreign object detection probability based on the thermal anomaly probability and the offset probability.

[0054] In some embodiments of this application, the foreign object detection probability is calculated by nonlinearly fusing the thermal anomaly probability and the offset probability.

[0055] Nonlinear fusion uses a small neural network (multilayer perceptron) to automatically learn the complex combination relationship between two probabilities (thermal anomaly probability and offset probability). Specifically, two probability values, Pr1 (thermal anomaly probability) and Pr2 (offset probability), are input. Pr1 and Pr2 are fed into a fully connected network with one hidden layer (8 neurons), and then pass through the ReLU activation function (introducing nonlinearity). The output is finally passed through a Softmax layer to obtain the final foreign object detection probability.

[0056] S104. Adjust the wireless charging mode according to the probability of foreign object detection.

[0057] In some embodiments of this application, the wireless charging mode includes at least one of the following: safety mode, alert mode, and blocking mode.

[0058] In some embodiments of this application, wireless charging is performed normally when entering safe mode.

[0059] In some embodiments of this application, when entering alert mode, the sampling frequency of power data is reduced and the sampling frequency of the original infrared thermal image is increased.

[0060] In some embodiments of this application, when entering alert mode, the input power is reduced to 50% of the original input power.

[0061] In some embodiments of this application, after acquiring each new frame of raw infrared thermal image, the raw infrared thermal image and the power data collected within the corresponding time window are synchronously encapsulated to form a complete input sample, and a composite neural network is triggered to perform calculations for foreign object detection in wireless charging.

[0062] It is understandable that the above scheme can make the frequency of acquiring a frame of raw infrared thermal image the same as the frequency of triggering a single operation of the composite neural network. Thus, after entering the alert mode, the frequency of calling the composite neural network can be increased by increasing the sampling frequency of the raw infrared thermal image.

[0063] In some embodiments of this application, when entering blocking mode, the inverter power supply is cut off and wireless charging is stopped.

[0064] In some embodiments of this application, when it is determined that the inverter power supply needs to be switched off within 50ms, the inverter power supply is switched off. 50ms is a preferred value after a trade-off, ensuring absolute power disconnection before the metallic foreign object reaches a dangerous temperature while minimizing the risk of malfunction. In actual products, the timeout can be adjusted within the range of 20ms to 1s depending on safety level and cost requirements, but it is recommended not to exceed 1s to ensure compliance with mainstream safety certifications.

[0065] In some embodiments of this application, step S104 may specifically include the following steps S104a, S104b and S104c: S104a: When the probability of foreign object detection is less than the first threshold, control the wireless charging system to enter the safety mode; S104b: When the probability of foreign object detection is greater than or equal to the first threshold and less than the second threshold, control the wireless charging system to enter the alarm mode; S104c: When the probability of foreign object detection is greater than or equal to the second threshold, control the wireless charging system to enter the blocking mode.

[0066] It should be noted that the first threshold is smaller than the second threshold.

[0067] Understandably, by setting a first threshold and a second threshold and comparing them, wireless charging can proceed normally when the probability of foreign object detection is low (the probability of foreign objects in the wireless charging system is small); when the probability of foreign object detection is high (the wireless charging system may contain metal pads with a diameter of about 5mm), the foreign object detection frequency can be increased in a timely manner, thereby shortening the response delay, maintaining real-time vigilance against dangerous foreign objects, ensuring that subsequent risks can be immediately blocked, and avoiding frequent interruptions to normal charging; and when the probability of foreign object detection is very high, charging can be stopped in time to ensure the safety of wireless charging. Through the above threshold settings and the corresponding proposed wireless charging modes, adaptive dynamic monitoring of wireless charging safety can be achieved.

