Online car-hailing safety monitoring method and system

Abstract

The invention relates to the technical field of safety monitoring and alarming, in particular to an online car-hailing safety monitoring method and system. The method comprises the steps of collecting monitoring data of an online hailed car, obtaining a sensor fluctuation factor of each sensor in each non-driving time period based on a fluctuation condition and a noise influence degree of the monitoring data in the non-driving time period, evaluating the sensor fluctuation factor based on an accurate evaluation coefficient, and further obtaining a fusion weight of each sensor, and the data fusion module is used for fusing the monitoring data of the online car-hailing vehicle as the weight of each sensor in the data fusion of the weighted average method, carrying out safety monitoring on the online car-hailing vehicle based on the fused monitoring data, and giving an alarm for an abnormal condition. According to the invention, the monitoring effect of the running state of the online car-hailing car is improved, and the alarm efficiency of the abnormal condition of the online car-hailing car is improved.

Description

Technical Field

[0001] The present application relates to the field of security monitoring and alarm technology, and in particular to a method and system for monitoring the safety of online ride-hailing services. Background Art

[0002] With the development of industrial technology, cars have been applied to people's daily lives, and the development of online ride-hailing has also been very rapid. Among them, the safety management of online ride-hailing is of great significance to road safety. Compared with private cars, online ride-hailing has a longer mileage and is affected by more factors of the external environment. The road conditions, climate, and aging and wear of the components of the online ride-hailing vehicle itself will cause the online ride-hailing vehicle to be unstable or abnormal, which will affect the driving safety and stability of the vehicle. In the traditional online ride-hailing safety monitoring method, the data collected by the online ride-hailing sensor will be affected by the performance of the sensor during the data collection process, which will lead to poor monitoring of the online ride-hailing operation status and failure to timely alarm for abnormal conditions of the online ride-hailing vehicle.

[0003] The weighted average method is a common and effective data fusion approach, used in multi-sensor data fusion and machine learning model fusion. Its core concept is to perform a weighted average of the prediction results from multiple data sources or models to achieve a more accurate and stable final result. Traditional data fusion algorithms using the weighted average method, when fusing ride-hailing data for further monitoring and alarming, typically average the weights of each sensor, without considering sensor performance and assigning different weights to different sensors. Therefore, the weight assignment method of data fusion algorithms based on the weighted average method needs to be further improved. Summary of the Invention

[0004] In order to solve the problem that during the process of data collection by online car-hailing sensors, the collected data will be affected by the performance of the sensors, resulting in poor monitoring of the online car-hailing operation status, the present application provides an online car-hailing safety monitoring method and system.

[0005] In the first aspect, the present application provides a method for monitoring the safety of online ride-hailing vehicles, which adopts the following technical solutions: A method for monitoring the safety of online ride-hailing vehicles comprises the following steps: collecting monitoring data from various sensors of the online ride-hailing vehicle, obtaining a sensor turbulence factor of each sensor during each non-driving period based on fluctuations in the monitoring data during non-driving periods and the degree of noise influence, analyzing similarities between a changing trend of a peak value of the monitoring data during driving periods and the sensor turbulence factor, and obtaining an accurate evaluation coefficient of the sensor; evaluating the sensor turbulence factor based on the accurate evaluation coefficient to obtain an evaluation turbulence coefficient, and further obtaining a fusion weight of each sensor; using the fusion weight as a weight for each sensor in data fusion using a weighted average method; fusing the monitoring data of the online ride-hailing vehicle; performing safety monitoring of the online ride-hailing vehicle based on the fused monitoring data, and generating an alarm for abnormal monitoring data; The sensor turbulence factor is calculated as follows: for each sensor in a non-driving period, the sensor fluctuation factor is obtained based on the difference between the peak point and the average of the sensor's monitoring data; the sensor noise significance factor is obtained based on the difference between the sensor's monitoring data and the predicted value, and the product of the fluctuation factor and the noise significance factor is used as the sensor turbulence factor of the sensor.

