Wireless charging method and system for electric bicycle

Abstract

The invention relates to the technical field of wireless charging, and discloses an electric bicycle wireless charging method and system, and the method comprises the steps: converting an access power supply into an alternating current with a preset frequency; the alternating current is connected to a transmitting coil to generate an alternating electromagnetic field; the electric bicycle entering the charging area is charged through the alternating electromagnetic field; performing real-time sensor detection on the environment in the charging area to obtain real-time sensor data; establishing an FOD detection model, obtaining a relationship between sensor data and foreign matters through deep learning of a historical sensor data set, and judging whether the foreign matters exist in the charging area or not according to real-time sensor data; controlling the power supply of the transmitting coil according to the foreign matter judgment result of the FOD detection model; and when the foreign matter is detected in the charging area, the power supply of the transmitting coil is cut off. According to the method and the system, the foreign matter detection accuracy is improved, the omission ratio is reduced, the foreign matter detection requirements in a complex environment can be met, and intelligent charging management is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless charging, and in particular to a wireless charging method and system for an electric bicycle. Background Art

[0002] In recent years, with the rapid development of renewable energy and the accelerated popularization of electric vehicles, electric bicycles have been widely used as a convenient and environmentally friendly means of transportation. In order to improve the convenience of electric bicycles and enhance user experience, more and more electric bicycle manufacturers and related technology companies have begun to explore wireless charging technology. Wireless charging technology can not only eliminate the cable connection and plugging in traditional charging methods, providing users with a more convenient charging experience; it can also reduce the wear of contact points while improving charging safety.

[0003] In existing wireless charging systems, the principle of electromagnetic induction is usually used to generate an alternating electromagnetic field through a transmitting coil, which then induces an alternating current in a receiving coil to achieve energy transmission. Although this technology has improved charging convenience, its safety is still an important research topic. Especially in public environments, there may be foreign objects (FOD) in the charging area, such as metal objects or other objects. These foreign objects may cause safety hazards such as short circuits, overheating or fire during charging, posing potential risks to users and devices.

[0004] To solve this problem, researchers have proposed a variety of FOD detection solutions, but most methods still have shortcomings such as low detection accuracy, high missed detection rate, slow response speed, and poor environmental adaptability. At the same time, traditional rule-based methods have poor adaptability in complex environments and are difficult to achieve real-time monitoring and rapid response. Summary of the invention

[0005] To this end, the technical problem to be solved by the present invention is to overcome the problems of low foreign object detection accuracy, slow response speed and poor environmental adaptability in the wireless charging process in the prior art, and to provide an electric bicycle wireless charging method and system to improve the accuracy of foreign object detection and reduce the missed detection rate, which can adapt to the foreign object detection needs in complex environments and realize intelligent charging management.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a wireless charging method for an electric bicycle, comprising: Convert the incoming power into an alternating current of a predetermined frequency; Connecting the alternating current to the transmitting coil to generate an alternating electromagnetic field; charging the electric bicycle entering the charging area through the alternating electromagnetic field; Performing real-time sensor detection on the environment in the charging area to obtain real-time sensor data; Establishing a FOD detection model, obtaining the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determining whether there are foreign objects in the charging area based on the real-time sensor data; The power supply of the transmitting coil is controlled according to the foreign object determination result of the FOD detection model; wherein, when a foreign object is detected in the charging area, the power supply of the transmitting coil is cut off.

[0007] In one embodiment of the present invention, constructing the FOD detection model includes: Extracting a data set related to metal foreign matter from the historical sensor data set to obtain a first sensor data set; performing deep learning on the first sensor data set based on a deep learning model to obtain a first foreign matter identifier; and, Extracting a data set related to plastic foreign matter from the historical sensor data set to obtain a second sensor data set; performing deep learning on the second sensor data set based on a deep learning model to obtain a second foreign matter identifier; and, Performing deep learning on the historical sensor data set based on a deep learning model to obtain a foreign object detector; The FOD detection model is constructed according to the first foreign object identifier, the second foreign object identifier and the foreign object detector.

[0008] In one embodiment of the present invention, the FOD detection model determines whether there is a foreign object in the charging area according to the real-time sensor data, including: The first foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; The second foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; The foreign object detector identifies whether there is a foreign object in the charging area according to the real-time sensor data; When any one of the first foreign object identifier, the second foreign object identifier and the foreign object detector identifies that a foreign object exists in the charging area according to the real-time sensor data, it is determined that a foreign object exists in the charging area.

