Foreign matter detection method for wireless charging system combining machine vision and auxiliary coil
Through the combination of the improved YOLOv8 network model and the plane-center symmetrical auxiliary coil, high-precision detection of metal foreign objects in the wireless charging system is achieved, and the problem of low detection accuracy of small foreign objects is solved, improving the system's safety and charging efficiency.
Patent Information
- Application Number
- CN202411058570.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-02
AI Technical Summary
When existing wireless charging systems detect metal foreign objects, especially small foreign objects, the detection accuracy is not high, which affects the system safety and charging efficiency.
Combined with the improved YOLOv8 network model and planar center symmetric auxiliary coil, the surface images of the emission coil are collected and analyzed through machine vision technology to achieve automated detection and identification of metal foreign matters, and real-time monitoring and feedback during charging.
It improves the detection accuracy and safety of the wireless charging system for metal foreign objects, ensures the stability and efficiency of the charging process, and can conduct comprehensive inspections before and after charging, avoiding charging interruptions caused by foreign objects.
Smart Images

Figure CN118953077A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of metal foreign body detection in a wireless charging system, and mainly to a foreign body detection method for a wireless charging system combining machine vision and an auxiliary coil. Background Art
[0002] In the wireless charging system of electric vehicles, the primary transmitting coil is placed on the ground and the secondary receiving coil is placed on the vehicle. There is usually a large air gap between the transmitting coil and the receiving coil, and there may be metal foreign objects such as coins and cans in this area, which will not only reduce the charging efficiency of the system, but also cause great safety hazards. Therefore, accurate detection of foreign objects is a necessary means to improve the safety of wireless charging systems for electric vehicles.
[0003] At present, the main methods for detecting foreign objects in wireless charging systems include auxiliary coil foreign object detection technology, system parameter foreign object detection technology, and machine learning foreign object detection technology. The auxiliary coil detection technology mainly involves laying the detection coil above the transmitting coil. When a metal foreign object approaches the detection area, the inductance and voltage of the detection coil will change, thereby realizing foreign object detection. A method for detecting metal foreign objects in a blind spot-free wireless charging system for electric vehicles with incomplete compensation of the detection coil published by Yang Ziyue, Xia Chenyang, Zhao Shuze, etc. improves the detection sensitivity by optimizing the incomplete compensation circuit parameters of the detection coil. The auxiliary coil detection technology requires a certain amount of space, and the detection accuracy of this detection technology will be affected to a certain extent when the transmitting coil and the receiving coil are not aligned.
[0004] The system parameter foreign body detection technology mainly detects certain system parameters such as coil quality factor, voltage, inductance, resonant frequency, etc., and judges whether there is a metal foreign body entering the system based on the change of parameters. An article proposed by JAFARI H et al. on detecting foreign bodies by measuring the resonant frequency and resonant current deviation of the primary coil proposed a discrimination mechanism that can effectively distinguish between the parameter changes caused by the entry of foreign bodies and the horizontal and vertical changes of the primary and secondary coils. The disadvantage of this detection technology is that when the metal foreign body entering the wireless charging system is small in size, the system parameters caused by it are not easy to be detected, thus affecting the system detection accuracy.
[0005] The metal foreign body detection method based on system parameters faces the problem of low detection accuracy when detecting tiny objects. Therefore, the method proposed in the present invention can realize foreign body detection in the whole process of wireless charging and the detection accuracy is higher than that of previous methods. Summary of the invention
[0006] In response to the above-mentioned technical problem of low monitoring accuracy when the metal object is small, the present technical solution provides a foreign body detection method for a wireless charging system that combines machine vision and an auxiliary coil, and combines an improved YOLOv8 network model and a plane center-symmetrical auxiliary coil for foreign body detection. The machine learning foreign body detection technology collects images of the transmitting coil surface to achieve automatic detection and identification of foreign bodies. The plane center-symmetrical auxiliary coil can accurately detect metal foreign bodies that have a greater impact on the wireless charging system, providing important protection for the safety of the electric vehicle wireless charging system. The above-mentioned problems are effectively solved.
