An intelligent sensing control system for ride protective equipment

Abstract

The application discloses an intelligent sensing control system for riding protective equipment, relates to the technical field of riding safety protection, and is used for solving the problem of inaccurate early intervention of inertial side slip; the application is composed of riding data collection, behavior event identification, behavior matching, risk evolution analysis and pre-response control; the system constructs a behavior graph with a plurality of source sensors, carries out feature difference value and structural consistency checking on an inertial side slip template, combines a trend change mode and a minimum continuous threshold value, triggers a pre-response (pre-inflation, prompt rhythm and optical signal adjustment) in an early risk stage, and carries out closed loop comparison on an execution state and a risk state; compared with a single threshold value triggering scheme, the application can reduce false positives and false negatives, improve response advance and intervention accuracy, reduce unnecessary triggering, and enhance adaptive reliability under complex roads.

Description

Technical Field

[0001] This invention relates to the field of riding safety protection technology, and more specifically, to an intelligent sensing and control system for riding protection equipment. Background Technology

[0002] With the increasing popularity of motorcycles, electric bicycles, and shared rides, cycling has become a growing part of urban and rural transportation systems. However, this has also brought about increasingly prominent traffic safety issues. Riders in open road environments lack the protection of the vehicle's external structure. Once an imbalance, skid, or collision occurs, injury is highly likely. This is especially true on wet or slippery roads, sharp bends, or in emergency situations where the probability of inertial skidding and rollovers increases significantly. Furthermore, accidents develop rapidly and have very little time to be controlled, making it difficult to effectively prevent injury through conventional methods of evasive maneuvering.

[0003] The shortcomings of existing technologies: Most existing riding protection equipment relies on a single trigger condition to activate protective actions, such as using collision sensors, tilt sensors, or acceleration thresholds to trigger airbags or warning systems. This approach often only reacts when the risk event is approaching or has entered an uncontrollable stage, resulting in a delayed response and difficulty in timely reducing accident risks. In addition, although some systems have introduced multi-source sensors, they lack the ability to fuse and analyze multi-dimensional data for complex riding behaviors and predict trends. They cannot identify progressively evolving high-risk behaviors such as inertial sideslip in the early stages, resulting in high rates of missed and false detections. Furthermore, they cannot achieve accurate pre-response before the risk has fully manifested. The lack of closed-loop verification and correction mechanisms between protective strategies and execution actions leads to the protective effect relying on a single judgment result, resulting in insufficient stability and reliability. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the following solution is proposed to solve the problem of inaccurate early intervention of inertial sideslip in the above-mentioned background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An intelligent sensing and control system for riding protective equipment includes a riding data acquisition module, a behavior event recognition module, a behavior matching module, a risk evolution analysis module, and a pre-response control module, with each module connected by a signal.

[0007] The riding data acquisition module is used to collect time-series signals including acceleration, angular velocity, speed and trajectory points based on multi-source sensor data integrated in the riding protective equipment, and to construct a sequence of riding behavior segments through a sliding time window mechanism;

[0008] The behavior event recognition module is used to perform temporal modeling on behavior fragment sequences, construct a behavior graph containing node states and state transition structures, extract action continuity features, and label the transition direction;

[0009] The behavior matching module performs structural matching between the current behavior map and the preset high-risk behavior template map of inertial sideslip. Based on the change magnitude of node states and the stability of the transfer path, it calculates the high-risk matching label of the behavior sequence.

[0010] The risk evolution analysis module is used to generate the current risk evolution path based on the high-risk matching label and determine whether the current risk evolution path meets the set trend change pattern. When the trend characteristics continue to exist and the conditions for entering the judgment interval are met, the pre-response control logic is triggered.

[0011] The pre-response control module, based on the execution of pre-response control logic, including the pre-inflation command to activate the protective device and the adjustment of the output rhythm and optical signal parameters of the prompting system, completes the early intervention and closed-loop control of the risk of inertial sideslip.

[0012] Furthermore, the riding data collection module includes:

[0013] The equipment collects continuous output signals of acceleration and angular velocity through its built-in inertial measurement unit, and synchronously acquires velocity data and trajectory point information through the velocity sensor and positioning module.

[0014] The acceleration, angular velocity, velocity and trajectory point signals are timestamped using a unified time reference to form a time-series data stream of each dimension arranged in the order of sampling time;

[0015] Set the duration and step interval of the sliding time window, and extract continuous data segments from the time-series data stream according to the set sliding rules;

[0016] Each segment of data is labeled as a riding behavior fragment, and the original order of the signals within the fragment in the time dimension is preserved to form a sequence of riding behavior fragments for behavior modeling.

[0017] Furthermore, the behavioral event recognition module includes:

[0018] Each behavior segment in the behavior segment sequence is defined as a node state, and the node state includes the acceleration features, angular velocity features, velocity change features and trajectory direction features within the segment.

[0019] Based on the adjacency relationship of behavioral segments in the time series, state transition connections are established between nodes, and each connection reflects the action change relationship between two adjacent segments;

[0020] Perform continuity analysis on the numerical changes between adjacent nodes to determine whether the action change is a stable continuation, a gradual transition or an abrupt transformation, and attach the analysis results as continuity features to the corresponding state transition connections.

[0021] Based on the changing trends of trajectory direction and angular velocity, the directional attribute of state transition is determined, and the transition direction is marked for each state transition connection.

[0022] Furthermore, based on the changing trends of trajectory direction and angular velocity, the directional attribute of the state transition is determined, and the transition direction is labeled for each state transition connection, including:

[0023] Calculate the trend of trajectory direction change from the trajectory point data of adjacent nodes, and determine whether the direction change is left deflection, right deflection or straight-line maintenance;

[0024] Extract the sign and magnitude of angular velocity changes from the angular velocity data of adjacent nodes to determine whether the angular velocity change is increasing, decreasing, or remaining stable.

