A cycling data processing method, device and terminal equipment

Abstract

The application provides a cycling data processing method and device and a terminal equipment, and is suitable for the technical field of data processing. The method comprises the following steps: obtaining cycling personnel movement rule information and cycling safety situation information according to cycling personnel identity attribute information, current cycling physiological information, current cycling movement situation information, historical cycling physiological information and historical cycling movement situation information; generating cycling personnel safety risk representation information according to the cycling personnel movement rule information and the cycling safety situation information; and generating cycling safety early warning information according to the cycling personnel safety risk representation information, current cycling environment information, historical cycling environment information, a cycling safety early warning information generation model and a cycling safety early warning strategy library. The application precisely adapts to individual differences of cycling personnel and dynamically changing cycling scenes, realizes personalized cycling safety early warning, improves the precision and efficiency of cycling data processing, and reduces cycling safety risks.

Description

Technical Field

[0001] This application belongs to the field of data processing technology, and in particular relates to cycling data processing methods, devices and terminal equipment. Background Technology

[0002] With the popularization of green travel concepts and the development of intelligent transportation technologies, the field of cycling safety early warning has entered a period of rapid development in multi-source perception and intelligent analysis. Currently, the industry has formed a perception system centered on millimeter-wave radar, multispectral vision, and inertial measurement units (IMUs) around proactive safety protection. Combined with edge computing and cloud collaborative architecture, it has achieved a technological evolution from passively recording accidents to proactively predicting risks.

[0003] Existing technologies typically employ a perception model using a single or a small combination of sensors, determining risk and triggering warnings based on fixed thresholds. However, the reliance on fixed thresholds for risk assessment makes it difficult to dynamically adapt to different riding scenarios and individual differences. Summary of the Invention

[0004] In view of this, embodiments of this application provide a cycling data processing method, apparatus, and terminal device, aiming to solve the problems of poor risk assessment flexibility, low data processing accuracy, and lack of personalized early warning strategies in the prior art.

[0005] The first aspect of this application provides a cycling data processing method, including:

[0006] Based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information, the cyclist's movement pattern information and cycling safety status information are obtained.

[0007] Based on the cyclist's movement pattern information and cycling safety situation information, generate cyclist safety risk representation information;

[0008] Based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library, cycling safety warning information is generated.

[0009] A second aspect of this application provides a cycling data processing device, comprising:

[0010] The information acquisition module is used to acquire cyclist identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple current cycling environment information, multiple historical cycling physiological information, multiple historical cycling movement status information, multiple historical cycling environment information, and a cycling safety warning information generation model.

[0011] The module for generating cyclist movement patterns and safety status information is used to obtain cyclist movement pattern information and cycling safety status information based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information.

[0012] The cyclist safety risk representation information generation module is used to generate cyclist safety risk representation information based on the cyclist's movement pattern information and cycling safety status information;

[0013] The cycling safety warning information generation module is used to generate cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library.

[0014] A third aspect of this application provides a terminal device, the terminal device including a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the computer program to implement the steps of the cycling data processing method described in the first aspect above.

[0015] A fourth aspect of this application provides a computer-readable storage medium, comprising: storing a computer program, which, when executed by a processor, implements the steps of the cycling data processing method described in the first aspect above.

[0016] The beneficial effects of this application embodiment compared with the prior art are: this application accurately adapts to the individual differences and dynamic changes in cycling scenarios of cyclists, realizes personalized cycling safety warnings, and constructs information on the movement patterns and cycling safety status of cyclists based on multi-dimensional cycling data, so that the generated cycling safety warning information is more in line with the current cycling safety needs, thereby improving the accuracy and efficiency of cycling data processing, reducing cycling safety risks, and providing reliable protection for the travel safety of cyclists. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment 1 of this application;

[0019] Figure 2 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment 2 of this application;

[0020] Figure 3 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment 3 of this application;

[0021] Figure 4 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment 4 of this application;

[0022] Figure 5 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment 5 of this application;

[0023] Figure 6 This is a schematic diagram illustrating the implementation flow of the cycling data processing method provided in Embodiment Six of this application;

[0024] Figure 7 This is a schematic diagram of the structure of the cycling data processing device provided in the embodiments of this application;

[0025] Figure 8 This is a schematic diagram of the terminal device provided in the embodiments of this application. Detailed Implementation

[0026] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0027] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0028] Figure 1 A flowchart illustrating the implementation of the cycling data processing method provided in Embodiment 1 of this application is shown, and is described in detail below:

[0029] Step S101: Obtain cyclist identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple current cycling environment information, multiple historical cycling physiological information, multiple historical cycling movement status information, multiple historical cycling environment information, and a cycling safety warning information generation model.

[0030] In this embodiment, the cyclist's identity attribute information can include the cyclist's age, gender, years of cycling experience, and cycling preference type. This information is used to match different individuals' cycling abilities and risk tolerance standards, such as the emergency response speed reference range corresponding to different years of cycling experience. This information can be manually entered by the cyclist through the interactive interface of the smart cycling terminal. Multiple current cycling physiological information items are real-time data reflecting the cyclist's physical state during cycling, used to determine whether the cyclist is in a state of fatigue, hypoxia, or other discomfort, thus affecting the assessment of cycling safety risks. Multiple current cycling movement status information items are real-time movement status data of the cyclist during cycling, used to determine the stability and rationality of cycling operations. Multiple current cycling environment information items are real-time environmental data surrounding the cycling scene, used to analyze safety hazards in the current cycling scene. Multiple historical cycling physiological information items are physiological data recorded during the cyclist's past cycling experiences, used to compare and analyze the changing patterns of their cycling physiological state, assisting in the optimization of personalized safety warning strategies. Situational information is a record of a cyclist's past riding status, used to uncover their riding habits and movement patterns; multiple historical riding environment information consists of environmental data from past riding experiences, used to analyze riding risk characteristics under different environments; the cycling safety warning information generation model can be a deep learning warning model based on multi-source data fusion, used to dynamically generate accurate safety warning information based on multi-dimensional cycling data. This can be achieved by training a multi-source cycling data-warning strategy dataset, combined with a CNN feature extraction model and a Transformer attention mechanism model. Multi-dimensional data correlation analysis is introduced during the training process to improve the accuracy and real-time performance of the warning model.

[0031] In this embodiment, the current cycling physiological information includes current cycling heart rate, current cycling blood oxygen saturation, and current cycling body surface sensation information. The current cycling heart rate refers to the cyclist's real-time heart rate during cycling, a crucial indicator for determining fatigue or overexertion. A heart rate consistently above 120 beats per minute indicates fatigue. This information can be collected in real-time by a photoelectric heart rate sensor built into the smart cycling helmet. The current cycling blood oxygen saturation refers to the cyclist's real-time blood oxygen content, reflecting their oxygen supply status. A blood oxygen saturation below 95% indicates hypoxia. This information can be synchronously collected and transmitted to the cycling data processing system via a smart bracelet. The current cycling body surface sensation information refers to the cyclist's real-time body surface temperature and comfort level during cycling, which fluctuates with cycling intensity and environmental changes. For example, a body surface temperature above 37.5℃ and a feeling of heat may affect the cyclist's concentration. This information can be obtained in real-time by a temperature sensor built into the smart cycling apparel.

[0032] In this embodiment, the current cycling motion status information includes current straight-line cycling status information, current cornering cycling status information, and current gear shifting start / stop status information. The current straight-line cycling status information refers to the real-time status data of the cyclist while cycling in a straight line, including straight-line speed, straight-line stability, and vehicle lean angle. For example, a straight-line speed fluctuation greater than 5 km / h lasting for more than 3 seconds is identified as an unstable straight-line state. This current straight-line cycling status information can be collected in real-time by the inertial measurement unit (IMU) built into the bicycle. The current cornering cycling status information refers to the information of the cyclist when performing a cornering maneuver. Real-time status data includes steering angle, steering speed, and steering smoothness. For example, if the steering angle is too large and the steering speed is too fast, it is identified as a steering risk state. The current riding cornering steering status information can be collected in collaboration between the vehicle's steering sensor and vision sensor. The current riding gear shifting start-stop status information refers to the real-time status data when the rider performs gear shifting, starting, or stopping operations, including gear shift range, start-stop response speed, and start-stop smoothness. For example, if the speed drop is greater than 10km / h during an emergency stop, it is identified as a start-stop risk state. The current riding gear shifting start-stop status information can be collected in real time by the vehicle's gear shift sensor and speed sensor.

