Motorcycle safety guarding early warning system

Abstract

The present invention pertains to the technical field of motorcycles. Disclosed is a motorcycle safety guarding early warning system, comprising: a data acquisition module, used for acquiring real-time operating data of a motorcycle; a risk coefficient fitting module, used for inputting the acquired real-time operating data into a pre-constructed accident prediction model to output a risk coefficient; a fuzzy probability output module, which uses a fuzzy logic algorithm to evaluate the risk coefficient, and calculates the probability of an accident; and an early warning module, which uses a preset safety threshold, and triggers early warning when the probability of accident occurrence is greater than or equal to the safety threshold. Thus, too frequent false alarms are avoided while the sensitivity of early warning is ensured, improving the applicability and reliability.

Description

Motorcycle safety guardian early warning system TECHNICAL FIELD

[0001] The present application relates to the technical field of motorcycles, more particularly, the present application relates to a motorcycle safety guardian early warning system. BACKGROUND

[0002] The patent with the application publication number CN115798182A discloses an intelligent safety management method and system for a motorcycle, which obtains a matching operation level by matching the operation level based on the monitored operation data; constructs road condition features based on continuous environmental image acquisition results; performs matching evaluation based on the matching operation level and the road condition features, and generates first warning information based on the matching evaluation difference level; detects the posture of the motorcycle user through a sensor device, and generates second warning information based on the posture detection results and the matching evaluation difference level; and performs motorcycle operation safety warning according to the above warning information, which can solve the technical problem that the safety control of the motorcycle in the prior art depends on traffic rules and the driving ability and quality of the driver, resulting in weak effectiveness of motorcycle safety management, and achieves the technical effects of improving the controllability of motorcycle driving and improving the effectiveness of motorcycle safety management.

[0003] However, the prior art still lacks an effective collection and monitoring mechanism for real-time operation data of the motorcycle, cannot obtain the operation state of the vehicle in a timely manner, and is difficult to predict potential accident risks. The existing accident prediction method is often too simple, cannot fully consider various influencing factors, and has low prediction accuracy. The traditional accident early warning system usually triggers the warning using a fixed threshold, lacks a dynamic adjustment mechanism for different road conditions and environments, and has the risk of false positives or false negatives. The prior art cannot effectively integrate human experience and knowledge and language expression into the accident risk assessment process, resulting in a lack of explainability and rationality of the assessment results.

[0004] In view of this, the present application provides a motorcycle safety guardian early warning system to solve the above problems.

[0005] SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a motorcycle safety guardian early warning system, comprising: a data acquisition module for acquiring real-time operation data of a motorcycle;

[0007] a danger coefficient fitting module for inputting the acquired real-time operation data into a pre-constructed accident prediction model and outputting a danger coefficient;

[0008] a fuzzy probability output module for evaluating the danger coefficient using a fuzzy logic algorithm and calculating an accident probability;

[0009] An early warning module is used to preset a safety threshold, when the accident probability is greater than or equal to the safety threshold, the early warning is triggered; each module is connected through wired and / or wireless mode.

[0010] Further, the motorcycle real-time running data includes vehicle speed, acceleration, steering angle and throttle opening.

[0011] Further, the vehicle speed acquisition mode includes: installing a Hall sensor or an encoder on the motorcycle wheel, calculating the vehicle speed by detecting the wheel speed;

[0012] The acceleration acquisition mode includes:

[0013] A three-axis accelerometer sensor is installed on the motorcycle body, the acceleration of the motorcycle is calculated by measuring the acceleration values in three directions, including the front-back direction, the left-right direction and the up-down direction.

[0014] Further, the calculation formula for calculating the acceleration of the motorcycle is:

[0015] Wherein, ax is the acceleration value of the motorcycle in the front-back direction, ay is the acceleration value of the motorcycle in the left-right direction, and az is the acceleration value of the motorcycle in the up-down direction.

[0016] Further, the steering angle acquisition mode includes: installing an angle sensor on the steering wheel or steering system of the motorcycle, and acquiring the steering angle by detecting the steering angle of the steering wheel or steering system.

