A method for automatically determining the accurate start and end times of loading and unloading goods of a freight truck

Abstract

The present invention provides a method for automatically determining the accurate start and end times of loading and unloading of freight trucks, comprising the following steps: constructing an original data queue and setting an original data sliding time window, constructing a mean data queue and setting a mean data sliding time window, constructing a slope data queue and setting a slope data sliding time window, constructing an action event queue, processing the action event queue, and judging the authenticity of freight truck action events. The present invention analyzes the data uploaded by the vehicle load sensor through four data queues and three sliding time windows, and combines the spectrum analysis method. First, all possible loading and unloading event units are identified, then the continuous loading and unloading event units are assembled into loading and unloading events, and finally the authenticity of the loading and unloading events is judged to filter out false events caused by interference jitter; it can widely adapt to various different loading and unloading scenarios, effectively cope with data jitter and short-term loss, and can accurately and timely identify the start and end time points of loading and unloading of freight trucks.

Description

Technical Field

[0001] The present invention belongs to the technical field of automatic monitoring of freight transportation, and particularly relates to a method for automatically determining the accurate start and end times of loading and unloading of freight trucks. Background Art

[0002] In order to meet the needs of freight transportation enterprises to automatically identify the accurate start and end time points of loading and unloading, vehicle load sensors are generally installed on freight trucks. The sensors can detect the load, speed, and geographical longitude and latitude information of the vehicle in real time, and report them to the background system regularly through the mobile data network. The background system needs to identify the true and accurate start and end time points of loading and unloading of the freight truck in a timely manner based on the data reported by the load sensor, so that managers can monitor and query on the background interface.

[0003] The algorithms in the industry for identifying the accurate start and end times of loading and unloading of freight trucks mainly rely on the slope of the load curve reported by the load sensor rising or falling within a certain period of time, and then comprehensively judge in combination with the vehicle speed value. However, due to the complex scenarios of loading and unloading of freight trucks, in some scenarios, loading and unloading are very fast, while in some scenarios, loading and unloading are slow, and there may also be short pauses during the operation. In addition, the data of the load sensor fluctuates greatly due to the shaking of the carriage or electromagnetic environment interference, and the data uploaded through the mobile data network may also be lost. This makes it difficult for the background algorithm to analyze. If the slope threshold for judging loading and unloading is too large, there may be missed judgments, and if the slope threshold for judging loading and unloading is too small, interference events may be misjudged as loading and unloading. Summary of the Invention

[0004] Aiming at the above problems, the purpose of the present invention is to provide a method for automatically determining the accurate start and end times of loading and unloading of freight trucks. Through this method, it can widely adapt to various different loading and unloading scenarios, effectively cope with data jitter and short-term loss, and can more accurately and timely identify the start and end time points of loading and unloading of freight trucks.

[0005] The purpose of the present invention is achieved through the following technical solutions: A method for automatically determining the accurate start and end times of loading and unloading of freight trucks, including the following steps: S1. Construct an original data queue and set an original data sliding time window: Each original data reported by the vehicle load sensor constitutes an original data event. All original data events within a fixed duration t are sorted according to the data timestamp to form a rectangular original data queue; the original data in each original data event includes a timestamp, a load value, a speed value, and longitude and latitude.

[0006] Set an original data sliding time window on the original data queue. The horizontal length of the original data sliding time window is a fixed duration t1; the sliding step of the original data sliding time window is T, and T is the time interval for the vehicle load sensor to report data.

[0007] S2. Construct a mean data queue and set a sliding time window for mean data: Each time the sliding time window of the original data slides one step, calculate the mean timestamp, mean load, mean speed, and mean longitude and latitude using the original data within the sliding time window of the original data at this time. The above mean data forms a new mean data event. Sort all mean data events within a fixed duration t in timestamp order to form a mean data queue.

[0008] Set a sliding time window for mean data on the mean data queue. The horizontal length of the sliding time window for mean data is a fixed duration t2; the sliding step of the sliding time window for mean data is one mean data event; S3. Construct a slope data queue and set a sliding time window for slope data: Each time the sliding time window for mean data slides one step, perform a linear regression on the mean load using the mean data within the sliding time window for mean data at this time. The slope of the regression line is used as the slope of the load curve for this sliding time window of mean data. At the same time, calculate the quadratic mean speed for the mean speed within the sliding time window for mean data at this time, and record the start and end timestamps of this sliding time window for mean data. The above data forms a new slope data event. Sort all slope data events within a fixed duration t in time order to form a slope data queue.

[0009] Set a sliding time window for slope data on the slope data queue. The horizontal length of the sliding time window for slope data is a fixed duration t3; the sliding step of the sliding time window for mean data is one slope data event.

[0010] S4. Construct an action event queue: Each time the sliding time window for slope data slides one step, calculate the mean slope using the slope of the load curve within the sliding time window for slope data at this time, and further smooth the mean slope. At the same time, calculate the cubic mean speed for the quadratic mean speed within the sliding time window for slope data again, and record the start and end timestamps, start and end mean loads, and start longitude and latitude of this sliding time window for slope data. Generate a new truck action event using the above data. Sort all truck action events within a fixed duration t in timestamp order to form an action event queue for the truck; the types of the truck action events include running events, loading events, unloading events, and stationary events.