[0033] In this embodiment, the current cycling environment information includes current road conditions, current ambient temperature and humidity, and surrounding traffic and pedestrian flow. The current road conditions refer to real-time data on the road the cyclist is currently on, including road surface smoothness, obstacle distribution, road gradient, and lane width. For example, raised obstacles with a diameter greater than 5cm are identified as road condition risks. This road condition information can be collected collaboratively by an onboard visual sensor and millimeter-wave radar. The current ambient temperature and humidity information refers to the current road conditions of the cyclist's current location. Real-time temperature and humidity data of the surrounding environment, such as temperatures above 35℃ or below 0℃, and humidity above 80%, may affect the cyclist's operation and road conditions. The current cycling environment's meteorological temperature and humidity information can be collected in real time by the vehicle's meteorological sensor. The current cycling environment's surrounding traffic and pedestrian information can refer to the real-time number of vehicles, vehicle speeds, number of pedestrians, and pedestrian movement trajectories around the cyclist. If the safe distance between the surrounding vehicles and the cyclist is less than 1.5 meters, it is identified as a traffic risk. The current cycling environment's surrounding traffic and pedestrian information can be collected in collaboration with the vehicle's millimeter-wave radar and visual sensors.

[0034] In this embodiment, historical cycling physiological information includes historical cycling heart rate information, historical cycling blood oxygen saturation information, and historical cycling surface sensation information. These information can refer to the heart rate, blood oxygen saturation, and surface sensation data recorded during each of the cyclist's past rides, reflecting the changing patterns of their cycling physiological state. This information can be used to compare the current cycling physiological state and train a personalized cycling safety warning information generation model. The historical cycling physiological information can be collected in advance by the system before formal operation, based on the cyclist's past cycling physiological data. After each ride, the system overlays and updates the current cycling physiological information with the existing historical cycling physiological information to generate and store a continuous historical record.

[0035] In this embodiment, historical cycling motion status information includes historical straight-line riding status information, historical curve turning status information, and historical gear shifting and start-stop status information. The historical straight-line riding status information, historical curve turning status information, and historical gear shifting and start-stop status information can refer to the cyclist's records of straight-line riding, curve turning, and gear shifting and start-stop statuses during each past ride. This information is used to analyze the cyclist's riding habits and motion patterns, assisting in the judgment of the current cycling motion status and the optimization of early warning strategies. Historical cycling motion status information can be recorded synchronously with the current cycling motion status information, continuously collected through various onboard sensors, automatically accumulated and dynamically updated after each ride, forming historical data storage.

[0036] In this embodiment, historical cycling environment information includes historical cycling road condition information, historical cycling environment weather temperature and humidity information, and historical cycling environment surrounding traffic and pedestrian flow information. Among them, historical cycling road condition information, historical cycling environment weather temperature and humidity information, and historical cycling environment surrounding traffic and pedestrian flow information can refer to the road conditions, weather temperature and humidity, and surrounding traffic and pedestrian flow data of each previous ride by the cyclist. This information is used to analyze the cycling risk characteristics under different environments and assist in the risk assessment of the current cycling environment. Historical cycling environment information can be continuously recorded by various onboard environmental sensing sensors. After each ride, the system automatically accumulates and dynamically updates the information as historical data storage.

[0037] Step S102: Based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information, obtain the cyclist's movement pattern information and cycling safety status information.

[0038] In this embodiment, the cyclist's identity attribute information can be first one-hot encoded to generate a basic feature vector, serving as the core identifier of individual differences among cyclists. Then, combining multiple current cycling physiological and dynamic status information, threshold discrimination and a random forest algorithm are used to determine the cyclist's current cycling state and physical endurance in real time. Next, based on multiple historical cycling physiological and dynamic status information, and combined with standard cycling parameters corresponding to the cyclist's identity attribute information, a dynamic cycling motion pattern feature space is constructed using an LSTM+Transformer hybrid temporal architecture. This space includes key features such as cycling speed variation patterns, cycling operation habits, and physiological state fluctuation patterns. A collaborative filtering algorithm is used to calculate the similarity between current and historical cycling data. Referring to the motion patterns and safety status of highly similar historical samples, the judgment of the current cycling state is optimized, thereby obtaining cyclist motion pattern information and cycling safety status information. The cyclist motion pattern information can be the inherent patterns of cycling speed, cycling operation, and physiological state changes formed by the cyclist over a long period. The cycling safety status information can be a description of the safety level corresponding to the cyclist's physical state and movement operation during the current cycling process, such as safe state, low-risk state, medium-risk state, and high-risk state.

[0039] Step S103: Generate cyclist safety risk representation information based on the cyclist's movement pattern information and cycling safety status information.

[0040] In this embodiment, it is understood that both the cyclist's movement pattern information and the cycling safety status information are closely related to cycling safety risks. The cyclist's movement pattern information and cycling safety status information can be matched based on a pre-set logical matrix, where the pre-set logical matrix can be manually configured. For example, if the cyclist's movement pattern information is "uniform and stable cycling with minimal physiological fluctuations" and the cycling safety status information is "safe state," then it corresponds to "low risk"; if the cyclist's movement pattern information is "frequent speed changes and turns with large physiological fluctuations" and the cycling safety status information is "moderate risk state," then it corresponds to "medium-high risk"; if the cyclist's movement pattern information is "drastic fluctuations in cycling speed and obvious fatigue cycling characteristics" and the cycling safety status information is "high risk state," then it corresponds to "extremely high risk." By combining historical data on similar motion patterns and the occurrence of risks in safety situation combinations, such as the probability of risk occurrence under the past "frequent speed changes + moderate risk state", the mapping rules can be dynamically calibrated. Then, a structured safety risk representation information containing the current safety risk level, risk-related factors, and risk development trend can be output as the core basis for generating cycling safety warning information.

[0041] Step S104: Generate cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and preset cycling safety warning strategy library.

[0042] In this embodiment, the preset cycling safety warning strategy library can be a set of strategies that include warning methods, warning content, and warning priorities corresponding to different risk levels and different cycling scenarios. The preset cycling safety warning strategy library can be preset by humans. The cycling safety warning information generation model can be a multi-source data fusion warning model (MDF-APM). This model simultaneously inputs cyclist safety risk representation information, multiple current cycling environment information sets, and multiple historical cycling environment information sets into the MDF-APM model. A BERT pre-trained model encodes various data types to generate multi-dimensional feature vectors. After processing with a self-attention mechanism, contextual features are extracted to reflect the matching relationship between the current safety risk and the cycling environment. These features are then used as control conditions to search and match a pre-defined cycling safety warning strategy library. Combined with warning effect data from historical cycling environments, the warning strategies are dynamically optimized to generate cycling safety warning information that accurately matches the current cycling scenario and current safety risk. The cycling safety warning information can include warning level, warning content, warning prompt method, and warning response suggestions, such as "Medium-high risk, obstacles ahead and heavy traffic nearby, please slow down and pay attention."

[0043] The cycling data processing method provided in this application accurately adapts to individual differences and dynamic changes in cycling scenarios, enabling personalized cycling safety warnings. It constructs information on cyclists' movement patterns and cycling safety status based on multi-dimensional cycling data, making the generated cycling safety warning information more in line with current cycling safety needs. This improves the accuracy and efficiency of cycling data processing, reduces cycling safety risks, and provides reliable protection for cyclists' travel safety.

[0044] Figure 2 The flowchart illustrating the implementation of the cycling data processing method provided in Embodiment 2 of this application is shown. The difference between this method and Embodiment 1 is that step S104 specifically includes:

[0045] Step S201: Perform feature vector transformation processing on the multiple current cycling environment information and multiple historical cycling environment information to generate multiple current cycling environment feature vectors and multiple historical cycling environment feature vectors.