[0068] For example, assuming the first threshold is 0.3 and the second threshold is 0.7: 1) In one cycle of the composite neural network, the probability of thermal anomaly is 0.18, the probability of offset is 0.22, and the probability of foreign object detection is 0.2. Since the probability of foreign object detection is less than the first threshold, the wireless charging system determines that the current mode is safe. 2) In the next cycle of the composite neural network, the probability of thermal anomaly is 0.64, the probability of offset is 0.58, and the probability of foreign object detection is 0.61. Since the first threshold is less than the probability of foreign object detection and less than the second threshold, the wireless charging system enters a warning mode, automatically limiting the input power of the wireless charging system to 50% of the rated power, and simultaneously reducing the sampling frequency of the original infrared thermal image from 50Hz. Increased to 100Hz, the frequency of the composite neural network is increased accordingly to monitor the temperature rise more promptly; 3) If foreign objects continue to exist in subsequent cycles of the composite neural network, the local temperature of the wireless charging area will rise further. At this time, assuming the foreign object detection probability is 0.82, since the foreign object detection probability is greater than the second threshold, the wireless charging system determines that there is a dangerous metal foreign object and immediately sends a shutdown command to the inverter control module within 50ms to immediately cut off the wireless charging energy output, thereby avoiding the safety risk caused by the foreign object overheating.

[0069] It is understandable that by setting a threshold for the probability of detecting foreign objects and adjusting the wireless charging system to different wireless charging modes based on the judgment results, foreign objects can be monitored in a timely and efficient manner, and power can be cut off in time to protect safety when dangerous metal foreign objects are present.

[0070] In some embodiments of this application, the method provided in this application further includes the following step S105: S105. Input the power data and thermal energy distribution matrix into the composite neural network to perform foreign object detection and obtain the foreign object detection probability; wherein, the composite neural network includes a thermal anomaly analysis subnetwork, an offset analysis subnetwork and a fusion decision subnetwork.

[0071] In some embodiments of this application, a thermal anomaly analysis subnetwork is used to take the thermal energy distribution matrix as input and output the thermal anomaly probability.

[0072] In some embodiments of this application, the thermal anomaly analysis subnetwork includes an input layer, a convolutional block, a global average pooling layer, and a fully connected output layer.

[0073] The input layer of the thermal anomaly analysis subnetwork receives the thermal energy distribution matrix, with an input size of (H, W, 1), where H is the image height of the infrared thermal image corresponding to the thermal energy distribution matrix, and W is the image width of the infrared thermal image corresponding to the thermal energy distribution matrix.

[0074] The thermal anomaly analysis subnetwork may include one or more convolutional blocks, each of which includes a two-dimensional convolutional layer, a batch normalization layer, a max pooling layer, and a ReLU activation function.

[0075] Example 1: Assume the thermal anomaly analysis subnetwork includes four convolutional blocks, namely convolutional block a, convolutional block b, convolutional block c, and convolutional block d; where, 1) For convolutional block a, the kernel size of its two-dimensional convolutional layer is 3×3, the number of channels is 16, the stride is 1, the padding is same, and the output size is (H, W, 16). The batch normalization layer has an output size of (H, W, 16), the activation function 1 is the ReLU activation function, and its output size is (H, W, 16). The pooling window size of the max pooling layer is 2×2, the stride is 2, and the output size is (H / 2, W / 2, 16). 2) For convolutional block b, the kernel size of its two-dimensional convolutional layer is 3×3, the number of channels is 32, the stride is 1, the padding is same, and the output size is (H / 2, W / 2, 32). The output size of the batch normalization layer is (H / 2, W / 2, 32), the activation function is ReLU activation function, and its output size is (H / 2, W / 2, 32). The pooling window size of the max pooling layer is 2×2, the stride is 2, and the output size is (H / 4, W / 4, 32). 3) For convolutional block c, the kernel size of its two-dimensional convolutional layer is 3×3, the number of channels is 64, the stride is 1, the padding is same, and the output size is (H / 4, W / 4, 64). The output size of the batch normalization layer is (H / 4, W / 4, 64), the activation function is ReLU activation function, and its output size is (H / 4, W / 4, 64). The pooling window size of the max pooling layer is 2×2, the stride is 2, and the output size is (H / 8, W / 8, 64). 4) For convolutional block d, the kernel size of its two-dimensional convolutional layer is 3×3, the number of channels is 128, the stride is 1, the padding is same, and the output size is (H / 8, W / 8, 128). The output size of the batch normalization layer is (H / 8, W / 8, 128), the activation function is ReLU, and its output size is (H / 8, W / 8, 128).

[0076] Based on this, the global average pooling layer of the thermal anomaly analysis subnetwork can convert the feature map of convolutional block d into a 128-dimensional feature vector with an output size of 128. The fully connected output layer of the thermal anomaly analysis subnetwork has a 128-dimensional input and outputs a single neuron with the sigmoid activation function. Its output size is (1,), and its output thermal anomaly probability P r1 ∈ [0,1].