[0025] The trajectory direction change trend and angular velocity change results are combined and mapped to generate corresponding direction attribute categories, including left turn, right turn, straight stable, left acceleration turn, right acceleration turn, etc.

[0026] The generated directional attribute category is assigned to the corresponding state transition connection as the transition direction label of the connection.

[0027] Furthermore, the behavior matching module includes:

[0028] The preset high-risk behavior template diagram of inertial sideslip is invoked. The high-risk behavior template diagram of inertial sideslip contains node states and their connection relationships arranged in chronological order. Each node state corresponds to the reference range of the characteristic values ​​of each sensor during the inertial sideslip process.

[0029] The node states in the current behavior graph are matched one-to-one with the node states in the template graph in chronological order, and the feature difference value of each pair of matched nodes is calculated.

[0030] Perform a structural consistency check on the connection relationships between matching nodes to determine whether the state transition path is consistent with the connection path of the template graph, and evaluate the stability of the transition path;

[0031] When the node feature difference values ​​are all within the allowable range set by the template graph and the stability of the transfer path reaches the set threshold, the behavior graph is marked as a high-risk matching state, a high-risk matching label is generated, and it is attached to the current behavior sequence.

[0032] Furthermore, the specific calculation process for characteristic difference values ​​and structural consistency checks includes:

[0033] The calculation process of feature difference value is as follows: obtain the numerical range of each sensor feature in the current behavior map node state and the reference numerical range of the corresponding node state in the template map respectively, calculate the absolute value of the difference in value of each feature and the consistency of the direction of change, determine whether the absolute value of the difference is less than the preset allowable fluctuation range, and when all features meet the allowable fluctuation range, the node is determined as a feature matching node.

[0034] The structural consistency check process is as follows: In the behavior graph, trace the connection paths between nodes in chronological order, record the direction attributes and continuity indicators of each transition connection in the path, and compare the direction change patterns of the path with the direction change patterns in the template graph one by one. If the direction attributes and continuity indicators of each transition connection match the template graph... Figure 1 If the path is consistent and there are no interruptions or reverse transfers in the path, then the path structure consistency requirement is met.

[0035] When both the feature difference value determination and the structural consistency check meet the set conditions, the corresponding node and path combination is marked as a structural matching unit.

[0036] Furthermore, the risk evolution analysis module includes:

[0037] The nodes marked with high-risk matching labels and their adjacent nodes are arranged sequentially in time series to form the current risk evolution path. The path includes the sensor feature values, directional attributes and state transition relationships of each node.

[0038] Analyze the trend of characteristic value changes at each node in the risk evolution path to determine whether the direction and magnitude of changes in characteristics such as acceleration, angular velocity, velocity and trajectory direction meet the requirements of the preset trend change pattern.

[0039] In the process of determining the trend change pattern, the cumulative time of the feature changes of continuous nodes in the path is detected to determine whether the duration of the trend feature has reached the set minimum duration threshold.

[0040] When the trend change pattern meets the preset requirements, the duration of the trend feature reaches the minimum duration threshold, and the state transition direction sequence in the path is consistent with the preset direction change pattern, the current risk evolution path is determined to enter the judgment interval.

[0041] Trigger the pre-response control logic and send an execution pre-response command to the protection control process.

[0042] Furthermore, the pre-response control module includes:

[0043] Receive the pre-response command and parse it into the corresponding protective device control parameters, prompt system control parameters and optical signal control parameters;

[0044] Send pre-inflation control commands to riding protective equipment and continuously monitor airbag pressure and inflation rate;

[0045] Adjust the output rhythm of the prompting system, shorten the interval between sound or vibration prompts to within the set value, and add inertial side slip risk warning information to the prompt content to improve rider attention;

[0046] Modify the control parameters of the optical signal, including increasing the flicker frequency of the light source, increasing the light intensity, or changing the color of the light source;

[0047] During the execution of pre-response control actions, the working status data of the protective devices and the warning system are continuously collected and compared with the results of risk status changes.

[0048] The technical effects and advantages of the intelligent sensing and control system for riding protective equipment of the present invention are as follows:

[0049] This invention constructs a closed-loop intelligent perception and control system that integrates riding data acquisition, behavioral event recognition, behavior matching, risk evolution analysis, and pre-response control. This system enables continuous monitoring, accurate identification, and early intervention of high-risk riding behaviors such as inertial sideslip. The system constructs a riding behavior map with temporal continuity based on multi-source sensors and performs feature difference value calculation and structural consistency checks with the high-risk behavior template map of inertial sideslip. This allows pre-response control to be triggered when the risk is still in its early stages of evolution. Compared with traditional protection methods that rely solely on a single sensor threshold, this invention can significantly reduce false alarm and false negative rates, achieve advance control of risk response, and avoid the lag and insufficient protection caused by riders activating protective devices only after the risk has fully manifested. This improves the overall timeliness of protection and the accuracy of intervention.

[0050] Furthermore, this invention introduces a feature difference value judgment mechanism, a state transition path structure consistency judgment mechanism, and a risk trend change pattern comparison mechanism. This ensures that risk identification is not only based on single-point outliers but also combines the dynamic evolution relationship and directional attribute sequence of multi-dimensional features, thereby achieving stability and traceability in risk assessment. Through a risk level control parameter lookup table, the pre-response control module can convert the risk identification results into an executable parameter set including pre-inflation target pressure, inflation rate range, prompt rhythm, and optical signal adjustment scheme. During execution, it monitors the matching degree between control actions and risk states in real time, supporting anomaly correction and closed-loop recording. This collaborative mechanism enables the system to complete early protection and prompts with minimal intervention, avoiding unnecessary frequent triggering that could interfere with riders, while also improving the service life of the protective device and the system's adaptability in complex road environments. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the structure of an intelligent sensing and control system for riding protective equipment according to the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] In order to achieve the above objectives, Figure 1 A schematic diagram of the intelligent sensing and control system for riding protective equipment of the present invention is given, which specifically includes a riding data acquisition module, a behavior event recognition module, a behavior matching module, a risk evolution analysis module, and a pre-response control module, and the modules are connected by signals.