[0007] The present invention is achieved through the following technical solutions: A foreign body detection method for a wireless charging system combining machine vision and an auxiliary coil comprises the following steps: Step 1: Create a data set, select the data set of the electric vehicle wireless charging system, and divide the labeled images in the data set into a training set, a validation set, and a test set; Step 2: Improve the YOLOv8 detection model to obtain the improved YOLOv8-PDSW network model; the specific operation method is: Step 2.1: Replace the network structure of the YOLOv8 model with the PP-LCNet network; Step 2.2: In the YOLOv8n backbone network, introduce the second to last and third to last C2f modules in the backbone network into the deformable convolutional network DCN, and name the modules they belong to C2f-DCN; Step 2.3: Add SENet attention mechanism after the YOLOv8 upsampling stage structure and after the first C2f in the downsampling stage; Step 2.4: Introduce the Wise-IoU loss function to replace the original DF Loss and CIoU Loss loss functions; Step 3: Before the electric vehicle is wirelessly charged, the video device is used to capture images and transmit the images to the computer terminal. The YOLOv8-PDSW network model is used on the computer terminal to detect foreign objects, and the detection results are returned to the charging coil controller; if it is determined that there is a foreign object, charging is prohibited, and charging can continue after the foreign object is removed; if there is no foreign object, charging starts; Step 4: When the electric vehicle starts charging, a planar center-symmetrical auxiliary coil is used to detect metal foreign objects, and the detection results are returned to the charging coil controller; if the planar center-symmetrical auxiliary coil detects metal foreign objects, charging is stopped immediately; if no metal foreign objects are detected, charging continues until charging is completed.
[0008] Furthermore, in the data set of the electric vehicle wireless charging system described in step one, the samples in the data set include coins, cans, nuts, screws, etc., and the number of data of all foreign objects accounts for the same proportion in the data set; the data set is divided into a training set, a validation set, and a test set in a ratio of 6:2:2 by random sampling, and the training set and the test set are labeled using a data labeling tool.
[0009] Furthermore, the operation flow of the SENet attention mechanism described in step 2.3 is: S1: The input feature map is converted into a channel-level description, and global average pooling is performed on each channel to capture global context information; S2: learns the dependencies between channels through two fully connected layers and generates a weight for each channel; S3: Use channel weights to adjust the original feature map. The feature map of each channel is multiplied by its corresponding weight, thereby weakening or strengthening certain feature channels.
[0010] Furthermore, the operation flow of the Wise-IoU loss function in step 2.4 is as follows: calculating the IoU score between the true box and the predicted box, measuring the degree of overlap between the two, calculating the distance between the center points of the bounding boxes of the true box and the predicted box, determining the maximum distance between the two, and calculating the area between the two boxes, which is used to measure the relationship between the boxes.
[0011] Furthermore, the improved YOLOv8-PDSW network model is convolutionally operated by CBS, DCN is a deformable convolution module, SPPF is a spatial pyramid pooling module, SENet is an attention mechanism module, Upsample is an upsampling module, Concat is a feature connection module, and Detect is a detection head.
[0012] Furthermore, the YOLOv8-PDSW network model is used in the computer terminal to detect foreign objects as described in step 3. The specific operation steps include: Step 3.1: Preprocess the image to be detected, adjust the image size, and perform data enhancement operations; Step 3.2: Pass the processed image to the PP-LCNet network lightweight network, and extract features of the input image through the convolution layer and pooling layer; Step 3.3: Perform bounding box prediction, use the anchor box on each feature map to predict the bounding box of the foreign object target, regress and classify the anchor box, and obtain the location and category of the foreign object target; Step 3.4: According to the location information and category of the bounding box, the metal foreign body target is identified and marked in the image.
[0013] Furthermore, the plane center-symmetrical auxiliary coil described in step four is based on the principle of electromagnetic induction, and the alternating magnetic field will generate a certain induced voltage inside the auxiliary coil; and from the symmetry of the magnetic field of the primary transmitting coil, it can be seen that when the wireless charging system operates normally, a set of symmetrical auxiliary coils A1 and B1 have the same induced voltage, and the difference in induced voltage between the two is 0; when a metal foreign body falls on the auxiliary coil A1 or B1, the induced voltage across the auxiliary coil A1 or the auxiliary coil B1 will change, thereby generating an induced voltage difference, and by comparing the changes in the induced voltage difference, rapid detection of the metal foreign body can be achieved.
[0014] Furthermore, the auxiliary coils A1 and B1 are represented as a coil pair. If the auxiliary coils A1 and B1 are symmetrically arranged, when the metal foreign matter is in the middle of the auxiliary coils A1 and B1, and the metal foreign matter covers the same area in these two auxiliary coils, the induced voltage difference is still 0, which will form a detection blind spot; therefore, the auxiliary coil A1 is placed in the upper left corner of the plane, and the auxiliary coil B1 is placed in the lower right corner of the plane; the auxiliary coils A2 and B2 and other auxiliary coil structures are constructed in the same way.
[0015] Furthermore, the auxiliary coils A1 and B1 are a pair of coils, the induced voltages of the two coils are in opposite directions, and the induced voltages are equal in magnitude under the same structure; the corresponding output voltage is U A1 and U B1 , when there is no metal foreign matter, |U A1 |=|U B1 |, when a metal foreign body falls on the auxiliary coil A1, there is |U A1 |>|U B1 |or|U A1 |<|U B1 |.