[0011] S5. Processing of the action event queue: Whenever a new truck action event is generated, first check the action event queue. If the action event queue is empty and the new event is a loading event or an unloading event, then put the new event into the action event queue; otherwise, if there is an old event in the action event queue, process it according to the following steps: S51. If the action types of the old event and the new event are the same, it means that two adjacent events are in the same continuous loading or unloading process. Just extend the end timestamp of the old event to the end timestamp of the new event, and update the end load value of the old event with the end load value of the new event.

[0012] S52. If the new event is a running event, remove the event from the queue and make a true / false judgment.

[0013] S53. If the new event is a stationary event and the stationary time exceeds the threshold t4, then remove the event from the queue and make a true / false judgment.

[0014] S54. If the action type of the new event is opposite to that of the old event, then remove the event from the queue and make a true / false judgment, and put the new event into the queue.

[0015] S6. True / false judgment of truck action events.

[0016] After that, the final loading and unloading events are obtained.

[0017] Make a true / false judgment on the truck action events removed from the action event queue in steps S52 - S54.

[0018] The t is equal to 0.5 hours to 8 hours; T is equal to 30 seconds to 60 seconds; t1 is equal to 3 minutes; t2 is equal to 5 - 8 minutes; t3 is equal to 3 minutes; t4 is equal to 6 minutes.

[0019] The present invention further includes S7. Further identifying false events by using spectrum analysis. The steps are as follows: Take the ratio of the maximum amplitude in the low - frequency band to the maximum amplitude in the high - frequency band. If the ratio is greater than the threshold n, it can be judged as a true event; if the ratio is less than the threshold n, then a larger threshold is required to make a more stringent identification of the events in step S6.

[0020] The present invention analyzes the data uploaded by the vehicle load sensor through four data queues and three sliding time windows, and combines the spectrum analysis method. First, use a smaller slope threshold to identify all possible loading and unloading event units, then assemble the continuous loading and unloading event units into loading and unloading events, and finally make a true / false judgment on the loading and unloading events to filter out false events caused by interference jitter; this method can widely adapt to various different loading and unloading scenarios, can effectively cope with data jitter and short - term loss, and can more accurately and timely identify the start and end time points of truck loading and unloading. Description of the Drawings

[0021] The invention will be further described in detail below with reference to the accompanying drawings.

[0022] Figure 1 It is a relationship diagram of a data queue and a sliding time window. Specific embodiments

[0023] The following specific embodiments illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0024] During the loading and unloading process, the vehicle load sensor will detect the vehicle's load, speed, and geographical longitude and latitude information in real time, and report it to the background system regularly through the mobile data network. The background system needs to identify the true and accurate start and end time points of the truck's loading and unloading in a timely manner based on the data reported by the load sensor, so that the management personnel can monitor and query on the background interface.

[0025] As Figure 1 shown, a method for automatically determining the accurate start and end times of truck loading and unloading according to the present invention includes the following steps: S1. Construct an original data queue and set an original data sliding time window: S11. Construct an original data queue: Each original data event reported by the vehicle load sensor to the background system constitutes an original data event. All original data events within a fixed duration t (t is equal to 2 hours) are sorted according to the data timestamp to form a rectangular original data queue (the horizontal length of the original data queue is equal to the fixed duration t). The horizontal direction of the original data queue is the complete data composed of all original data events, and the vertical direction is the original data corresponding to each original data event; the original data in each original data event includes a timestamp, a load value, a speed value, and longitude and latitude (that is, each original data queue is obtained by arranging multiple original data events horizontally from front to back in the order of the time timestamp, and each original data event includes a timestamp, a load value, a speed value, and longitude and latitude arranged vertically).

[0026] S12. Set an original data sliding time window: On the original data queue, set an original data sliding time window that can continuously slide backward on the original data queue in the order of the timestamp. The horizontal length of the original data sliding time window is a fixed duration t1 (t1 is equal to 3 minutes), and the vertical width is greater than or equal to the vertical width of the original data queue (to ensure that the window can completely display all the original data of each original data event); the sliding step of the original data sliding time window is T (T is equal to 60 seconds), and T is the time interval for the vehicle load sensor to report data.

[0027] S2. Construct a mean data queue and set a sliding time window for mean data: S21. Construct a mean data queue: Each time the original data sliding time window slides one step, calculate the mean timestamp, mean load, mean speed, and mean longitude and latitude using the original data within the original data sliding time window at this time. The above mean data forms a new mean data event (that is, a new mean data event is formed each time the original data sliding time window slides one step). Sort all mean data events within a fixed duration t in timestamp order to form a (rectangular) mean data queue; horizontally, the mean data queue is the complete data composed of all mean data events, and vertically, it is the mean data corresponding to each mean data event (that is, each mean data queue is obtained by arranging multiple mean data events horizontally from front to back in timestamp order, and each mean data event includes the mean timestamp, mean load, mean speed, and mean longitude and latitude arranged vertically).

