Dangerous chemical vehicle early warning method, device and equipment based on residence point detection and storage medium

Abstract

The invention provides a hazardous chemical vehicle early warning method, device and equipment based on residence point detection and a storage medium, and relates to the field of spatio-temporal data mining. According to the invention, a space-time density clustering method is adopted to preliminarily screen possible resident areas. In order to prevent a low-speed area of normal road driving from being mistakenly judged as a target resident area, the method comprises the following steps: clustering resident areas; the method comprises the following steps: firstly, constructing seven features of a maximum distance ratio, an average included angle degree, a ratio of the number of included angles greater than zero, an average speed, a speed standard deviation, a speed skewness and total time in a resident subsequence for an original trajectory data set based on feature derivation, and training a labeled resident area in combination with a classification algorithm, and judging whether the new resident subsequence is a normal resident area or not. According to the invention, peripheral search is carried out on the residence point through a crawler technology, and rapid early warning can be carried out on high-risk area residence, illegal rework enterprises and illegal residence vehicles.

Description

Technical Field

[0001] The present invention relates to the technical field of spatiotemporal data mining, and in particular to a hazardous chemicals vehicle early warning method, device, equipment and storage medium based on residence point detection. Background Art

[0002] Hazardous chemicals are a general term for flammable, explosive, and highly corrosive chemicals. Hazardous chemical transport vehicles are like a mobile source of danger. Abnormal residence may pose a great threat to people's property and public safety. With the rapid development of positioning systems, vehicles generate massive real-time trajectory data. Based on spatiotemporal data and through data mining technology, residence points can be identified, and targeted warnings can be made in combination with actual scenarios, which is of great significance for preventing safety risks.

[0003] In actual scenarios, for hazardous chemical vehicles, the goal is often to detect illegal parking locations, such as illegal factories, gas stations, service areas, etc. The current algorithm not only identifies parking points with target location stay characteristics, but also includes low-speed areas of normal road driving. Low-speed areas are often caused by congestion, traffic lights, etc., and are not the target of attention.

[0004] In addition, the current residence point recognition algorithms for hazardous chemicals vehicles mostly remain at the stage of identifying residence areas. They neither screen out the real residence areas nor combine map search technology to detect whether the residence areas are abnormal residences. Abnormal residences are often the business focus of hazardous chemicals vehicle detection. Summary of the invention

[0005] Purpose of the invention: To propose a hazardous chemicals vehicle early warning method, device, equipment and storage medium based on residence point detection to solve the problem of abnormal residence warning of hazardous chemicals vehicles.

[0006] In the first aspect, a hazardous chemicals vehicle early warning method based on stationary point detection is proposed, comprising the following steps:

[0007] Acquire an original trajectory data set of a hazardous chemicals transport vehicle, wherein the original trajectory data set includes a plurality of original trajectory points;

[0008] Based on the original trajectory data set, the instantaneous speed between the current moment and the previous moment is calculated, and the drift points whose instantaneous speed is greater than a predetermined value are filtered out from a plurality of original trajectory points;

[0009] Based on the spatiotemporal density clustering algorithm, several resident points in the original trajectory points are identified, and several resident points in the predetermined area are clustered to obtain a resident area; multiple resident areas are used to construct a resident subsequence, and multiple eigenvalues ​​of the resident subsequence are calculated to form a eigenvector;

[0010] Combined with the actual monitoring indicators, normal and abnormal residence areas are manually marked from several residence areas, which are used as output targets;

[0011] Select a classification algorithm model, use the feature vector as the input feature set, and divide it into a training set and a test set according to a predetermined ratio; use the training set to train the classification algorithm model, use the test set to evaluate the effect of the classification algorithm model, and output the trained model;

[0012] Based on the trained model, predict whether the latest resident subsequence after clustering preliminary screening is a resident area: if so, retain the current subsequence, and calculate the average longitude and latitude in the resident area as the current resident center point, otherwise do not retain it;

[0013] The data generated by each vehicle every day forms the trajectory of the vehicle on that day. The above steps are repeated for each trajectory until all trajectory points are detected;

[0014] Access the public map service API to build a warning scenario. If it is detected that the current vehicle trajectory meets the warning scenario, a warning message will be given.