[0054] The riding data acquisition module is used to collect time-series signals including acceleration, angular velocity, speed and trajectory points based on multi-source sensor data integrated in the riding protective equipment, and to construct a sequence of riding behavior segments through a sliding time window mechanism;

[0055] The behavior event recognition module is used to perform temporal modeling on behavior fragment sequences, construct a behavior graph containing node states and state transition structures, extract action continuity features, and label the transition direction;

[0056] The behavior matching module performs structural matching between the current behavior map and the preset high-risk behavior template map of inertial sideslip. Based on the change magnitude of node states and the stability of the transfer path, it calculates the high-risk matching label of the behavior sequence.

[0057] The risk evolution analysis module is used to generate the current risk evolution path based on the high-risk matching label and determine whether the current risk evolution path meets the set trend change pattern. When the trend characteristics continue to exist and the conditions for entering the judgment interval are met, the pre-response control logic is triggered.

[0058] The pre-response control module, based on the execution of pre-response control logic, including the pre-inflation command to activate the protective device and the adjustment of the output rhythm and optical signal parameters of the prompting system, completes the early intervention and closed-loop control of the risk of inertial sideslip.

[0059] The riding data acquisition module is used to collaboratively utilize the multi-source sensors integrated in the equipment to obtain complete and structured basic data on riding behavior, and to construct a sequence of riding behavior segments with temporal continuity. The specific implementation steps include the following:

[0060] The inertial measurement unit (IMU) built into the riding protective gear collects acceleration and angular velocity signals generated during riding in real time. The acceleration signal reflects the rider's linear motion changes on the three-dimensional coordinate axis, while the angular velocity signal reflects the rider's rotation, tilting, or posture changes. Both signals are continuously output and acquired according to a fixed sampling period, such as once every 10 milliseconds, to ensure that the data has sufficient time resolution.

[0061] The system integrates a speed sensor and positioning module to synchronously acquire speed data and trajectory point information during riding. Speed ​​data reflects the current velocity, while trajectory point information represents the current spatial coordinates, typically achieved via GPS or GNSS. The positioning module should possess differential augmentation or assisted positioning capabilities to improve the integrity of position information in obstructed environments. A unified time reference is used to align acceleration, angular velocity, velocity, and trajectory point signals with timestamps. This time reference can be provided by the main control processor within the equipment. The main control processor allocates a unified timestamp to all data sampling processes via a synchronized system clock and arranges different modal data into a unified sampling order based on the timestamps, forming a time-series data stream with a clear temporal order. In practice, if a certain modal signal is missing, the most recent valid value from the previous sampling period can be used to fill the gap, ensuring the integrity and continuity of the data timing.

[0062] Define a sliding time window to extract behavioral segments from the time-series data stream. The duration of the sliding time window should be set according to the response speed of actual riding behavior changes, for example, it can be set to 1 second; the step interval is the time step in which the sliding window moves forward each time, such as 200 milliseconds.

[0063] For example, during a ride downhill into a left turn on an urban road, the system collects acceleration and angular velocity signals every ten milliseconds using a unified time base. Simultaneously, the speed sensor and positioning module synchronously acquire speed data and trajectory point information. A sliding time window is set to last 1 second with a step interval of 200 milliseconds. Continuous data segments are sequentially extracted from the time-series data stream according to the sliding rules. Each data segment is marked as a riding behavior fragment, and the original time sequence of all sampling points within the fragment is retained, forming a riding behavior fragment sequence for behavior modeling. When the positioning signal is briefly missing at the tunnel entrance, the timestamp is aligned according to the rule of filling with the most recent valid value from the previous sampling period, ensuring the continuity and consistency of the time-series data stream. A valid value refers to a sampling point that simultaneously meets the following conditions: complete timestamp, sensor self-test passed, and value falls within the sensor's physical range and the environmental safety range. Samples that do not meet any of these conditions are marked as invalid samples and are not used for filling.

[0064] It should be noted that during the application process, the sliding time window advances continuously at the set step intervals. Within each window period, a complete data segment containing acceleration, angular velocity, velocity and trajectory points is extracted from the time-series data stream as an independent behavior data block.

[0065] For each data segment extracted from the sliding time window, the system marks it as a riding behavior segment. During the marking process, the original time sequence of all sampling points in the data segment and the corresponding signal data of various types must be preserved to ensure that each segment has internal consistency in time sequence. All generated behavior segments are arranged in sequence to finally form a structured riding behavior segment sequence with time evolution characteristics.

[0066] The behavior event recognition module includes the following implementation steps:

[0067] Each behavior segment in the behavior segment sequence is defined as a node state, which consists of four types of data: acceleration features, angular velocity features, velocity change features, and trajectory direction features within the segment.

[0068] Acceleration features are used to describe the changes in linear motion acceleration within the segment. By collecting triaxial acceleration data at each sampling point during the segment's duration and arranging them in chronological order, the maximum acceleration value, minimum acceleration value, and acceleration variation range within the segment are calculated.

[0069] Angular velocity characteristics are used to reflect the dynamic characteristics of rotational motion within the segment. They are obtained by acquiring the three-axis angular velocity signals of the gyroscope and analyzing the trend of angular velocity changes over time.

[0070] The velocity change characteristics are obtained by comparing the linear velocity values ​​at the beginning and end of the segment and combining them with the fluctuation of the velocity curve at intermediate sampling points, reflecting the acceleration / deceleration state and the magnitude of the velocity change.