[0077] After acquiring the original infrared thermal image, the thermal energy distribution matrix corresponding to the original infrared thermal image is input into the thermal anomaly analysis subnetwork, which can learn the local hot spot features and temperature gradient abrupt change features in the infrared thermal image.

[0078] In this way, the thermal anomaly analysis subnetwork can be used to sense the spatial hot spots in the infrared thermal image of the wireless charging area, thereby outputting the thermal anomaly probability to characterize the spatial hot spot features.

[0079] In some embodiments of this application, an offset analysis subnetwork is used to take power data and the average radiance of an infrared thermal image as input and output an offset probability.

[0080] In some embodiments of this application, the offset analysis subnetwork may include an input layer, a fully connected layer, and an output layer.

[0081] In some embodiments of this application, the input size of the input layer of the offset analysis subnetwork is 2, and the input layer can receive a two-dimensional feature vector [P_in, L_avg], where P_in is the power data mentioned above and L_avg is the average radiance mentioned above.

[0082] In some embodiments of this application, the offset analysis subnetwork may have multiple fully connected layers.

[0083] Example 2, assume that the offset analysis subnetwork includes three fully connected layers, namely fully connected layer e, fully connected layer f and fully connected layer g; wherein, 1) the fully connected layer e may include 64 neurons and a ReLU activation function with an output size of 64; 2) The fully connected layer f can include 32 neurons, with the ReLU activation function and an output size of 32; 3) The fully connected layer g can include 16 neurons, with the ReLU activation function and an output size of 16. Based on this, the output layer of the offset analysis subnetwork can include one neuron, with the Sigmoid activation function, an output size of (1,), and an output offset probability P. r2 ∈ [0,1].

[0084] Understandably, by setting the fully connected layers of the offset analysis subnetwork to multiple layers, the input power data and average radiance can be perceived from the perspective of power consumption offset, thereby improving the accuracy of foreign object detection.

[0085] In some embodiments of this application, a fusion decision subnetwork is used to take the thermal anomaly probability and the offset probability as inputs and output the foreign object detection probability.

[0086] In some embodiments of this application, the fusion decision subnetwork may include an input layer, a fully connected layer, and an output layer.

[0087] In some embodiments of this application, the fully connected layers of the fusion decision subnetwork may be one or more.

[0088] In some embodiments of this application, the fusion decision subnetwork can receive a fusion feature vector containing the aforementioned thermal anomaly probability and offset probability.

[0089] In some embodiments of this application, the output layer of the fusion decision subnetwork may include two neurons, one neuron corresponding to the absence of foreign objects and the other neuron corresponding to the presence of foreign objects.

[0090] Example 3, combining Examples 1 and 2, shows that the input layer of the fusion decision subnetwork can receive a fusion feature vector [P]. r1 ,P r2 The fully connected layer of the fusion decision subnetwork can include 8 neurons, with ReLU activation function and an output size of 8. The output layer of the fusion decision subnetwork has 2 neurons corresponding to neurons with foreign objects and neurons without foreign objects, respectively, with Softmax activation function and an output size of 2. The output layer uses the probability corresponding to the neurons with foreign objects as the probability of foreign object detection.

[0091] In some embodiments of this application, after obtaining the thermal anomaly probability and offset probability based on the acquired power data and the original infrared thermal image, the fusion decision sub-network can fuse the thermal anomaly probability and offset probability, thereby integrating spatial hot spot features and power consumption offset features, effectively distinguishing real foreign objects from environmental noise, and improving the robustness of foreign object detection.

[0092] In some embodiments of this application, the composite neural network is trained through the following steps S106a, S106b, and S106c: S106a: Construct an initial training dataset including first power data, first raw infrared thermal image, second power data, and second raw infrared thermal image; In some embodiments of this application, the first power data and the first original infrared thermal image are power data and original infrared thermal images collected without foreign objects, and the second power data and the second original infrared thermal image are power data and original infrared thermal images collected with foreign objects. S106b: Perform data preprocessing and augmentation on the initial training dataset to obtain the training dataset; S106c: Establish the network architecture of the composite neural network. The input of the composite neural network is the training dataset, and the output of the composite neural network is the foreign object detection probability. Train the composite neural network.