[0006] This application divides the driving process of online ride-hailing vehicles into driving periods and non-driving periods. By integrating the fluctuation and noise characteristics of the non-driving period, the sensor turbulence factor is calculated, the reliability of the sensor is quantified, and the problem of environmental factors interfering with the sensor in the static state of the vehicle is solved; the peak change trend of the monitoring data in the driving period is combined with the sensor turbulence factor to calculate the evaluation turbulence coefficient, accurately evaluate the sensor turbulence factor, and solve the problem of environmental factors interfering with the sensor in the dynamic state; finally, the fusion weight of each sensor is obtained based on the turbulence evaluation coefficient, and the weight in the data fusion of the weighted average method is adjusted to improve the accuracy of the fusion of the monitoring data of each sensor, and reduce the problem of poor online ride-hailing monitoring effect caused by unreasonable sensor weights.

[0007] Furthermore, the method for obtaining the fluctuation factor is: Calculate the mean of all monitoring data of each sensor in each non-driving period as the average coefficient of the sensor data in each non-driving period; For the monitoring data of each sensor in each non-driving period, a peak detection algorithm is used to obtain the peak point of all monitoring data in the non-driving period; The absolute value of the difference between each peak point and the average coefficient of the sensor in the non-driving period is calculated, and the average of all normalized absolute values ​​of the difference is used as the fluctuation factor of each sensor in each non-driving period.

[0008] This application focuses on significant abnormal fluctuations by calculating the difference between the peak point and the mean, thereby avoiding steady-state vibration interference of the sensor of the online car-hailing vehicle during non-driving periods.

[0009] Furthermore, the method for obtaining the noise significance factor is: For the monitoring data collected by each sensor in each non-driving period, a time series model is used to predict it; Obtain the predicted value of the monitoring data collected by the sensor in each non-driving period, take the absolute value of the difference between the mean of all predicted values ​​and each monitoring data as the first difference, calculate the standard deviation of the normalized first difference of all monitoring data to obtain the noise significance factor of each sensor in each non-driving period.

[0010] This application dynamically quantifies the noise intensity by using the difference between the time series prediction value and the monitoring data. Compared with the traditional noise quantification based only on the existing monitoring data, it improves the environmental adaptability when evaluating the performance of the sensor.

[0011] Furthermore, the method for obtaining the accurate evaluation coefficient is: Obtaining a turbulence assessment factor for each sensor during each driving period based on a change trend near a peak point of monitoring data of the sensor during the driving period; For each sensor, the sequence composed of the sensor turbulence factors in all non-driving periods is taken as the first sequence, the sequence composed of the sensor turbulence assessment factors in all driving periods is taken as the second sequence, and the similarity between the first sequence and the second sequence is taken as the sensor accuracy assessment coefficient.

[0012] This application calculates the turbulence assessment factor based on the tangent slope of the peak point of the fitting curve. Compared with the traditional evaluation that directly uses the monitoring data of the driving period, it is less affected by random noise, highlights the real trend changes more, and the evaluation effect using accurate evaluation coefficients is more real and accurate.

[0013] Furthermore, the method for obtaining the turbulence assessment factor is: For each sensor, a curve fitting algorithm is used to obtain a fitting curve of its monitoring data during the driving period; For the fitting curve of the monitoring data of each sensor in each driving period, the peak detection algorithm is used to obtain the peak points on the fitting curve. With each peak point as the center, the absolute value of the tangent slope of the peak point on the fitting curve is used as the first absolute value, and the mean of all first absolute values ​​on the fitting curve is used as the turbulence assessment factor of each sensor in each driving period.

[0014] Furthermore, the similarity is the Pearson correlation coefficient between the first sequence and the second sequence.

[0015] Furthermore, the method for obtaining the evaluation turbulence coefficient is: Calculate the average of the sensor turbulence factors of each sensor in all non-driving periods to obtain the sensor turbulence coefficient; For each sensor, the product of the turbulence coefficient and the accurate evaluation coefficient is taken as the evaluation turbulence coefficient of the sensor.

[0016] This application combines the turbulence coefficient and the accuracy assessment coefficient to integrate dynamic indicators and evaluate them, avoiding the evaluation bias of a single evaluation criterion.

[0017] Furthermore, the method for obtaining the fusion weight is: Based on the autocorrelation of the monitoring data of the sensor in each driving period, the accuracy coefficient of each sensor is obtained; The calculation formula of the fusion weight is: Where, is the fusion weight of each sensor, is the precision coefficient of each sensor, is the evaluation turbulence coefficient of each sensor, is the preset initial fusion weight of each sensor.