[0071] Trajectory orientation features are obtained by projecting positioning data onto a horizontal plane and analyzing the overall directional offset and directional stability of the motion trajectory within a segment.

[0072] After defining the node states, state transition connections are established between nodes based on the adjacency relationships of behavioral segments in the time series. Each connection reflects the action change relationship between two adjacent segments. When the connection is generated, it is bound to the corresponding start node and target node identifiers. Then, a continuity analysis is performed on the numerical changes between adjacent nodes, that is, the four types of feature values ​​corresponding to the two nodes are compared one by one, and the change amplitude and change rate of each type of feature are calculated. If the change amplitude of all features is within the preset stable range, it is determined to be a stable continuation; if the overall change amplitude of features is small but the change rate shows a slow trend, it is determined to be a gradual transition; if the change amplitude of one or more types of features exceeds the mutation threshold and the change rate is significantly higher than the normal range, it is determined to be a mutation transition, and the continuity analysis result is attached to the corresponding state transition connection as a continuous feature.

[0073] Based on this, the directional attribute of state transition is determined according to the trend of trajectory direction and angular velocity change. That is, it is determined whether the motion direction of adjacent segments is consistent, whether a turn or a reversal occurs, and combined with the positive and negative trend and magnitude of angular velocity change, it is labeled as a clear directional type such as straight stable, leftward transition, rightward transition or reverse transition. This ensures that each state transition connection has a directional label, thereby forming a complete behavioral map containing node state, feature change information and directional attributes, providing a basic data structure for subsequent behavioral pattern analysis and risk assessment.

[0074] Based on the changing trends of trajectory direction and angular velocity, the directional attribute of state transition is determined, and the transition direction is labeled for each state transition connection. The specific steps are as follows:

[0075] When it is necessary to determine the directional attribute of the state transition based on the changing trend of the trajectory direction and angular velocity, the system first acquires the trajectory point data of adjacent nodes. The trajectory point data is a continuous spatial coordinate point sequence collected at a uniform time interval. By calculating the spatial coordinate difference between each pair of adjacent nodes in sequence, the instantaneous direction vector of each trajectory segment is obtained. Then, the angle between adjacent direction vectors and their positive and negative signs are compared to determine whether the changing trend of the trajectory direction belongs to one of three types: left deflection, right deflection, or straight-line maintenance. Left deflection indicates that the current trajectory direction has rotated counterclockwise relative to the previous direction vector, right deflection indicates that the current trajectory direction has rotated clockwise relative to the previous direction vector, and straight-line maintenance indicates that the change in angle is close to zero and within the preset straight-line tolerance range.

[0076] After determining the trajectory direction trend, the angular velocity data of adjacent nodes is extracted from the output of the inertial measurement unit corresponding to the trajectory point data. This angular velocity data is the instantaneous angular velocity value measured at the same timestamp, and the unit is degrees per second. In this example, the angular velocity is measured in degrees / second. If it is measured in radians per second, it is converted according to a fixed ratio.

[0077] By calculating the difference in angular velocity between adjacent nodes and comparing their signs and magnitudes, the change in angular velocity is classified into one of three types: increasing, decreasing, or remaining stable. An increasing angular velocity indicates that the current steering speed is rising; a decreasing angular velocity indicates that the steering speed is decreasing; and a stable angular velocity indicates that the change in steering speed is within a preset stability threshold range.

[0078] Subsequently, the trajectory direction change trend and the angular velocity change results are combined and mapped to generate the corresponding direction attribute category. The rules for this combined mapping are as follows:

[0079] When the trajectory deflects to the left and the angular velocity increases, the direction attribute category is "accelerating to the left"; when the trajectory deflects to the left and the angular velocity decreases, the direction attribute category is "decelerating to the left"; when the trajectory deflects to the right and the angular velocity increases, the direction attribute category is "accelerating to the right"; when the trajectory deflects to the right and the angular velocity decreases, the direction attribute category is "decelerating to the right"; when the trajectory remains straight and the angular velocity change is stable, the direction attribute category is "stable straight".

[0080] Depending on the different combinations, the directional attribute categories may also include, but are not limited to, defined types such as left turn, right turn, straight-line stability, left acceleration turn, and right acceleration turn. All generated directional attribute categories will be assigned to their corresponding state transition connections as transition direction annotation information for that connection and stored in the state transition data structure for subsequent path prediction, risk identification, and control decision invocation.

[0081] The behavior matching module includes the following implementation steps:

[0082] First, the system invokes a high-risk inertial sideslip behavior template diagram stored in the system behavior template library. This template diagram consists of multiple node states arranged chronologically and their connections. Node states describe the characteristic value ranges of various sensors (including accelerometers, angular velocity sensors, attitude sensors, and position sensors) during inertial sideslip. These characteristic value ranges were obtained through statistical analysis of a large amount of inertial sideslip experimental sample data during the system development phase and were confirmed after expert verification. Each node state in the template diagram is accompanied by a timestamp and connection path information between it and the preceding and following nodes. The connection path represents the sequence and logical constraints of the transition from one node state to another.

[0083] The node states arranged in chronological order in the current behavior graph are sequentially matched one by one with the corresponding node states in the inertial sideslip high-risk behavior template graph. During the matching process, the system extracts the sensor feature values of the current node state and compares them item by item with the set reference feature value range in the template node, calculates the feature difference value for each pair of matching nodes. The calculation of this feature difference value uses the absolute difference method, that is, the sensor feature value of the current node is subtracted from the reference value of the template node, and the absolute value of the obtained result is used as the difference measure for this feature item.

[0084] For nodes with multi-dimensional feature combinations, calculate the difference values of each feature item separately and compare them item by item with their respective allowable fluctuation ranges; when the absolute values of the differences of all features are less than their respective allowable fluctuation ranges, and the consistency of the change directions of all features is the same, the node is determined as a feature matching node.