[0017] The throttle opening acquisition mode includes: installing a displacement sensor on the throttle system of the motorcycle, and acquiring the throttle opening by detecting the displacement of the throttle pull wire.

[0018] Further, the construction method of the accident prediction model includes:

[0019] Collecting the running data in the historical process in the experimental environment as the historical running data, the data type of the historical running data is the same as that of the motorcycle real-time running data;

[0020] The collected historical running data is preprocessed, the preprocessing method includes removing outliers, filling missing values and data standardization;

[0021] Features are extracted from the preprocessed historical running data, including acceleration first derivative, acceleration second derivative and the product of vehicle speed and steering angle;

[0022] The preprocessed historical operation data and the extracted features are organized into an input tensor or vector for training an accident prediction model, the accident prediction model adopts one of a deep neural network, a random forest, and a support vector machine, according to the input data, the accident prediction model will perform inference calculation, and output a value of a risk coefficient, the loss function of the accident prediction model is defined as a cross-entropy loss function, if the value of the loss function does not change in the process of continuous P times training, the construction of the accident prediction model is completed.

[0023] Further, the calculation method of the accident probability comprises:

[0024] The risk coefficient output by the accident prediction model is fuzzed into a fuzzy set, n fuzzy language values are defined in advance, and a membership function is set for each fuzzy language value, according to the numerical value of the risk coefficient, the value of the membership function on each fuzzy language value is calculated, and is recorded as membership;

[0025] A fuzzy rule base is established, which contains a series of fuzzy rules in the form of IF-THEN, the fuzzy rule associates the fuzzy language value corresponding to the risk coefficient with the fuzzy language value of the accident probability;

[0026] For the input risk coefficient, according to the membership on each fuzzy language value, the fuzzy reasoning mechanism is used to find the matching fuzzy rule in the rule base, and the membership of the fuzzy language value corresponding to the accident probability is calculated;

[0027] The fuzzy set obtained by reasoning is converted into an accident probability value through the centroid method, that is, the final output, to obtain the accident probability.

[0028] Further, the expression of the membership function is: Wherein, c is the center of the bell-shaped curve, a is the control curve width coefficient, b is the control curve steepness coefficient, and x is the input variable value.

[0029] Further, the preset method of the safety threshold comprises:

[0030] The road conditions and environment of motorcycle riding are classified, which are divided into expressway, urban road, rural road, rainy and snowy weather and night, for each type of road conditions and environment, a risk model is established;

[0031] The expression of the risk model of any type of road conditions and environment is: R=f(x1,x2,x3), wherein x1 is a road condition coefficient, x2 is a road visibility, x3 is a road traffic flow, R is a risk value of the type of road conditions and environment, and the value range is [0, 1], and f() is a function obtained by historical data training fitting;

[0032] The accident probability L is combined with the risk value R to obtain a dynamically adjusted safety threshold T=L*(1+R); when R is equal to 0, it represents an ideal road condition environment, and when R is greater than 0, the safety threshold will be increased y according to the risk value, and y is a fine tuning coefficient.

[0033] The technical effects and advantages of the motorcycle safety guardian early warning system of the application are as follows:

[0034] The motorcycle real-time running data is collected by various sensors, the vehicle state is comprehensively monitored, the data basis for accident prediction is provided, the advanced technology such as deep learning is used to construct an accident prediction model, the characteristics are learned from historical data, the accuracy of accident prediction is improved, the fuzzy logic algorithm is used to integrate human language expression and experience knowledge, the probability of accident occurrence is calculated, the explainability and rationality of the evaluation result are improved, the safety threshold is dynamically adjusted according to different road conditions and environments, the warning sensitivity is ensured, and the excessive frequent false alarm is avoided, and the practicability and reliability are improved. BRIEF DESCRIPTION OF DRAWINGS

[0035] Fig. 1 is a schematic diagram of the motorcycle safety guardian early warning system of the application;

[0036] Fig. 2 is a schematic diagram of the motorcycle safety guardian early warning method of the application. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0038] Embodiment 1

[0039] Please refer to Fig. 1, the motorcycle safety guardian early warning system described in the embodiment includes:

[0040] The data acquisition module is used for collecting real-time running data of the motorcycle;

[0041] The danger coefficient fitting module is used for inputting the collected real-time running data into the pre-constructed accident prediction model, and outputting a danger coefficient;

[0042] The fuzzy probability output module uses a fuzzy logic algorithm to evaluate the danger coefficient and calculate an accident probability;

[0043] An early warning module is used to preset a safety threshold, when the accident probability is greater than or equal to the safety threshold, the early warning is triggered; the modules are connected through wired and / or wireless mode to realize data transmission between the modules.