[0009] In one embodiment of the present invention, after obtaining the real-time sensor data in the charging area, the method further includes: Performing amplitude feature extraction on the real-time sensor data to obtain amplitude features; Calculating the mean, variance, maximum and minimum values ​​of the amplitude feature, and converting the calculation results into a first eigenvector; Extracting frequency domain features from the real-time sensor data to obtain frequency domain features; Calculating the absolute value, skewness and kurtosis of the frequency domain feature, and converting the calculation results into a second eigenvector; Merging the first feature vector and the second feature vector into a real-time data feature vector; The FOD detection model determines whether there is a foreign object in the charging area based on the real-time data feature vector.

[0010] In one embodiment of the present invention, the sensor array comprises: an infrared sensor array, through which real-time temperature data is obtained, and the FOD detection model identifies whether there is a foreign object in the charging area by analyzing the real-time temperature data; An electromagnetic sensor array, through which real-time electromagnetic wave data is obtained, and the FOD detection model identifies whether there is a foreign object in the charging area by analyzing the real-time electromagnetic wave data; An ultrasonic sensor array is used to obtain real-time sound wave data, and the FOD detection model identifies the location of foreign matter by analyzing the real-time sound wave data.

[0011] In one embodiment of the present invention, extracting a data set related to metal foreign matter from the historical sensor data set to obtain a first sensor data set includes: Preprocessing the historical sensor data set; Performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the metal material; Converting the result of feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; Perform K-means clustering on the feature vector to obtain K clusters; Calculating the cosine similarity between each of the K clusters and the cluster seed to obtain K first cosine similarities; If the first cosine similarity is not less than the cosine similarity threshold, the historical sensor data corresponding to the cluster having the cosine similarity is added to the first sensor data set.

[0012] In one embodiment of the present invention, extracting a data set related to plastic foreign matter from the historical sensor data set to obtain a second sensor data set includes: Preprocessing the historical sensor data set; Performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the plastic material; Converting the result of feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; Perform K-means clustering on the feature vector to obtain K clusters; Calculating the cosine similarity between each of the K clusters and the cluster seed to obtain K second cosine similarities; If the second cosine similarity is not less than the cosine similarity threshold, the historical sensor data corresponding to the cluster having the cosine similarity is added to the second sensor data set.

[0013] In one embodiment of the present invention, performing K-means clustering on the feature vector includes: S1: Initialize the cluster centers of the K clusters S2: Calculate the distance between the data points of the feature vector and the center of each cluster, and assign each data point to the nearest center point; S3: updating the cluster center according to all assigned feature vectors; S4: Set a stop condition, and iterate the steps S2-S3 repeatedly according to the stop condition.

[0014] In a second aspect, in order to solve the above technical problems, the present invention provides an electric bicycle wireless charging system, comprising: A power converter connected to a power source to convert the power source into an alternating current of a predetermined frequency; A transmitting coil is connected to the power converter and generates an alternating electromagnetic field after being connected to the alternating current; the electric bicycle entering the charging area is charged by the alternating electromagnetic field; a sensor array for acquiring real-time sensor data within the charging area; A FOD detection model that obtains the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determines whether there is a foreign object in the charging area based on the real-time sensor data; A controller controls the power converter according to the foreign object determination result of the FOD detection model, and cuts off the power supply of the transmitting coil when a foreign object is detected in the charging area.

[0015] In one embodiment of the present invention, the FOD detection model includes: A first foreign body identifier, comprising a first input layer, a first feature extraction layer and a first output layer; wherein the real-time sensor data is connected to the first input layer, the first feature extraction layer performs feature extraction on the real-time sensor data; and the first output layer outputs a first recognition result according to the feature extraction result of the first feature extraction layer; A second foreign body identifier, comprising a second input layer, a second feature extraction layer and a second output layer; wherein the real-time sensor data is connected to the second input layer, and the second feature extraction layer extracts features from the real-time sensor data; and the second output layer outputs a second recognition result according to the feature extraction result of the second feature extraction layer; A foreign body detector, comprising a third input layer, a third feature extraction layer, a logic determination layer and a third output layer; wherein the real-time sensor data, the first recognition result and the second recognition result are all connected to the third input layer; the third feature extraction layer performs feature extraction on the data connected to the third input layer; the logic determination layer performs detection and determination on the feature extraction result of the third feature extraction layer based on the determination logic; and the third output layer outputs a foreign body detection result according to the result of the detection and determination; The determination logic includes determining that there is a foreign object in the charging area when any one of the first recognition result, the second recognition result and the analysis result according to the third feature extraction layer identifies that there is a foreign object in the charging area.