[0015] In a further embodiment of the first aspect, the original trajectory data set includes the sequence ID, vehicle ID, longitude and latitude coordinates, and time data of the current transport vehicle; based on the longitude and latitude coordinates between the current time and the previous time, the distance difference d is converted i , thus calculating the instantaneous speed between the current moment and the previous moment t i Indicates the time difference between the current moment and the previous moment; trajectory points with instantaneous speed greater than 150km / h are regarded as drift points and are removed.

[0016] In a further embodiment of the first aspect, there are n position points in the resident subsequence, and the maximum distance proportion x1, average angle degree x2, proportion of angles greater than zero x3, average speed x4, speed standard deviation x5, speed skewness x6, and total time x7 of the resident subsequence are calculated respectively, and the above seven eigenvalues ​​constitute a eigenvector.

[0017] Among them, the calculation method of the maximum distance proportion x1 is as follows:

[0018]

[0019] Where D max Indicates the maximum distance between two positions in the resident subsequence, d i Indicates the distance difference between the current moment and the previous moment.

[0020] The calculation method of the average angle x2 is as follows:

[0021]

[0022] angle i =(arccos(P / (P1P2*P2P3)) / π)*180

[0023] The calculation method for the proportion of angles greater than zero x3 is as follows:

[0024]

[0025] In the formula, angle i represents the i-th angle. For the i-th angle, it is composed of three longitude and latitude points, namely (lat1, lng1), (lat2, lng2), and (lat3, lng3), and the angle range is 0-180°; P represents the dot product of the three-dimensional Cartesian coordinates corresponding to the three coordinate points; P1P2 represents the distance between the first coordinate point and the second coordinate point; P2P3 represents the distance between the second coordinate point and the third coordinate point.

[0026] The average speed x4 is calculated as follows:

[0027]

[0028] The speed standard deviation x5 is calculated as follows:

[0029]

[0030] In the formula, v i represents the instantaneous speed between the current moment and the previous moment; n is the number of position points in the resident subsequence;

[0031] The velocity skewness x6 is calculated as follows:

[0032]

[0033] In the formula, v i represents the instantaneous speed between the current moment and the previous moment; n is the number of position points in the resident subsequence;

[0034] The total time x7 is calculated as follows:

[0035] x7=t n -t1

[0036] In the formula, t n is the time of the nth position point in the resident subsequence; t1 is the time of the starting position point in the resident subsequence.

[0037] In a further embodiment of the first aspect, after accessing the public map service API, search parameters are configured, and the search parameters include the nearest companies, gas stations, and service areas.

[0038] In a further embodiment of the first aspect, the warning scenario includes:

[0039] Warning scenario 1: If the residence point is near a flammable and explosive area such as a gas station or a crowded area such as a service area, then mark the high-risk area for residence;

[0040] Warning scenario 2: Statistics on the parking areas in the past month. If the number of times all vehicles stop at a certain enterprise in a month exceeds the parking threshold, and the parking point belongs to a closed or suspended enterprise, the enterprise will be marked as illegally resuming work;

[0041] Warning scenario three: Count the number of abnormal stays in the past month. If the number of abnormal stays of a vehicle exceeds the abnormal stay threshold within a month, the vehicle will be marked as illegally parked.

[0042] According to a second aspect of the present invention, a hazardous chemical vehicle warning device based on residence point detection is proposed, which includes a data acquisition unit, a first calculation unit, a second calculation unit, a model training unit, a judgment unit, a driving trajectory output unit, and a warning unit.