[0046] In this embodiment, multiple current cycling environment information includes current cycling road conditions, current cycling environment weather temperature and humidity information, and current cycling environment surrounding traffic and pedestrian flow information. Multiple historical cycling environment information includes historical cycling road conditions, historical cycling environment weather temperature and humidity information, and historical cycling environment surrounding traffic and pedestrian flow information. Feature vector conversion processing can be implemented using the CNN-LSTM feature extraction model. This model is a pre-set deep learning model that can accurately extract key features from environmental information and complete vector conversion. Specifically, the data from multiple current and historical cycling environment information sources are first converted into structured numerical values. For example, the road surface smoothness in the current cycling road condition information is set to a value of 0-10 (0 represents extremely uneven, 10 represents completely smooth); the temperature in the current cycling environment's temperature and humidity information is set to a real-time value of -20℃ to 45℃, and the humidity is set to a real-time value of 0%-100%; the number of vehicles in the current cycling environment's surrounding traffic and pedestrian flow information is set to a value of 0-50 vehicles / 100 meters, and the number of pedestrians is set to a value of 0-30 people / 100 meters. The numerical conversion standard for historical cycling environment information is consistent with that of the current cycling environment information. These structured numerical values ​​are then normalized, mapping them uniformly to the 0-1 range to avoid the impact of numerical differences between different dimensions of data on feature extraction. The normalization process uses the Min-Max normalization method, where the preset normalization parameters can be manually preset, specifically taking a minimum value of 0 and a maximum value of 1. Then, the normalized structured numerical values ​​are input into the CNN-LSTM feature extraction model. The CNN model extracts spatial features of various environmental information, such as the spatial distribution features of road obstacles and the spatial location features of surrounding traffic. Then, the LSTM model extracts temporal features of environmental information, such as the temporal features of changes in weather temperature and humidity in historical cycling environments and the temporal features of road condition fluctuations. This generates multiple current cycling environment feature vectors corresponding to each current cycling environment information, as well as multiple historical cycling environment feature vectors corresponding to each historical cycling environment information. The dimension of each feature vector is set to 128, which can be preset by the user, specifically set to 128, to ensure that the feature vector can comprehensively represent the core features of environmental information.

[0047] Step S202: Based on the multiple current cycling environment feature vectors, multiple historical cycling environment feature vectors, and a preset cycling safety warning strategy library, the cycling safety warning information generation model is iteratively trained to obtain the trained cycling safety warning information generation model.

[0048] In this embodiment, the preset cycling safety warning strategy library can be manually preset, specifically including four warning levels: low risk, medium risk, medium-high risk, and extremely high risk. Each level corresponds to different cycling scenarios and warning strategies. For example, a low-risk scenario corresponds to "voice prompt to pay attention," a medium-risk scenario corresponds to "flashing lights + voice prompt to slow down," a medium-high risk scenario corresponds to "vibration prompt + flashing lights + voice warning," and an extremely high risk scenario corresponds to "emergency vibration + high-frequency lights + urgent voice warning + simultaneous push to the bound terminal." The preset warning level classification parameters can be manually preset, specifically taking values ​​of low risk (risk value 0-0.2), medium risk (risk value 0.2-0.4), medium-high risk (risk value 0.4-0.7), and extremely high risk (risk value 0.7-1.0). The cycling safety warning information generation model can be a manually preset deep learning model, specifically using a Transformer-UNet hybrid model. This model can achieve correlation analysis of multi-source environmental features and accurate training of the warning model. The training process is as follows: First, multiple current cycling environment feature vectors are concatenated with multiple historical cycling environment feature vectors to generate a fused environment feature matrix. The number of rows in the fused matrix equals the total number of current and historical environment feature vectors, and the number of columns is 128 (consistent with the feature vector dimension). Then, warning labels corresponding to each set of fused environment feature matrices are extracted from a pre-defined cycling safety warning strategy library. For example, a set of fused environment feature matrices might correspond to a warning label of "medium-high risk, obstacles ahead and dense traffic around," which serves as a supervision signal for model training. Finally, the fused environment feature matrix and the corresponding warning labels are input into the Transformer-UNet hybrid model. The Transformer model captures the correlation between different environment feature vectors, such as the current cycling road condition features and historical similar road conditions. The correlation between features, the correlation between current surrounding traffic features and current weather temperature and humidity features, is analyzed. The UNet model is used to deeply mine and reconstruct the features, optimizing the accuracy of feature extraction. Then, the error between the model's output warning prediction and the preset warning label is calculated using the cross-entropy loss function. The preset cross-entropy loss function parameters can be manually set, specifically a learning rate of 0.001, 100 iterations, and a batch size of 32. The parameters of the Transformer-UNet hybrid model are then updated using the backpropagation algorithm, and the above training process is repeated. After each iteration, the validation set error is calculated. Iteration can be stopped when the validation set error is less than 0.05, thus obtaining the trained cycling safety warning information generation model, ensuring that the model can accurately capture the correlation between environmental features and warning strategies.

[0049] Step S203: Generate cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment feature vectors, and the trained cycling safety warning information generation model.

[0050] In this embodiment, the cyclist safety risk representation information is first encoded using a BERT pre-trained model (a pre-defined deep learning model). This model converts the structured safety risk representation information, which includes the current safety risk level, risk-related factors, and risk development trends, into a 128-dimensional safety risk feature vector. This vector maintains the same dimensionality as multiple current cycling environment feature vectors, facilitating subsequent feature fusion. The safety risk feature vector is then fused with these multiple current cycling environment feature vectors, generating a comprehensive feature vector through feature concatenation. This comprehensive feature vector has a dimension of 256 (128-dimensional safety risk feature vector + 128-dimensional current environment feature vector). The pre-defined feature weight parameters during the fusion process can be manually set, specifically 0.6 for the safety risk feature vector and 0.4 for the current cycling environment feature vector, highlighting the synergistic impact of safety risks and the real-time environment. The comprehensive feature vectors are then input into the trained cycling safety warning information generation model. Based on a pre-defined cycling safety warning strategy library, the trained model analyzes and processes the comprehensive feature vectors, combining the correlation between environmental features and warning strategies learned during training. It determines the warning level, content, method, and response suggestions for the current cycling scenario, thereby generating structured cycling safety warning information. For example, if the cyclist's safety risk is described as "medium-high risk," and multiple current cycling environment feature vectors represent "a raised obstacle (6cm in diameter) 50 meters ahead, a safe distance of 1.2 meters between surrounding vehicles and the cyclist, and an ambient temperature of 36℃," the trained cycling safety warning information generation model will generate a warning message such as "Medium-high risk, a raised obstacle ahead and close traffic flow, high ambient temperature leading to fatigue, please reduce speed to below 10km / h, pay attention to surrounding vehicles, and wear protective gear." This ensures that the generated warning information accurately matches the current safety risk and real-time cycling environment, providing targeted safety guidance for cyclists.

[0051] The cycling data processing method provided in this application transforms multiple current cycling environment information and multiple historical cycling environment information into feature vectors, which are then used to iteratively train a cycling safety warning information generation model. This enables the trained model to accurately capture the correlation between environmental features and cycling safety risks, and to generate warning information that is more relevant to the real-time cycling scenario by combining it with the cyclist's safety risk representation information. This improves the accuracy and efficiency of cycling data processing, optimizes the effect of personalized cycling safety warnings, and provides more reliable protection for the travel safety of cyclists.

[0052] Figure 3 The flowchart illustrating the implementation of the cycling data processing method provided in Embodiment 3 of this application is shown. Its difference from Embodiment 2 described above lies in:

[0053] The preset cycling safety warning strategy library includes multiple preset cycling environment benchmark feature vectors and multiple preset cycling safety warning strategy information; among them, the preset cycling environment benchmark feature vectors and the preset cycling safety warning strategy information correspond one-to-one.

[0054] Step S203 specifically includes:

[0055] Step S301: Perform matching processing based on the multiple current cycling environment feature vectors and multiple preset cycling environment benchmark feature vectors to generate multiple current matched cycling environment benchmark feature vectors.