[0018] Furthermore, the method for obtaining the precision coefficient is: Calculating the autocorrelation coefficient of the sequence composed of the monitoring data of each sensor in each driving period as the correlation trend of each sensor in each driving period; For each sensor, the mean of its related trends in all driving periods is taken as the accuracy coefficient of each sensor.

[0019] Secondly, this application provides a ride-hailing safety monitoring system that adopts the following technical solutions: A system for monitoring the safety of online ride-hailing vehicles includes a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, a method for monitoring the safety of online ride-hailing vehicles as described above is implemented.

[0020] The above-mentioned online car-hailing safety monitoring method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor. Thus, a system is made based on the memory and the processor for easy use.

[0021] This application has the following technical effects: The present application fuses the data of online-hailing vehicles based on the performance of each sensor on the online-hailing vehicle, thereby conducting safety monitoring of the online-hailing vehicle; for each non-driving period, based on the fluctuation of the data collected by each sensor and the difference between the data prediction value and the actual value, a sensor turbulence factor of each sensor in each non-driving period is constructed, reflecting the stability and noise level of the sensor, and the sensor turbulence factor is evaluated based on the change of the peak value of the sensor data during the driving period of the online-hailing vehicle to obtain an evaluated turbulence coefficient; finally, based on the evaluated turbulence coefficient and combined with the autocorrelation of the sensor data during the driving of the online-hailing vehicle, the weight of each sensor in the data fusion algorithm of the weighted average method is obtained, and the monitoring data collected by all sensors in each type of monitoring data is fused, so as to conduct safety monitoring of the online-hailing vehicle based on the fused data, thereby solving the problem that the collected data will be affected by the performance of the sensor and affect the effect of the monitoring data fusion, resulting in poor monitoring effect of the online-hailing vehicle's operating status, and failure to timely alarm of abnormal conditions of the online-hailing vehicle, thereby improving the monitoring effect of the online-hailing vehicle's operating status and improving the efficiency of alarming of abnormal conditions of the online-hailing vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a method flow chart of a method for monitoring the safety of online ride-hailing services in this application. DETAILED DESCRIPTION

[0023] An embodiment of the present application discloses a method for monitoring the safety of an online-hailing vehicle. The method collects monitoring data of the online-hailing vehicle, obtains a sensor turbulence factor of each sensor in each non-driving period based on the fluctuation of the monitoring data in the non-driving period and the degree of noise influence, analyzes the similarity between the changing trend of the peak value of the monitoring data in the driving period and the sensor turbulence factor in the non-driving period, obtains an accurate evaluation coefficient of the sensor, evaluates the sensor turbulence factor based on the accurate evaluation coefficient to obtain an evaluation turbulence coefficient, and then obtains the fusion weight of each sensor, which is used as the weight of each sensor in the data fusion of the weighted average method, fuses the monitoring data of the online-hailing vehicle, conducts safety monitoring of the online-hailing vehicle based on the fused monitoring data, and issues an alarm for abnormal monitoring data.

[0024] Reference Figure 1 , a method for monitoring the safety of online car-hailing includes steps S1 to S3.

[0025] S1: Collect monitoring data from various sensors of online ride-hailing vehicles.

[0026] In order to effectively monitor the driving process of each online car-hailing vehicle and promptly issue an alarm for abnormal situations, this application uses multiple sensors to obtain monitoring data of the online car-hailing vehicle. In one embodiment of this application, the monitoring data of the online car-hailing vehicle obtained is humidity data, temperature data, vehicle speed data, rotation speed data and acceleration data. The sensors selected are ADXL345 acceleration sensor and DHT12 temperature and humidity sensor. The implementer can select other types of online car-hailing vehicle monitoring data and other types of sensors for monitoring based on actual conditions. The monitoring data of each type of online car-hailing vehicle mentioned above is monitored using multiple sensors. In one embodiment of this application, the number of sensors used for the monitoring data of each type of online car-hailing vehicle is 3. The implementer can select the number of other sensors based on actual conditions.

[0027] S2: Based on the fluctuation of the monitoring data during the non-driving period and the degree of noise influence, obtain the sensor turbulence factor of each sensor during each non-driving period, analyze the similarity between the changing trend of the monitoring data peak during the driving period and the sensor turbulence factor, and obtain the accurate evaluation coefficient of the sensor; evaluate the sensor turbulence factor based on the accurate evaluation coefficient to obtain the evaluation turbulence coefficient.