[0044] Further, the motorcycle real-time running data includes vehicle speed, acceleration, steering angle and throttle opening degree.

[0045] The vehicle speed acquisition mode includes: installing a Hall sensor or an encoder on the motorcycle wheel, and calculating the vehicle speed by detecting the wheel speed.

[0046] The acceleration acquisition mode includes:

[0047] A three-axis accelerometer sensor is installed on the motorcycle body, and the acceleration data of the motorcycle is calculated by measuring the acceleration values in three directions, including front and back, left and right, and up and down.

[0048] The steering angle acquisition mode includes: installing an angle sensor (such as a rotary encoder or an absolute value encoder) on the steering wheel or steering system of the motorcycle, and acquiring the steering angle by detecting the steering wheel or steering system.

[0049] The throttle opening degree acquisition mode includes: installing a displacement sensor (such as a potentiometer or a Hall sensor) on the throttle system of the motorcycle, and acquiring the throttle opening degree by detecting the displacement of the throttle pull wire.

[0050] All sensors are connected to the main controller through wired or wireless mode, and the collected real-time running data is transmitted to the main controller for subsequent data processing and analysis.

[0051] Further, the construction method of the accident prediction model includes:

[0052] The running data in the historical process is collected in the experimental environment, which is recorded as historical running data, and the data type of the historical running data is the same as that of the motorcycle real-time running data.

[0053] The collected historical running data is preprocessed, and the preprocessing method includes removing outliers, filling missing values and data standardization, to ensure the integrity and consistency of the data.

[0054] Features are extracted from the preprocessed historical running data, including acceleration first derivative, acceleration second derivative (i.e. jerk) and the product of vehicle speed and steering angle.

[0055] The pre-processed historical running data and the extracted features are organized into input tensors or vectors for training an accident prediction model, which adopts one of deep neural networks, random forests, and support vector machines. According to the input data, the accident prediction model will perform inference calculation and output a value of a risk coefficient. The loss function of the accident prediction model is defined as a cross-entropy loss function. If the value of the loss function does not change in the process of continuous P times of training, the construction of the accident prediction model is completed.

[0056] The risk coefficient is a floating-point value between 0 and 1, and the closer to 1, the higher the risk of an accident.

[0057] Over time, it may be necessary to retrain the accident prediction model using newly collected data to adapt to changes in roads, vehicles, and drivers, and to maintain the accuracy of the model.

[0058] Further, the calculation method of the accident probability includes:

[0059] The risk coefficient output by the accident prediction model is fuzzified into a fuzzy set. N fuzzy language values are defined in advance, such as "extremely low risk", "low risk", "medium risk", "high risk", and "extremely high risk". Each fuzzy language value is set with a membership function. According to the value of the risk coefficient, the value of the membership function of each fuzzy language value is calculated, denoted as membership.

[0060] The expression of the membership function is: Where c is the center of the bell-shaped curve, a is the control curve width coefficient, b is the control curve steepness coefficient, and x is the input variable value.

[0061] A fuzzy rule base is established, which contains a series of fuzzy rules in the form of IF-THEN. The fuzzy rule associates the fuzzy language value corresponding to the risk coefficient with the fuzzy language value of the accident probability, for example:

[0062] IF risk coefficient is extremely low risk THEN accident probability is extremely small;

[0063] The antecedent (IF part) of the rule describes the risk coefficient, and the consequent (THEN part) describes the corresponding accident probability. The fuzzy rule base is learned based on historical data.

[0064] For the input risk coefficient, according to its membership in each fuzzy language value, the fuzzy reasoning mechanism (such as Mamdani reasoning or Sugeno reasoning) is used to find the matching fuzzy rule in the rule base and calculate the membership of the fuzzy language value corresponding to the accident probability.