[0016] The above technical solution of the present invention has the following beneficial effects compared with the prior art: The electric bicycle wireless charging method and system described in the present invention realize automatic detection of foreign objects in the charging area of ​​the electric bicycle, and intelligently manage the charging of the electric bicycle according to the foreign object detection results, effectively improving the accuracy of foreign object detection and reducing the missed detection rate, and can adapt to the foreign object detection needs in complex environments and realize intelligent charging management. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein: Figure 1 A schematic diagram of a flow chart of a wireless charging method for an electric bicycle in a preferred embodiment of the present invention; Figure 2 A schematic diagram of a process for constructing a FOD detection model in a preferred embodiment of the present invention; Figure 3 A schematic diagram of a process of analyzing real-time sensor data by an FOD detection model in a preferred embodiment of the present invention to determine whether there is a foreign object in a charging area; Figure 4 A schematic diagram of a process for obtaining a first sensor data set in a preferred embodiment of the present invention; Figure 5 A schematic structural diagram of a wireless charging system for an electric bicycle in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention. Embodiment 1

[0019] Reference Figure 1 As shown, an embodiment of the present invention discloses a wireless charging method for an electric bicycle, which specifically includes the following steps: S100: converting the input power into an alternating current of a predetermined frequency; during the conversion process, the predetermined frequency is an important parameter, and the frequency range usually selected will depend on the specific requirements of wireless charging (for example, the efficiency of electromagnetic induction, the charging distance, etc.); this predetermined frequency is usually between tens of kilohertz and hundreds of kilohertz.

[0020] S200: Connect the alternating current to the transmitting coil to generate an alternating electromagnetic field; charge the electric bicycle that enters the charging area through the alternating electromagnetic field; when the transmitting coil is connected to the alternating current, an alternating electromagnetic field is generated, and the alternating electromagnetic field diffuses to the surrounding space in the form of electromagnetic waves; wherein the frequency of the electromagnetic field is the same as the predetermined frequency. The wireless charging system has a specific charging area, and when the electric bicycle enters the area, the receiving coil on the electric bicycle (usually similar to the transmitting coil, using the same material and design) will be excited by the alternating electromagnetic field, and under the action of the alternating electromagnetic field, an induced current will be generated inside the receiving coil on the electric bicycle, and the charging system inside the electric bicycle will rectify the induced current and use it to charge the battery of the electric bicycle.

[0021] S300: Perform real-time sensor detection on the environment in the charging area to obtain real-time sensor data; deploy multiple types of sensors (such as infrared sensors, ultrasonic sensors, electromagnetic sensors, etc.) in the charging area to monitor the environmental information in the charging area; these sensors can detect different types of foreign objects, temperature, distance and other information. Multiple types of sensors continuously monitor the environment and send the collected data to the central control unit in real time to provide multi-dimensional data for subsequent foreign object detection.

[0022] In a specific application scenario, this step first requires deploying various types of sensors in the charging area, preferably infrared sensors, electromagnetic sensors and ultrasonic sensors. The installation location and quantity of each type of sensor are selected according to the on-site conditions to form an array; real-time temperature data is obtained through the infrared sensor array; real-time electromagnetic wave data is obtained through the electromagnetic sensor array; real-time sound wave data is obtained through the ultrasonic sensor array; various types of sensors feed back real-time temperature data, real-time electromagnetic wave data and real-time sound wave data, providing a stable and accurate data basis for subsequent detection of whether there are foreign objects in the charging area.

[0023] S400: Establish an FOD detection model, obtain the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determine whether there are foreign objects in the charging area based on the real-time sensor data; the historical sensor data set of this step includes data on normal charging without foreign objects in the charging area and status data with foreign objects in the charging area; and annotate the historical sensor data to clarify which data samples correspond to the presence of foreign objects and which correspond to normal status; through data analysis, identify and select the features that have the greatest impact on model prediction, and obtain the relationship between sensor data and foreign objects; data preprocessing, such as normalization, missing value processing, and dimensionality reduction, can be performed to ensure data quality and validity. Select a suitable deep learning architecture (such as convolutional neural network CNN, recurrent neural network RNN ​​or improved model), divide the historical sensor data set into training set and test set, use the training set to train the model, adjust the model parameters to improve the recognition accuracy, evaluate the model performance through cross-validation method, ensure that the model has good performance on unseen data, so as to obtain a FOD detection model that meets the requirements, input real-time sensor data into the FOD detection model, and obtain relevant information about foreign objects in the charging area, such as whether there are foreign objects in the charging area, the type of foreign objects, the size of foreign objects and the location of foreign objects. Among them, deep learning can effectively learn the complex relationship between sensor data and foreign objects, and improve the precision and accuracy of foreign object detection; and as time goes by and data accumulates, the model can be continuously optimized to adapt to new environmental changes and foreign object characteristics, and improve the adaptability of the system.