[0028] Since online ride-hailing vehicles travel more mileage than private cars, their performance and components will degrade and age faster, resulting in unstable operating conditions. Traditionally, in the process of monitoring online ride-hailing data, multiple sensor data are used to monitor each type of online ride-hailing vehicle. In order to reduce the impact of redundant data, the online ride-hailing vehicle monitoring data needs to be fused. In the traditional weighted average data fusion algorithm, the setting of the weight of each sensor mainly relies on experience and is highly subjective, which will affect the fusion results of the online ride-hailing monitoring data, resulting in poor safety monitoring of online ride-hailing vehicles and the inability to issue alarms in abnormal situations.

[0029] Therefore, this application analyzes the situation of online car-hailing vehicles, sets different weights for different sensors, and obtains better data fusion effects, solving the problem that the monitoring data of online car-hailing vehicles is affected by the performance of the sensors, resulting in poor fusion results of the monitoring data of online car-hailing vehicles, which ultimately leads to poor safety monitoring of the operating status of online car-hailing vehicles and the inability to timely alarm for abnormal situations. Therefore, this application analyzes the monitoring data of online car-hailing vehicles collected by various sensors on the online car-hailing vehicles, further evaluates the performance of each sensor, and based on the evaluation results, assigns corresponding weights to each sensor in the data fusion algorithm based on the weighted average method.

[0030] Specifically, the time period when the online-hailing car is traveling is regarded as the driving period, and the time period when the online-hailing car stops traveling is regarded as the non-driving period.

[0031] Furthermore, in order to reflect the fluctuation of the monitoring data collected by each sensor under the influence of the environment during the non-driving period of the online car-hailing service, the mean of all monitoring data of each sensor in each non-driving period is calculated as the average coefficient of the data of each sensor in each non-driving period; further, for the monitoring data of each sensor in each non-driving period, a peak detection algorithm is used to obtain the peak point of all monitoring data in the non-driving period, and the absolute value of the difference between each peak point and the average coefficient of the sensor in the non-driving period is calculated, and the average of all normalized absolute values ​​of the difference is used as the fluctuation factor of each sensor in each non-driving period. It should be noted that during the non-driving period of the online car-hailing service, the sensor is not interfered with by more factors during the driving process, and the collected monitoring data can better reflect the performance of the sensor. When the fluctuation factor of the collected monitoring data is large, the performance of the sensor is poor; conversely, when the fluctuation factor of the collected monitoring data is small, the performance of the sensor is high. In one embodiment of the present application, the peak detection algorithm selected is the AMPD algorithm. The implementer can select other peak detection algorithms based on actual conditions.

[0032] Furthermore, to reflect the extent to which the monitoring data collected by sensors during the non-driving period of the online ride-hailing service is affected by noise, the monitoring data collected by each sensor in each non-driving period is predicted using the ARIMA model. The ARIMA model is a well-known technique and will not be described in detail here. For each sensor, the predicted value of the monitoring data collected by the sensor in each non-driving period is obtained, the absolute value of the difference between the mean of all preset values ​​and each monitoring data is calculated as the first difference, and the standard deviation of the normalized first difference of all monitoring data is calculated as the noise significance factor of each sensor in each non-driving period. It should be noted that the ARIMA model is a time series prediction model that can capture the changing trend of time series data and thus predict time series data. However, the ARIMA model is less effective in processing data with high noise. Therefore, when the monitoring data collected by the sensor is significantly affected by noise, the difference between the predicted data and the monitoring data collected by the sensor is large, and the value of the noise significance factor obtained in this case is large; conversely, the value of the noise significance factor obtained is small.

[0033] Specifically, since the fluctuation factor reflects the performance of the sensor in collecting data when it is less interfered by external environmental factors, and the noise significance factor reflects the influence of noise on the monitoring data collected by the sensor when it is less interfered by external environmental factors, in order to reflect the turbulence of each sensor when it is less interfered by external environmental factors, for each sensor data in each non-driving period, the product of the fluctuation factor and the noise significance factor is calculated as the sensor turbulence factor of the data collected by each sensor in each non-driving period.