[0093] In some embodiments of this application, the first power data includes power data collected under multiple input powers in the absence of foreign objects, and the first raw infrared thermal image includes raw infrared thermal images collected under multiple input powers. The multiple input powers may include at least two of the following: 10W, 15W, and 20W. For each input power, its corresponding first raw infrared thermal image corresponds to multiple infrared thermal image frames, which cover different locations and different shooting angles of the wireless charging area.

[0094] In some embodiments of this application, the foreign object tags on the first power data and the first raw infrared thermal image are marked as free of foreign objects.

[0095] In some embodiments of this application, the first power data and the first raw infrared thermal image also include power data and raw infrared thermal images collected when there is heat source interference in the wireless charging area but no metal foreign objects.

[0096] For example, the heat source interference could be interference from a hand approaching the wireless charging area or interference from other heat-generating devices.

[0097] In some embodiments of this application, the second power data includes power data collected under multiple input powers in the presence of foreign objects, and the second raw infrared thermal image includes raw infrared thermal images collected under multiple input powers. The multiple input powers may include at least two of the following: 10W, 15W, and 20W. For each input power, the corresponding second raw infrared thermal image corresponds to multiple infrared thermal image frames, which cover different locations within the wireless charging area.

[0098] In some embodiments of this application, different foreign object samples are placed in the wireless charging area in the presence of foreign objects. These foreign object samples may include metallic and non-metallic foreign objects. For each foreign object sample, its corresponding second power data and second raw infrared thermal image are repeatedly acquired at multiple input power levels.

[0099] In some embodiments of this application, the metallic foreign object may include metallic foreign objects of different materials. The material of the metallic foreign object may include one or more of the following: iron, copper, aluminum, stainless steel, or other materials; this application embodiment does not limit this. The metallic foreign object may include metallic foreign objects of different sizes. The size of the metallic foreign object may include one or more of the following: 1mm×1mm, 3mm×3mm, 5mm×5mm, 10mm×10mm. The metallic foreign object may include metallic foreign objects of different shapes. The shape of the metallic foreign object may include one or more of the following: sheet-like, block-like, strip-like; this application embodiment does not limit this.

[0100] In some embodiments of this application, the foreign object tags of the second power data and the second raw infrared thermal image are marked as having foreign objects.

[0101] In some embodiments of this application, step S106b can be specifically implemented by the following steps C1, C2, and C3: C1: Convert the pixel values ​​of the first original infrared thermal image into temperature values ​​to obtain the first thermal energy distribution matrix; convert the pixel values ​​of the second original infrared thermal image into temperature values ​​to obtain the second thermal energy distribution matrix. C2: Apply random geometric transformations to the first original infrared thermal image belonging to different image frames and the second original infrared thermal image belonging to different image frames, while keeping the corresponding first power data, second power data, first average radiance, and second average radiance unchanged. C3: Normalize the pixel values ​​of the thermal energy distribution matrix to the [0,1] interval, and standardize the first power data, the second power data, the first average radiance, and the second average radiance.

[0102] In this application, the following commonly used methods can be used to standardize the power data and average radiance: Z-score normalization and Min-Max normalization.

[0103] In some embodiments of this application, random geometric transformations may include at least one of the following: rotation ±10°, horizontal / vertical flipping, random cropping and scaling back to the original size, or other geometric transformations. This application does not limit these transformations.

[0104] In some embodiments of this application, the calculation process of the first thermal energy distribution matrix and the second thermal energy distribution matrix can be found in the detailed description of the calculation of the thermal energy distribution matrix above. To avoid repetition, it will not be repeated here.

[0105] In some embodiments of this application, the first average radiance is the average radiance of the first thermal energy distribution matrix, and the second average radiance is the average radiance of the second thermal energy distribution matrix. The calculation process of the first average radiance and the second average radiance can be found in the detailed description of the calculation of average radiance above. To avoid repetition, it will not be repeated here.

[0106] In some embodiments of this application, the training dataset is divided into multiple subsets, including a training set, a validation set, and a test set. The training set is used for model parameter learning, the validation set is used for hyperparameter tuning, early stopping judgment, and model selection, and the test set is used for performance evaluation.

[0107] In some embodiments of this application, the proportion of data from multiple subsets in the training dataset can be divided according to the subset type. In one example, the training set accounts for 70% of the data, the validation set accounts for 15%, and the test set accounts for 15%.