[0024] S500: Control the power supply of the transmitting coil according to the foreign object determination result of the FOD detection model; wherein, when a foreign object is detected in the charging area, the power supply of the transmitting coil is cut off. In this step, the trained and verified FOD detection model is deployed in the charging control system to ensure that the model can receive and process sensor data in real time. The FOD detection model analyzes the real-time sensor data and outputs the foreign object determination result. If a foreign object is detected in the charging area, the model will return a clear signal (such as "foreign object exists" or "no foreign object"). According to the determination result of the FOD model, the control system will make corresponding power supply decisions: when a foreign object is detected, the control system immediately cuts off the power supply of the transmitting coil to prevent the alternating electromagnetic field from causing potential danger to the foreign object or the surrounding environment; at the same time, the system can issue an alarm to notify the user, reminding him to clean up the foreign object before trying to charge; when no foreign object is detected, the normal power supply of the transmitting coil is maintained to allow the charging process to proceed normally.

[0025] The electric bicycle wireless charging method described in the present invention realizes automatic detection of foreign objects in the charging area of ​​the electric bicycle, and intelligently manages the charging of the electric bicycle according to the foreign object detection results, effectively improving the accuracy of foreign object detection and reducing the missed detection rate, and can adapt to the foreign object detection needs in complex environments and realize intelligent charging management.

[0026] In real charging scenarios, metal and plastic foreign objects pose the greatest safety risks to the charging area. For example, metal materials usually have good electrical conductivity. When metal foreign objects accidentally enter the charging area, they may come into contact with the transmitting coil or other live parts, causing a short circuit, which may lead to serious consequences such as fire, electric shock or equipment damage. In addition, during wireless charging, the transmitting coil generates an alternating electromagnetic field. The presence of metal foreign objects may change the distribution of the electromagnetic field, causing the system to work unstably, reducing charging efficiency, and even damaging the device or the charger itself. Although plastic is not a conductive material, some plastics may deform, melt or release harmful gases in high temperature or high energy environments. If the temperature is too high during charging, plastic foreign objects may also cause fires. In addition, during charging, plastic foreign objects will block airflow, and the heat generated by the charging facilities may not be dissipated in time, causing the equipment to overheat and increase the risk of fire.

[0027] In view of the great safety hazard posed by metal foreign objects and plastic foreign objects to the charging area, the wireless charging method of the embodiment of the present invention has been further improved by constructing a more efficient FOD detection model and taking a step-by-step approach to extract data sets related to metal foreign objects and plastic foreign objects from historical sensor data sets, and train identifiers for metal foreign objects and plastic foreign objects respectively. Through targeted model training, the recognition rate of metal foreign objects and plastic foreign objects can be significantly improved, the misjudgment rate can be reduced, and safety can be improved.

[0028] Specifically, Figure 2 As shown, constructing the FOD detection model includes: A data set related to metal foreign matter is extracted from the historical sensor data set to obtain a first sensor data set; deep learning is performed on the first sensor data set based on a deep learning model to obtain a first foreign matter identifier; the historical sensor data set is analyzed according to characteristics such as temperature, distance, and electromagnetic field changes detected by the sensor, and data samples marked as "metal foreign matter" are screened out, and these data are merged to form a first sensor data set. The first foreign matter identifier is obtained by training the historical sensor data related to metal foreign matter alone, and the recognition rate of the first foreign matter identifier for metal foreign matter is greatly improved.

[0029] A data set related to plastic foreign matter is extracted from the historical sensor data set to obtain a second sensor data set; deep learning is performed on the second sensor data set based on a deep learning model to obtain a second foreign matter identifier; and in the same manner as the first foreign matter identifier is obtained, the second foreign matter identifier is obtained by training only the historical sensors related to plastics, thereby greatly improving the recognition rate of plastic foreign matter.

[0030] The historical sensor data set is deep-learned based on a deep learning model to obtain a foreign object detector; the foreign object detector is trained on all historical sensor data (including metal foreign objects and plastic foreign objects) to obtain a universal foreign object detector.

[0031] The FOD detection model is constructed according to the first foreign object identifier, the second foreign object identifier and the foreign object detector. The first foreign object identifier, the second foreign object identifier and the universal foreign object detector are combined to form the final FOD detection model, and a decision-making mechanism is formulated. For example, the first foreign object identifier, the second foreign object identifier and the foreign object detector are used simultaneously for identification of sensor data. As long as one of them identifies a foreign object, it is determined that there is a foreign object in the charging area, which can ensure the accuracy of the detection, thereby improving the recognition rate and reducing the false positive rate. Of course, you can also use a universal foreign object detector for preliminary judgment, and then use a special metal and plastic identifier for further confirmation, which can ensure the accuracy of detection while improving detection efficiency.

