A method and system for fusion of environmental perception data of vehicle-mounted terminal

Abstract

The present invention relates to the technical field of intelligent sensing systems, and in particular to a method and system for fusing environmental perception data of an on-board terminal. The method comprises: performing cluster analysis based on a meteorological vector time series to obtain date clusters of various meteorological conditions; determining a travel deviation factor of a bus in each preset time period of each day in the date cluster according to a positioning parameter time subsequence, and determining a dust environment accumulation degree in each dimension at each moment of each preset time period of each day in the date cluster; determining an interpolation order determination coefficient of a dust parameter time subsequence in each dimension of each preset time period of each day in combination with the travel deviation factor and the dust environment accumulation degree; and performing spline interpolation of different orders based on the interpolation order determination coefficient, thereby improving the accuracy of data interpolation results in the environmental perception data fusion process, facilitating obtaining new dust parameter time series in each dimension, and further performing data fusion processing.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent sensing systems, and in particular to a method and system for fusing environmental perception data of a vehicle-mounted terminal. Background Art

[0002] With the acceleration of urbanization, urban atmospheric particulate matter pollution is becoming increasingly serious and has become a significant factor affecting urban environmental quality and resident health. The frequent occurrence of urban smog can be largely attributed to the increase in atmospheric particulate matter. As one of the main sources of urban atmospheric particulate matter, road dust pollution has attracted considerable attention. In urban environments, due to the large number of transportation vehicles such as ride-hailing services and buses, long transportation times, and wide coverage areas, urban road dust is characterized by its regularity, cyclical nature, rapid pollution rate, and comprehensive coverage. This makes road dust a significant source of urban air pollution, urgently requiring monitoring for effective control.

[0003] The application of on-board mobile dust online monitoring systems for monitoring road dust pollution provides a new approach to precise dust control. This requires integrating and analyzing different types of dust data. When processing data from different sensors, especially in vehicle-mounted terminals, different sampling principles lead to differences in the data parameters collected by different sensors. To effectively integrate and analyze this data, interpolation is often required to ensure that the data from different sensors are aligned on the timeline.

[0004] The existing spline interpolation fitting algorithm used for dust data interpolation processing has a fixed difference order that cannot adapt to the interference of space and traffic exhaust accumulation during daily driving of vehicle terminals. The accumulation of traffic exhaust causes the acquired dust data to have certain nonlinear performance, further affecting the accuracy of the dust data interpolation results, which is not conducive to the fusion of environmental perception data of vehicle terminals. Summary of the Invention

[0005] In order to solve the technical problem of low accuracy of data interpolation results in the process of environmental perception data fusion of the above-mentioned vehicle-mounted terminal, the purpose of the present invention is to provide a method and system for environmental perception data fusion of a vehicle-mounted terminal. The technical solutions adopted are as follows:

[0006] One embodiment of the present invention provides a method for fusing environmental perception data of a vehicle-mounted terminal, the method comprising the following steps:

[0007] Obtain the meteorological vector time series, positioning parameter time series, and dust parameter time series of several dimensions corresponding to the bus's onboard terminal during a preset historical period; perform cluster analysis based on the meteorological vector time series to obtain date clusters for various meteorological conditions;

[0008] For any date cluster, determine the bus travel deviation factor for each preset time period of each day in the date cluster according to the positioning parameter time subsequence corresponding to each preset time period of each day in the date cluster;

[0009] Determine the trend deviation value of the dust parameter in each dimension at each moment of each day in the date cluster, determine any moment of any day in the date cluster as the pending moment, and select the target positioning moment that meets the preset distance condition of the pending moment from all moments of the remaining days in the date cluster except the day;

[0010] The dust accumulation degree in each dimension at the pending time is determined based on the difference in dust parameters in the same dimension, positioning distance, and trend deviation value in the same dimension between the pending time and each target positioning time that meets the preset distance conditions at the pending time;

[0011] According to the bus travel deviation factor in each preset time period of each day in the date cluster and the dust environment accumulation degree in each dimension at each moment, the interpolation order determination coefficient of the dust parameter time subseries in each dimension in each preset time period of each day is determined;

[0012] Based on the interpolation order determination coefficient, spline interpolation of different orders is performed on the time series of dust parameters in each dimension to obtain new time series of dust parameters in each dimension, and then data fusion processing is performed.

[0013] Furthermore, the cluster analysis based on the meteorological vector time series to obtain the date clusters of various meteorological conditions includes:

[0014] Add the meteorological vectors belonging to the same day in the meteorological vector time series to obtain each representative meteorological vector;

[0015] The optimal number of clusters is determined, and then all representative meteorological vectors are clustered to obtain date clusters, which are recorded as date clusters of various meteorological conditions.

[0016] Furthermore, the method of determining the bus travel deviation factor for each preset divided time period of each day in the date cluster according to the positioning parameter time subsequence corresponding to each preset divided time period of each day in the date cluster includes:

[0017] According to the same preset time division method, each day in the date cluster is divided into time periods corresponding to each day;

[0018] In the positioning parameter time subsequence corresponding to each preset divided time period of each day, the positioning distance between adjacent moments is calculated, and all positioning distances corresponding to the same preset divided time period are added together to obtain the vehicle travel distance for each preset divided time period;

[0019] Based on the positioning distance of each unit time period in each preset divided time period of each day, calculate the positioning distance difference between any day and the same unit time period of the same preset divided time period of the remaining days in the date cluster except that day;

[0020] The travel deviation factor of the bus in each preset divided time period is determined based on the vehicle travel distance in each preset divided time period of each day and the positioning distance difference sequence corresponding to the same preset divided time period of any day and the remaining days in the date cluster except that day.