[0065] The fuzzy set obtained by reasoning is converted into an accident probability value, i.e. the final output, by using the centroid method, to obtain the accident probability;

[0066] The advantage of the fuzzy logic algorithm is that it can well handle uncertainty and subjectivity, and integrate human experience and knowledge and language expression into it. Through reasonable design of the fuzzy rule base, the fuzzy logic algorithm can well map the risk coefficient to the accident probability, and provide a basis for subsequent early warning triggering.

[0067] Further, the preset manner of the safety threshold comprises:

[0068] The road conditions and environment for motorcycle riding are classified, which are divided into expressway, urban road, rural road, rainy and snowy weather and night. For each type of road conditions and environment, a risk model is established for quantifying the influence of the road conditions and environment on the accident probability. The risk model is trained based on historical data.

[0069] The expression of the risk model of any type of road conditions and environment is R = f(x1, x2, x3), wherein x1 is a road surface condition coefficient, x2 is a road visibility, x3 is a road traffic flow, R is a risk value of the type of road conditions and environment, and the value range is [0, 1], and f() is a function obtained by fitting historical data training.

[0070] The accident probability L is combined with the risk value R to obtain a dynamically adjusted safety threshold T = L·(1+R); when R is equal to 0, it represents an ideal road condition and environment, and when R is greater than 0, the safety threshold will be increased y according to the risk value, y is a fine tuning coefficient, and the value is very small; so that the early warning system is more sensitive in the high-risk road condition and environment.

[0071] The early warning can more accurately dynamically adjust the safety threshold according to the real-time road conditions and environment, while ensuring the sensitivity of the early warning, and also avoids excessive frequent false alarms, thereby improving the practicability and reliability of the system.

[0072] In this embodiment, the real-time running data of the motorcycle is collected by various sensors, the vehicle state is comprehensively monitored, the data basis for accident prediction is provided, the advanced technology such as deep learning is used to construct an accident prediction model, the characteristics are learned from the historical data, the accuracy of accident prediction is improved, the fuzzy logic algorithm is used to integrate human language expression and experience knowledge, the probability of accident occurrence is calculated, the explainability and rationality of the evaluation result are improved, the safety threshold is dynamically adjusted for different road conditions and environment, while ensuring the sensitivity of the early warning, and also avoiding excessive frequent false alarms, thereby improving the practicability and reliability.

[0073] Embodiment 2

[0074] Please refer to Figure 2, the embodiment not described in detail part see example 1 described content, provide motorcycle safety guardian early warning method, including:

[0075] S1, collecting real-time running data of motorcycle;

[0076] S2, input the collected real-time running data into the pre-constructed accident prediction model, output the risk coefficient;

[0077] S3, evaluate the risk coefficient by fuzzy logic algorithm, and calculate the accident probability;

[0078] S4, preset safety threshold, when the accident probability is greater than or equal to the safety threshold, the early warning is triggered.

[0079] Example 3

[0080] [According to the rules 91 correction 22.07.2024] the embodiment discloses an electronic device, including memory, processor and computer program stored in memory and running on processor, the processor executes the computer program to realize the operation mode of the motorcycle safety guardian early warning method provided above.

[0081] Because the electronic device introduced in this embodiment is the electronic device used to implement the motorcycle safety guardian early warning method in this embodiment, based on the motorcycle safety guardian early warning method introduced in this embodiment, those skilled in the art can understand the specific implementation mode of the electronic device of this embodiment and its various forms, so the method how to realize the method in this embodiment is not introduced in detail. As long as those skilled in the art implement the electronic device used to implement the motorcycle safety guardian early warning method in this embodiment, it belongs to the scope of protection of this application.

[0082] The above formulas are dimensionless values, and the formulas are obtained by collecting a large amount of data to simulate the latest real situation, and the preset parameters and threshold values in the formula are set by those skilled in the art according to the actual situation.