[0034] It should be noted that during non-driving periods, online ride-hailing vehicles are less disturbed by the external environment and are in a relatively stable state. At this time, under normal circumstances, the smaller the volatility of the data of each sensor, and the data collected by the sensor is not affected by other uncertain factors during driving, the smaller the noise in the data collected by the sensor should be, and the smaller the value of the noise significance factor obtained at this time. At this time, the smaller the value of the sensor turbulence factor obtained, the higher the stability of the sensor and the lower the noise level; conversely, the larger the value of the sensor turbulence factor obtained, the lower the stability of the sensor and the higher the noise level.

[0035] The sensor turbulence factor is an analysis of the sensor turbulence based on the sensor data during the non-driving period. Furthermore, in order to evaluate the overall performance of the online ride-hailing vehicle during the driving period and the non-driving period, the accuracy of the sensor turbulence factor is evaluated based on the sensor data during the driving period of the online ride-hailing vehicle.

[0036] Specifically, for each sensor, a curve fitting algorithm is used to obtain a fitting curve of its monitoring data during the driving period. In one embodiment of the present application, the curve fitting algorithm selected is the least squares method. The implementer can select other curve fitting algorithms based on actual conditions.

[0037] For the fitting curve of the monitoring data of each sensor in each driving period, a peak detection algorithm is used to obtain the peak points on the fitting curve, and with each peak point as the center, the absolute value of the slope of the tangent of the peak point on the fitting curve is calculated as the first absolute value, and the average of all first absolute values ​​on the fitting curve is used as the turbulence assessment factor of each sensor in each driving period; the peak detection algorithm selected in one embodiment of the present application is the AMPD algorithm, and the implementer can select other peak detection algorithms based on actual conditions.

[0038] For the sensors on the online ride-hailing vehicles, the sensors are less subject to interference during the period when the online ride-hailing vehicles are stationary. However, the possibility of being interfered with by the environment is greater when the online ride-hailing vehicles are driving. Therefore, it is necessary to evaluate the sensor turbulence factor obtained during the non-driving period based on the changes in the sensor data during the driving process of the online ride-hailing vehicles.

[0039] When the first absolute value of the peak point is large, it means that the data obtained by the sensor during the driving of the online car-hailing vehicle will be greatly affected by the environment, and the data obtained by the sensor will have a large difference from other surrounding data due to interference; when the value of the sensor turbulence factor of each sensor on the online car-hailing vehicle is large, it means that the stability of the sensor is low. During the time period when the online car-hailing vehicle is driving, the stability of the data collected by the sensor is poor, and the value of the turbulence assessment factor during the corresponding driving period is large.

[0040] Therefore, based on the above analysis, the changing trend of the sensor turbulence factor is the same as that of the turbulence assessment factor. For each sensor, the sequence consisting of the sensor turbulence factor during all non-driving periods is used as the first sequence, and the sequence consisting of the sensor turbulence assessment factor during all driving periods is used as the second sequence. The similarity between the first and second sequences is used as the sensor accuracy assessment coefficient. In one embodiment of the present application, the Pearson correlation coefficient is selected as the measure of the similarity between the first and second sequences. Implementers may select other methods based on actual circumstances.

[0041] At this point, the accuracy evaluation coefficient of each sensor on the online car-hailing vehicle is obtained, and the sensor turbulence factor of the sensor in each non-driving period is evaluated. The larger the value of the accuracy evaluation coefficient, the more accurate the value of the sensor turbulence factor is, and the more it can reflect the sensor turbulence situation; conversely, the less accurate the value of the sensor turbulence factor is, the more difficult it is to reflect the sensor turbulence situation.

[0042] Furthermore, for each sensor, the average value of the sensor turbulence factor of the sensor in all non-driving periods is used as the turbulence coefficient of each sensor, and the evaluated turbulence coefficient is obtained based on the accurate evaluation coefficient. The calculation formula is: Where, is the evaluation turbulence coefficient of each sensor; is the turbulence coefficient of each sensor, is the accuracy evaluation coefficient of each sensor. It should be noted that the accuracy evaluation coefficient reflects the accuracy of the turbulence coefficient. When the sensor turbulence factor on the online car-hailing vehicle accurately evaluates the sensor turbulence, the obtained accuracy evaluation coefficient is larger, and the obtained evaluation turbulence coefficient is larger; conversely, the obtained evaluation turbulence coefficient is smaller. Among them, the evaluation turbulence coefficient accurately reflects the stability and noise level of the sensor in collecting online car-hailing data. The larger the value of the evaluation turbulence coefficient, the worse the sensor stability and the higher the noise level. At this time, when fusing the vehicle status data of the online car-hailing vehicle, in order to reduce the impact of the poor sensor stability and high noise level, it is necessary to give the sensor a lower weight. Conversely, when the value of the evaluation turbulence coefficient is smaller, it means that the sensor stability is better and the noise level is lower. At this time, when fusing the vehicle status data of the online car-hailing vehicle, it is necessary to give the sensor a higher weight.