[0021] Furthermore, the bus travel deviation factor for each preset divided time period is determined based on the vehicle travel distance for each preset divided time period of each day and the positioning distance difference sequence corresponding to any day and the remaining days in the date cluster except for that day in the same preset divided time period, including:

[0022] Where w i represents the travel deviation factor of the bus in the i-th preset division period, i represents the serial number of the preset division period, norm represents the linear normalization function, Δl i Represents the difference in vehicle travel distance between any day and the remaining days in the date cluster except that day in the i-th preset division period, ∑Δl i represents the cumulative value of the difference in driving distance of all vehicles corresponding to the i-th preset division period, ε i represents the standard deviation of the positioning distance difference sequence corresponding to any day and the remaining days in the date cluster except this day in the i-th preset division period, ∑ε i Represents the cumulative value of all standard deviations corresponding to the i-th preset division period.

[0023] Furthermore, determining the trend deviation value of the dust parameter in each dimension at each moment of each day in the date cluster includes:

[0024] For any preset division period of each day in the date cluster, the dust parameter subsequence under any dimension of the preset division period is fitted to obtain the deviation value before and after fitting corresponding to the dust parameter under the dimension at each moment in the preset division period;

[0025] Obtain the deviation value of the dust parameter in the dimension at the moment before and after any moment, and take the average of the deviation values ​​of the two moments as the trend deviation value of the dust parameter in the dimension at the moment.

[0026] Furthermore, the step of selecting the target positioning moments that meet the preset distance condition of the pending moment from all moments of each day in the date cluster except the day includes:

[0027] For any preset time period, determine the positioning distance between the pending moment of the preset time period of any day in the date cluster and its adjacent moments before and after, and use the minimum value of the two positioning distances as the distance determination threshold;

[0028] Select the remaining positioning parameter subsequences of the preset time periods for each day in the date cluster, and calculate the positioning distance between each moment in the positioning parameter subsequence and the pending moment as a distance evaluation index;

[0029] Compare the distance evaluation index with the distance judgment threshold, and take the time corresponding to the distance evaluation index being less than the distance judgment threshold as the target positioning time that meets the preset distance condition of the pending time.

[0030] Furthermore, the determination of the dust environment accumulation degree in each dimension at the pending time based on the difference in dust parameters in the same dimension between the pending time and each target positioning time meeting the preset distance condition at the pending time, the positioning distance, and the difference in trend deviation value in the same dimension includes:

[0031] Where c o represents the dust accumulation degree under the target dimension at the undetermined time of day o. Any dimension is determined as the target dimension, J represents the number of target positioning moments, j represents the sequence number of the target positioning moment, and a o represents the dust parameter under the target dimension at the undetermined time, a oj represents the dust parameter under the target dimension at the jth target positioning time in the oth day that meets the preset distance condition at the pending time, (a o -a oj ) represents the target dimension dust parameter difference between the pending time and the jth target positioning time in the oth day that meets the preset distance condition of the pending time, k oj Indicates (a o -a oj ) weight coefficient, exp represents the exponential function with the natural constant e as the base;

[0032] norm represents the linear normalization function, Δm ojrepresents the positioning distance between the pending time on the oth day and the jth target positioning time on the oth day that meets the preset distance condition of the pending time, Δf oj It represents the target dimension trend deviation difference between the pending time on day o and the j-th target positioning time on day o that meets the preset distance condition of the pending time.

[0033] Furthermore, the interpolation order determination coefficient of the dust parameter time subsequence in each dimension in each preset time period of each day is determined based on the bus travel deviation factor in each preset time period of each day in the date cluster and the dust environment accumulation degree in each dimension at each moment, including:

[0034] For any preset time period of each day in the date cluster, obtain the dust environment accumulation degree subsequences under each dimension of the preset time period, perform similarity analysis on the dust environment accumulation degree subsequence under the target dimension and the dust environment accumulation degree subsequences under other dimensions except the target dimension, and obtain the dust accumulation commonality under the target dimension of the preset time period of each day;

[0035] Based on the travel deviation factor of buses in each preset time period and the commonality of dust accumulation in each dimension, the interpolation order determination coefficient of the dust parameter time subseries in each dimension of each preset time period is determined; among them, the travel deviation factor is negatively correlated with the interpolation order determination coefficient, and the commonality of dust accumulation is positively correlated with the interpolation order determination coefficient.

[0036] Furthermore, the interpolation order determination coefficient of the dust parameter time subsequence in each dimension in each preset time period of each day is determined based on the bus travel deviation factor and the dust accumulation commonality in each dimension in each preset time period of each day, including:

[0037] g oin =1-softmax[w oi ×(1-d oin )]; where g oin represents the interpolation order determination coefficient of the dust parameter time subsequence in the nth dimension of the i-th preset division period on the o-th day, softmax represents the maximum and minimum value normalization function, w oi represents the bus travel deviation factor in the i-th preset division period on the o-th day, d oin It represents the commonality of dust accumulation in the nth dimension of the i-th preset division period on the o-th day.

[0038] Another embodiment of the present invention provides an environment perception data fusion system for a vehicle-mounted terminal, including a processor and a memory, wherein the processor is configured to process instructions stored in the memory to implement an environment perception data fusion method for the vehicle-mounted terminal.

[0039] The present invention has the following beneficial effects:

[0040] The present invention provides a method and system for fusing environmental perception data of a vehicle-mounted terminal. By collecting a time series of meteorological vectors in a preset historical period, the preset historical period is divided into date clusters of various meteorological conditions. This method can avoid the influence of meteorological factors on dust parameters to a certain extent, and improve the accuracy of subsequent interpolation analysis results of dust parameters. Based on the positioning parameters of the same date clusters collected in the preset historical period, a travel deviation factor is determined to characterize the degree of influence of exhaust accumulation caused by traffic congestion on buses. The larger the travel deviation factor, the more the dust parameter is affected by local accumulation. In order to analyze the influence of dust data on exhaust accumulation in space The influence of the trend deviation value and the target positioning time is first determined, and then the dust environment accumulation degree, that is, the spatial accumulation performance, is analyzed based on the trend deviation value and the target positioning time; combined with the travel deviation factor and the dust environment accumulation degree, an adaptive interpolation order determination coefficient is determined for the dust parameter time subsequence of each dimension under each preset time period of each day. The dust parameter interpolation processing realized by the interpolation order determination coefficient can greatly improve the data accuracy of the dust parameters collected by the actual vehicle terminal, while avoiding the problem of inaccurate interpolation caused by nonlinearity, which is conducive to the environmental perception data fusion of the vehicle terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a flowchart of a method for fusing environmental perception data of a vehicle-mounted terminal according to an embodiment of the present invention;