[0032] Reference Figure 3 As shown, the FOD detection model determines whether there is a foreign object in the charging area according to the real-time sensor data, including: The first foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; the second foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; the foreign object detector identifies whether there is a foreign object in the charging area according to the real-time sensor data; when any one of the first foreign object identifier, the second foreign object identifier and the foreign object detector identifies that there is a foreign object in the charging area according to the real-time sensor data, it is determined that there is a foreign object in the charging area.

[0033] In specific application scenarios, the first foreign object identifier focuses on identifying metal foreign objects; the second foreign object identifier focuses on identifying plastic foreign objects; the general foreign object detector is designed to identify various types of foreign objects, including but not limited to metal and plastic, and applies deep learning algorithms to analyze all real-time sensor data, and uses a comprehensive model to determine whether there are foreign objects. If any identifier (the first foreign object identifier, the second foreign object identifier, or the foreign object detector) determines that there is a foreign object in the charging area based on real-time sensor data, the system will determine that there is a foreign object in the area. This redundant design improves the reliability of detection: even if one identifier mistakenly determines that there is no foreign object, as long as other identifiers detect foreign objects, the system can still respond to the corresponding safety mechanism. Therefore, by integrating the first foreign object identifier, the second foreign object identifier, and the foreign object detector, the FOD detection model can comprehensively and effectively identify foreign objects in the charging area and ensure the safety of the charging process. This solid detection mechanism provides further safety protection for the field of wireless charging.

[0034] Specifically, after obtaining the real-time sensor data in the charging area, it also includes: Perform amplitude feature extraction on the real-time sensor data to obtain amplitude features; calculate the mean, variance, maximum and minimum values ​​of the amplitude features, and convert the calculation results into a first feature vector; perform frequency domain feature extraction on the real-time sensor data to obtain frequency domain features; calculate the absolute value, skewness and kurtosis of the frequency domain features, and convert the calculation results into a second feature vector; merge the first feature vector and the second feature vector into a real-time data feature vector; the FOD detection model determines whether there is a foreign object in the charging area according to the real-time data feature vector.

[0035] On the basis of the above embodiments, the processing and feature extraction of real-time sensor data are the key to ensure that the FOD detection model can effectively and accurately identify foreign objects. The amplitude feature refers to the absolute change characteristics of the signal in the time domain. A filter is used to remove noise, and then the amplitude feature is extracted from the real-time sensor data; and the statistics of the amplitude feature are calculated: Mean: calculate the average value of the amplitude feature; Variance: calculate the variance of the amplitude feature to reflect the discreteness of the data; Maximum Value: obtain the maximum value in the amplitude feature; Minimum Value: obtain the minimum value in the amplitude feature. The above four statistics (mean, variance, maximum value and minimum value) are combined to form the first eigenvector [mean, variance, maximum value, minimum value]. The frequency domain feature is the characteristic analyzed after converting the time signal to the frequency domain through Fourier transform, which is usually used to reveal the periodic components in the signal. Perform Fourier transform on real-time sensor data to obtain frequency domain signals and extract amplitude information in the frequency domain; Absolute value (Magnitude): calculate the absolute value of the frequency domain feature, corresponding to the intensity of each frequency component of the signal; Skewness: used to measure the asymmetry of the frequency domain feature distribution; Kurtosis: used to measure the steepness of the frequency domain feature distribution. The above three statistics (absolute value, skewness and kurtosis) are combined to form the second eigenvector [absolute value, skewness, kurtosis]. By extracting and integrating amplitude features and frequency domain features of real-time sensor data; this feature-based analysis method not only improves the accuracy of foreign object detection by comprehensively considering the statistical characteristics and spectral characteristics of the signal, but also enhances the robustness of the model in complex environments, providing a guarantee for the safety of the wireless charging system.

[0036] Reference Figure 4 As shown, a data set related to metal foreign matter is extracted from the historical sensor data set to obtain a first sensor data set, including preprocessing the historical sensor data set; performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the metal material; converting the result of the feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; performing K-means clustering on the feature vector to obtain K clusters; performing cosine similarity calculation with a clustering seed on each of the K clusters to obtain K first cosine similarities; if the first cosine similarity is not less than a cosine similarity threshold, adding the historical sensor data corresponding to the cluster with the cosine similarity to the first sensor data set.