[0108] In some embodiments of this application, the training dataset includes data collected from different charging devices, and for each subset of the training dataset, the data covers all charging devices corresponding to the training dataset.

[0109] In some embodiments of this application, the network architecture of the composite neural network includes three sub-networks: a thermal anomaly analysis sub-network, an offset analysis sub-network, and a fusion decision sub-network. It should be noted that the inputs and outputs of these three sub-networks can be found in the above descriptions of the thermal anomaly analysis sub-network, offset analysis sub-network, and fusion decision sub-network; to avoid repetition, they will not be repeated here.

[0110] In some embodiments of this application, the above-mentioned thermal anomaly analysis subnetwork is a two-dimensional convolutional neural network (CNN), the above-mentioned offset analysis subnetwork belongs to a multi-layer perceptron (MLP), and the above-mentioned fusion decision subnetwork belongs to an MLP.

[0111] In some embodiments of this application, the composite neural network is trained on a binary classification task. The output layer consists of two neurons, and its output consists of two neurons. The two neurons correspond to the foreign object categories of "with foreign object" and "without foreign object". After Softmax, the composite neural network takes the probability corresponding to the foreign object category of "with foreign object" as the foreign object detection probability.

[0112] In some embodiments of this application, the training objective of the composite neural network is to minimize the difference between the foreign object category and the foreign object label corresponding to the foreign object detection probability.

[0113] In some embodiments of this application, the loss function of the composite neural network adopts the binary cross-entropy loss function, the expression of which is shown in Formula 1: L = -[y·log(G) + (1-y)·log(1-G)] (Formula 1) Where y is the foreign object label (0 indicates no foreign object, 1 indicates foreign object), and G is the foreign object detection probability output by the composite neural network.

[0114] In some embodiments of this application, the training process of the composite neural network includes: 1) Organize training samples A batch of training samples is drawn from the training dataset, with each training sample being a triplet. In each iteration, a batch of training samples (batch size set to 32) is randomly drawn from the training set and input into the composite neural network.

[0115] For example, a training sample can be represented as {T_energy0, P_in0, L_avg0, label}; where T_energy0 is the normalized first thermal energy distribution matrix or the second thermal energy distribution matrix, P_in0 is the normalized first power data or the second power data, L_avg0 is the normalized first average radiance or the second average radiance, and label is the foreign object label.

[0116] 2) Optimizer and Learning Rate Scheduling The AdamW optimizer was selected, with an initial learning rate of 1e-3 and weight decay set to 1e-4. A cosine annealing strategy was used for learning rate scheduling, with a period of 30 epochs and a minimum learning rate of 1e-6. After each epoch, the learning rate was reset after decaying to its minimum value using a cosine function.

[0117] 3) Entering the composite neural network process The process consists of two phases: In Phase 1, only the offset analysis subnetwork and the fusion decision subnetwork are trained. To accelerate convergence and stabilize the initial training, all convolutional layers of the hot anomaly analysis subnetwork are frozen, and only the offset analysis subnetwork and the fusion decision subnetwork are trained. The learning rate is set to 1e-3, and the training is performed for 10 epochs to allow these branches to quickly adapt to the task. In Phase 2, all network layers are unfrozen, and the entire network is fine-tuned end-to-end using a smaller learning rate of 1e-4 until the early stopping condition is triggered.

[0118] 4) Early Termination Mechanism and Model Selection After each epoch, the classification accuracy is evaluated on the validation set. If the validation accuracy does not improve for 20 consecutive epochs, training is stopped and the model with the highest validation accuracy is rolled back as the final model of the composite neural network.

[0119] This application provides a multi-dimensional information fusion sensing method for detecting foreign objects in wireless charging. First, power data from the wireless charging transmitter and the original infrared thermal image of the wireless charging area are acquired. The power data and the original infrared thermal image are processed to obtain a thermal anomaly probability and a shift probability. The shift probability characterizes the degree of deviation between the ratio of the power data and the average radiance of the infrared thermal image relative to a desired ratio. The desired ratio is determined based on desired power data and desired radiance. The desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects. The thermal anomaly probability characterizes the degree of deviation between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects. The foreign object detection probability is calculated based on the thermal anomaly probability and the shift probability. The wireless charging mode is adjusted according to the foreign object detection probability. The wireless charging mode includes at least one of the following: a safety mode, a warning mode, and a blocking mode. In the above scheme, since the thermal anomaly probability and offset probability are first obtained based on power data and the original infrared thermal image, and then the foreign object detection probability obtained by fusing the thermal anomaly probability and offset probability is used to adjust the wireless charging mode, it is possible to perceive the impact of foreign objects on the power consumption of wireless charging through power data, and to perceive the local overheating phenomenon of wireless charging caused by foreign objects through the original infrared thermal image. Thus, it is possible to perform spatial and power consumption dual-dimensional perception and fusion during foreign object detection, thereby improving the accuracy of foreign object detection in wireless charging.