[0043] Figure 2 A flowchart of the steps for determining the target positioning time in an embodiment of the present invention;

[0044] Figure 3 4 is a flowchart of step S5 in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] To further illustrate the technical means and effects employed by the present invention to achieve its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementations, structures, features, and effects of the technical solutions proposed by the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The application scenarios targeted by the present invention may be:

[0048] Different sensors installed on vehicle terminals have different parameters for collecting environmental perception data and different sampling principles, which makes the acquired data mismatched in time series. The conventional spline interpolation fitting algorithm has a fixed interpolation order when performing data interpolation, which cannot adapt to the interference of space and traffic exhaust during daily driving of vehicle terminals, making it difficult to perform environmental perception data fusion processing.

[0049] An embodiment of the present invention provides a method for fusing environmental perception data of a vehicle terminal, such as Figure 1 As shown, the following steps are included:

[0050] S1, obtain the meteorological vector time series, positioning parameter time series, and dust parameter time series of several dimensions corresponding to the bus's onboard terminal in a preset historical period; perform cluster analysis based on the meteorological vector time series to obtain date clusters of various meteorological conditions.

[0051] The above step S1 can be implemented through steps S11 to S12 (not shown):

[0052] S11, obtaining a meteorological vector time series, a positioning parameter time series, and a dust parameter time series of several dimensions corresponding to the onboard terminal of the bus in a preset historical period.

[0053] Here, the preset historical period can be set to the seven days closest to the current day. To reduce the amount of data analysis, this embodiment only collects environmental perception data during bus operation hours. Environmental perception data includes dust parameters, positioning parameters, and meteorological parameters. The sampling frequency is typically limited by different environmental perception data and sensors, meaning that the collection frequency of different environmental perception data is typically different. Of course, implementers can also customize the size of the preset historical period based on specific needs.

[0054] Specifically, the bus's onboard mobile dust monitoring system uses a pump-sampling method to collect environmental sensor data. This includes: PM2.5 and PM10 collected by particulate matter sensors; NO2, CO, SO2, O3, and TVOC collected by gas sensors; and meteorological parameters such as temperature, humidity, atmospheric pressure, wind speed, and wind direction, which form a meteorological vector. Simultaneously, the bus's real-time GPS (Global Positioning System) location records, i.e., positioning parameters, and the bus's latitude and longitude coordinates, are obtained using the onboard terminal. Ultimately, a time series of meteorological vectors, a time series of positioning parameters, and a time series of dust parameters in several dimensions are generated.

[0055] To facilitate subsequent data processing, this embodiment first collectively refers to the multi-dimensional monitoring data collected by the particulate matter sensor and gas sensor as dust parameters. The dust parameters and meteorological parameters are then pre-normalized and subjected to conventional denoising or outlier cleaning. Data pre-processing methods such as data normalization and denoising are conventional techniques and fall outside the scope of this invention, so they will not be elaborated on here.

[0056] S12, performing cluster analysis based on the meteorological vector time series to obtain date clusters of various meteorological conditions.

[0057] First, it's important to note that buses' onboard terminals are affected by the vehicle's movement. Different types of environmental perception data are collected during driving, and therefore typically require highly sensitive sensors. However, buses may experience varying road conditions, traffic conditions, and weather conditions while in motion. Furthermore, buses, due to their nature, have fixed stops, aside from traffic jams and other situations. Therefore, these driving behaviors can cause drastic changes in the environmental perception data collected by highly sensitive sensors.

[0058] In this embodiment, the dust parameters collected by the vehicle terminal are significantly affected by meteorological parameters. Therefore, before performing reference analysis, it is necessary to distinguish them based on meteorological vectors. That is, based on the meteorological vector time series, dates with similar meteorological conditions are clustered into the same cluster. The specific implementation steps include:

[0059] S221, adding meteorological vectors belonging to the same day in the meteorological vector time series to obtain representative meteorological vectors.

[0060] S222: Determine the optimal number of clusters, and then perform clustering processing on all representative meteorological vectors to obtain date clusters, which are recorded as date clusters of various meteorological conditions.

[0061] When determining each date cluster, the meteorological vectors for the same day are summed to obtain a representative meteorological vector. This allows each day to be used as sample data, with each representative meteorological vector serving as a distance metric for each dimension. The silhouette coefficient can be used to determine the optimal number of clusters. K-means clustering is then used to cluster all representative meteorological vectors into the optimal number of date clusters, each representing a group of buses under different weather conditions. The implementation of the silhouette coefficient and K-means clustering are both prior art and fall outside the scope of this invention, so they will not be elaborated on here.

[0062] So far, this embodiment has obtained the positioning parameter time series, the dust parameter time series of several dimensions, and the date clusters of various meteorological conditions determined based on the meteorological vector time series.

[0063] S2: determining the travel deviation factor of the bus in each preset divided time period of each day in the date cluster according to the positioning parameter time subsequence corresponding to each preset divided time period of each day in the date cluster.

[0064] Here, the travel deviation factor refers to the regularity of the bus's driving behavior under similar weather conditions. The stronger the regularity of the vehicle's driving behavior, the less interference the bus is subjected to under similar weather conditions.

[0065] In this embodiment, meteorological conditions will directly interfere with dust parameters, and different meteorological conditions have a certain impact on traffic. Therefore, when performing dust parameter and positioning parameter analysis later, the data in the same date cluster are analyzed. In the subsequent step analysis, this embodiment uses any date cluster as an example for illustration, that is, the parameters of the subsequent analysis are all data in the same date cluster.