[0056] In this embodiment, the preset cycling environment baseline feature vector can be manually preset. It is a standard feature vector generated in advance based on typical environmental data of different cycling scenarios. Specifically, it has 128 dimensions, which is consistent with the dimensions of multiple current cycling environment feature vectors, facilitating matching calculations. The preset cycling environment baseline feature vector can be generated by clustering massive cycling environment data using the K-means clustering algorithm. The number of clusters is preset to 50, which can be manually preset and specifically set to 50. The matching process can be implemented using the cosine similarity algorithm. The similarity threshold corresponding to this algorithm can be manually preset and specifically set to 0.8. That is, when the cosine similarity between multiple current cycling environment feature vectors and a preset cycling environment baseline feature vector is greater than or equal to 0.8, the two are considered to be successfully matched. Specifically, each current cycling environment feature vector is first compared with all preset cycling environment benchmark feature vectors using cosine similarity calculation. This yields a similarity value between each current cycling environment feature vector and each preset cycling environment benchmark feature vector. Then, preset cycling environment benchmark feature vectors with a similarity value greater than or equal to 0.8 are selected. If multiple preset cycling environment benchmark feature vectors meet the criteria, the one with the highest similarity value is selected as the matching result corresponding to the current cycling environment feature vector. This generates multiple current matched cycling environment benchmark feature vectors, ensuring that each current cycling environment feature vector can match the most suitable preset benchmark feature.

[0057] Step S302: Perform pairing processing based on the multiple current matching post-riding environment baseline feature vectors and multiple preset riding safety warning strategy information to obtain multiple current matching post-riding safety warning strategy information.

[0058] In this embodiment, the preset cycling safety warning strategy information can be manually preset and corresponds one-to-one with multiple preset cycling environment benchmark feature vectors. Each preset cycling environment benchmark feature vector corresponds to a unique preset cycling safety warning strategy information. The specific correspondence can be stored in advance in a preset cycling safety warning strategy library. For example, a preset cycling environment benchmark feature vector representing "smooth road surface, sparse traffic, and suitable weather" corresponds to the preset cycling safety warning strategy information of "low risk, voice prompt to pay attention to road conditions"; a preset cycling environment benchmark feature vector representing "road surface with obstacles, dense traffic, and high temperature environment" corresponds to the preset cycling safety warning strategy information of "medium to high risk, vibration prompt + voice warning, slow down". During the pairing process, the preset cycling safety warning strategy information corresponding to each currently matched cycling environment benchmark feature vector in the preset cycling safety warning strategy library is first obtained. Then, the currently matched cycling environment benchmark feature vector is bound one-to-one with its corresponding preset cycling safety warning strategy information, thereby generating multiple currently matched cycling safety warning strategy information to ensure that each currently matched benchmark feature can correspond to an accurate warning strategy.

[0059] Step S303: Perform splicing processing on the multiple current cycling environment feature vectors and multiple current paired cycling safety warning strategy information to obtain multiple spliced ​​vectors of current cycling environment and warning strategy features.

[0060] In this embodiment, multiple currently paired cycling safety warning strategy information is first encoded. A BERT pre-trained model (a pre-defined deep learning model) is used to convert each currently paired cycling safety warning strategy information into a 128-dimensional warning strategy feature vector, maintaining the same dimension as multiple current cycling environment feature vectors for easy concatenation. Then, each current cycling environment feature vector is concatenated with its corresponding 128-dimensional warning strategy feature vector. The resulting concatenated vector of current cycling environment and warning strategy features has a dimension of 256 (128-dimensional current cycling environment feature vector + 128-dimensional warning strategy feature vector). The concatenation order parameter can be pre-defined, specifically "current cycling environment feature vector first, then warning strategy feature vector," ensuring consistency and standardization of the concatenated vectors. This generates multiple concatenated vectors of current cycling environment and warning strategy features, achieving deep fusion of current environment features and corresponding warning strategy features.

[0061] Step S304: Perform matching processing based on the multiple historical cycling environment feature vectors and multiple preset cycling environment benchmark feature vectors to generate multiple historical matched cycling environment benchmark feature vectors.

[0062] In this embodiment, the preset cycling environment baseline feature vector can be manually preset, specifically with 128 dimensions. The threshold of the cosine similarity algorithm used for matching can also be manually preset, specifically with a value of 0.8, and the clustering number parameter is 50. Specifically, each historical cycling environment feature vector is compared with all preset cycling environment baseline feature vectors using cosine similarity calculation to obtain the similarity value between each historical cycling environment feature vector and each preset cycling environment baseline feature vector. Then, preset cycling environment baseline feature vectors with similarity values ​​greater than or equal to 0.8 are selected. If multiple preset cycling environment baseline feature vectors meet the conditions, the one with the highest similarity value is selected as the matching result corresponding to that historical cycling environment feature vector, thereby generating multiple historical matched cycling environment baseline feature vectors. This ensures that historical environment features can also match the fitting preset baseline features, providing historical dimension support for subsequent model training.

[0063] Step S305: Perform pairing processing based on the multiple historical matched cycling environment baseline feature vectors and multiple preset cycling safety warning strategy information to obtain multiple historical matched cycling safety warning strategy information.

[0064] In this embodiment, the preset cycling safety warning strategy information can be manually preset and corresponds one-to-one with multiple preset cycling environment benchmark feature vectors. The correspondence is stored in a preset cycling safety warning strategy library. Specifically, the preset cycling safety warning strategy information corresponding to each historically matched cycling environment benchmark feature vector in the preset cycling safety warning strategy library is obtained. Then, the historically matched cycling environment benchmark feature vectors are bound one-to-one with their corresponding preset cycling safety warning strategy information to generate multiple historically matched cycling safety warning strategy information, thereby achieving accurate matching between historical environmental benchmark features and warning strategies.

[0065] Step S306: Based on the multiple historical cycling environment feature vectors and multiple historical paired cycling safety warning strategy information, perform splicing processing to obtain multiple historical cycling environment and warning strategy feature splicing vectors.

[0066] In this embodiment, when encoding multiple historical paired cycling safety warning strategy information, the BERT pre-trained model (a pre-defined deep learning model) is still used to convert it into a 128-dimensional warning strategy feature vector, consistent with the dimension of multiple historical cycling environment feature vectors. Then, each historical cycling environment feature vector and its corresponding 128-dimensional warning strategy feature vector are concatenated and spliced ​​in the order of "historical cycling environment feature vector first, then warning strategy feature vector." The resulting spliced ​​historical cycling environment and warning strategy feature vector has a dimension of 256, thus generating multiple spliced ​​historical cycling environment and warning strategy feature vectors. This achieves deep fusion of historical environment features and corresponding warning strategy features, providing complete historical data support for model iterative training.

[0067] Step S307: Based on the multiple current cycling environment and warning strategy feature splicing vectors and multiple historical cycling environment and warning strategy feature splicing vectors, the cycling safety warning information generation model is iteratively trained to obtain the trained cycling safety warning information generation model.

[0068] In this embodiment, the cycling safety warning information generation model can be a pre-defined deep learning model, specifically employing a Transformer-UNet hybrid model to ensure the coherence and consistency of model training. The training process is as follows: First, multiple concatenated vectors of current cycling environment and warning strategy features are merged with multiple concatenated vectors of historical cycling environment and warning strategy features to generate a training dataset. The number of samples in the training dataset equals the total number of concatenated vectors of the two classes, with each sample being a 256-dimensional feature vector. Then, warning labels corresponding to each concatenated vector are extracted from a pre-defined cycling safety warning strategy library as supervisory signals for model training. For example, a concatenated vector of current cycling environment and warning strategy features might correspond to a warning label of "medium-high risk, slow down." Finally, the training dataset and the corresponding warning labels are input into the Transformer-UNet hybrid model. The Transformer model captures the correlation between current and historical concatenated vectors, such as the correlation between a current environment-warning strategy combination and historical similar combinations. The Transformer-UNet hybrid model is trained by deeply mining and reconstructing features using the UNet model to optimize feature extraction accuracy. The error between the model's output warning prediction and the preset warning label is then calculated using the cross-entropy loss function. The parameters of the preset cross-entropy loss function can be manually set, specifically a learning rate of 0.001, 100 iterations, and a batch size of 32. The parameters of the Transformer-UNet hybrid model are then updated using the backpropagation algorithm. This training process is repeated, with the validation set error calculated after each iteration. Iteration stops when the validation set error is less than 0.05 (this error threshold can be manually set, specifically 0.05). This results in a trained cycling safety warning information generation model, ensuring that the model accurately captures the correlation between the combined features of the environment and the warning strategy and the warning result.