[0016] Furthermore, the detection circuit of the plane center-symmetrical auxiliary coil includes an amplifier circuit, a filter circuit and a voltage acquisition circuit; the output end of the center-symmetrical auxiliary coil is connected to the amplifier circuit, and the voltage value obtained by electromagnetic induction is small. The voltage is amplified by the operational amplifier circuit and then input into the filter circuit; the transmitting frequency of the transmitting coil in the plane center-symmetrical auxiliary coil is 85KHZ, and the filter circuit selects the signal passing through 85KHZ and suppresses the other frequency signals; the voltage acquisition circuit performs voltage acquisition on the filtered circuit to obtain the induced voltage value of the auxiliary detection. Beneficial Effects
[0017] The invention proposes a wireless charging system foreign body detection method combining machine vision and auxiliary coil, which has the following beneficial effects compared with the prior art: (1) The present invention combines the improved YOLOv8 network model and the plane center symmetric auxiliary coil to detect foreign objects. The machine learning foreign object detection technology collects the surface image of the transmitting coil to realize the automatic detection and identification of foreign objects; the plane center symmetric auxiliary coil can accurately detect the metal foreign objects that have a greater impact on the wireless charging system, providing an important guarantee for the safety of the electric vehicle wireless charging system. Before the wireless charging stage of the electric vehicle, the improved YOLOv8 is used to detect foreign objects to ensure the charging safety before the wireless charging of the electric vehicle and avoid repeated starting and stopping of the vehicle. When the electric vehicle starts charging, the plane center symmetric auxiliary coil is used to detect foreign objects, which can accurately detect whether there are micro-metal foreign objects affecting the wireless charging system; and the detection results can be quickly fed back. The combination of the two detection methods can not only realize the detection of the electric vehicle wireless charging system when it is in working state, but also can be detected in advance before the system works, so that the detection of foreign objects in the whole stage of the wireless charging of the electric vehicle can be realized, so as to ensure the charging safety and improve the charging efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 It is a schematic diagram of the overall process of the present invention.
[0019] Figure 2 This is a structural block diagram of the improved YOLOv8 network model in the present invention.
[0020] Figure 3 Schematic diagram of the accuracy curve of the original YOLOv8 network model.
[0021] Figure 4 It is a schematic diagram of the accuracy curve of the improved YOLOv8 network model in the present invention.
[0022] Figure 5 Schematic diagram of the precision-recall curve of the original YOLOv8 network model.
[0023] Figure 6 It is a precision-recall curve of the improved YOLOv8 network model in the present invention.
[0024] Figure 7 This is a verification diagram of the experimental detection effect of the improved YOLOv8 network model in the present invention.
[0025] Figure 8 It is a structural schematic diagram of the plane center-symmetrical auxiliary coil in the present invention.
[0026] Fig. 9 It is a schematic diagram of the auxiliary coil symmetrically placed on the center of the plane where the metal foreign matter falls in the present invention.
[0027] Fig.10The voltage waveform diagram is generated when a metal foreign object such as a one-yuan coin falls on the detection coil. DETAILED DESCRIPTION
[0028] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Under the premise of not departing from the design concept of the present invention, various modifications and improvements made by ordinary persons in the art to the technical solutions of the present invention should all fall within the protection scope of the present invention. Example
[0029] like Figure 1 As shown, a foreign body detection method for a wireless charging system combining machine vision and an auxiliary coil includes the following steps: Step 1: Create a data set, select the data set of electric vehicle wireless charging system, and divide the labeled images in the data set into training set, verification set and test set.
[0030] In the dataset of the electric vehicle wireless charging system, the samples in the dataset include coins, cans, nuts, screws, etc., and the number of data on all foreign objects accounts for the same proportion in the dataset; the dataset is divided into training set, validation set and test set in a ratio of 6:2:2 by random sampling, and the training set and test set are labeled using data annotation tools.
[0031] The training set is used to train the improved model. During the training process, the training error is gradient-decreased, and the training weight parameters can be adjusted through experiments. The validation set is used to adjust the model's hyperparameters and conduct a preliminary evaluation of the model's capabilities. The test set is used to detect the model's detection accuracy and evaluate generalization capabilities.
[0032] Step 2: Improve the YOLOv8 detection model to obtain the improved YOLOv8-PDSW network model; the specific operation method is: Step 2.1: Replace the network structure of the YOLOv8 model with the PP-LCNet network.
[0033] PP-LCNet is a lightweight convolutional neural network based on a deep neural network mathematical core library, designed to improve computing efficiency and performance on mobile devices. It is more suitable for practical application scenarios because of its advantages such as low computing cost and high efficiency.