[0066] The above step S2 can be implemented through steps S21 to S24 (not shown):

[0067] S21 , dividing each day in the date cluster according to the same preset time division method to obtain preset division time periods corresponding to each day.

[0068] It should be noted that when buses travel along pre-set routes, the peak travel times are almost the same every day, but the actual road conditions are different. Therefore, even if the bus driver leaves on time every day, the actual road conditions and the number of surrounding vehicles within the same time range will vary. These differences lead to different interference from the gas environment in the sensor within the same time range every day, such as the local accumulation of exhaust gas due to congestion around the bus. Therefore, in order to analyze environmental perception data within a local time range, the daily data collection period is divided into time periods. The specific implementation steps may include:

[0069] In this embodiment, the preset time division method can be set to be divided once every hour. The daily data collection period is divided into one time period per hour, and the preset division time periods corresponding to each day in the date cluster can be obtained. For example, if the environmental perception data subsequence corresponding to the preset division time period from 2:00 PM to 3:00 PM on a certain day in the date cluster is selected, the environmental perception data subsequence corresponding to the preset division time period from 2:00 PM to 3:00 PM on the remaining days in the date cluster can also be selected.

[0070] S22 , calculating positioning distances between adjacent moments in the positioning parameter time subsequence corresponding to each preset divided time period of each day, and adding up all positioning distances corresponding to the same preset divided time period to obtain the vehicle travel distance for each preset divided time period.

[0071] Here, the vehicle travel distance is obtained by adding the Euclidean distances between the positioning parameters at adjacent moments, which is used to subsequently analyze the vehicle travel distance deviation in the same preset divided time period every day to determine the travel deviation factor.

[0072] In this embodiment, first, based on the positioning frequency and the preset division time period, the positioning parameters of each moment of each day in each preset division time period in the same date cluster are obtained to form a positioning parameter time subsequence, and each preset division time period has its corresponding positioning parameter time subsequence; secondly, according to the time sequence and using the Euclidean distance calculation method, the positioning distance of each unit time period in the preset division time period is calculated respectively, and the cumulative calculation is performed, that is, the distance between the first moment and the second moment, and the distance between the second moment and the third moment. The calculation is repeated until all the positioning distances in the preset division time period are obtained, and all the positioning distances are added together to obtain the vehicle travel distance corresponding to the preset division time period.

[0073] S23 , calculating the positioning distance difference between any day and the same unit time period in the same preset divided time period of the remaining days in the date cluster except the day according to the positioning distance of each unit time period in each preset divided time period of each day.

[0074] Here, the positioning distance difference refers to the absolute value of the difference between the positioning distances of any two days in the same date cluster in the same unit time period. The positioning distance difference is used to analyze the changes in the positioning distance difference to determine the travel deviation factor.

[0075] In this embodiment, an analysis is performed using any day as an example to obtain the positioning distances of each unit time period of any day in the date cluster and each day except that day in the same preset time period. The absolute value of the difference between the positioning distances of the same unit time period in the same preset time period on different days is calculated, that is, the positioning distance difference. The positioning distance differences can then be sorted in chronological order.

[0076] S24, determining the bus travel deviation factor for each preset divided time period based on the vehicle travel distance for each preset divided time period of each day and the positioning distance difference sequence corresponding to any day and the remaining days in the date cluster except for that day in the same preset divided time period.

[0077] In this embodiment, the traffic interference corresponding to the same preset divided time period is usually similar. Therefore, the traffic interference situation can be evaluated by the difference in the vehicle driving distance of the bus in the same preset divided time period on different days. If the difference in vehicle driving distance is large, and the positioning distance difference in the same unit time period has poor regularity, it means that the bus driving in the preset divided time period is more disturbed, and the interference to the dust parameters collected on the on-board terminal is relatively strong.

[0078] As an example, the calculation formula for the bus travel deviation factor in the i-th preset division period of any day in any date cluster can be:

[0079] Where w i represents the travel deviation factor of the bus in the i-th preset division period, i represents the serial number of the preset division period, norm represents the linear normalization function, Δl i Represents the difference in vehicle travel distance between any day and the remaining days in the date cluster except that day in the i-th preset division period, ∑Δl i represents the cumulative value of the difference in driving distance of all vehicles corresponding to the i-th preset division period, ε i represents the standard deviation of the positioning distance difference sequence corresponding to any day and the remaining days in the date cluster except this day in the i-th preset division period, ∑ε i Represents the cumulative value of all standard deviations corresponding to the i-th preset division period.

[0080] In the calculation formula of the travel deviation factor, each bus in each preset time period of each day in the date cluster has its corresponding travel deviation factor. This embodiment takes the i-th preset time period of any day as an example for analysis. The travel deviation factor of other preset time periods can be determined by referring to the same calculation method; norm(∑Δl i) represents the normalization of the accumulated value of the difference in driving distance of all vehicles corresponding to the i-th preset divided time period. It is one of the important factors in determining the travel deviation factor. The travel deviation factor is quantified by analyzing the difference in driving distance of a certain day and the remaining days in the date cluster in the i-th preset divided time period. The greater the difference in driving distance between a certain day and other days under the same meteorological conditions in the same time period, the greater the interference effect on the driving process on that day, and the greater the driving deviation factor; norm(∑ε i ) represents the normalization of the accumulated values ​​of all standard deviations corresponding to the i-th preset division period. It is quantified by analyzing the degree of change in the positioning distance difference between a certain day and the remaining days in the date cluster in the i-th preset division period. The larger the standard deviation of the positioning distance difference sequence corresponding to the same preset division period on different days, the more irregular the positioning distance difference change, the greater the possibility of driving deviation, and the greater the interference effect on the dust parameters of the i-th preset division period of the day.

[0081] So far, this embodiment has obtained the travel deviation factor of the bus in each preset divided time period of each day in the date cluster.