[0069] The cycling data processing method provided in this application enables the trained model to accurately learn the correspondence between environmental features and early warning strategies, and to generate more targeted cycling safety early warning information that is more in line with real-time cycling scenarios by combining the cyclist's safety risk representation information. This improves the accuracy and efficiency of cycling data processing, optimizes the effect of personalized cycling safety early warning, and provides more reliable protection for the travel safety of cyclists.

[0070] Figure 4 The flowchart illustrating the implementation of the cycling data processing method provided in Embodiment 4 of this application is shown. The difference between this method and Embodiment 2 is that step S203 specifically includes:

[0071] Step S401: Based on the preset cycling safety warning parameter feature extraction rules, perform parameter feature extraction processing on the preset cycling safety warning strategy library to obtain multiple cycling safety warning parameter feature information.

[0072] In this embodiment, the preset cycling safety warning parameter feature extraction rules can be manually preset. This is a standardized process for extracting key parameter features from a preset cycling safety warning strategy library. Specifically, it includes four core extraction dimensions: warning level parameters, warning prompting method parameters, warning response suggestion parameters, and warning priority parameters. Each extraction dimension corresponds to a specific parameter value range. For example, the warning level parameter can be set to low risk, medium risk, medium-high risk, and extremely high risk; the warning prompting method parameter can be set to voice prompt, flashing light, vibration prompt, emergency vibration + high-frequency light + urgent voice warning + simultaneous push to the bound terminal; and the warning priority parameter can be set to levels 1-5 (level 1 is the lowest, level 5 is the highest). These parameter value ranges can be manually preset. The preset cycling safety warning strategy library can also be manually preset and contains multiple warning strategies corresponding to different cycling scenarios. Each warning strategy includes the aforementioned four core parameters. During the extraction process, according to the preset cycling safety warning parameter feature extraction rules, the warning level parameters, warning prompt method parameters, warning response suggestion parameters, and warning priority parameters are extracted one by one from each warning strategy in the preset cycling safety warning strategy library. These parameters are organized into a structured parameter set, thereby generating multiple cycling safety warning parameter feature information. Each cycling safety warning parameter feature information corresponds to the core parameter of a warning strategy, ensuring that the extracted parameter features are comprehensive and accurate.

[0073] Step S402: Perform vector transformation processing on the feature information of the multiple cycling safety warning parameters to obtain multiple cycling safety warning parameter feature vectors.

[0074] In this embodiment, vector transformation processing can be implemented using a CNN-LSTM feature extraction model. This model is a pre-defined deep learning model that can convert structured parameter features into high-dimensional feature vectors. Specifically, firstly, multiple cycling safety warning parameter features are numerically processed, converting non-numerical parameters into structured values. For example, the warning level parameter is assigned values ​​of low risk = 1, medium risk = 2, medium-high risk = 3, and extremely high risk = 4. The warning priority parameter directly uses values ​​from 1 to 5. The warning prompting method parameter is assigned values ​​of voice prompt = 1, flashing light = 2, vibration prompt = 3, and emergency vibration + high-frequency light + urgent voice warning + simultaneous push to the bound terminal = 4. The warning response suggestion parameter is assigned values ​​from 1 to 3 based on the complexity of the suggestion (level 1 is the simplest, level 3 is the most complex). These assignment standards can be pre-defined. Then, the numerically transformed parameters are normalized using the Min-Max normalization method, mapping the values ​​uniformly to the range of 0-1. The pre-defined normalization parameters can be pre-defined, specifically taking a minimum value of 0 and a maximum value of 1. The normalized parameters are then input into the CNN-LSTM feature extraction model. The CNN model extracts the spatial correlation features of the parameter features, and the LSTM model extracts the logical correlation features of the parameter features, thereby generating multiple cycling safety warning parameter feature vectors. The dimension of each feature vector is set to 128, which can be preset by the user, specifically 128, to ensure that the feature vector can fully represent the core features of the warning parameters.

[0075] Step S403: Merge the multiple current cycling environment feature vectors and multiple cycling safety warning parameter feature vectors to obtain multiple cycling environment and safety warning parameter feature vectors.

[0076] In this embodiment, the merging process is implemented using feature concatenation. The preset merging order parameter can be manually set, specifically "multiple current cycling environment feature vectors first, followed by multiple cycling safety warning parameter feature vectors," ensuring consistency of the merged vectors. Each current cycling environment feature vector is 128-dimensional, and each cycling safety warning parameter feature vector is also 128-dimensional. After concatenating and merging them, the generated cycling environment and safety warning parameter feature vectors have a dimension of 256 (128-dimensional current cycling environment feature vector + 128-dimensional cycling safety warning parameter feature vector). During the merging process, it is necessary to ensure that each current cycling environment feature vector is matched and merged with the corresponding cycling safety warning parameter feature vector. That is, the current cycling environment feature vector representing a certain cycling environment is merged with the cycling safety warning parameter feature vector adapted to that environment, thereby generating multiple cycling environment and safety warning parameter feature vectors. This achieves deep fusion of current cycling environment features and warning parameter features, providing more comprehensive feature support for subsequent model training.

[0077] Step S404: Based on the multiple cycling environment and safety warning parameter feature vectors and multiple historical cycling environment feature vectors, the cycling safety warning information generation model is iteratively trained to obtain the trained cycling safety warning information generation model.

[0078] In this embodiment, the cycling safety warning information generation model can be a pre-defined deep learning model, specifically using the Transformer-UNet hybrid model, consistent with the models in Embodiments 2 and 3, to ensure the continuity of model training. The training process is as follows: First, multiple cycling environment and safety warning parameter feature vectors are concatenated with multiple historical cycling environment feature vectors to generate a fusion training feature matrix. The number of rows in the fusion training feature matrix equals the total number of the two types of feature vectors, and the number of columns is 256 dimensions (consistent with the dimension of the cycling environment and safety warning parameter feature vectors). Multiple historical cycling environment feature vectors need to be expanded to 256 dimensions using zero-padding. The zero-padding dimension parameter can be pre-defined, specifically 128 dimensions (expanding from 128 dimensions to 256 dimensions). Then, warning labels corresponding to each set of fusion training features are extracted from a pre-defined cycling safety warning strategy library as supervision signals for model training. For example, a set of fusion training features might correspond to the warning label "medium risk, flashing lights + voice prompt to slow down". Then, the fused training feature matrix and corresponding warning labels are input into the Transformer-UNet hybrid model. The Transformer model captures the correlation between the cycling environment and warning parameter features, as well as historical environment features, such as the matching relationship between the current environment and warning parameters, and the correlation between the current environment and historical similar environments. The UNet model performs in-depth feature mining and reconstruction to optimize the accuracy of feature extraction. Next, the error between the warning prediction result output by the model and the preset warning label is calculated. The cross-entropy loss function is used to calculate the error value. The preset cross-entropy loss function parameters can be manually preset, specifically with a learning rate of 0.001, 100 iterations, and a batch size of 32. Then, the parameters of the Transformer-UNet hybrid model are updated through the backpropagation algorithm, and the above training process is repeated. After each iteration, the validation set error is calculated. When the validation set error is less than 0.05 (this error threshold can be manually preset, specifically with a value of 0.05), the iteration stops. This results in a trained cycling safety warning information generation model, ensuring that the model can accurately capture the correlation between environmental features, warning parameter features, and warning results.