[0120] Based on the multi-dimensional information fusion sensing method for detecting foreign objects in wireless charging according to the above embodiments, this application also provides a multi-dimensional information fusion sensing device for detecting foreign objects in wireless charging, such as... Figure 2 As shown, Figure 2 This is a schematic diagram of the structure of a wireless charging foreign object detection device with multi-dimensional information fusion sensing provided in an embodiment of this application. The wireless charging foreign object detection device 2 with multi-dimensional information fusion sensing includes: an acquisition module 201, a processing module 202, and an adjustment module 203.

[0121] The acquisition module 201 is used to acquire power data of the wireless charging transmitter and the original infrared thermal image of the wireless charging area; Processing module 202 is used to process power data and raw infrared thermal images to obtain thermal anomaly probability and offset probability. The offset probability characterizes the degree of deviation between the ratio of power data and the average radiance of the infrared thermal image relative to a desired ratio. The desired ratio is determined based on desired power data and desired radiance. The desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects. The thermal anomaly probability characterizes the degree of deviation between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects. The foreign object detection probability is calculated based on the thermal anomaly probability and the offset probability. The adjustment module 203 is used to adjust the wireless charging mode according to the foreign object detection probability; wherein the wireless charging mode includes at least one of the following: safety mode, alert mode and blocking mode.

[0122] In some embodiments of this application, the processing module 202 is used to convert the original infrared thermal image to obtain a thermal energy distribution matrix; obtain the thermal anomaly probability based on the thermal energy distribution matrix; obtain the average radiance based on the thermal energy distribution matrix; and obtain the offset probability based on the power data and the average radiance.

[0123] In some embodiments of this application, the processing module 202 is further configured to preprocess the original infrared thermal image to obtain an infrared thermal image; and convert the pixel values ​​in the infrared thermal image into temperature values ​​to obtain a thermal energy distribution matrix.

[0124] In some embodiments of this application, the processing module 202 is further configured to obtain spatial discontinuity features based on the thermal energy distribution matrix; wherein, the spatial discontinuity features include local hot spot features and temperature gradient abrupt change features; obtain thermal anomaly probability based on the spatial discontinuity features; calculate the statistical average of all elements in the above thermal energy distribution matrix, and use the above statistical average as the above average radiance.

[0125] In some embodiments of this application, the wireless charging foreign object detection device with multi-dimensional information fusion sensing provided in this application further includes an input module 204; wherein... The input module 204 is used to input power data and thermal energy distribution matrix into the composite neural network for foreign object detection and to obtain the foreign object detection probability. The composite neural network includes a thermal anomaly analysis subnetwork, a displacement analysis subnetwork, and a fusion decision subnetwork. The thermal anomaly analysis subnetwork takes the thermal energy distribution matrix as input and outputs the thermal anomaly probability. The displacement analysis subnetwork takes the power data and the average radiance of the infrared thermal image as input and outputs the displacement probability. The fusion decision subnetwork takes the thermal anomaly probability and the displacement probability as input and outputs the foreign object detection probability.

[0126] In some embodiments of this application, the processing module 202 is further configured to perform nonlinear fusion of the thermal anomaly probability and the offset probability to obtain the foreign object detection probability.

[0127] In some embodiments of this application, the adjustment module 203 is further configured to control the wireless charging system to enter a safe mode when the foreign object detection probability obtained by the processing module 202 is less than a first threshold; control the wireless charging system to enter a warning mode when the foreign object detection probability obtained by the processing module 202 is greater than or equal to the first threshold and less than a second threshold; and control the wireless charging system to enter a blocking mode when the foreign object detection probability obtained by the processing module 202 is greater than or equal to the second threshold; wherein the first threshold is less than the second threshold.