[0034] The Darknet-53 module of the YOLOv8 backbone network has a deeper network structure. Although it can improve the model's ability to express image features to a certain extent, the complexity of its module leads to a longer model reasoning time. Therefore, this embodiment uses the PP-LCNet module in YOLOv8n to improve the backbone network to achieve a lightweight and fast detection method.
[0035] Step 2.2: In the YOLOv8n backbone network, introduce the second to last and third to last C2f modules in the backbone network into the deformable convolutional network DCN, and name the modules they belong to C2f-DCN.
[0036] Deformable convolution can adaptively adjust the shape and size of the receptive field according to the irregular shape of the object, further improving the robustness of CNN. Traditional convolution operations can only extract features from regular rectangular receptive fields, but in the actual wireless charging process, the foreign objects entering are often irregular, so deformable convolution is needed to identify foreign objects with irregular shapes. When sampling, deformable convolution is more in line with the shape of foreign objects and is more robust. In addition, in the wireless charging system of electric vehicles, some smaller objects are easy to fall into it. If traditional convolution is still used, it may cause some small objects to be difficult to detect. Therefore, if Figure 2 As shown in the figure, the second to last and third to last C2f modules of the backbone network are replaced with deformable convolution modules to improve the detection effect of small target foreign objects.
[0037] Step 2.3: Add SENet attention mechanism after the YOLOv8 upsampling stage structure and after the first C2f in the downsampling stage.
[0038] The SENet network can automatically learn the importance of different channel features. The SENet attention mechanism enables the model to better focus on foreign object features and improve the accuracy of foreign object detection. SENet represents a channel attention strategy that can automatically learn the importance of different channel features. In simple terms, it can enhance important features and suppress unimportant features. Figure 2 As shown in the figure, the SENet attention mechanism is added after the upsampling stage structure and after the first C2f in the downsampling stage to improve the detection effect of foreign objects.
[0039] The operation process of SENet attention mechanism is: S1: The input feature map is converted into a channel-level description, and global average pooling is performed on each channel to capture global context information; S2: learns the dependencies between channels through two fully connected layers and generates a weight for each channel; S3: Use channel weights to adjust the original feature map. The feature map of each channel is multiplied by its corresponding weight, thereby weakening or strengthening certain feature channels.
[0040] Step 2.4: Introduce the Wise-IoU loss function to replace the original DF Loss and CIoU Loss loss functions.
[0041] YOLOv8n uses DF Loss and CIoU Loss to calculate the bounding box regression loss. Among them, CIoU uses a monotonic focusing mechanism, but does not consider the balance between simple samples and difficult samples. In addition, the traditional loss function only considers the intersection of the predicted box and the real box during detection, but ignores the classification information. Therefore, in order to solve the deviation of the detection result caused by the traditional IoU loss function, this embodiment introduces the Wise-IoU loss function.
[0042] The operation process of the Wise-IoU loss function is as follows: calculate the IoU score between the true box and the predicted box, measure the degree of overlap between the two, calculate the distance between the center points of the bounding boxes of the true box and the predicted box, determine the maximum distance between the two, and calculate the area between the two boxes. The role of this area is to measure the relationship between the boxes.
[0043] like Figure 2 As shown in the figure, the improved YOLOv8-PDSW network model is composed of CBS for convolution operation, DCN is a deformable convolution module, SPPF is a spatial pyramid pooling module, SENet is an attention mechanism module, Upsample is an upsampling module, Concat is a feature connection module, and Detect is a detection head.
[0044] In order to verify the superiority of the improved YOLOv8-PDSW network model, the inventors conducted simulation verification and analysis on it. The specific process is as follows: 1. Create a data set: Since there is currently no public and comprehensive dataset of the required type on the Internet, a self-made dataset was created using network resources and manual photography. The dataset includes four categories, and coins, cans, screws, and nuts are selected as metal foreign objects, with a total of 1,500 images.
[0045] The data set was randomly extracted and divided into training set (1050 images), validation set (300 images), and test set (150 images) in a ratio of 7:2:1. The training set was used to train the improved model. During the training process, the training error was gradient-decreased and the trainable weight parameters were obtained. The validation set was used to adjust the model's hyperparameters and conduct a preliminary evaluation of the model's capabilities. The test set was used to detect the model's detection accuracy and evaluate its generalization ability. 2. Evaluation index setting
[0046] This embodiment uses precision (Precision), parameters (Parameters / M), mean average precision (meanAverage Precision, mAP), F1-Score and floating point operations (FLOPs) to evaluate the improved model.