[0083] The above is only the preferred embodiment of the present application, the protection scope of the present application is not limited to the above-mentioned examples, any technical scheme belonging to the idea of the present application is also within the protection scope of the present application. It should be pointed out that for ordinary technical users in this technical field, some improvements and decorations without departing from the principle of the present application are also considered as the protection scope of the present application.

Claims

1. A safety guardian warning system for a motorcycle, characterized in that, The motorcycle real-time running data includes vehicle speed, acceleration, steering angle and throttle opening degree. The vehicle speed acquisition mode includes: installing a Hall sensor or an encoder on the motorcycle wheel, and calculating the vehicle speed by detecting the wheel speed. The acceleration acquisition mode includes: The calculation formula for calculating the motorcycle acceleration is: The steering angle acquisition mode includes: installing an angle sensor on the steering wheel or steering system of the motorcycle, and acquiring the steering angle by detecting the steering wheel or steering system. The throttle opening degree acquisition mode includes: installing a displacement sensor on the throttle system of the motorcycle, and acquiring the throttle opening degree by detecting the displacement of the throttle pull wire.

2. The motorcycle safety guardian warning system of claim 1, wherein, The construction method of the accident prediction model includes:

3. The motorcycle safety guardian warning system of claim 2, wherein, The historical running data collected in the experimental environment is denoted as historical running data, and the data type of the historical running data is the same as that of the motorcycle real-time running data. The collected historical running data is preprocessed, and the preprocessing method includes removing outliers, filling missing values and data standardization. Features are extracted from the preprocessed historical running data, including acceleration first derivative, acceleration second derivative and the product of vehicle speed and steering angle.

4. The motorcycle safety guardian warning system of claim 3, wherein, The preprocessed historical running data and the extracted features are organized into input tensors or vectors for training the accident prediction model. The accident prediction model adopts one of deep neural network, random forest and support vector machine, and according to the input data, the accident prediction model will perform inference calculation to output a risk coefficient value.

5. The motorcycle safety guardian warning system of claim 4, wherein, The loss function of the accident prediction model is defined as cross-entropy loss function, and if the value of the loss function does not change in the continuous P times of training, the construction of the accident prediction model is completed. The accident probability calculation method includes:

6. The motorcycle safety guardian warning system of claim 5, wherein, The risk coefficient output by the accident prediction model is fuzzified into a fuzzy set, n fuzzy language values are defined in advance, and a membership function is set for each fuzzy language value. The value of the membership function of the risk coefficient on each fuzzy language value is calculated, denoted as membership degree. ​ ​ ​ 7. The motorcycle safety guardian warning system of claim 6, wherein, ​ ​ A fuzzy rule base is established, which contains a series of fuzzy rules in the form of IF-THEN, and the fuzzy rules associate the fuzzy language value of the risk coefficient with the fuzzy language value of the accident probability; For the input risk coefficient, according to the membership degree of the risk coefficient on each fuzzy language value, a matching fuzzy rule is found in the rule base by using a fuzzy reasoning mechanism, and the membership degree of the fuzzy language value of the accident probability is calculated; The fuzzy set obtained by reasoning is converted into an accident probability value by using the centroid method, i.e. the final output, to obtain the accident probability.

8. The motorcycle safety guardian warning system of claim 7, wherein, The expression of the membership function is: Wherein, c is the center of the bell curve, a is the control curve width coefficient, b is the control curve steepness coefficient, and x is the input variable value.

9. The motorcycle safety guardian warning system of claim 8, wherein, The preset mode of the safety threshold comprises: Classify the road conditions and environment of motorcycle riding, and divide them into highways, urban roads, rural roads, rainy and snowy weather and night, and establish a risk model for each type of road condition and environment; The expression of the risk model of any type of road condition and environment is R=f(x1, x2, x3), wherein x1 is a road surface condition coefficient, x2 is a road visibility, x3 is a road traffic flow, R is a risk value of the type of road condition and environment, and the value range of R is [0, 1], and f() is a function obtained by training and fitting historical data; The accident probability L is combined with the risk value R to obtain a dynamically adjusted safety threshold T=L·(1+R); when R is equal to 0, it represents an ideal road condition and environment, and when R is greater than 0, the safety threshold will be increased y according to the risk value, and y is a fine tuning coefficient.

Images