[0043] S3: Obtain the fusion weight of each sensor, use the fusion weight as the weight of each sensor in the data fusion of the weighted average method, fuse the monitoring data of the online car-hailing vehicle, monitor the safety of the online car-hailing vehicle based on the fused monitoring data, and issue an alarm for abnormal monitoring data.

[0044] Furthermore, for the sequence composed of the monitoring data of each sensor in each driving period, the autocorrelation coefficient of the sequence is calculated as the correlation trend of each sensor in each driving period. It should be noted that during the driving of the online car-hailing service, the correlation trend reflects the correlation between the monitoring data of the sensor at different time points, and during the driving of the online car-hailing service, there is a correlation between the monitoring data obtained by the sensor. The greater the correlation, the more accurate the monitoring data obtained by the sensor.

[0045] Based on the above analysis, when the performance of the sensor is better, the sensor with better performance should be given a larger weight when fusing the sensor monitoring data. In this case, the value of the obtained correlation trend is larger and the value of the evaluation turbulence factor is smaller. Therefore, the weight of each sensor is assigned based on the correlation trend and the evaluation turbulence factor, specifically: For each sensor, the average of its related trends in all driving periods is used as the accuracy coefficient of each sensor. Furthermore, the weight of each sensor in each monitoring data is obtained based on the accuracy coefficient and turbulence coefficient. The calculation formula is: Where, is the fusion weight of each sensor, is the precision coefficient of each sensor, is the evaluation turbulence coefficient of each sensor, is the preset initial fusion weight of each sensor. In one embodiment of the present application, the preset initial fusion weight of each sensor is the same value. The implementer may select other values ​​based on actual conditions.

[0046] It should be noted that when the performance of the sensor is better, the obtained accuracy coefficient is larger and the evaluation turbulence coefficient is smaller, the sensor should be given a larger fusion weight at this time, and the value of the obtained fusion weight is larger; conversely, the value of the obtained fusion weight is smaller.

[0047] At this point, the fusion weight of each sensor in each type of monitoring data is obtained. For all sensors corresponding to each type of monitoring data, the data fusion algorithm of the weighted average method is used to fuse the monitoring data obtained by each sensor to obtain more accurate fused online car-hailing monitoring data. Among them, the data fusion algorithm of the weighted average method is a well-known technology and will not be described in detail in this application; further, for each type of monitoring data, the fused monitoring data is obtained, and the online car-hailing is safety monitored based on the fused monitoring data. When the monitoring data exceeds the normal range, an alarm device is used to sound an alarm to complete the safety monitoring of the online car-hailing; among them, the normal range of the monitoring data can be selected based on the actual situation of the online car-hailing, and will not be described in detail in this application.

[0048] An embodiment of the present application also discloses an online car-hailing safety monitoring system, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, an online car-hailing safety monitoring method according to the present application is implemented.

[0049] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0050] The above are all preferred embodiments of the present application, and are not intended to limit the scope of protection of the present application. Therefore, any equivalent changes made based on the structure, shape, and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A method for monitoring the safety of online ride-hailing vehicles, characterized in that: The method includes the following steps: collecting monitoring data of various sensors of an online car-hailing vehicle, obtaining a sensor turbulence factor of each sensor in each non-driving period based on the fluctuation of the monitoring data in the non-driving period and the degree of influence of noise, analyzing the similarity between the changing trend of the peak value of the monitoring data in the driving period and the sensor turbulence factor, and obtaining an accurate evaluation coefficient of the sensor; evaluating the sensor turbulence factor based on the accurate evaluation coefficient to obtain an evaluation turbulence coefficient, and then obtaining a fusion weight of each sensor, using the fusion weight as the weight of each sensor in the data fusion of the weighted average method, fusing the monitoring data of the online car-hailing vehicle, performing safety monitoring on the online car-hailing vehicle based on the fused monitoring data, and issuing an alarm for abnormal monitoring data; The sensor turbulence factor is calculated as follows: for each sensor in a non-driving period, the sensor fluctuation factor is obtained based on the difference between the peak point and the average of the sensor's monitoring data; the sensor noise significance factor is obtained based on the difference between the sensor's monitoring data and the predicted value, and the product of the fluctuation factor and the noise significance factor is used as the sensor turbulence factor of the sensor.