[0047] Precision refers to the ratio of the number of correctly detected samples to the total number of detected samples. Parameter quantity refers to the size of the model's parameters. In the mean average precision (mAP), m stands for mean. mAP@0.5 refers to the average accuracy of all categories when the IoU reaches 0.5 in the target detection task. Floating point calculation number refers to the number of floating point operations per second, which is one of the criteria for considering the amount of calculation of network models. (III) Simulation analysis
[0048] The experimental simulation was carried out under the 64-bit Windows 11 operating system, the GPU was RTX3090, the video memory was 24GB, the CPU was Intel Core i7-10700k, the host memory was 96GB, the programming language was Python 3.9.13, the CUDA version was 12.2.146, the batch size was 10, the number of iterations was 150, and the training was based on the deep learning framework Pytorch 1.11.0.
[0049] 3.1 Analysis of evaluation index results of different models In order to verify the difference and superiority of the improved algorithm in this embodiment and other target detection algorithms, accuracy, recall rate, mAP@0.5, and floating point operations are used as evaluation indicators. The improved model YOL0v8-PDSW in this embodiment is compared with YOL0v5, YOL0v7, YOLOv8n, and YOLOv8s algorithms under the same conditions. The accuracy curves of YOL0v8n and YOL0v8-PDSW are shown in Figure 1. Figure 3 and Figure 4 The simulation results of various evaluation indicators are shown in Table 1.
[0050] according to Figure 3 and Figure 4 In comparison, the accuracy of the original model before improvement was 92.6%, and the accuracy of the model after improvement was 95.7%. The accuracy of the improved model was 3.1% higher than that of the original model.
[0051] Table 1
[0052] According to the simulation results in Table 1, the unimproved YOLOv8n algorithm integrates the advantages of previous single-target detection algorithms, and exceeds the previous YOLOv5 and YOLOv7 algorithms in terms of accuracy, mAP@0.5%, and F1-Score. In addition, the YOLOv8n algorithm is smaller than the YOLOv5 and YOLOv7 algorithms in terms of parameter quantity and floating-point calculation number. Although the YOLOv8s algorithm is higher than the YOLOv8n algorithm in terms of mAP@0.5%, its accuracy is lower than the YOLOv8n algorithm, and it is much larger than the YOLOv8n in terms of parameter quantity and floating-point calculation number.
[0053] 3.2 Comparative analysis of the results of adding different improvement modules In order to verify the effectiveness of each improved module of the algorithm in this embodiment, the YOL0v8n model is used as the baseline model, and the accuracy, recall rate, mAP@0.5%, and floating-point calculation number are used as evaluation indicators. The results are compared through the permutations and combinations of different improved modules. The results are shown in Table 2.
[0054]
[0055] In order to better verify the impact of the improved modules and the combination of improved modules on the performance improvement of the original YOLOv8n algorithm, this experiment designed a comparative analysis. The simulation results are shown in Table 2.
[0056] Among them, √ means that the module has been added, and × means that the module has not been added. After using the PP-LCNet lightweight network to improve the original YOLOv8n backbone network, the accuracy of the original algorithm has dropped by 3.4%, mAP@0.5% has dropped by 1.8%, and the number of model parameters and floating-point calculations have been significantly reduced. Adding DCN deformable convolution and SE attention mechanism to the original YOLOv8n algorithm model has significantly improved the accuracy and mAP@0.5% has been slightly improved, but because it will increase the complexity of the model, the number of parameters and floating-point calculations will increase.
[0057] By adding the SE attention mechanism and the Wise-IoU loss function to the original YOLOv8n algorithm, the accuracy is improved to a certain extent, effectively improving the bounding box regression ability of the model, but the mAP@0.5% is slightly reduced, and the number of parameters and floating-point calculations of the model will increase. In summary, compared with the original YOLOv8n algorithm model, the improved YOLOv8-PDSW model studied in this embodiment has an accuracy improvement of 3.1%, an mAP@0.5 improvement of 0.6%, and a decrease in the number of parameters and floating-point calculations of 26% and 13%, respectively, indicating the effectiveness of the improved algorithm proposed in this article. The precision-recall curves of YOLOv8n and YOLOv8-PDSW are shown below. Figure 5 and Figure 6 shown.
[0058] Depend on Figure 5 and Figure 6 By comparison, the overall mAP@0.5% of the improved algorithm increased by 0.6%, remained basically stable in the coin and nut categories, and improved in the screw and can categories.
[0059] 3.3 Experimental test results verification like Figure 7As shown in the figure, the original YOLOv8n algorithm model is not particularly effective when facing foreign bodies of different sizes and complex backgrounds. In order to verify that the improved YOLOv8-PDSW algorithm model still has good detection results when facing these problems, this paper conducts experimental comparison and verification. The confidence level indicates the possibility that the detection system judges the detected object as a foreign body. The greater the confidence level, the greater the possibility that the system judges it as a foreign body. From the comparison between Experiment 1 and Experiment 2, it can be seen that in Experiment 1, the detection confidence level of the YOLOv8n-PDSW model increased from 0.5 to 0.7 compared with the original YOLOv8 model, and the improved model is more likely to judge it as a foreign body than the original model. In Experiment 2, the detection confidence level of the YOLOv8n-PDSW model is higher than that of the original YOLOv8n model. In summary, the effectiveness of the improved model can be obtained.