[0082] S3, determining the trend deviation value of the dust parameter in each dimension at each moment of each day in the date cluster, and determining any moment of any day in the date cluster as the pending moment, and selecting the target positioning moment that meets the preset distance condition of the pending moment from all moments of the remaining days in the date cluster except the day.

[0083] It should be noted that since buses have fixed routes in the city, the dust parameters they actually collect are not only affected by the accumulation of exhaust gas during traffic, but are also affected by the spatial routes. For example, when a bus passes through sections with large gas emissions such as industrial parks, the dust parameters accumulated in the current environment are indeed the actual ones in the current environment, rather than the exhaust gas accumulation during traffic congestion.

[0084] The above step S3 can be implemented through steps S31 to S32 (not shown):

[0085] S31, determining the trend deviation value of the dust parameter in each dimension at each time of each day in the date cluster.

[0086] Here, the dust parameter of any dimension is taken as PM2.5 concentration. The deviation value refers to the absolute value of the difference between the actual dust parameter and the fitted dust parameter of the ideal smooth distribution. The trend deviation value refers to the numerical value of the two deviation values ​​before and after a certain moment. The steps for determining the trend deviation value may include:

[0087] S311 , for any preset time period of each day in the date cluster, fitting a subsequence of dust parameters in any dimension of the preset time period is performed to obtain deviation values ​​before and after fitting corresponding to the dust parameters in the dimension at each moment in the preset time period.

[0088] In this embodiment, a least-squares fitting method is performed on a subsequence of dust parameters for any dimension within a preset time period to obtain a fitted dust parameter value for that dimension at each moment in the preset time period. The absolute difference between the actual dust parameter for that dimension at each moment and the fitted dust parameter value is calculated as the deviation value for that dimension at that moment. The implementation of the least-squares fitting method is prior art and falls outside the scope of this invention, so it will not be elaborated on here.

[0089] S312: Obtain the deviation value of the dust parameter in the dimension at the moment before and after any moment, and take the average of the deviation values ​​of the two moments as the trend deviation value of the dust parameter in the dimension at the moment.

[0090] It should be noted that after determining the deviation value by calculating the difference in PM2.5 concentration before and after fitting at a certain moment, the smaller the deviation value, the more consistent the PM2.5 concentration at that moment is with the spatial changes of buses. The spatial accumulation of PM2.5 concentration does not show a more obvious single mutation, but the PM2.5 concentration of the entire preset time period tends to increase or decrease. After averaging the deviation values ​​of the adjacent moments, the situation of sudden outliers can be further avoided. The sudden outlier here refers to the abnormal situation of the data collected by the sensor caused by local exhaust accumulation when traffic congestion occurs.

[0091] S32: Determine any time of any day in the date cluster as the pending time, and select target positioning times that meet the preset distance condition of the pending time from all times of each day in the date cluster except the day.

[0092] Here, the target positioning moment refers to the moment that meets the preset distance condition of the pending moment, which is used for subsequent dust environment accumulation analysis. The preset distance condition refers to the distance judgment threshold less than that determined by the pending moment. Each moment in the positioning parameter sequence of each day in the same date cluster has its corresponding target positioning moments. Taking any preset division period as an example, the target positioning moment can be determined by Figure 2 Steps S321 to S323 shown implement:

[0093] S321, determining the positioning distance between the pending moment of a preset divided time period of any day in the date cluster and its adjacent preceding and following moments, and taking the minimum value of the two positioning distances as a distance determination threshold.

[0094] In this embodiment, the previous and next moments adjacent to the pending moment are obtained, and the positioning distances between the pending moment and the previous and next moments are calculated based on the positioning parameters. The two positioning distances are compared, and the smallest positioning distance is used as the distance determination threshold for the pending moment.

[0095] S322 , selecting the remaining positioning parameter subsequences of the preset time periods of each day in the date cluster, and calculating the positioning distance between each moment in the positioning parameter subsequence and the pending moment as a distance evaluation index.

[0096] In this embodiment, the remaining days in the date cluster cluster refer to the remaining days except any day in the date cluster cluster described in step S321. The preset division period is the preset division period to which the pending moment belongs. The distance evaluation index corresponding to each moment in the preset division period to which the pending moment of each remaining day in the date cluster cluster belongs is obtained, which is used for comparative analysis with the subsequent distance judgment threshold.

[0097] S323 , comparing the distance evaluation index with the distance determination threshold, and taking the time corresponding to when the distance evaluation index is less than the distance determination threshold as the target positioning time that meets the preset distance condition of the pending time.

[0098] It should be noted that the purpose of determining the target positioning time corresponding to the pending time is to determine the time in the remaining days of the same date cluster that is close to the pending time, which can be used to analyze the difference in dust concentration in similar spaces.

[0099] So far, this embodiment has obtained several target positioning moments corresponding to the pending moments.

[0100] S4, determining the dust environment accumulation degree in each dimension at the pending time based on the difference in dust parameters in the same dimension between the pending time and each target positioning time that meets the preset distance conditions at the pending time, the positioning distance, and the difference in trend deviation value in the same dimension.

[0101] Here, the dust accumulation degree refers to the dust accumulation caused by the driving space factors at a certain moment. The greater the dust accumulation degree, the greater the dust parameters at that moment are affected by the vehicle driving space factors. For example, the dust parameters at that moment are greatly affected by the emission gases from the industrial park on the driving route.

[0102] In this embodiment, a greater dust accumulation indicates the presence of an industrial park along the route. Consequently, the dust parameters at locations near the industrial park should be relatively large, manifesting as consistently large or relatively stable dust parameters for multiple consecutive days. A smaller dust accumulation indicates that the amplitude of the dust parameter at the time of determination should be relatively low. Therefore, large or abnormally sudden changes in dust parameters are typically caused by localized exhaust gas accumulation. For example, a weak dust accumulation can be represented by 1, 1, 8, 1, 1, while a strong dust accumulation can be represented by 3, 4, 8, 3, 2. Therefore, for the same dust parameter value of 8, a weak dust accumulation is more abnormal.