[0079] The cycling data processing method provided in this application enables the trained model to accurately learn the synergistic effects of warning parameters and environmental features, and to generate cycling safety warning information that is more relevant to real-time cycling scenarios, more targeted and reasonable, by combining the cyclist's safety risk representation information. This improves the accuracy and efficiency of cycling data processing, optimizes the effect of personalized cycling safety warnings, and provides a more solid guarantee for the travel safety of cyclists.

[0080] Figure 5 The flowchart illustrating the implementation of the cycling data processing method provided in Embodiment 5 of this application is shown. The difference between this method and Embodiment 1 is that, after step S104, the method further includes:

[0081] Step S501: Perform feature extraction processing on the cycling safety warning information to obtain cycling safety warning feature information; the cycling safety warning feature information includes cycling warning level feature information, cycling warning push timeliness feature information, and cycling warning prompt type information.

[0082] In this embodiment, feature extraction can be implemented using a CNN-based feature extraction model. This model is a pre-defined deep learning model capable of accurately extracting core features from cycling safety warning information. Specifically, the cycling safety warning information is first converted into structured text data, clearly distinguishing three core dimensions: cycling warning level features, cycling warning push timeliness features, and cycling warning prompt type information. Then, the structured text data is feature extracted using a CNN-based feature extraction model. The cycling warning level features are extracted as low risk, medium risk, medium-high risk, and extremely high risk, specifically taking values ​​of low risk (risk value 0-0.2), medium risk (risk value 0.2-0.4), medium-high risk (risk value 0.4-0.7), and extremely high risk (risk value 0.7-1.0). The values ​​can be preset by the user. The timing feature information for cycling warning pushes corresponds to the extraction of timing identifiers for instant push and delayed push (delay 1-3 seconds), specifically instant push (0-second delay) and delayed push (1 second, 2 seconds, 3 seconds). These values ​​can be preset by the user. The cycling warning prompt type information corresponds to the extraction of type identifiers for voice prompts, flashing lights, vibration prompts, and emergency vibration + high-frequency light + urgent voice warning + simultaneous push to the bound terminal. Specifically, the values ​​are voice prompts, flashing lights, vibration prompts, and emergency vibration + high-frequency light + urgent voice warning + simultaneous push to the bound terminal. These values ​​can be preset by the user. Then, the extracted three types of features are normalized and converted into standardized feature data to generate cycling safety warning feature information, ensuring that the extracted features accurately represent the core attributes of cycling safety warning information.

[0083] Step S502: Based on the preset mapping relationship between cycling safety warning features and cycling terminal vibration intensity, cycling terminal vibration intensity information is generated according to the cycling safety warning feature information and cycling warning push timeliness feature information.

[0084] In this embodiment, the mapping relationship between the preset cycling safety warning features and the vibration intensity of the cycling terminal can be preset manually. It is a multi-dimensional rule matrix constructed based on historical warning data and cyclist feedback. The specific mapping relationship is as follows: when the cycling warning level is low risk and the cycling warning push time is instant, the corresponding cycling terminal vibration intensity is weak (vibration frequency 5Hz, amplitude 0.3mm); when the cycling warning level is medium risk and the cycling warning push time is instant, the corresponding cycling terminal vibration intensity is medium (vibration frequency 8Hz, amplitude 0.5mm); When the warning level is medium-high risk and the cycling warning push is instant, the corresponding vibration intensity of the cycling terminal is medium-high (vibration frequency 12Hz, amplitude 0.8mm); when the warning level is extremely high risk and the cycling warning push is instant, the corresponding vibration intensity of the cycling terminal is high (vibration frequency 15Hz, amplitude 1.0mm); if the cycling warning push is delayed, regardless of the warning level, the vibration intensity is reduced by 2Hz and 0.2mm from the corresponding level. The specific values ​​of the vibration frequency and amplitude can be preset. Specifically, the cycling warning level feature information is first extracted from the cycling safety warning feature information, combined with the cycling warning push timeliness feature information, and then a search and matching is performed in the preset mapping relationship between cycling safety warning features and cycling terminal vibration intensity to determine the corresponding cycling terminal vibration intensity parameters. These parameters are then converted into electrical signal commands to generate cycling terminal vibration intensity information. This information can be sent to cycling terminals such as smart cycling helmets and smart cycling gloves to control them to vibrate at a specified intensity, achieving coordinated adaptation with the warning level and push timeliness.

[0085] Step S503: Based on the preset mapping relationship between cycling safety warning features and cycling warning light control, cycling warning light control information is generated according to the cycling warning level feature information and cycling warning prompt type information.

[0086] In this embodiment, the preset mapping relationship between cycling safety warning features and cycling warning light control can be preset manually, covering the correspondence between cycling warning level, cycling warning type and light brightness, light color, and light flashing frequency. The specific mapping rules are as follows: When the cycling warning level is low risk and the cycling warning type is voice prompt, the corresponding light control information is low brightness (10 lux), warm white (3000K), and constant light (flashing frequency 0Hz); when the cycling warning level is medium risk and the cycling warning type is flashing light, the corresponding light control information is medium brightness (30 lux), warm white (3000K), and constant light (flashing frequency 0Hz). When the cycling warning level is medium to high risk and the cycling warning prompt type is vibration prompt + light flashing, the corresponding light control information is medium to high brightness (50 lux), orange (2000K), and high frequency flashing (2Hz); when the cycling warning level is extremely high risk and the cycling warning prompt type is emergency vibration + high frequency light + urgent voice warning + synchronous push to the bound terminal, the corresponding light control information is high brightness (80 lux), red (1800K), and extremely high frequency flashing (3Hz). The specific values ​​of the above light brightness, color, and flashing frequency can be preset by humans. The preset mapping relationships between cycling safety warning features and cycling terminal vibration intensity, as well as between preset cycling safety warning features and cycling indicator light control, can all be the optimal mapping relationships obtained through experiments. The specific construction process is as follows: First, core warning features (warning level, prompt type, push timeliness) are extracted from the preset cycling safety warning strategy library. Simultaneously, gradient combinations are set for cycling terminal vibration parameters (vibration frequency gradient of 5-15Hz, amplitude gradient of 0.3-1.0mm) and cycling indicator light parameters (brightness gradient of 10-80lux, hue gradient of 1800-3000K, flashing frequency gradient of 0-3Hz). Then, the operational feedback and risk avoidance effects of cyclists under different warning-vibration-light combinations are collected. Based on the goal that cyclists can quickly identify warnings and avoid risks in a timely manner, the initial mapping rules are determined through statistical analysis, and then optimized and improved to form the final mapping relationship. Specifically, the system first extracts the characteristic information of the cycling warning level and the type of cycling warning prompt. It then matches these information with the preset mapping relationship between cycling safety warning characteristics and cycling prompt light control. Next, it determines the corresponding light brightness, light color, and light flashing frequency parameters. These parameters are then converted into control signals to generate cycling prompt light control information. This information can be sent to the vehicle lighting equipment and the lighting module of the smart cycling helmet to achieve precise control of the lights.

[0087] The cycling data processing method provided in this application embodiment realizes the coordinated control of cycling safety warnings and multimodal prompts, improves the recognition and timeliness of cycling warnings, accurately matches the current cycling safety risk level and warning needs, enables cyclists to quickly perceive warning information and take timely avoidance measures, thereby reducing cycling safety risks and improving the comprehensiveness and adaptability of cycling safety protection.

[0088] Figure 6 The flowchart illustrating the implementation of the cycling data processing method provided in Embodiment Six of this application is shown. The difference between this method and Embodiment Five is that, after step S503, the method further includes:

[0089] Step S601: Obtain multiple post-warning cycling physiological information, multiple post-warning cycling motion status information, and multiple post-warning cycling environmental information.