[0060] in conclusion This embodiment adopts an improved YOLOv8 model. Based on the YOLOv8 model, the improved model replaces the backbone network in the model with a PP-LCNet lightweight network to reduce the number of model parameters and floating-point calculations, introduces deformable convolution DCN, enhances the ability to identify irregular shaped foreign objects, adds SE attention mechanism, improves the feature fusion ability of the model, and uses Wise-IoU as the loss function to improve the bounding box regression ability of the network model. Compared with the original model, the improved model of this embodiment improves the detection accuracy by 3.1%, reduces the number of model parameters by 26%, and reduces the number of floating-point calculations by 13%. It can be concluded that the method proposed in this embodiment can meet the requirements of electric vehicle wireless charging system for foreign object detection and has practical use value.
[0061] Step 3: Before wireless charging of the electric vehicle, the video device is used to capture images and transmit the images to the computer terminal. The YOLOv8-PDSW network model is used on the computer terminal to detect foreign objects, and the detection results are returned to the charging coil controller. If it is determined that there is a foreign object, charging is prohibited. Charging can continue after the foreign object is removed. If there is no foreign object, charging begins.
[0062] The specific operation steps of using the YOLOv8-PDSW network model to detect foreign objects on a computer terminal include: Step 3.1: Preprocess the image to be detected, adjust the image size, and perform data enhancement operations; Step 3.2: Pass the processed image to the PP-LCNet network lightweight network, and extract features of the input image through the convolution layer and pooling layer; Step 3.3: Perform bounding box prediction, use the anchor box on each feature map to predict the bounding box of the foreign object target, regress and classify the anchor box, and obtain the location and category of the foreign object target; Step 3.4: According to the location information and category of the bounding box, the metal foreign body target is identified and marked in the image.
[0063] Step 4: When the electric vehicle starts to charge, a planar centrally symmetrical auxiliary coil is used to detect metal foreign bodies, and the detection results are returned to the charging coil controller; the detection circuit of the planar centrally symmetrical auxiliary coil includes an amplifier circuit, a filter circuit and a voltage acquisition circuit; the output end of the centrally symmetrical auxiliary coil is connected to the amplifier circuit, and the voltage value obtained by electromagnetic induction is small. The voltage is amplified by the operational amplifier circuit and then input into the filter circuit; the transmitting frequency of the transmitting coil in the planar centrally symmetrical auxiliary coil is 85KHZ, and the filter circuit selects the signal passing through 85KHZ and suppresses the remaining frequency signals; the voltage acquisition circuit collects the voltage of the filtered circuit to obtain the induced voltage value of the auxiliary detection.
[0064] The detection principle of the planar center-symmetrical detection coil is based on the principle of electromagnetic induction. When a metal foreign body enters the wireless charging system, it will cause a change in the magnetic field of the wireless charging system, thereby generating an induced electromotive force. When the metal foreign body falls on the auxiliary coil, the equivalent inductance and equivalent resistance of the center-symmetrical auxiliary coil will change accordingly, and the output induced voltage of the center-symmetrical auxiliary coil will also change. The equivalent inductance and equivalent resistance of metal foreign bodies of different types and sizes are different, which affects the degree of change of the center-symmetrical auxiliary coil parameters to different degrees. The more drastic the change in the center-symmetrical auxiliary coil parameters, the easier it is to detect.
[0065] The planar central symmetrical auxiliary coil is based on the principle of electromagnetic induction. The alternating magnetic field will generate a certain induced voltage inside the auxiliary coil. From the symmetry of the magnetic field of the primary transmitting coil, it can be seen that when the wireless charging system operates normally, a set of symmetrical auxiliary coils A1 and B1 have the same induced voltage, and the difference between the induced voltages is 0. When a metal foreign body falls on the auxiliary coil A1 or B1, the induced voltage across the auxiliary coil A1 or B1 will change, thereby generating an induced voltage difference. By comparing the changes in the induced voltage difference, an induced voltage difference is generated. By comparing the changes in the induced voltage difference, rapid detection of metal foreign bodies can be achieved.
[0066] like Figure 8 As shown, the plane center-symmetrical auxiliary coil is composed of 5 groups of coils, namely A1, B1, A2, B2, A3, B3, A4, B4, A5, B5. All coils are close to the surface of the transmitting coil. The plane center-symmetrical auxiliary coil, the transmitting coil, and the receiving coil all use rectangular coils.