[0103] After obtaining the trend deviation value of the dust parameters in each dimension at each moment and the target positioning moments corresponding to each moment, analyze the dust parameters, positioning parameters and trend deviation values ​​of the dust parameters in each dimension between the pending moment and the target positioning moments that meet the preset distance conditions of the pending moment, calculate the difference in dust parameters in the same dimension, positioning distance and trend deviation value difference in the same dimension between the pending moment and any target positioning moment, and then determine the dust environment accumulation degree in each dimension at the pending moment.

[0104] As an example, the calculation formula for the dust accumulation degree in the target dimension at the undetermined time on day o can be:

[0105] Where c o represents the dust accumulation degree under the target dimension at the undetermined time of day o. Any dimension is determined as the target dimension, J represents the number of target positioning moments, j represents the sequence number of the target positioning moment, and a o represents the dust parameter under the target dimension at the undetermined time, a oj represents the dust parameter under the target dimension at the jth target positioning time in the oth day that meets the preset distance condition at the pending time, (a o -a oj ) represents the target dimension dust parameter difference between the pending time and the jth target positioning time in the oth day that meets the preset distance condition of the pending time, k oj Indicates (a o -a oj ), exp represents the exponential function with the natural constant e as the base.

[0106] In the calculation formula of dust accumulation degree, It represents the standard deviation of the difference in dust parameters between the pending time and each target positioning time that meets the preset distance conditions of the pending time. The larger the standard deviation, the greater the difference between the dust parameters in the target dimension corresponding to the pending time and the dust parameters at similar locations in the remaining daily clusters on the same date. The weaker the spatial dust accumulation at the pending time, the more likely it is affected by local accumulation under the current traffic conditions. Therefore, it is necessary to perform negative correlation processing on the standard deviation, that is, to use exp(-). Of course, implementers can also use other negative correlation processing methods to achieve this. Referring to the calculation method of the dust environment accumulation degree in the target dimension at the pending time on day o, the dust environment accumulation degree in each dimension at each time of each day can be obtained.

[0107] The calculation formula of the weight coefficient can be:

[0108] norm represents the linear normalization function, Δm oj represents the positioning distance between the pending time on the oth day and the jth target positioning time on the oth day that meets the preset distance condition of the pending time, Δf oj It represents the target dimension trend deviation difference between the pending time on day o and the j-th target positioning time on day o that meets the preset distance condition of the pending time.

[0109] In the calculation formula of the weight coefficient, norm(Δm oj ) represents the spatial positioning distance feature, norm(Δf oj ) represents the trend deviation characteristics of the dust parameters in the trend change. The difference between the pending time and the i-th target positioning time is analyzed from the two dimensions of space and trend to analyze the credibility of the existence of spatial dust accumulation. norm(Δm oj ) and norm(Δf oj ) is larger, the less credible the spatial dust accumulation degree analyzed between the pending time and the i-th target positioning time is, and on the contrary, the greater the impact of local accumulation caused by traffic congestion.

[0110] So far, this embodiment has obtained the dust accumulation degree in each dimension at each moment in each preset divided time period of each day in the date cluster.

[0111] S5, determining the interpolation order determination coefficient of the dust parameter time subsequence in each dimension in each preset time period of each day according to the bus travel deviation factor in each preset time period of each day and the dust environment accumulation degree in each dimension at each moment.

[0112] Here, the interpolation order determination coefficient is used to determine the adaptive order in the spline interpolation fitting algorithm. The interpolation order determination coefficient is determined by analyzing two factors that affect dust parameters: local exhaust gas accumulation caused by traffic jams and exhaust gas accumulation generated by industrial parks along the driving route.

[0113] For any preset division period of each day in the date cluster, the above step S5 can be performed by Figure 3 The steps S51 to S52 shown implement:

[0114] S51, obtaining the dust environment accumulation degree subsequences under each dimension of the preset divided time period, performing similarity analysis on the dust environment accumulation degree subsequence under the target dimension and the dust environment accumulation degree subsequences under other dimensions except the target dimension, and obtaining the dust accumulation commonality under the target dimension of the preset divided time period for each day.

[0115] Here, dust accumulation commonality refers to the similarity between dust accumulation in one dimension and dust accumulation in other dimensions. The dust accumulation degree actually reflects whether dust accumulation caused by spatial changes exists at each moment. This dust accumulation is reflected in multiple dimensional parameters, not just a single dust parameter. Therefore, we construct subsequences based on the obtained dust accumulation degree and perform similarity analysis on the subsequences of dust accumulation degrees in each dimension to determine the dust accumulation commonality.

[0116] As an example, the calculation formula for the dust accumulation commonality in the u-th dimension of the preset divided time period can be:

[0117] Where, d u represents the commonality of dust accumulation under the u-th dimension of the preset division period, softmax represents the maximum and minimum value normalization function, U represents the number of dimensions of dust parameters, h represents the serial number of other dimensions except the u-th dimension, D uh represents the DTW distance between the dust environment accumulation degree subsequence under the u-th dimension and the dust environment accumulation degree subsequence under the h-th dimension in the preset divided time period, It represents the similarity between the dust environment accumulation degree subsequence under the u-th dimension of the preset divided time period and the dust environment accumulation degree subsequence under the h-th dimension. +0.1 is to avoid the situation where the denominator of the fraction is zero.

[0118] In the calculation formula of dust accumulation commonality, D uhThe larger the value is, the worse the similarity between the dust environment accumulation degree subsequence in the u-th dimension of the preset divided time period and the dust environment accumulation degree subsequence in the h-th dimension is, and the smaller the dust accumulation commonality in the u-th dimension is; the larger the dust accumulation commonality is, the closer the dust parameters in the u-th dimension are to the accumulation performance trend in space, and the dust parameters in the u-th dimension are relatively less affected by the local accumulation interference caused by traffic congestion; referring to the calculation method of the dust accumulation commonality in the u-th dimension, the dust accumulation commonality in each dimension of each preset divided time period of each day can be obtained.