[0043] The data acquisition unit is used to acquire an original trajectory data set of a hazardous chemicals transport vehicle, wherein the original trajectory data set includes a plurality of original trajectory points;

[0044] The first calculation unit calculates the instantaneous speed between the current moment and the previous moment based on the original trajectory data set, and filters out the drift points whose instantaneous speed is greater than a predetermined value from a plurality of original trajectory points;

[0045] The second calculation unit identifies a number of resident points in the original trajectory points based on the spatiotemporal density clustering algorithm, and clusters the resident points in the predetermined area to obtain a resident area; the multiple resident areas construct a resident subsequence, and calculates a number of eigenvalues ​​of the resident subsequence to form a eigenvector;

[0046] The model training unit is used to select a classification algorithm model, take the feature vector as an input feature set, and divide it into a training set and a test set according to a predetermined ratio; use the training set to train the classification algorithm model, use the test set to evaluate the effect of the classification algorithm model, and output the trained model;

[0047] The judgment unit predicts whether the latest resident subsequence after preliminary clustering screening is a resident area based on the trained model: if so, the current subsequence is retained, and the average longitude and latitude in the resident area are calculated as the current resident center point, otherwise, it is not retained;

[0048] The driving trajectory output unit is used to monitor the data generated by each vehicle every day to form the trajectory of the vehicle on that day until all trajectory points are detected;

[0049] The warning unit is used to access the public map service API and build a warning scenario. If it is detected that the current vehicle trajectory meets the warning scenario, a warning message is given.

[0050] According to a third aspect of the present invention, an electronic device is provided, which includes a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the hazardous chemical vehicle early warning method based on residence point detection as described in the first aspect is implemented.

[0051] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, in which at least one executable instruction is stored. When the executable instruction is executed on an electronic device, the electronic device executes the hazardous chemicals vehicle early warning method based on residence point detection as described in the first aspect.

[0052] Beneficial effects:

[0053] (1) The present invention realizes two functions of residence area identification and residence area filtering through data mining technology. First, the possible residence areas are preliminarily identified from the vehicle trajectory data through the spatiotemporal density clustering method. Compared with the residence point detection algorithm based on geometric characteristics, this method obtains the maximum density connected area through the density reachable relationship, effectively avoiding the problem of the same area being divided into two or more continuous residence points. Secondly, low-speed areas are usually caused by congestion, traffic lights and other situations, and all current residence point detection algorithms cannot identify low-speed areas. Even the essential idea of ​​most algorithms is to identify low-speed areas as residence areas. For this reason, the present invention further screens out the residence areas with business value generated by the loading and unloading of hazardous chemicals vehicles through feature derivation and training classification algorithms. At the same time, the 7 features derived by the present invention take into account multiple dimensions such as distance, angle, speed, time, etc., and most of them adopt the proportion or average method to reduce the impact of data problems such as inconsistent sample size and inconsistent time interval.

[0054] (2) The present invention provides a systematic early warning analysis solution, which uses crawler technology to search the surrounding areas of the residence points, obtains the location names corresponding to the residence points, and designs three scenarios specifically, which can achieve rapid early warning of residence in high-risk areas, illegal resumption of work enterprises, and illegally stationed vehicles, thereby generating practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of a hazardous chemicals vehicle early warning method based on residence point detection in an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram showing that a low-speed zone during normal road driving was misjudged as a retention point using the traditional solution before improvement.

[0057] Figure 3 It is a schematic diagram of multiple consecutive stay points in the same area.

[0058] Figure 4 It is a schematic diagram of the angles of trajectory points.

[0059] Figure 5 This is a schematic diagram of the resident area results after clustering.

[0060] Figure 6 It is a schematic diagram of the resident area results after screening. DETAILED DESCRIPTION

[0061] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features known in the art are not described.

[0062] In order to solve the problem of abnormal residence warning of hazardous chemical vehicles, the present invention provides a set of systematic solutions. The residence point extraction method based on geometric characteristics is currently a more commonly used extraction method. The method is simple to operate and easy to understand, but after the starting point is fixed, the method may just reach the upper threshold limit in the residence area, resulting in the interior of the residence area being disassembled into two or more continuous residence center points. In order to avoid this situation, the present invention uses a spatiotemporal density clustering method to preliminarily screen possible residence areas. The density clustering method can obtain a sample set connected by the maximum density through a density reachable relationship. In order to avoid misjudging the low-speed area of ​​normal road driving as the target residence area, the present invention first constructs the maximum distance proportion, average angle number, proportion of angles greater than zero, average speed, speed standard deviation, speed skewness, and total time in the residence subsequence based on feature derivation for the original trajectory data set after clustering. The marked residence area is then trained in combination with the classification algorithm to determine whether the new residence subsequence is a normal residence area. In order to specifically identify whether the parking point of hazardous chemicals vehicles is an abnormal parking point, the present invention searches for the nearest location name of the parking point that meets the business conditions, such as companies, gas stations, service areas, etc., based on the surrounding search function of the map service, and makes abnormal markings based on the actual scenario.