[0090] In this embodiment, the multiple post-warning cycling physiological information are real-time physiological data of the cyclist during the continuous operation of cycling safety warning information push, cycling terminal vibration adjustment, and cycling indicator light adjustment. These include post-warning cycling heart rate information, post-warning cycling blood oxygen saturation information, and post-warning cycling surface sensation information. The post-warning cycling heart rate information is collected in real time by the photoelectric heart rate sensor built into the smart cycling helmet. The post-warning cycling blood oxygen saturation information is collected synchronously by the smart wristband and transmitted to the cycling data processing system. The post-warning cycling surface sensation information is obtained in real time by the temperature sensor built into the smart cycling suit. These information are used to determine whether the cyclist's physical condition has returned to a stable state after the warning and whether they are still in a state of fatigue or hypoxia. Multiple post-warning cycling status information are real-time motion data of the cyclist after the warning prompt. These include information on the straight-line cycling status, the curve-turning status, and the gear shifting and start-stop status. The straight-line cycling status information is collected in real time by the inertial measurement unit (IMU) built into the bicycle. The curve-turning status information is collected in collaboration with the onboard steering sensor and vision sensor. The gear shifting and start-stop status information is collected in real time by the gear shift sensor and speed sensor of the bicycle. These information are used to determine whether the cyclist has adjusted their cycling operation according to the warning prompt and whether the cycling status has become stable. Multiple post-warning cycling environment information sets are real-time environmental data of the surrounding environment after the warning is issued. These include road conditions, ambient temperature and humidity, and traffic and pedestrian flow. Road conditions are collected via a combination of onboard vision sensors and millimeter-wave radar; ambient temperature and humidity are collected in real-time by onboard weather sensors; and traffic and pedestrian flow are collected via a combination of onboard millimeter-wave radar and vision sensors. This data is used to analyze whether the cycling environment has changed after the warning and whether safety hazards have been mitigated. During data collection, the sampling frequency of all sensors can be preset, specifically 1 time per second, ensuring that the collected data accurately reflects the cycling status and environmental changes after the warning. The collected data is then structured and organized to generate multiple sets of post-warning cycling physiological information, multiple sets of post-warning cycling motion status information, and multiple sets of post-warning cycling environmental information.

[0091] Step S602: Send the multiple post-warning cycling physiological information, multiple post-warning cycling movement status information, and multiple post-warning cycling environmental information to the user terminal.

[0092] In this embodiment, the user terminal can be a smartphone, smart cycling terminal, or a linked family member's terminal, such as a mobile app or a smart cycling control screen. First, multiple sets of post-warning cycling physiological information, post-warning cycling movement status information, and post-warning cycling environmental information are formatted, converting various structured data into visual charts, such as cycling heart rate change curves, cycling speed fluctuation charts, and real-time environmental parameter display graphs. Data comparison information before and after the warning is also annotated, allowing users to intuitively view the warning effect. The preset transmission frequency can be manually preset, specifically sending real-time data every 5 seconds and a data summary report every 30 seconds. The transmission method uses a wireless communication module (such as Bluetooth or 5G) to ensure the real-time and stable transmission of data. Then, the formatted information is sent to the user terminal via the wireless communication module, allowing the cyclist to view the impact of the warning prompts on their cycling status and operations in real time, while also allowing the linked family member to monitor the cyclist's travel safety in real time.

[0093] In this embodiment, preferably, after sending multiple post-warning cycling physiological information, multiple post-warning cycling movement status information, and multiple post-warning cycling environmental information to the user terminal, it is also possible to determine whether the cyclist has adjusted their cycling state according to the warning prompts, whether the current cycling safety risk has decreased, and to regenerate cycling safety warning information based on these post-warning information, and adjust the vibration intensity information of the cycling terminal and the cycling indicator light control information according to the regenerated cycling safety warning information. Threshold discrimination and random forest algorithms can be used to determine the changes in the cycling safety risk level after the warning in real time. For example, if the risk level is determined to be medium to high before the warning, and after the warning, multiple post-warning cycling physiological information shows that the heart rate has returned to normal, multiple post-warning cycling movement status information shows that the cycling operation has become more stable, and multiple post-warning cycling environmental information shows that the safety hazard has been alleviated, then the risk is determined to have decreased to medium risk. In this case, cycling safety warning information corresponding to medium risk is regenerated, and the vibration intensity of the cycling terminal is adjusted to medium intensity and the cycling indicator light is adjusted to medium brightness yellow flashing based on this information. For example, if the risk level is determined to be extremely high before the warning, and after the warning, various post-warning information shows that the cyclist has not adjusted their operation in time and the safety hazard has not been alleviated, then... If the risk is still determined to be extremely high, the current cycling safety warning information, cycling terminal vibration intensity information, and cycling indicator light control information will be maintained. At the same time, the frequency of urgent voice prompts will be increased (once every 1 second) to urge cyclists to take timely evasive measures. For example, if the risk was determined to be medium before the warning, and after the warning, various post-warning information shows that the cycling status is stable and the safety hazard has been eliminated, then the risk is determined to be low. In this case, cycling safety warning information corresponding to low risk will be regenerated, and the cycling terminal vibration intensity will be adjusted to weak intensity, and the cycling indicator light will be adjusted to low brightness warm white constant light to avoid excessive warnings affecting the cycling experience.

[0094] The cycling data processing method provided in this application not only enables cyclists and their registered family members to monitor their cycling safety status in real time after receiving a warning, but also provides a reliable basis for continuously and dynamically adjusting cycling safety warning strategies, cycling terminal vibration intensity, and cycling indicator light parameters. This improves the accuracy and effectiveness of cycling safety warnings, achieves continuous optimization of cycling safety protection, reduces cycling safety risks, and provides stronger support for cyclists' travel safety.

[0095] Corresponding to the method in the above embodiments, Figure 7 A structural block diagram of the cycling data processing device provided in the embodiments of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown. Figure 7 The cycling data processing device in the example can be the execution subject of the cycling data processing method provided in the aforementioned embodiment 1.

[0096] Reference Figure 7 The cycling data processing device includes:

[0097] The information acquisition module 710 is used to acquire cyclist identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple current cycling environment information, multiple historical cycling physiological information, multiple historical cycling movement status information, multiple historical cycling environment information, and a cycling safety warning information generation model.

[0098] The cyclist movement pattern and safety status information generation module 720 is used to obtain cyclist movement pattern information and cycling safety status information based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information.

[0099] The cyclist safety risk representation information generation module 730 is used to generate cyclist safety risk representation information based on the cyclist's movement pattern information and cycling safety status information;

[0100] The cycling safety warning information generation module 740 is used to generate cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library.

[0101] The process by which each module in the cycling data processing device provided in this application implements its respective function can be found in the foregoing. Figure 1 The description of Embodiment 1 shown will not be repeated here.

[0102] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0103] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0104] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0105] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0106] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. It should also be understood that although the terms "first," "second," etc., are used in the text to describe various elements in some embodiments of this application, these elements should not be limited by these terms. These terms are merely used to distinguish one element from another. For example, a first table may be named a second table, and similarly, a second table may be named a first table, without departing from the scope of the various described embodiments. Both the first table and the second table are tables, but they are not the same table.

[0107] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0108] The cycling data processing method provided in this application can be applied to terminal devices such as mobile phones, tablets, wearable devices, in-vehicle devices, augmented reality / virtual reality devices, laptops, super mobile personal computers, netbooks, and personal digital assistants. This application does not impose any restrictions on the specific type of terminal device.

[0109] For example, the terminal device may be a station in a WLAN, a cellular phone, a cordless phone, a session initiation protocol phone, a wireless local loop station, a personal digital processing device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, an in-vehicle device, a vehicle networking terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a set-top box, a user premises equipment, and / or other devices for communication over a wireless system, as well as next-generation communication systems, such as mobile terminals in 5G networks or mobile terminals in future evolved public terrestrial mobile networks, etc.

[0110] As an example and not a limitation, when the terminal device is a wearable device, the term "wearable device" can also refer to any device that utilizes wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices worn directly on the body or integrated into a user's clothing or accessories. Wearable devices are not merely hardware devices; they achieve powerful functions through software support, data interaction, and cloud interaction. Broadly defined, wearable smart devices include those with comprehensive functions, large sizes, and the ability to perform complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those focused on a specific application function that require interaction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0111] Figure 8 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. For example... Figure 8As shown, the terminal device 8 of this embodiment includes: at least one processor 80 ( Figure 8 Only one is shown in the image), and a memory 81 is stored in which a computer program 82 can run on the processor 80. When the processor 80 executes the computer program 82, it implements the steps in the various cycling data processing method embodiments described above, for example... Figure 1 Steps S101 to S104 are shown. Alternatively, when the processor 80 executes the computer program 82, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 7 The functions of modules 710 to 740 are shown.