[0119] S52 , determining an interpolation order determination coefficient of a dust parameter time subsequence in each dimension in each preset divided time period of each day based on the bus travel deviation factor and the dust accumulation commonality in each dimension in each preset divided time period of each day.

[0120] Here, the larger the travel deviation factor, the stronger the interference of local accumulation caused by traffic congestion on the dust parameters, resulting in its temporal performance being inconsistent with the overall performance. When the driving deviation factor is used as the interpolation weight, a higher constraint coefficient is required, and the interpolation order determination coefficient should be smaller. The greater the commonality of dust accumulation, the stronger the interference of spatial accumulation caused by industrial parks on the dust parameters. When the commonality of dust accumulation is used as the interpolation weight, a higher constraint coefficient is not required, and the interpolation order determination coefficient should be larger.

[0121] As an example, the calculation formula for the interpolation order determination coefficient of the dust parameter time subsequence in the nth dimension in the i-th preset division period on the o-th day can be:

[0122] g oin =1-softmax[w oi ×(1-d oin )]; where g oin represents the interpolation order determination coefficient of the dust parameter time subsequence in the nth dimension of the i-th preset division period on the o-th day, softmax represents the maximum and minimum value normalization function, w oi represents the bus travel deviation factor in the i-th preset division period on the o-th day, d oin It represents the commonality of dust accumulation in the nth dimension of the i-th preset division period on the o-th day.

[0123] In the calculation formula of the interpolation order determination coefficient, the travel deviation factor is large, indicating that the current bus operation pattern is inconsistent with the norm, and the corresponding traffic congestion and other conditions are different. Therefore, the sensors installed in the on-board terminal may be affected by strong local dust gas accumulation interference, and the interpolation order determination coefficient should be smaller; the logic of the commonality of dust accumulation is exactly opposite to that of the travel deviation factor, so here a negative correlation is first performed and then multiplied by the travel deviation factor.

[0124] Thus, this embodiment obtains the interpolation order determination coefficient of the dust parameter time subsequence in each dimension of each preset divided time period of each day.

[0125] S6, based on the interpolation order determination coefficient, perform spline interpolation of different orders on the dust parameter time series of each dimension to obtain new dust parameter time series of each dimension, and then perform data fusion processing.

[0126] In this embodiment, after obtaining the interpolation order determination coefficient, all interpolation order determination coefficients are linearly mapped to 1 to 6, and the mapping results are rounded off to obtain the final determination order; based on the final determination order of the dust parameter time subsequence in each dimension of each preset time period of each day, spline interpolation of different orders is performed on the dust parameter time series of each dimension, so as to obtain new dust parameter time series of each dimension with a more uniform sampling frequency; the new dust parameter time series of each dimension is uploaded to the mobile air online detection system for data fusion processing.

[0127] Among them, the main menu bar in the cruise-type air online detection system interface can include: implementation monitoring, data analysis, report management, alarm management, point management, system management; real-time vehicle positioning, monitoring data, monitoring video, and data video overlay.

[0128] Another embodiment of the present invention provides an environment perception data fusion system for a vehicle-mounted terminal, including a processor and a memory, wherein the processor is configured to process instructions stored in the memory to implement an environment perception data fusion method for the vehicle-mounted terminal.

[0129] In summary, the present invention can greatly improve the data accuracy of dust gas collected by the actual vehicle-mounted terminal, while avoiding the problem of inaccurate interpolation caused by nonlinearity, and ultimately combine the system interface to integrate driving records and gas parameters to achieve the purpose of environmental perception fusion.

[0130] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for fusion of environmental perception data of a vehicle terminal, characterized in that: The following steps are involved: Obtain the meteorological vector time series, positioning parameter time series, and dust parameter time series of several dimensions corresponding to the bus's onboard terminal during a preset historical period; Cluster analysis is performed based on the meteorological vector time series to obtain date clusters of various meteorological conditions; For any date cluster, determine the bus travel deviation factor for each preset time period of each day in the date cluster according to the positioning parameter time subsequence corresponding to each preset time period of each day in the date cluster; Determine the trend deviation value of the dust parameter in each dimension at each moment of each day in the date cluster, determine any moment of any day in the date cluster as the pending moment, and select the target positioning moment that meets the preset distance condition of the pending moment from all moments of the remaining days in the date cluster except the day; The dust accumulation degree in each dimension at the pending time is determined based on the difference in dust parameters in the same dimension, positioning distance, and trend deviation value in the same dimension between the pending time and each target positioning time that meets the preset distance conditions at the pending time; According to the bus travel deviation factor in each preset time period of each day in the date cluster and the dust environment accumulation degree in each dimension at each moment, the interpolation order determination coefficient of the dust parameter time subseries in each dimension in each preset time period of each day is determined; Based on the interpolation order determination coefficient, spline interpolation of different orders is performed on the time series of dust parameters in each dimension to obtain new time series of dust parameters in each dimension, and then data fusion processing is performed.

2. The method for fusion of environmental perception data of a vehicle terminal according to claim 1, characterized in that: The cluster analysis based on the meteorological vector time series is performed to obtain date clusters of various meteorological conditions, including: Add the meteorological vectors belonging to the same day in the meteorological vector time series to obtain each representative meteorological vector; The optimal number of clusters is determined, and then all representative meteorological vectors are clustered to obtain date clusters, which are recorded as date clusters of various meteorological conditions.

3. The method for fusion of environmental perception data of a vehicle terminal according to claim 1, characterized in that: The method of determining the bus travel deviation factor for each preset divided time period of each day in the date cluster according to the positioning parameter time subsequence corresponding to each preset divided time period of each day in the date cluster includes: According to the same preset time division method, each day in the date cluster is divided into time periods corresponding to each day; In the positioning parameter time subsequence corresponding to each preset divided time period of each day, the positioning distance between adjacent moments is calculated, and all positioning distances corresponding to the same preset divided time period are added together to obtain the vehicle travel distance for each preset divided time period; Based on the positioning distance of each unit time period in each preset divided time period of each day, calculate the positioning distance difference between any day and the same unit time period of the same preset divided time period of the remaining days in the date cluster except that day; The travel deviation factor of the bus in each preset divided time period is determined based on the vehicle travel distance in each preset divided time period of each day and the positioning distance difference sequence corresponding to the same preset divided time period of any day and the remaining days in the date cluster except that day.