[0128] Based on the multi-dimensional information fusion sensing method for detecting foreign objects in wireless charging based on the above embodiments, this application also provides a multi-dimensional information fusion sensing device for detecting foreign objects in wireless charging, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of a wireless charging foreign object detection device based on multi-dimensional information fusion sensing, provided in an embodiment of this application. The device 3 includes a processor 301 and a memory 302. The memory 302 stores a computer program; the processor 301 retrieves and runs the computer program from the memory to execute a wireless charging foreign object detection method based on multi-dimensional information fusion sensing as described in the above embodiment.

[0129] In the embodiments of this application, the processor 301 described above can be at least one of the following: Application-Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that for different devices, the electronic device used to implement the above processor function can also be other types, and the embodiments of this application do not specifically limit it.

[0130] This application provides a computer-readable storage medium storing a computer program for implementing a wireless charging foreign object detection method based on multi-dimensional information fusion sensing as described in any of the above embodiments when executed by a processor.

[0131] For example, the program instructions corresponding to the wireless charging foreign object detection method of multi-dimensional information fusion sensing in this embodiment can be stored on storage media such as optical discs, hard disks, and USB flash drives. When the program instructions corresponding to the wireless charging foreign object detection method of multi-dimensional information fusion sensing in the storage medium are read or executed by an electronic device, the wireless charging foreign object detection of multi-dimensional information fusion sensing as described in any of the above embodiments can be realized.

[0132] Furthermore, in the embodiments of this application, the functional modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0133] If the integrated unit is implemented as a software functional module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to the prior art, or all or part 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.) or processor to execute all or part of the steps of the method of this embodiment. 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] It should be understood that the phrases "one embodiment," "an embodiment," or "some embodiments" mentioned throughout the specification mean that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment," "in one embodiment," or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the embodiments in this application are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. The descriptions of the various embodiments above tend to emphasize the differences between the various embodiments; their similarities or commonalities can be referred to mutually, and for the sake of brevity, these will not be repeated here.

[0135] The modules described above as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0136] In addition, each functional module in the various embodiments of this application can be integrated into one processing unit, or each module can be a separate unit, or two or more modules can be integrated into one unit; the integrated modules can be implemented in hardware or in the form of hardware plus software functional units.

[0137] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0138] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0139] The features disclosed in the several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0140] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0141] The above description is merely an implementation method of the embodiments of this application, but the protection scope of the embodiments 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 protection scope of this application.

Claims

1. A method for detecting foreign objects in wireless charging based on multi-dimensional information fusion sensing, characterized in that, The method includes: Acquire power data from the wireless charging transmitter and the original infrared thermal image of the wireless charging area; wherein, the original infrared thermal image is full-field data of the wireless charging area; The power data and the original infrared thermal image are processed to obtain a thermal anomaly probability and a shift probability. The shift probability characterizes the degree of deviation between the ratio of the power data and the average radiance of the infrared thermal image and a desired ratio. The desired ratio is determined based on desired power data and desired radiance, where the desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects. The thermal anomaly probability characterizes the degree of deviation between the spatial temperature of the infrared thermal image of the wireless charging area and the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects. The probability of foreign object detection is calculated based on the thermal anomaly probability and the offset probability. The wireless charging mode is adjusted according to the foreign object detection probability; wherein the wireless charging mode includes at least one of the following: safety mode, alert mode, and blocking mode; The step of processing the power data and the original infrared thermal image to obtain the thermal anomaly probability and the offset probability includes: The original infrared thermal image is converted to obtain a thermal energy distribution matrix; The probability of thermal anomaly is obtained based on the aforementioned thermal energy distribution matrix; The average radiance is obtained based on the thermal energy distribution matrix, and the offset probability is obtained based on the power data and the average radiance. The method further includes: The power data and the thermal energy distribution matrix are input into a composite neural network for foreign object detection to obtain the foreign object detection probability; wherein, the composite neural network includes a thermal anomaly analysis subnetwork, an offset analysis subnetwork, and a fusion decision subnetwork; The thermal anomaly analysis subnetwork is used to take the thermal energy distribution matrix as input and output the thermal anomaly probability. The offset analysis subnetwork is used to take the power data and the average radiance of the infrared thermal image as inputs and output the offset probability. The fusion decision subnetwork is used to take the thermal anomaly probability and the offset probability as inputs and output the foreign object detection probability.