[0085] Next, perform a structural consistency check on the connection relationships between the matching nodes, that is, determine whether the state transition path in the current behavior graph is exactly the same as the connection path in the template graph, and evaluate the stability degree of this transition path. The evaluation of the stability degree is based on the smoothness of the change of the sensor feature value during the state transition. If the change rate is within the preset fluctuation threshold and the continuous time period reaches the minimum duration requirement, it is considered that the transition path is stable;

[0086] After completing the above comparison and inspection, if the comprehensive feature difference values of all matching nodes are within the allowable range set in the template graph, and the stability degree of the transition path reaches the set stability threshold, the system will mark the current behavior graph as a high-risk matching state and generate a corresponding high-risk matching label. The high-risk matching label contains information such as the matching time, trigger node number, key feature items, and detailed difference values, which are used for subsequent linkage control by the risk response module. The generated high-risk matching label will be attached to the current behavior sequence to ensure that when the subsequent control logic receives the behavior sequence, it can accurately identify the high-risk state and execute the predetermined protection strategy.

[0087] For example, the behavior matching module detected and matched high-risk behaviors of inertial sideslip in a real-world motorcycle riding scenario. Specifically, the rider was driving a motorcycle at approximately 50 km / h onto a left-turn, slippery road surface. Upon entering the initial section of the curve, the riding data acquisition module, through the built-in inertial measurement unit, collected continuous signals showing a lateral acceleration of 0.85 m / s², an angular velocity of 5° / s, a velocity change characteristic value of 0, and a trajectory direction consistent with the road centerline. Upon entering the middle section of the curve, the lateral acceleration increased to 1.45 m / s², the angular velocity increased to 18° / s, the speed decreased by 3 km / h, and the trajectory direction deviated to the left. Towards the end of the curve, the lateral acceleration fluctuation increased to 2.1 m / s², the angular velocity jumped to 26° / s, the speed continued to decrease by 5 km / h, and the trajectory direction reached an angle of 18° with the road tangent.

[0088] The three time-series data segments are constructed as node states, and state transition paths between nodes are established in chronological order. The paths are: straight-line stability to left acceleration and turning, and then left turning. The behavior matching module calls a preset high-risk behavior template diagram for inertial sideslip. The reference feature range recorded in the template diagram is: lateral acceleration continuously increases and is in the range of 1.4 to 2.5 m / s², angular velocity is in the range of 15° / s to 30° / s and continuously increases, speed decreases by 3 to 8 km / h in a short period of time, the cumulative trajectory deflection angle exceeds 15°, and the path direction pattern is straight-line stability to left acceleration and turning, and then left turning. The behavior matching module matches each node state of the current behavior map with the template node state one by one according to the template node order, calculates the absolute value of the numerical difference of lateral acceleration, angular velocity, velocity change characteristic representative value, and trajectory deflection angle, judges the consistency of the change direction of each feature, and confirms that the absolute value of the difference is less than the preset allowable fluctuation range. In this embodiment, the lateral acceleration difference between node 1 and the first node of the template is 0.05 m / s², and the angular velocity difference is 0° / s. The lateral acceleration and angular velocity differences between node 2 and the middle node of the template are both 0. The feature differences between node 3 and the last node of the template also meet the range requirements. Furthermore, the behavior matching module performs a structural consistency check on the state transition paths between nodes, confirming that the direction attribute sequence is consistent with the template and that there are no interruptions or reverse transitions in the path. Finally, the system marks the behavior map as a high-risk matching state, generates a high-risk inertial sideslip label, and attaches it to the current behavior sequence.

[0089] The specific calculation process for characteristic difference values ​​and structural consistency checks includes:

[0090] The calculation process of feature difference values ​​is carried out according to the following steps: First, the numerical range of each sensor feature in the current behavior map node state and the reference numerical range of the corresponding node state in the template map are obtained respectively. The reference numerical range is a closed interval formed by the lower limit and the upper limit value obtained by statistically analyzing the inertial sideslip samples during the sample acquisition stage, and is solidified in the template map through the calibration process.

[0091] For each sensor feature, the relative position of the current feature representative value and the reference value range is used as the basis for measuring the difference: when the current feature representative value falls within the reference value range, the absolute value of the difference of the feature is recorded as zero.

[0092] When the current representative value of a feature is lower than the reference lower limit, the absolute value of the difference between the current representative value of a feature and the reference lower limit is used as the absolute value of the difference for that feature.

[0093] When the current representative value of a feature is higher than the upper limit of the reference range, the absolute value of the difference between the current representative value and the upper limit of the reference range is used as the absolute value of the difference for that feature. The current representative value of a feature is obtained by taking the median of all sampling points of the same feature within the segment corresponding to the state of that node after denoising, in order to avoid the influence of extreme values.

[0094] The system also determines the consistency of the change direction of each feature: by comparing the smoothed representative value of the feature at the start and end times of the node's state segment, if the end representative value is greater than the start representative value and the template node records an upward trend for that feature, then the direction is consistent.

[0095] If the ending representative value is less than the starting representative value and the template node records show a downward trend, then the direction is consistent;

[0096] If the difference between the two does not exceed the minimum significant change and the template node records a stable trend, then the directions are consistent;

[0097] The minimum significant change is set as a fixed threshold based on the combined level of sensor quantization noise and environmental noise during factory calibration. The system pre-sets an allowable fluctuation range for each feature. This allowable fluctuation range is the maximum acceptable distance between the current representative value of the feature and the boundary of the reference interval in normal, non-risk samples. It is obtained statistically and stored in the parameter table. When the absolute value of the difference of all features is less than their respective allowable fluctuation range, and the direction of change of all features is consistent, the node is determined to be a feature matching node. If any feature does not meet either of the two conditions, the node is not determined to be a feature matching node, and the reason for the mismatch is marked in the record for traceability.