[0063] This embodiment discloses a hazardous chemical vehicle early warning method based on residence point detection. The technical solution of the present invention includes two parts. The first part is to identify and screen the normal residence area of ​​hazardous chemical vehicles based on data mining methods. The second part is to identify the nearest location name corresponding to the residence center point based on the surrounding search, and mark the abnormality in combination with the business scenario.

[0064] The first part is to identify vehicle parking areas based on data mining methods. First, possible parking areas are found based on the spatiotemporal density clustering method. Then, the parking areas are screened based on feature derivation and classification algorithms, filtering out low-speed areas caused by traffic jams, traffic lights, etc. The specific steps include the following:

[0065] S1. Obtain the original trajectory dataset of hazardous chemicals transport vehicles, including sequence ID, vehicle ID, longitude, latitude, and time data.

[0066] S2. Calculate the distance difference d between the current moment and the previous moment respectively i , time difference t i and speed v i , the distance difference can be calculated based on the two longitude and latitude coordinates through Vincenty (ellipsoid model), Haversine (spherical model) and other formulas. Finally, the drift points with a speed greater than 150km / h are filtered.

[0067] S3. Based on the spatiotemporal density clustering algorithm, the original trajectory points are identified for residence points, and the longest distance, total distance, total time, and spatial distribution of the suspected residence areas obtained by clustering are analyzed exploratoryly. In combination with the business scenarios and the exploratory analysis results, the distance, time and other parameters of the spatiotemporal density algorithm are adjusted to the optimal level.

[0068] S4. Construct seven features in the resident subsequence: the maximum distance ratio, the average angle ratio, the ratio of angles greater than zero, the average speed, the speed standard deviation, the speed skewness, and the total time. Low-speed areas caused by traffic jams, traffic lights, etc. often have the characteristics of a larger maximum distance ratio, a smaller average angle, a smaller number of angles greater than zero, and a more balanced speed distribution. Assuming there are n position points in the resident subsequence, the calculation method of each feature is as follows:

[0069] 1) Maximum distance ratio

[0070]

[0071] Where D max Indicates the maximum distance between two positions in the resident subsequence, d i Indicates the distance difference between the current moment and the previous moment.

[0072] 2) Average angle

[0073]

[0074] angle i =(arccos(P / (P1P2*P2P3)) / π)*180

[0075] P=(P 11 -P21 )*(P 31 -P 21 )+(P 12 -P 22 )*(P 32 -P 22 )+(P 13 -P 23 )*(P 33 -P 23 )

[0076]

[0077] P 11 =r*cos(lat1*π / 180)*cos(lng1*π / 180)

[0078] P 12 =r*cos(lat1*π / 180)*sin(lng1*π / 180)

[0079] P 13 =r*sin(lng1*π / 180)

[0080] P 21 =r*cos(lat2*π / 180)*cos(lng2*π / 180)

[0081] P 22 =r*cos(lat2*π / 180)*sin(lng2*π / 180)

[0082] P 23 =r*sin(lng2*π / 180)

[0083] P 31 =r*cos(lat3*π*180)*cos(lng3*π / 180)

[0084] P 32 =r*cos(lat3*π / 180)*sin(lng3*π / 180)

[0085] P 33 =r*sin(lng3*π / 180)

[0086] r=6371

[0087] Among them, angle i It represents the i-th angle. For the i-th angle, it consists of three longitude and latitude points, namely (lat1, lng1), (lat2, lng2), (lat3, lng3), and the angle range is 0-180°. r represents the radius of the earth in kilometers.