2. A method for monitoring the safety of online ride-hailing services according to claim 1, characterized in that: The method for obtaining the fluctuation factor is: Calculate the mean of all monitoring data of each sensor in each non-driving period as the average coefficient of the sensor data in each non-driving period; For the monitoring data of each sensor in each non-driving period, a peak detection algorithm is used to obtain the peak point of all monitoring data in the non-driving period; The absolute value of the difference between each peak point and the average coefficient of the sensor in the non-driving period is calculated, and the average of all normalized absolute values ​​of the difference is used as the fluctuation factor of each sensor in each non-driving period.

3. A method for monitoring the safety of online ride-hailing services according to claim 1, characterized in that: The method for obtaining the noise significance factor is: For the monitoring data collected by each sensor in each non-driving period, a time series model is used to predict it; Obtain the predicted value of the monitoring data collected by the sensor in each non-driving period, take the absolute value of the difference between the mean of all predicted values ​​and each monitoring data as the first difference, calculate the standard deviation of the normalized first difference of all monitoring data to obtain the noise significance factor of each sensor in each non-driving period.

4. A method for monitoring the safety of online ride-hailing services according to claim 1, characterized in that: The method for obtaining the accurate evaluation coefficient is: Obtaining a turbulence assessment factor for each sensor during each driving period based on a change trend near a peak point of monitoring data of the sensor during the driving period; For each sensor, the sequence composed of the sensor turbulence factors in all non-driving periods is taken as the first sequence, the sequence composed of the sensor turbulence assessment factors in all driving periods is taken as the second sequence, and the similarity between the first sequence and the second sequence is taken as the sensor accuracy assessment coefficient.

5. A method for monitoring the safety of online ride-hailing services according to claim 4, characterized in that: The method for obtaining the turbulence assessment factor is: For each sensor, a curve fitting algorithm is used to obtain a fitting curve of its monitoring data during the driving period; For the fitting curve of the monitoring data of each sensor in each driving period, the peak detection algorithm is used to obtain the peak points on the fitting curve. With each peak point as the center, the absolute value of the tangent slope of the peak point on the fitting curve is used as the first absolute value, and the mean of all first absolute values ​​on the fitting curve is used as the turbulence assessment factor of each sensor in each driving period.

6. A method for monitoring the safety of online ride-hailing services according to claim 4, characterized in that: The similarity is the Pearson correlation coefficient between the first sequence and the second sequence.

7. A method for monitoring the safety of online ride-hailing services according to claim 1, characterized in that: The method for obtaining the evaluation turbulence coefficient is: Calculate the average of the sensor turbulence factors of each sensor in all non-driving periods to obtain the sensor turbulence coefficient; For each sensor, the product of the turbulence coefficient and the accurate evaluation coefficient is taken as the evaluation turbulence coefficient of the sensor.

8. A method for monitoring the safety of online ride-hailing vehicles according to claim 1, characterized in that: The method for obtaining the fusion weight is: Based on the autocorrelation of the monitoring data of the sensor in each driving period, the accuracy coefficient of each sensor is obtained; The calculation formula of the fusion weight is: Where, is the fusion weight of each sensor, is the precision coefficient of each sensor, is the evaluation turbulence coefficient of each sensor, is the preset initial fusion weight of each sensor.

9. A method for monitoring the safety of online ride-hailing services according to claim 8, characterized in that: The method for obtaining the precision coefficient is: Calculating the autocorrelation coefficient of the sequence composed of the monitoring data of each sensor in each driving period as the correlation trend of each sensor in each driving period; For each sensor, the mean of its related trends in all driving periods is taken as the accuracy coefficient of each sensor.

10. A safety monitoring system for online car-hailing, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for monitoring the safety of an online car-hailing service according to any one of claims 1 to 9 is implemented.

Images