[0098] The structural consistency check process traces the connection paths between nodes in the behavior graph in chronological order and compares the path attributes with the template graph records one by one.

[0099] First, read candidate path segments of the same length as the template graph from the current behavior graph, and record the directional attribute and continuity identifier of each transition connection in the candidate path segment; the directional attribute is the category label generated above based on the trajectory direction and angular velocity change trend, and the continuity identifier is one of three types: stable continuation, gradual transition or abrupt change;

[0100] The system performs a one-to-one comparison of the direction attribute sequence of the candidate path segment with the direction attribute sequence of the corresponding path in the template graph. If the direction attributes are completely consistent at all corresponding positions, the direction attribute verification passes; otherwise, the direction attribute verification fails and the consistency check is terminated. Subsequently, the system compares the continuity identifier sequence of the candidate path segment with the continuity identifier sequence of the corresponding path in the template graph item by item. If the continuity identifiers at all corresponding positions are consistent, and there are no missing transition connections between any nodes within the candidate path segment, the path is considered uninterrupted. Simultaneously, it checks for reversed time sequence, i.e., whether the timestamp of any transition connection is earlier than its starting node timestamp or later than its target node timestamp. If such a situation exists, it is determined to be a reverse transition and is considered inconsistent. Only when the direction attribute verification and continuity identifier verification pass, and no interruption or reverse transition is found, is the path structure consistency determined to meet the requirements.

[0101] When the feature difference value determination and the structural consistency check meet the set conditions at the same time, the system marks the combination of the corresponding feature matching node and its connected path segment as a structural matching unit, and generates a structural matching unit record in the behavior graph that includes the matching range, matching time, the type of verification passed, and the empty items of the failed items.

[0102] The risk evolution analysis module includes the following implementation steps:

[0103] The nodes that have been marked with high-risk matching tags by the behavior matching module are organized into time series. The nodes that are immediately adjacent to each other in time are included in the same continuous path. If necessary, the path can be extended forward and backward according to the preset maximum path coverage time until the minimum path length required for judgment is reached, thus forming the current risk evolution path.

[0104] The current risk evolution path stores a list of nodes in chronological order. Each node contains sensor feature values ​​(representative values ​​of acceleration, angular velocity, velocity change, and trajectory direction), directional attributes, and state transition relationships with adjacent nodes.

[0105] Meanwhile, to ensure the comparability of different modal features, the median of the sensor feature values ​​within each node segment is taken after denoising processing as the representative feature value of that node; the directional attribute is the discrete category obtained above based on the trajectory direction and angular velocity change trend; the state transition relationship includes the transition direction from the starting node to the target node and the continuity feature identifier.

[0106] The system analyzes the trend of feature values ​​at each node along the path. Specifically, it compares the representative values ​​of the same feature at adjacent nodes to determine whether the value increases, decreases, or remains stable from the previous node to the next. It distinguishes between stable and unstable changes based on a preset minimum significant change threshold. Simultaneously, based on preset amplitude grading boundaries, the amplitude of change is categorized into small, medium, and large amplitudes. The trend change pattern is a set of rules extracted and solidified during the system's sample training phase. This set explicitly defines the joint evolution relationship of various features when inertial sideslip risk occurs. For example, angular velocity continuously increases and reaches a medium amplitude or higher; trajectory direction continuously deflects and the deflection angle cumulatively increases; and velocity change features change from decreasing to stable or fluctuating slightly within a short period.

[0107] Check each node along the path to see if the above-mentioned joint evolution relationship holds simultaneously, thereby determining whether the direction and magnitude of change of features such as acceleration, angular velocity, velocity, and trajectory direction meet the requirements of the preset trend change pattern.

[0108] After determining the trend change pattern, the risk evolution analysis module performs cumulative time detection on continuous node segments in the path that satisfy the trend change pattern:

[0109] The duration of the continuous segment is calculated using the timestamps of its first and last nodes and compared with a pre-set minimum duration threshold. The minimum duration threshold is determined based on statistical results of the difference in duration between normal riding and confirmed inertial sideslip events in the sample data, and is fixed as a parameter during device factory calibration. If the duration reaches or exceeds the minimum duration threshold, the state transition direction sequence corresponding to the continuous segment is simultaneously compared one-to-one with a preset direction change pattern, where the direction change pattern is a finite sequence obtained through sample induction.

[0110] For example, if the path is "straight and stable, then accelerates to the left and then turns to the left" or "straight and stable, then accelerates to the right and then turns to the right," and the comparison results show that the two paths are completely consistent in order and category, and there is no interruption or reverse shift in the time sequence within the path, the module determines that the current risk evolution path has entered the judgment interval. After entering the judgment interval, the risk evolution analysis module immediately triggers the pre-response control logic and sends an execution pre-response command to the protection control process. This execution pre-response command carries pre-inflation control parameters, prompt system control parameters, and optical signal control parameters in the form of structured parameters. After receiving the command, the protection control process executes subsequent pre-inflation, prompt rhythm adjustment, and optical signal adjustment operations according to the parameter content, thereby completing early intervention before the inertial sideslip risk fully manifests and laying the data and timing foundation for closed-loop control.

[0111] The pre-response control module includes the following implementation steps:

[0112] After receiving the pre-response instruction sent by the risk evolution analysis module, the instruction is first parsed, and the control intent carried therein is mapped into three sets of parameters: protective device control parameters, prompting system control parameters, and optical signal control parameters.

[0113] The control parameters of the protective device include at least the pre-inflation target pressure, maximum inflation duration, allowable inflation rate range and safe shutdown conditions. The above parameters are derived from the factory calibration table and the risk level-control parameter comparison table. The risk level-control parameter comparison table is established during the sample training phase based on the correspondence between different risk levels and expected response lead, and is written into the storage area during equipment initialization.