[0037] In specific application scenarios, preprocessing includes denoising and enhancing sensor data to ensure the accuracy of feature extraction; as well as removing missing values ​​and outliers to ensure data quality; normalizing the data to make data of different dimensions comparable. Feature extraction is performed on the preprocessed historical sensor data set to obtain feature data with identification capabilities; feature extraction can include features such as amplitude features, frequency domain features, and statistical features, which can reflect the characteristics of metal materials, such as signal energy and frequency characteristics. The extracted feature data is integrated into feature vectors, and each sample corresponds to a feature vector to ensure that the feature vector can effectively support the calculation of cosine similarity. The K-means clustering algorithm is performed on the feature vector. The goal of K-means clustering is to divide the feature vector into K clusters so that the similarity within each cluster is as high as possible, and the difference between clusters is as large as possible; the selection of a suitable K value can be determined by methods such as the elbow rule. The clustering seed is a factor that is certain to be related to metal foreign matter; the clustering seed here is determined based on historical experience data; first, the clustering seed is converted into data that can be operated with the feature vector, and the cosine similarity between the clustering seed and K clusters is calculated respectively to obtain the first cosine similarity; wherein, cluster analysis is performed first, and then the cosine similarity calculation with the clustering seed is performed based on the clustering analysis result, which can reduce the amount of calculation and improve the operation efficiency; illustratively, the clustering seed and the feature vector in the same cluster after clustering are calculated once for cosine similarity to obtain the first cosine similarity. The first cosine similarity is used to describe the similarity between a cluster and the clustering seed. The larger the first cosine similarity, the more similar the cluster is to the clustering seed, and the more relevant the historical sensor data corresponding to the cluster is to the metal foreign matter.

[0038] Extracting a data set related to plastic foreign matter from the historical sensor data set to obtain a second sensor data set includes preprocessing the historical sensor data set; performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the plastic material; converting the result of the feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; performing K-means clustering on the feature vector to obtain K clusters; performing cosine similarity calculation on each cluster in the K clusters with a clustering seed to obtain K second cosine similarities; if the second cosine similarity is not less than a cosine similarity threshold, adding the historical sensor data corresponding to the cluster with the cosine similarity to the second sensor data set.

[0039] The method for obtaining the second sensor data set is the same as the first sensor data set. The two schemes are the same and have the same technical effects, so they will not be described in detail here.

[0040] Reference Figure 5As shown, in one embodiment of the present invention, K-means clustering is performed on the feature vector, including: S1: Initialize the cluster centers of the K clusters; S2: Calculate the distance between the data points of the feature vector and the center of each cluster, and assign each data point to the nearest center point; S3: updating the cluster center according to all assigned feature vectors; S4: Setting a stop condition, and iterating the steps S2-S3 according to the stop condition.

[0041] On the basis of the above embodiment, it is first necessary to select K initial cluster centers. The selection of the initial cluster centers can be divided according to the expected value or manually set according to the detection experience. For each data point of the feature vector, the distance between it and the K cluster centers is calculated. Usually, the Euclidean distance or other similarity metrics can be used to assign the data point to the cluster center with the nearest distance, that is, the data point is assigned to the cluster that minimizes the distance. For each cluster, the average value (center of mass) of all data points in the cluster is calculated, and this average value is used as the new cluster center. This step will update the position of the cluster center, and the iteration is repeated until the stopping condition is met. Common stopping conditions include reaching the maximum number of iterations, the cluster center no longer changes significantly, or reaching a certain convergence threshold. When the K-means algorithm stops iterating, each data point is assigned to a cluster and the center point of the cluster is determined. Embodiment 2

[0042] Reference Figure 5 As shown, the embodiment of the present invention discloses a wireless charging system for an electric bicycle, comprising: A power converter connected to a power source to convert the power source into an alternating current of a predetermined frequency; A transmitting coil is connected to the power converter and generates an alternating electromagnetic field after being connected to the alternating current; the electric bicycle entering the charging area is charged by the alternating electromagnetic field; a sensor array for acquiring real-time sensor data within the charging area; A FOD detection model that obtains the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determines whether there is a foreign object in the charging area based on the real-time sensor data; A controller controls the power converter according to the foreign object determination result of the FOD detection model, and cuts off the power supply of the transmitting coil when a foreign object is detected in the charging area.

[0043] The electric bicycle wireless charging system described in the embodiment of the present invention can effectively realize the wireless charging of the electric bicycle, and the technical effects that can be achieved are as described in the above embodiment, which will not be repeated here.