[0112] The terminal device 8 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor 80 and a memory 81. Those skilled in the art will understand that... Figure 8 This is merely an example of terminal device 8 and does not constitute a limitation on terminal device 8. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal device may also include input transmission devices, network access devices, buses, etc.

[0113] The processor 80 may be a central processing unit, or it may be other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0114] In some embodiments, the memory 81 may be an internal storage unit of the terminal device 8, such as a hard disk or memory of the terminal device 8. The memory 81 may also be an external storage device of the terminal device 8, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc., equipped on the terminal device 8. Furthermore, the memory 81 may include both internal and external storage units of the terminal device 8. The memory 81 is used to store operating systems, applications, bootloaders, data, and other programs, such as the program code of computer programs. The memory 81 can also be used to temporarily store data that has been sent or will be sent.

[0115] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0116] This application also provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, it causes the terminal device to implement the steps in any of the above method embodiments.

[0117] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0118] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the various method embodiments above.

[0119] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0120] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0121] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0123] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A cycling data processing method, characterized in that, include: A model is generated by acquiring cyclist identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple current cycling environment information, multiple historical cycling physiological information, multiple historical cycling movement status information, multiple historical cycling environment information, and cycling safety warning information. Based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information, the cyclist's movement pattern information and cycling safety status information are obtained. Based on the cyclist's movement pattern information and cycling safety situation information, generate cyclist safety risk representation information; Based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library, cycling safety warning information is generated.

2. The cycling data processing method as described in claim 1, characterized in that, The current cycling physiological information includes current cycling heart rate information, current cycling blood oxygen saturation information, and current cycling body surface sensation information; The current cycling motion status information includes current straight-line cycling status information, current curve turning status information, and current gear shift start / stop status information; The current cycling environment information includes current road conditions, current weather temperature and humidity, and surrounding traffic and pedestrian information. The historical cycling physiological information includes historical cycling heart rate information, historical cycling blood oxygen saturation information, and historical cycling body surface sensation information. The historical cycling motion status information includes historical cycling straight-line travel status information, historical cycling curve turning status information, and historical cycling gear shift start / stop status information. The historical cycling environment information includes road conditions, weather, temperature and humidity, and traffic and pedestrian flow around the historical cycling environment.

3. The cycling data processing method as described in claim 1, characterized in that, The step of generating cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library specifically includes: The multiple current cycling environment information and multiple historical cycling environment information are processed by feature vector transformation to generate multiple current cycling environment feature vectors and multiple historical cycling environment feature vectors. Based on the multiple current cycling environment feature vectors, multiple historical cycling environment feature vectors, and a preset cycling safety warning strategy library, the cycling safety warning information generation model is iteratively trained to obtain the trained cycling safety warning information generation model. Based on the cyclist safety risk representation information, multiple current cycling environment feature vectors, and the trained cycling safety warning information generation model, cycling safety warning information is generated.

4. The cycling data processing method as described in claim 3, characterized in that, The preset cycling safety warning strategy library includes multiple preset cycling environment benchmark feature vectors and multiple preset cycling safety warning strategy information; among them, the preset cycling environment benchmark feature vectors and the preset cycling safety warning strategy information correspond one-to-one. The step of iteratively training the cycling safety warning information generation model based on the multiple current cycling environment feature vectors, multiple historical cycling environment feature vectors, and a preset cycling safety warning strategy library to obtain the trained cycling safety warning information generation model specifically includes: The matching process is performed based on the multiple current cycling environment feature vectors and multiple preset cycling environment benchmark feature vectors to generate multiple current matched cycling environment benchmark feature vectors. The pairing process is performed based on the multiple current matching post-riding environment baseline feature vectors and multiple preset riding safety warning strategy information to obtain multiple current matching post-riding safety warning strategy information. The current cycling environment feature vectors and the current paired cycling safety warning strategy information are spliced ​​together to obtain multiple spliced ​​vectors of current cycling environment and warning strategy features. The matching process is performed based on the multiple historical cycling environment feature vectors and multiple preset cycling environment benchmark feature vectors to generate multiple historical matched cycling environment benchmark feature vectors. The matching process is performed based on the multiple historical matching cycling environment baseline feature vectors and multiple preset cycling safety warning strategy information to obtain multiple historical matching cycling safety warning strategy information. The multiple historical cycling environment feature vectors and multiple historical paired cycling safety warning strategy information are spliced ​​together to obtain multiple historical cycling environment and warning strategy feature splicing vectors. Based on the spliced ​​vectors of multiple current cycling environment and early warning strategy features, as well as multiple spliced ​​vectors of multiple historical cycling environment and early warning strategy features, the cycling safety early warning information generation model is iteratively trained to obtain the trained cycling safety early warning information generation model.

5. The cycling data processing method as described in claim 3, characterized in that, The step of iteratively training the cycling safety warning information generation model based on the multiple current cycling environment feature vectors, multiple historical cycling environment feature vectors, and a preset cycling safety warning strategy library to obtain the trained cycling safety warning information generation model specifically includes: Based on the preset cycling safety warning parameter feature extraction rules, the preset cycling safety warning strategy library is processed to extract parameter features, resulting in multiple cycling safety warning parameter feature information. Based on the feature information of the multiple cycling safety warning parameters, vector transformation processing is performed to obtain multiple cycling safety warning parameter feature vectors; The multiple current cycling environment feature vectors and multiple cycling safety warning parameter feature vectors are merged to obtain multiple cycling environment and safety warning parameter feature vectors. Based on the feature vectors of multiple cycling environments and safety warning parameters, as well as multiple historical cycling environment feature vectors, the cycling safety warning information generation model is iteratively trained to obtain the trained cycling safety warning information generation model.

6. The cycling data processing method as described in claim 1, characterized in that, After the step of generating cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library, the method further includes: The cycling safety warning information is processed by feature extraction to obtain cycling safety warning feature information; the cycling safety warning feature information includes cycling warning level feature information, cycling warning push timeliness feature information, and cycling warning prompt type information. Based on the preset mapping relationship between cycling safety warning features and cycling terminal vibration intensity, cycling terminal vibration intensity information is generated according to the cycling safety warning feature information and cycling warning push timeliness feature information. Based on the preset mapping relationship between cycling safety warning features and cycling warning light control, cycling warning light control information is generated according to the cycling warning level feature information and cycling warning prompt type information.

7. The cycling data processing method as described in claim 6, characterized in that, After the step of generating cycling warning light control information based on the preset mapping relationship between cycling safety warning features and cycling warning light control, and according to the cycling warning level feature information and cycling warning type information, the method further includes: Acquire physiological information, motor status information, and environmental information of cycling after multiple warnings; The multiple post-warning cycling physiological information, multiple post-warning cycling movement status information, and multiple post-warning cycling environmental information are sent to the user terminal.

8. A cycling data processing device, characterized in that, include: The information acquisition module is used to acquire cyclist identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple current cycling environment information, multiple historical cycling physiological information, multiple historical cycling movement status information, multiple historical cycling environment information, and a cycling safety warning information generation model. The module for generating cyclist movement patterns and safety status information is used to obtain cyclist movement pattern information and cycling safety status information based on the cyclist's identity attribute information, multiple current cycling physiological information, multiple current cycling movement status information, multiple historical cycling physiological information, and multiple historical cycling movement status information. The cyclist safety risk representation information generation module is used to generate cyclist safety risk representation information based on the cyclist's movement pattern information and cycling safety status information; The cycling safety warning information generation module is used to generate cycling safety warning information based on the cyclist safety risk representation information, multiple current cycling environment information, multiple historical cycling environment information, cycling safety warning information generation model, and a preset cycling safety warning strategy library.

9. A terminal device, characterized in that, The terminal device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images