[0088] 3) The proportion of angles greater than zero

[0089]

[0090] 4) Average speed

[0091]

[0092] v i Indicates the speed between the current time and the previous time.

[0093] 5) Speed ​​standard deviation

[0094]

[0095] 6) Speed ​​skewness

[0096]

[0097] 7) Total time

[0098] x7=t n -t1

[0099] S5. Label the residence area based on the actual situation to determine whether it is a normal residence area. "0" represents a normal residence area and "1" represents an abnormal residence area. This is used as the output target. At the same time, the above 7 features are used as the input feature set, and are divided into training sets and test sets according to a certain ratio. Next, first select a commonly used classification algorithm model, and select the algorithm model with the highest score as the optimal model based on the k-fold cross-validation method. The scoring indicators can be more comprehensive indicators such as F1 and AUC to avoid the problem of accuracy failure caused by the small number of samples in the abnormal residence area. Secondly, based on the grid search algorithm and combined with k-fold cross-validation, adjust the parameters of the above optimal model to determine the optimal parameters. Finally, based on the optimal model after the above adjustment, train the training set, and evaluate the model effect in combination with the test set.

[0100] S6. Based on the model trained in S5, predict whether the latest resident subsequence after preliminary clustering screening is a resident area. If so, retain the current subsequence, and calculate the average longitude and latitude in the resident area as the current resident center point, otherwise do not retain it.

[0101] S7. The data generated by each vehicle every day forms the trajectory of the vehicle on that day. The above steps are repeated for each trajectory until all trajectory points are detected.

[0102] The second part is to identify the nearest corresponding location name of the residence point based on the surrounding search, and perform anomaly detection in combination with the business scenario. Specifically, it includes the following steps:

[0103] S1. Obtain permission to use public map service functions, such as Amap, Tencent, Baidu, etc., and enter the surrounding search function.

[0104] S2. Design a crawler program based on the interface parameters, which includes initiating access requests and performing searches and returning results.

[0105] S3. Configure search parameters based on actual scenarios, such as searching for the nearest company, gas station, service area, etc.

[0106] S4. Design the following early warning scenarios: First, if the residence point is near a flammable and explosive area such as a gas station or a crowded area such as a service area, mark it as a high-risk area; second, count the residence areas in the past month. If the number of times all vehicles stay in a certain enterprise exceeds the residence threshold within a month, and the residence point belongs to a closed or suspended enterprise, mark the enterprise that has resumed work illegally; third, count the number of abnormal residences in the past month. If the number of abnormal residences of a vehicle exceeds the abnormal residence threshold within a month, mark the illegally parked vehicle.

[0107] This embodiment also discloses a hazardous chemical vehicle warning device based on residence point detection, which is used to automatically execute the process of the hazardous chemical vehicle warning method based on residence point detection disclosed in the above embodiment. Specifically, the warning device is composed of a data acquisition unit, a first calculation unit, a second calculation unit, a model training unit, a judgment unit, a driving trajectory output unit, and a warning unit.

[0108] The data acquisition unit is used to acquire an original trajectory data set of a hazardous chemicals transport vehicle, wherein the original trajectory data set includes a plurality of original trajectory points.

[0109] The first calculation unit calculates the instantaneous speed between the current moment and the previous moment based on the original trajectory data set, and filters out drift points whose instantaneous speed is greater than a predetermined value from a plurality of original trajectory points.

[0110] The second calculation unit identifies several resident points in the original trajectory points based on the spatiotemporal density clustering algorithm, and clusters several resident points in a predetermined area to obtain a resident area; multiple resident areas construct a resident subsequence, and calculate multiple eigenvalues ​​of the resident subsequence to form a eigenvector.

[0111] The model training unit is used to select a classification algorithm model, take the feature vector as the input feature set, and divide it into a training set and a test set according to a predetermined ratio; use the training set to train the classification algorithm model, use the test set to evaluate the effect of the classification algorithm model, and output the trained model.

[0112] The judgment unit predicts whether the latest resident subsequence after preliminary clustering screening is a resident area based on the trained model: if so, the current subsequence is retained, and the average longitude and latitude in the resident area are calculated as the current resident center point, otherwise it is not retained.