[0044] Furthermore, the FOD detection model includes: A first foreign body identifier, comprising a first input layer, a first feature extraction layer and a first output layer; wherein the real-time sensor data is connected to the first input layer, the first feature extraction layer performs feature extraction on the real-time sensor data; and the first output layer outputs a first recognition result according to the feature extraction result of the first feature extraction layer; A second foreign body identifier, comprising a second input layer, a second feature extraction layer and a second output layer; wherein the real-time sensor data is connected to the second input layer, and the second feature extraction layer extracts features from the real-time sensor data; and the second output layer outputs a second recognition result according to the feature extraction result of the second feature extraction layer; A foreign body detector, comprising a third input layer, a third feature extraction layer, a logic determination layer and a third output layer; wherein the real-time sensor data, the first recognition result and the second recognition result are all connected to the third input layer; the third feature extraction layer performs feature extraction on the data connected to the third input layer; the logic determination layer performs detection and determination on the feature extraction result of the third feature extraction layer based on the determination logic; and the third output layer outputs a foreign body detection result according to the result of the detection and determination; The determination logic includes determining that there is a foreign object in the charging area when any one of the first recognition result, the second recognition result and the analysis result according to the third feature extraction layer identifies that there is a foreign object in the charging area.

[0045] The FOD monitoring model described in the embodiment of the present invention includes a first foreign object identifier that focuses on identifying possible metal foreign objects; a second foreign object identifier that focuses on identifying possible plastic foreign objects; and a foreign object detector that comprehensively considers the information from the first and second foreign object identifiers to perform more comprehensive foreign object detection. When any of the first recognition result, the second recognition result, or the analysis result of the third feature extraction layer identifies that there is a foreign object in the charging area, the system will determine that there is indeed a foreign object in the charging area. The design of this mechanism enhances the redundancy and reliability of the system, ensuring that even if a certain identifier makes a misjudgment, the effective detection of other parts can still ensure safety. The FOD monitoring model combines multiple foreign object identifiers and an integrated foreign object detector, and uses feature extraction and logic judgment mechanisms to effectively improve the wireless charging system's ability to detect foreign objects in the charging area, providing a strong guarantee for the safe charging of electric bicycles. This structure is not only flexible, but also can adapt to various types of foreign object identification needs.

[0046] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0047] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0048] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0050] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A wireless charging method for an electric bicycle, characterized in that: include, Convert the incoming power into an alternating current of a predetermined frequency; Connecting the alternating current to the transmitting coil to generate an alternating electromagnetic field; The electric bicycle entering the charging area is charged by the alternating electromagnetic field; Performing real-time sensor detection on the environment in the charging area to obtain real-time sensor data; Establishing a FOD detection model, obtaining the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determining whether there are foreign objects in the charging area based on the real-time sensor data; The power supply of the transmitting coil is controlled according to the foreign object determination result of the FOD detection model; wherein, when a foreign object is detected in the charging area, the power supply of the transmitting coil is cut off.

2. The wireless charging method for an electric bicycle according to claim 1, characterized in that: Constructing the FOD detection model includes: Extracting a data set related to metal foreign matter from the historical sensor data set to obtain a first sensor data set; Performing deep learning on the first sensor data set based on a deep learning model to obtain a first foreign body identifier; and Extracting a data set related to plastic foreign matter from the historical sensor data set to obtain a second sensor data set; Performing deep learning on the second sensor data set based on a deep learning model to obtain a second foreign body identifier; and Performing deep learning on the historical sensor data set based on a deep learning model to obtain a foreign object detector; The FOD detection model is constructed according to the first foreign object identifier, the second foreign object identifier and the foreign object detector.

3. The wireless charging method for an electric bicycle according to claim 2, characterized in that: The FOD detection model determines whether there is a foreign object in the charging area according to the real-time sensor data, including: The first foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; The second foreign object identifier identifies whether there is a foreign object in the charging area according to the real-time sensor data; The foreign object detector identifies whether there is a foreign object in the charging area according to the real-time sensor data; When any one of the first foreign object identifier, the second foreign object identifier and the foreign object detector identifies that a foreign object exists in the charging area according to the real-time sensor data, it is determined that a foreign object exists in the charging area.

4. The wireless charging method for an electric bicycle according to any one of claims 1 to 3, characterized in that: After obtaining real-time sensor data within the charging area, it also includes: Performing amplitude feature extraction on the real-time sensor data to obtain amplitude features; Calculating the mean, variance, maximum and minimum values ​​of the amplitude feature, and converting the calculation results into a first eigenvector; Extracting frequency domain features from the real-time sensor data to obtain frequency domain features; Calculating the absolute value, skewness and kurtosis of the frequency domain feature, and converting the calculation results into a second eigenvector; Merging the first feature vector and the second feature vector into a real-time data feature vector; The FOD detection model determines whether there is a foreign object in the charging area based on the real-time data feature vector.