[0067] Auxiliary coils A1 and B1 are represented as a set of coil pairs. If the auxiliary coils A1 and B1 are symmetrically arranged, when the metal foreign matter is in the middle of the auxiliary coils A1 and B1, and the metal foreign matter covers the same area in these two auxiliary coils, the difference in induced voltage is still 0, which will form a detection blind spot; therefore, the auxiliary coil A1 is placed in the upper left corner of the plane, and the auxiliary coil B1 is placed in the lower right corner of the plane; the auxiliary coils A2 and B2 and other auxiliary coils form a similar structure.
[0068] The auxiliary coils A1 and B1 form a pair of coils. The induced voltages of the two coils are in opposite directions. Under the same structure, the induced voltages are equal. The corresponding output voltage is U A1 and U B1 , when there is no metal foreign matter, |U A1 |=|U B1 |, when a metal foreign body falls on the auxiliary coil A1, there is |U A1 |>|U B1 |or|U A1 |<|U B1 |.
[0069] When a metal foreign object enters the wireless charging system and falls between the auxiliary coil A1 and the auxiliary coil B5, Fig. 9 As shown, the black object represents a metal foreign body. The upper left auxiliary coil A1 and the lower right auxiliary coil B1 form a group of auxiliary coils. The upper right auxiliary coil B5 and the lower left auxiliary coil A5 form a group of auxiliary coils. The induced voltage of the auxiliary coil A1 will change, while the induced voltage of the auxiliary coil B1 remains unchanged. The induced voltage of the auxiliary coil B5 will change, while the induced voltage of the auxiliary coil A5 remains unchanged. At this time, there is an induced voltage difference, and the output of the differential circuit is not 0, which means that there is a metal foreign body in the system. The calculation formula is:
[0070] in, Represents the induced voltage difference formed by the auxiliary coil A1 and the auxiliary coil B1, represents the induced voltage of auxiliary coil A1, represents the induced voltage of the auxiliary coil B1, represents the induced voltage difference formed by the auxiliary coil A5 and the auxiliary coil B5, represents the induced voltage of auxiliary coil A5, Represents the induced voltage of the auxiliary coil B5.
[0071] When t=0.01s, a 1-yuan coin metal foreign body enters the wireless charging system coupling mechanism. When the 1-yuan coin metal foreign body falls on the auxiliary coil A1, Fig.10The voltage waveform of the auxiliary coil A1 is shown. Due to the certain error in the circuit design, the detection system has an output voltage of about 0.1v. The output voltage threshold of the detection system for judging whether a metal foreign body exists is set to 0.1v. Fig.10 It can be seen that there is no metal foreign matter in the coupling mechanism before 0.01s. When the 1 yuan coin foreign matter exists, the output voltage of the detection system becomes 1.1v, which is higher than the set voltage threshold. It can be determined that there is metal foreign matter in the coupling mechanism of the wireless charging system.
[0072] If the plane center symmetrical auxiliary coil detects a metal foreign body, charging will be stopped immediately; if no metal foreign body is detected, charging will continue until charging is completed.
[0073] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the present invention.
Claims
1. A foreign body detection method for a wireless charging system combining machine vision and an auxiliary coil, characterized in that: Includes steps: Step 1: Create a data set, select the data set of the electric vehicle wireless charging system, and divide the labeled images in the data set into a training set, a validation set, and a test set; Step 2: Improve the YOLOv8 detection model to obtain the improved YOLOv8-PDSW network model; the specific operation method is: Step 2.1: Replace the network structure of the YOLOv8 model with the PP-LCNet network; Step 2.2: In the YOLOv8n backbone network, introduce the second to last and third to last C2f modules in the backbone network into the deformable convolutional network DCN, and name the modules they belong to C2f-DCN; Step 2.3: Add SENet attention mechanism after the YOLOv8 upsampling stage structure and after the first C2f in the downsampling stage; Step 2.4: Introduce the Wise-IoU loss function to replace the original DF Loss and CIoU Loss loss functions; Step 3: Before the electric vehicle is wirelessly charged, the video device is used to capture images and transmit the images to the computer terminal. The YOLOv8-PDSW network model is used on the computer terminal to detect foreign objects, and the detection results are returned to the charging coil controller; if it is determined that there is a foreign object, charging is prohibited, and charging can continue after the foreign object is removed; if there is no foreign object, charging starts; Step 4: When the electric vehicle starts charging, a planar central symmetrical auxiliary coil is used to detect metal foreign bodies, and the detection results are returned to the charging coil controller; If the plane center symmetrical auxiliary coil detects a metal foreign body, charging will be stopped immediately; if no metal foreign body is detected, charging will continue until charging is completed.