[0113] The driving trajectory output unit is used to monitor the data generated by each vehicle every day to form the trajectory of the vehicle on that day until all trajectory points are detected.

[0114] The warning unit is used to access the public map service API and build a warning scenario. If it is detected that the current vehicle trajectory meets the warning scenario, a warning message is given.

[0115] The hazardous chemicals vehicle warning device is an electronic device that automatically executes the process of the hazardous chemicals vehicle warning method based on the stationary point detection disclosed in the above embodiment. The electronic device includes a processor, a memory, a communication interface and a communication bus. The processor, the memory and the communication interface communicate with each other through the communication bus. The memory is used to store at least one executable instruction, which enables the processor to execute Figure 1 The specific process of the hazardous chemicals vehicle early warning method based on residence point detection disclosed in the above-mentioned embodiments 1 to 2 will not be repeated here.

[0116] In addition, in the actual application process of the hazardous chemicals vehicle warning method based on residence point detection disclosed in the above embodiment, its operation logic can be written into a set of executable instructions, and the executable instructions can be written into a storage medium and run on an electronic device.

[0117] More specific examples of the computer-readable storage medium mentioned in this embodiment may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device.

[0118] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes may be made to it in form and detail without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A hazardous chemicals vehicle early warning method based on stationary point detection, characterized in that: The steps include: Acquire an original trajectory data set of a hazardous chemicals transport vehicle, wherein the original trajectory data set includes a plurality of original trajectory points; Based on the original trajectory data set, the instantaneous speed between the current moment and the previous moment is calculated, and the drift points whose instantaneous speed is greater than a predetermined value are filtered out from a plurality of original trajectory points; Based on the spatiotemporal density clustering algorithm, several residence points in the original trajectory points are identified, and several residence points in the predetermined area are clustered to obtain the residence area; The plurality of resident regions construct a resident subsequence, and a plurality of eigenvalues ​​of the resident subsequence are calculated. Composition feature vector; Combined with the actual monitoring indicators, normal and abnormal residence areas are manually marked from several residence areas, which are used as output targets; Select a classification algorithm model, use the feature vector as the input feature set, and divide it into a training set and a test set according to a predetermined ratio; use the training set to train the classification algorithm model, use the test set to evaluate the effect of the classification algorithm model, and output the trained model; Based on the trained model, predict whether the latest resident subsequence after clustering preliminary screening is a resident area: if so, retain the current subsequence, and calculate the average longitude and latitude in the resident area as the current resident center point, otherwise do not retain it; The data generated by each vehicle every day forms the trajectory of the vehicle on that day. The above steps are repeated for each trajectory until all trajectory points are detected; Access the public map service API to build a warning scenario. If it is detected that the current vehicle trajectory meets the warning scenario, a warning message will be given.

2. The method for early warning of hazardous chemicals vehicles based on residence point detection according to claim 1 is characterized in that: The original trajectory data set includes the sequence ID, vehicle ID, longitude and latitude coordinates, and time data of the current transport vehicle; Based on the latitude and longitude coordinates between the current time and the previous time, the distance difference d is calculated i , thus calculating the instantaneous speed between the current moment and the previous moment t i Indicates the time difference between the current moment and the previous moment; The trajectory points with instantaneous speed greater than 150 km / h are regarded as drift points and eliminated.

3. The method for early warning of hazardous chemicals vehicles based on residence point detection according to claim 2 is characterized in that: There are n position points in the resident subsequence, and the maximum distance proportion x1, average angle x2, proportion of angles greater than zero x3, average speed x4, speed standard deviation x5, speed skewness x6, and total time x7 of the resident subsequence are calculated respectively. The above seven eigenvalues ​​constitute a eigenvector.

4. The method for early warning of hazardous chemicals vehicles based on residence point detection according to claim 3 is characterized in that: The maximum distance ratio x1 is calculated as follows: Where D max Indicates the maximum distance between two positions in the resident subsequence, d i Indicates the distance difference between the current moment and the previous moment.