2. The method according to claim 1, wherein converting the original infrared thermal image to obtain a thermal energy distribution matrix comprises: The original infrared thermal image is preprocessed to obtain an infrared thermal image; The pixel values ​​in the infrared thermal image are converted into temperature values ​​to obtain the thermal energy distribution matrix.

3. The method according to claim 1, characterized in that, The process of obtaining the thermal anomaly probability based on the thermal energy distribution matrix includes: Spatial discontinuity features are obtained based on the thermal energy distribution matrix; wherein, the spatial discontinuity features include local hot spot features and temperature gradient abrupt change features; The thermal anomaly probability is obtained based on the spatial discontinuity characteristics. The process of obtaining the average radiance based on the thermal energy distribution matrix includes: The statistical average of all elements in the thermal energy distribution matrix is ​​calculated, and the statistical average is used as the average radiance.

4. The method according to claim 1, characterized in that, The calculation of the foreign object detection probability based on the thermal anomaly probability and the offset probability includes: The probability of foreign object detection is obtained by nonlinearly fusing the thermal anomaly probability and the offset probability.

5. The method according to claim 1, wherein adjusting the wireless charging mode according to the foreign object detection probability comprises: When the foreign object detection probability is less than the first threshold, the wireless charging system is controlled to enter the safe mode; When the foreign object detection probability is greater than or equal to the first threshold and less than the second threshold, the wireless charging system is controlled to enter the alert mode. When the foreign object detection probability is greater than or equal to the second threshold, the wireless charging system is controlled to enter the blocking mode; Wherein, the first threshold is less than the second threshold.

6. A wireless charging foreign object detection device based on multi-dimensional information fusion sensing, characterized in that, The device includes: an acquisition module, a processing module, and an adjustment module. The acquisition module is used to acquire power data of the wireless charging transmitter and the original infrared thermal image of the wireless charging area; wherein, the original infrared thermal image is the full-field data of the wireless charging area; The processing module is used to process the power data and the original infrared thermal image acquired by the acquisition module to obtain the thermal anomaly probability and offset probability of the infrared thermal image; wherein, the offset probability is used to characterize the degree of deviation of the ratio between the power data and the average radiance of the infrared thermal image relative to a desired ratio; the desired ratio is determined based on desired power data and desired radiance, wherein the desired power data is the power data of the wireless charging transmitter when there are no foreign objects, and the desired radiance is the average radiance of the infrared thermal image of the wireless charging area when there are no foreign objects; the thermal anomaly probability is used to characterize the degree of deviation of the spatial temperature of the infrared thermal image of the wireless charging area relative to the spatial temperature of the infrared thermal image of the wireless charging area when there are no foreign objects; and the foreign object detection probability is calculated based on the thermal anomaly probability and the offset probability obtained by the processing module. The adjustment module is used to adjust the wireless charging mode according to the foreign object detection probability calculated by the processing module; wherein the wireless charging mode includes at least one of the following: safety mode, alert mode, and blocking mode; The processing module is further configured to convert the original infrared thermal image to obtain a thermal energy distribution matrix; obtain a thermal anomaly probability based on the thermal energy distribution matrix; obtain an average radiance based on the thermal energy distribution matrix, and obtain a shift probability based on the power data and the average radiance; input the power data and the thermal energy distribution matrix into a composite neural network for foreign object detection to obtain a foreign object detection probability; wherein, the composite neural network includes a thermal anomaly analysis subnetwork, a shift analysis subnetwork, and a fusion decision subnetwork; the thermal anomaly analysis subnetwork is configured to take the thermal energy distribution matrix as input and output the thermal anomaly probability; the shift analysis subnetwork is configured to take the power data and the average radiance of the infrared thermal image as input and output the shift probability; the fusion decision subnetwork is configured to take the thermal anomaly probability and the shift probability as input and output the foreign object detection probability.

7. A wireless charging foreign object detection device based on multi-dimensional information fusion sensing, characterized in that, include: Processor and memory, of which, The memory is used to store computer programs; The processor is configured to call and run the computer program from the memory to perform the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores executable instructions for causing a processor to execute, thereby implementing the method of any one of claims 1 to 5.

Images