4. The method for fusion of environmental perception data of a vehicle terminal according to claim 3, characterized in that: The method of determining the bus travel deviation factor for each preset time period based on the vehicle travel distance for each preset time period of each day and the positioning distance difference sequence corresponding to the same preset time period of any day and the remaining days in the date cluster except for that day includes: Where w i represents the travel deviation factor of the bus in the i-th preset division period, i represents the serial number of the preset division period, norm represents the linear normalization function, Δl i Represents the difference in vehicle travel distance between any day and the remaining days in the date cluster except that day in the i-th preset division period, ∑Δl i represents the cumulative value of the difference in driving distance of all vehicles corresponding to the i-th preset division period, ε i represents the standard deviation of the positioning distance difference sequence corresponding to any day and the remaining days in the date cluster except this day in the i-th preset division period, ∑ε i Represents the cumulative value of all standard deviations corresponding to the i-th preset division period.

5. The method for fusion of environmental perception data of a vehicle terminal according to claim 3, characterized in that: The determination of the trend deviation value of the dust parameter in each dimension at each moment of each day in the date cluster includes: For any preset division period of each day in the date cluster, the dust parameter subsequence under any dimension of the preset division period is fitted to obtain the deviation value before and after fitting corresponding to the dust parameter under the dimension at each moment in the preset division period; Obtain the deviation value of the dust parameter in the dimension at the moment before and after any moment, and take the average of the deviation values ​​of the two moments as the trend deviation value of the dust parameter in the dimension at the moment.

6. The method for fusion of environmental perception data of a vehicle terminal according to claim 1, characterized in that: The step of selecting the target positioning moment that meets the preset distance condition of the pending moment from all the moments of each day in the date cluster except the current day includes: For any preset time period, determine the positioning distance between the pending moment of the preset time period of any day in the date cluster and its adjacent moments before and after, and use the minimum value of the two positioning distances as the distance determination threshold; Select the remaining positioning parameter subsequences of the preset time periods for each day in the date cluster, and calculate the positioning distance between each moment in the positioning parameter subsequence and the pending moment as a distance evaluation index; Compare the distance evaluation index with the distance judgment threshold, and take the time corresponding to the distance evaluation index being less than the distance judgment threshold as the target positioning time that meets the preset distance condition of the pending time.

7. The method for fusion of environmental perception data of a vehicle terminal according to claim 6, characterized in that: The method of determining the dust environment accumulation degree in each dimension at the pending time based on the difference in dust parameters in the same dimension between the pending time and each target positioning time meeting the preset distance condition at the pending time, the positioning distance, and the difference in trend deviation value in the same dimension includes: Where c o represents the dust accumulation degree under the target dimension at the undetermined time of day o. Any dimension is determined as the target dimension, J represents the number of target positioning moments, j represents the sequence number of the target positioning moment, and a o represents the dust parameter under the target dimension at the undetermined time, a oj represents the dust parameter under the target dimension at the jth target positioning time in the oth day that meets the preset distance condition at the pending time, (a o -a oj ) represents the target dimension dust parameter difference between the pending time and the jth target positioning time in the oth day that meets the preset distance condition of the pending time, k oj Indicates (a o -a oj ) weight coefficient, exp represents the exponential function with the natural constant e as the base; norm represents the linear normalization function, Δm oj represents the positioning distance between the pending time on the oth day and the jth target positioning time on the oth day that meets the preset distance condition of the pending time, Δf oj It represents the difference in target dimension trend deviation between the pending time on day o and the j-th target positioning time on day o that meets the preset distance condition of the pending time.

8. The method for fusion of environmental perception data of a vehicle-mounted terminal according to claim 7, characterized in that: The method of determining the interpolation order determination coefficient of the dust parameter time subsequence in each dimension in each preset time period of each day according to the bus travel deviation factor in each preset time period of each day and the dust environment accumulation degree in each dimension at each moment includes: For any preset time period of each day in the date cluster, obtain the dust environment accumulation degree subsequences under each dimension of the preset time period, perform similarity analysis on the dust environment accumulation degree subsequence under the target dimension and the dust environment accumulation degree subsequences under other dimensions except the target dimension, and obtain the dust accumulation commonality under the target dimension of the preset time period of each day; Based on the travel deviation factor of buses in each preset time period and the commonality of dust accumulation in each dimension, the interpolation order determination coefficient of the dust parameter time subseries in each dimension of each preset time period is determined; among them, the travel deviation factor is negatively correlated with the interpolation order determination coefficient, and the commonality of dust accumulation is positively correlated with the interpolation order determination coefficient.

9. The method for fusion of environmental perception data of a vehicle-mounted terminal according to claim 8, characterized in that: The interpolation order determination coefficient of the dust parameter time subsequence in each dimension in each preset time period of each day is determined based on the bus travel deviation factor in each preset time period of each day and the dust accumulation commonality in each dimension, including: g oin =1-softmax[w oi ×(1-d oin )]; where g oin represents the interpolation order determination coefficient of the dust parameter time subsequence in the nth dimension of the i-th preset division period on the o-th day, softmax represents the maximum and minimum value normalization function, w oi represents the bus travel deviation factor in the i-th preset division period on the o-th day, d oin It represents the commonality of dust accumulation in the nth dimension of the i-th preset division period on the o-th day.

10. An environmental perception data fusion system for a vehicle-mounted terminal, characterized in that: The method comprises a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement an environmental perception data fusion method for a vehicle-mounted terminal as described in any one of claims 1 to 9.

Images