[0114] The prompting system control parameters include at least the prompting method, output rhythm, and prompting duration. The prompting method can be selected between sound prompts and vibration prompts. The output rhythm is represented by the time interval between two adjacent prompts. The set values ​​are given by the risk level-control parameter lookup table. When executed, the prompting content is clearly set as an inertial sideslip risk warning message so that the rider can intuitively identify the current risk situation.

[0115] The optical signal control parameters include at least three items: flicker frequency, light intensity, and light source color. The flicker frequency is represented by the number of times the light source is turned on and off per unit time, the light intensity is represented by the rated brightness level of the light source, and the light source color is represented by a value from a fixed set of colors supported by the device. These three items are written into the register configuration area of ​​the optical signal drive unit after instruction parsing.

[0116] After parameter parsing is completed, a pre-inflation control command is sent to the riding protective equipment according to the control parameters of the protective device. The inflation mechanism is started and the sensor outputs of airbag pressure and inflation rate are continuously read under a fixed sampling period. The instantaneous pressure and instantaneous inflation rate obtained in each sampling period are compared with the pre-inflation target pressure and the allowable inflation rate range one by one: when the instantaneous pressure is not lower than the pre-inflation target pressure in a continuous detection period, or when the maximum inflation duration is reached, the inflation stop logic is triggered; when the instantaneous inflation rate continuously exceeds the allowable inflation rate range, the safety shutdown condition is triggered and the abnormal reason code is recorded.

[0117] At the same time, the pre-response control module adjusts the output rhythm of the prompting system according to the prompting system control parameters, sets the interval time of the sound prompt or vibration prompt to a fixed interval not greater than the set value, and adds inertial side slip risk warning information to the prompt content; it modifies the working state of the optical signal according to the optical signal control parameters, sets the flashing frequency to the value specified by the instruction, switches the light intensity to the specified level, and switches the light source color to the specified color.

[0118] During the execution of the pre-response control action, the pre-response control module continuously collects the working status data of the protective device and the warning system, and compares and analyzes it with the results of changes in the risk status. The working status data includes at least the continuous reading of the airbag pressure, the continuous reading of the working current of the inflation mechanism, the calculation result of the instantaneous inflation rate, the actual trigger time sequence of the sound or vibration warning, the actual opening and closing time sequence of the optical signal, and the brightness level record;

[0119] The risk status change results are output by the risk evolution analysis module according to a unified time base, and include at least the risk level labels and state transition direction sequences within a continuous time period. The comparative analysis is performed in a timestamp-aligned manner: the working status data and risk status change results within the same time period are compared time by time to determine whether the pre-inflation is completed during the risk level increase phase, whether the triggering rhythm of the prompting system and optical signals is consistent with the output rhythm specified in the risk level-control parameter comparison table, and whether the above control actions exit to the standby state as required within several consecutive detection cycles after the risk level drops to the basic safety level.

[0120] If the comparison results show that any control action fails to meet the parameter requirements within the corresponding risk stage, the pre-response control module immediately outputs a correction command. The correction command is reissued to the corresponding execution unit in the form of parameter overwrite and the correction time point is recorded. If the comparison results show that all control actions meet the parameter requirements within the corresponding risk stage, the pre-response control module records the pre-response cycle as completed and writes the record back to the event log for subsequent closed-loop analysis and strategy optimization.

[0121] For example, during nighttime cycling, when the detected risk of inertial sideslip exceeds a preset trigger threshold, the pre-response control module receives a pre-response command and parses it to obtain a modification scheme for the optical signal control parameters:

[0122] The flashing frequency of the light source was increased from once per second to three times per second, thereby enhancing the visual attention of riders and other road users.

[0123] The light intensity was increased from the original setting of 80 lumens to 120 lumens to enhance the visibility distance of the light signal in low-light environments.

[0124] The light source color was switched from the default white to a highly warning amber color to create a risk warning signal in the traffic vision system;

[0125] The adjustment values ​​of these parameters are determined by the system during the design phase based on road light environment testing and human eye reaction time studies, and are stored in the optical signal control parameter library. During operation, they are called by the pre-response control module according to instructions.

[0126] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0127] Those skilled in the art will recognize that the modules 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.

[0128] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0130] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent sensing and control system for riding protective equipment, characterized in that: It includes a riding data acquisition module, a behavior event recognition module, a behavior matching module, a risk evolution analysis module, and a pre-response control module, and the modules are connected by signals. The riding data acquisition module is used to collect time-series signals including acceleration, angular velocity, speed and trajectory points based on multi-source sensor data integrated in the riding protective equipment, and to construct a sequence of riding behavior segments through a sliding time window mechanism; The behavior event recognition module is used to perform temporal modeling on behavior fragment sequences, construct a behavior graph containing node states and state transition structures, extract action continuity features, and label the transition direction; The behavior matching module performs structural matching between the current behavior map and the preset high-risk behavior template map of inertial sideslip. Based on the change magnitude of node states and the stability of the transfer path, it calculates the high-risk matching label of the behavior sequence. The risk evolution analysis module is used to generate the current risk evolution path based on the high-risk matching label and determine whether the current risk evolution path meets the set trend change pattern. When the trend characteristics continue to exist and the conditions for entering the judgment interval are met, the pre-response control logic is triggered. The risk evolution path is obtained by arranging the nodes marked with high-risk matching labels and their adjacent nodes in a time series. The risk evolution path includes the sensor characteristic values, directional attributes and state transition relationships of each node. The pre-response control module, based on the execution of pre-response control logic, including the pre-inflation command to activate the protective device and the adjustment of the output rhythm and optical signal parameters of the prompting system, completes the early intervention and closed-loop control of the risk of inertial sideslip.