2. According to claim 1, a foreign body detection method for a wireless charging system combining machine vision and an auxiliary coil is characterized in that: In the data set of the electric vehicle wireless charging system described in step 1, the samples in the data set include coins, cans, nuts, screws, etc., and the number of data of all foreign objects accounts for the same proportion in the data set; the data set is divided into a training set, a validation set, and a test set in a ratio of 6:2:2 by random sampling, and the training set and the test set are labeled using a data labeling tool.
3. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 1, characterized in that: The operation flow of the SENet attention mechanism described in step 2.3 is: S1: The input feature map is converted into a channel-level description, and global average pooling is performed on each channel to capture global context information; S2: learns the dependencies between channels through two fully connected layers and generates a weight for each channel; S3: Use channel weights to adjust the original feature map. The feature map of each channel is multiplied by its corresponding weight, thereby weakening or strengthening certain feature channels.
4. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 1, characterized in that: The operation flow of the Wise-IoU loss function in step 2.4 is as follows: calculate the IoU score between the true box and the predicted box, measure the degree of overlap between the two, calculate the distance between the center points of the bounding boxes of the true box and the predicted box, determine the maximum distance between the two, and calculate the area between the two boxes. The role of this area is to measure the relationship between the boxes.
5. A method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 1, 3 or 4, characterized in that: The improved YOLOv8-PDSW network model is convolutionally operated by CBS, DCN is a deformable convolution module, SPPF is a spatial pyramid pooling module, SENet is an attention mechanism module, Upsample is an upsampling module, Concat is a feature connection module, and Detect is a detection head.
6. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 1, characterized in that: The specific steps of using the YOLOv8-PDSW network model to detect foreign objects on a computer terminal as described in step 3 include: Step 3.1: Preprocess the image to be detected, adjust the image size, and perform data enhancement operations; Step 3.2: Pass the processed image to the PP-LCNet network lightweight network, and extract features of the input image through the convolution layer and pooling layer; Step 3.3: Perform bounding box prediction, use the anchor box on each feature map to predict the bounding box of the foreign object target, regress and classify the anchor box, and obtain the location and category of the foreign object target; Step 3.4: According to the location information and category of the bounding box, the metal foreign body target is identified and marked in the image.
7. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 1, characterized in that: The plane center symmetrical auxiliary coil described in step 4 is based on the principle of electromagnetic induction. The alternating magnetic field will generate a certain induced voltage inside the auxiliary coil. And from the symmetry of the magnetic field of the primary transmitting coil, it can be seen that when the wireless charging system operates normally, a set of symmetrical auxiliary coils A1 and B1 have the same induced voltage, and the difference between the induced voltages is 0. When a metal foreign body falls on the auxiliary coil A1 or B1, the induced voltage across the auxiliary coil A1 or B1 will change, thereby generating an induced voltage difference. By comparing the changes in the induced voltage difference, the metal foreign body can be quickly detected.
8. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 7, characterized in that: The auxiliary coils A1 and B1 are represented as a coil pair, the auxiliary coil A1 is placed at the upper left corner of the plane, and the auxiliary coil B1 is placed at the lower right corner of the plane; the auxiliary coils A2 and B2 and other auxiliary coils form a structure similarly.
9. A method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 7 or 8, characterized in that: The auxiliary coils A1 and B1 are a pair of coils. The induced voltages of the two coils are in opposite directions. Under the same structure, the induced voltages are equal. The corresponding output voltage is U A1 and U B1 , when there is no metal foreign matter, |U A1 |=|U B1 |, when a metal foreign body falls on the auxiliary coil A1, there is |U A1 |>|U B1 |or|U A1 |<|U B1 |.
10. The method for detecting foreign matter in a wireless charging system combining machine vision and an auxiliary coil according to claim 7, characterized in that: The detection circuit of the plane center-symmetrical auxiliary coil includes an amplifier circuit, a filter circuit and a voltage acquisition circuit; the output end of the center-symmetrical auxiliary coil is connected to the amplifier circuit, and the voltage value obtained by electromagnetic induction is small. The voltage is amplified by the operational amplifier circuit and then input into the filter circuit; the transmitting frequency of the transmitting coil in the plane center-symmetrical auxiliary coil is 85KHZ, and the filter circuit selects the signal passing through 85KHZ and suppresses the other frequency signals; the voltage acquisition circuit collects the voltage of the filtered circuit to obtain the induced voltage value of the auxiliary detection.
Citation Information
Patent Citations
Wireless charging, alignment and metal foreign body detection system of electric automobile and method thereof
CN105235545A
Foreign matter detection method and system for assisting wireless charging of new energy automobile
CN117853823A
Unmanned aerial vehicle laser charging tracking system using improved YOLOv8
CN117922864A
Underwater wireless charging platform docking bin angle and foreign matter identification method
CN118298288A
Automated defect classification and detection
US20230343078A1
Cited By
Data processing method and device based on image recognition and cone crusher
CN121438298A