5. The method for early warning of hazardous chemicals vehicles based on residence point detection according to claim 3 is characterized in that: The calculation method of the average angle x2 is as follows: angle i =(arccos(P / (P1P2*P2P3)) / π)*180 The calculation method for the proportion of angles greater than zero x3 is as follows: In the formula, angle i represents the i-th angle. For the i-th angle, it is composed of three longitude and latitude points, namely (lat1, lng1), (lat2, lng2), and (lat3, lng3), and the angle range is 0-180°; P represents the dot product of the three-dimensional Cartesian coordinates corresponding to the three coordinate points; P1P2 represents the distance between the first coordinate point and the second coordinate point; P2P3 represents the distance between the second coordinate point and the third coordinate point.

6. The method for early warning of hazardous chemicals vehicles based on residence point detection according to claim 3 is characterized in that: The average speed x4 is calculated as follows: The speed standard deviation x5 is calculated as follows: In the formula, v i represents the instantaneous speed between the current moment and the previous moment; n is the number of position points in the resident subsequence; The velocity skewness x6 is calculated as follows: In the formula, v i represents the instantaneous speed between the current moment and the previous moment; n is the number of position points in the resident subsequence; The total time x7 is calculated as follows: x7=t n -t1 Where, t n is the time of the nth position point in the resident subsequence; t1 is the time of the starting position point in the resident subsequence.

7. The hazardous chemicals vehicle early warning method based on residence point detection according to any one of claims 1 to 6, characterized in that: After accessing the public map service API, it also includes: configuring search parameters, including the nearest companies, gas stations, and service areas; The warning scenarios include: Warning scenario 1: If the residence point is near a flammable and explosive area such as a gas station or a crowded area such as a service area, then mark the high-risk area for residence; Warning scenario 2: Statistics on the parking areas in the past month. If the number of times all vehicles stop at a certain enterprise in a month exceeds the parking threshold, and the parking point belongs to a closed or suspended enterprise, the enterprise will be marked as illegally resuming work; Warning scenario three: Count the number of abnormal stays in the past month. If the number of abnormal stays of a vehicle exceeds the abnormal stay threshold within a month, the vehicle will be marked as illegally parked.

8. A hazardous chemicals vehicle early warning device based on stationary point detection, characterized in that: include: Data acquisition unit; the data acquisition unit is used to obtain an original trajectory data set of a hazardous chemicals transport vehicle, wherein the original trajectory data set includes a number of original trajectory points; A first calculation unit; the first calculation unit calculates the instantaneous speed between the current moment and the previous moment based on the original trajectory data set, and filters out drift points whose instantaneous speed is greater than a predetermined value from a plurality of original trajectory points; The second computing unit identifies a number of residence points in the original trajectory points based on a spatiotemporal density clustering algorithm, and clusters the residence points in a predetermined area to obtain a residence area; The plurality of resident regions construct a resident subsequence, and a plurality of eigenvalues ​​of the resident subsequence are calculated. Composition feature vector; Model training unit; the model training unit is used to select a classification algorithm model, take the feature vector as an input feature set, and divide it into a training set and a test set according to a predetermined ratio; use the training set to train the classification algorithm model, use the test set to evaluate the effect of the classification algorithm model, and output the trained model; A judgment unit; the judgment unit predicts whether the latest resident subsequence after preliminary clustering screening is a resident area based on the trained model: if so, the current subsequence is retained, and the average longitude and latitude in the resident area are calculated as the current resident center point, otherwise, it is not retained; The driving track output unit is used to monitor the data generated by each vehicle every day to form the track of the vehicle on that day until all track points are detected; Warning unit: The warning unit is used to access the public map service API and build a warning scenario. If it is detected that the current vehicle trajectory meets the warning scenario, a warning message is given.

9. An electronic device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the hazardous chemicals vehicle early warning method based on residence point detection as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The storage medium stores at least one executable instruction. When the executable instruction is executed on the electronic device, the electronic device executes the hazardous chemicals vehicle early warning method based on residence point detection as described in any one of claims 1 to 7.

Images