2. The intelligent sensing and control system for riding protective equipment according to claim 1, characterized in that: The riding data acquisition module includes: The equipment collects continuous output signals of acceleration and angular velocity through its built-in inertial measurement unit, and synchronously acquires velocity data and trajectory point information through the velocity sensor and positioning module. The acceleration, angular velocity, velocity and trajectory point signals are timestamped using a unified time reference to form a time-series data stream of each dimension arranged in the order of sampling time; Set the duration and step interval of the sliding time window, and extract continuous data segments from the time-series data stream according to the set sliding rules; Each segment of data is labeled as a riding behavior fragment, and the original order of the signals within the fragment in the time dimension is preserved to form a sequence of riding behavior fragments for behavior modeling.

3. The intelligent sensing and control system for riding protective equipment according to claim 2, characterized in that: The behavior event recognition module includes: Each behavior segment in the behavior segment sequence is defined as a node state, and the node state includes the acceleration features, angular velocity features, velocity change features and trajectory direction features within the segment. Based on the adjacency relationship of behavioral segments in the time series, state transition connections are established between nodes, and each connection reflects the action change relationship between two adjacent segments; Perform continuity analysis on the numerical changes between adjacent nodes to determine whether the action change is a stable continuation, a gradual transition or an abrupt transformation, and attach the analysis results as continuity features to the corresponding state transition connections. Based on the changing trends of trajectory direction and angular velocity, the directional attribute of state transition is determined, and the transition direction is marked for each state transition connection.

4. The intelligent sensing and control system for riding protective equipment according to claim 3, characterized in that: Based on the changing trends of trajectory direction and angular velocity, the directional attribute of state transitions is determined, and the transition direction is labeled for each state transition connection, including: Calculate the trend of trajectory direction change from the trajectory point data of adjacent nodes, and determine whether the direction change is left deflection, right deflection or straight-line maintenance; Extract the sign and magnitude of angular velocity changes from the angular velocity data of adjacent nodes to determine whether the angular velocity change is increasing, decreasing, or remaining stable. The trajectory direction change trend and angular velocity change results are combined and mapped to generate corresponding direction attribute categories, including left turn, right turn, straight stable, left acceleration turn, and right acceleration turn; The generated directional attribute category is assigned to the corresponding state transition connection as the transition direction label of the connection.

5. The intelligent sensing and control system for riding protective equipment according to claim 4, characterized in that: The behavior matching module includes: The preset high-risk behavior template diagram of inertial sideslip is invoked. The high-risk behavior template diagram of inertial sideslip contains node states and their connection relationships arranged in chronological order. Each node state corresponds to the reference range of the characteristic values ​​of each sensor during the inertial sideslip process. The node states in the current behavior graph are matched one-to-one with the node states in the template graph in chronological order, and the feature difference value of each pair of matched nodes is calculated. Perform a structural consistency check on the connection relationships between matching nodes to determine whether the state transition path is consistent with the connection path of the template graph, and evaluate the stability of the transition path; When the node feature difference values ​​are all within the allowable range set by the template graph and the stability of the transfer path reaches the set threshold, the behavior graph is marked as a high-risk matching state, a high-risk matching label is generated, and it is attached to the current behavior sequence.

6. The intelligent sensing and control system for riding protective equipment according to claim 5, characterized in that: The specific calculation process for characteristic difference values ​​and structural consistency checks includes: The calculation process of feature difference value is as follows: obtain the numerical range of each sensor feature in the current behavior map node state and the reference numerical range of the corresponding node state in the template map respectively, calculate the absolute value of the difference in value of each feature and the consistency of the direction of change, determine whether the absolute value of the difference is less than the preset allowable fluctuation range, and when all features meet the allowable fluctuation range, the node is determined as a feature matching node. The process of structural consistency check is as follows: In the behavior graph, the connection path between nodes is traced in chronological order, the direction attribute and continuity identifier of each transition connection in the path are recorded, and the direction change pattern of the path is compared with the direction change pattern in the template graph one by one. If the direction attribute and continuity identifier of each transition connection are consistent with the template graph, and there is no interruption or reverse transition in the path, then the path structure consistency is determined to meet the requirements. When both the feature difference value determination and the structural consistency check meet the set conditions, the corresponding node and path combination is marked as a structural matching unit.

7. The intelligent sensing and control system for riding protective equipment according to claim 6, characterized in that: The risk evolution analysis module includes: Analyze the trend of characteristic value changes at each node in the risk evolution path to determine whether the direction and magnitude of changes in acceleration, angular velocity, velocity, and trajectory direction characteristics meet the requirements of the preset trend change pattern. In the process of determining the trend change pattern, the cumulative time of the feature changes of continuous nodes in the path is detected to determine whether the duration of the trend feature has reached the set minimum duration threshold. When the trend change pattern meets the preset requirements, the duration of the trend feature reaches the minimum duration threshold, and the state transition direction sequence in the path is consistent with the preset direction change pattern, the current risk evolution path is determined to enter the judgment interval. Trigger the pre-response control logic and send an execution pre-response command to the protection control process.

8. The intelligent sensing and control system for riding protective equipment according to claim 7, characterized in that: The pre-response control module includes: Receive the pre-response command and parse it into the corresponding protective device control parameters, prompt system control parameters and optical signal control parameters; Send pre-inflation control commands to riding protective equipment and continuously monitor airbag pressure and inflation rate; Adjust the output rhythm of the prompting system, shorten the interval between sound or vibration prompts to within the set value, and add inertial side slip risk warning information to the prompt content to improve rider attention; Modify the control parameters of the optical signal, including increasing the flicker frequency of the light source, increasing the light intensity, or changing the color of the light source; During the execution of pre-response control actions, the working status data of the protective devices and the warning system are continuously collected and compared with the results of risk status changes.

Images