5. The wireless charging method for an electric bicycle according to claim 4, characterized in that: The sensor array comprises: an infrared sensor array, through which real-time temperature data is obtained, and the FOD detection model identifies whether there is a foreign object in the charging area by analyzing the real-time temperature data; An electromagnetic sensor array, through which real-time electromagnetic wave data is obtained, and the FOD detection model identifies whether there is a foreign object in the charging area by analyzing the real-time electromagnetic wave data; An ultrasonic sensor array is used to obtain real-time sound wave data, and the FOD detection model identifies the location of foreign matter by analyzing the real-time sound wave data.

6. The wireless charging method for an electric bicycle according to claim 2, characterized in that: Extracting a data set related to metal foreign matter from the historical sensor data set to obtain a first sensor data set, include, Preprocessing the historical sensor data set; Performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the metal material; Converting the result of feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; Perform K-means clustering on the feature vector to obtain K clusters; Calculating the cosine similarity between each of the K clusters and the cluster seed to obtain K first cosine similarities; If the first cosine similarity is not less than the cosine similarity threshold, the historical sensor data corresponding to the cluster having the cosine similarity is added to the first sensor data set.

7. The wireless charging method for an electric bicycle according to claim 2, characterized in that: Extracting a data set related to plastic foreign matter from the historical sensor data set to obtain a second sensor data set, include, Preprocessing the historical sensor data set; Performing feature extraction on the preprocessed historical sensor data set; wherein the feature data obtained after the feature extraction is related to the plastic material; Converting the result of feature extraction into a feature vector; wherein the obtained feature vector supports cosine similarity calculation; Perform K-means clustering on the feature vector to obtain K clusters; Calculating the cosine similarity between each of the K clusters and the cluster seed to obtain K second cosine similarities; If the second cosine similarity is not less than the cosine similarity threshold, the historical sensor data corresponding to the cluster having the cosine similarity is added to the second sensor data set.

8. The wireless charging method for an electric bicycle according to claim 6 or 7, characterized in that: Performing K-means clustering on the feature vector includes: S1: Initialize the cluster centers of the K clusters S2: Calculate the distance between the data points of the feature vector and the center of each cluster, and assign each data point to the nearest center point; S3: updating the cluster center according to all assigned feature vectors; S4: Setting a stop condition, and iterating the steps S2-S3 according to the stop condition.

9. A wireless charging system for an electric bicycle, characterized in that: include, A power converter connected to a power source to convert the power source into an alternating current of a predetermined frequency; A transmitting coil is connected to the power converter and generates an alternating electromagnetic field after being connected to the alternating current; the electric bicycle entering the charging area is charged by the alternating electromagnetic field; a sensor array for acquiring real-time sensor data within the charging area; A FOD detection model that obtains the relationship between sensor data and foreign objects through deep learning of historical sensor data sets, and determines whether there is a foreign object in the charging area based on the real-time sensor data; A controller controls the power converter according to the foreign object determination result of the FOD detection model, and cuts off the power supply of the transmitting coil when a foreign object is detected in the charging area.

10. The electric bicycle wireless charging system according to claim 9, characterized in that: The FOD detection model includes, A first foreign body identifier, comprising a first input layer, a first feature extraction layer and a first output layer; wherein the real-time sensor data is connected to the first input layer, the first feature extraction layer performs feature extraction on the real-time sensor data; and the first output layer outputs a first recognition result according to the feature extraction result of the first feature extraction layer; A second foreign body identifier, comprising a second input layer, a second feature extraction layer and a second output layer; wherein the real-time sensor data is connected to the second input layer, and the second feature extraction layer extracts features from the real-time sensor data; and the second output layer outputs a second recognition result according to the feature extraction result of the second feature extraction layer; A foreign body detector, comprising a third input layer, a third feature extraction layer, a logic determination layer and a third output layer; wherein the real-time sensor data, the first recognition result and the second recognition result are all connected to the third input layer; the third feature extraction layer performs feature extraction on the data connected to the third input layer; the logic determination layer performs detection and determination on the feature extraction result of the third feature extraction layer based on the determination logic; and the third output layer outputs a foreign body detection result according to the result of the detection and determination; The determination logic includes determining that there is a foreign object in the charging area when any one of the first recognition result, the second recognition result and the analysis result according to the third feature extraction layer identifies that there is a foreign object in the charging area.

Images