[0028] The purpose of the original data sliding time window is mainly to smooth the original data and initially eliminate data jitter.

[0029] S21. Set a sliding time window for mean data: On the mean data queue, set a sliding time window for mean data that can continuously slide backward on the mean data queue in timestamp order. The horizontal length of the mean data sliding time window is a fixed duration t2 (t2 is equal to 6 minutes), and the vertical width is greater than the vertical width of the mean data queue (to ensure that the window can completely display all the original data of each mean data event); the sliding step of the mean data sliding time window is one mean data event; the purpose of the mean data sliding time window is to calculate the change slope of the load curve within a certain time using the mean data. The length of the mean data sliding time window cannot be set too small, otherwise it is easily affected by data jitter and unable to judge the trend change of the slope.

[0030] S3. Construct a slope data queue and set a sliding time window for slope data: S31. Construct a slope data queue: Each time the sliding time window of mean data slides one step, use the mean data within the sliding time window of mean data at this time to perform a linear regression on the load mean. The slope of the regression line is used as the slope of the load curve for this sliding time window of mean data. This slope is equal to the ratio of the change in the load mean of the regression line within a certain time (less than or equal to t2) to the change in the mean of timestamps. At the same time, calculate the quadratic velocity mean for the velocity mean within the sliding time window of mean data at this time, and record the start and end timestamps of this sliding time window of mean data. The above data forms a new slope data event (that is, a new slope data event is formed each time the sliding time window of mean data slides one step). Sort all slope data events within a fixed duration t in chronological order to form a slope data queue; horizontally, the slope data queue is the complete data composed of all slope data events, and vertically, it is the data corresponding to each slope data event (that is, each slope data queue is obtained by arranging multiple slope data events horizontally from front to back in chronological order of timestamps. Each mean data event includes the slope of the load curve, quadratic velocity mean, start and end timestamps arranged vertically).

[0031] S32. Set a sliding time window for slope data: On the slope data queue, set a sliding time window for slope data that can continuously slide backward on the slope data queue in chronological order of timestamps. The horizontal length of the sliding time window for slope data is a fixed duration t3 (t3 is equal to 3 minutes), and the vertical width is greater than the vertical width of the slope data queue (to ensure that the window can completely display all the original data of each slope data event); the sliding step of the sliding time window of mean data is one slope data event.

[0032] S4. Construct an action event queue: Each time the sliding time window of slope data slides one step, use the slope of the load curve within the sliding time window of slope data at this time to calculate the mean slope, and further smooth the mean slope. At the same time, calculate the cubic velocity mean for the quadratic velocity mean within the sliding time window of slope data again, and record the start and end timestamps, start and end load means, and start longitude and latitude of this sliding time window of slope data. Generate a new truck action event using the above data (that is, a new truck action event is formed each time the sliding time window of slope data slides one step). Sort all truck action events within a fixed duration t in chronological order to form an action event queue for the truck; horizontally, the action event queue is the complete data composed of all truck action events, and vertically, it is the data corresponding to each truck action event (that is, each action event queue is obtained by arranging multiple truck action events horizontally from front to back in chronological order of timestamps. Each truck action event includes the mean slope, cubic velocity mean, start and end timestamps, start and end load means, and start longitude and latitude arranged vertically); the truck action events include running events, loading events, unloading events, and stationary events.

[0033] The steps for identifying the truck action events are as follows: S41. If the average speed (referring to the average of three speeds) exceeds the threshold s, it is judged as a running event (indicating that the truck is in the process of traveling and transporting), otherwise, step S42 or step S43 is executed.

[0034] S42. If the slope of the load curve (referring to the average slope of each truck action event) is positive and greater than the threshold k1, it is judged as a loading event (indicating that the truck is in the process of loading); the k1 is equal to 15.

[0035] S43. If the slope of the load curve (referring to the average slope of each truck action event) is negative and less than the threshold k2, it is judged as an unloading event (indicating that the truck is in the process of unloading); the k2 is equal to -15.

[0036] S44. If none of the above situations exist (that is, the average speed is zero and the slope of the load curve is 0), it is judged as a stationary event (indicating that the truck is not performing any actions).

[0037] The threshold k1 in step S42 and the threshold k2 in step S43 above cannot be set too large, otherwise slow loading and unloading events cannot be detected. The specific values need to be determined according to the dimension of the load sensor; the loading and unloading events generated in this step are only a preliminary identified action unit, and the final loading and unloading events need to be further judged in subsequent steps; S5. Processing of the action event queue: The purpose of processing the action event queue is to: merge continuous loading or unloading events (including short pauses), and identify the time point when loading or unloading ends.

[0038] Whenever a new truck action event (hereinafter referred to as a new event, and an old event conversely) is generated, first check the action event queue. If the action event queue is empty and the new event is a loading event or an unloading event, the new event is put into the action event queue; otherwise, if there is already an old event in the action event queue, it is processed according to the following steps: S51. If the action types of the old event and the new event are the same, it means that two adjacent events are in the same continuous loading or unloading process. Only extend the end timestamp of the old event to the end timestamp of the new event, and update the end load value of the old event with the end load value of the new event.

[0039] S52. If the new event is a running event, remove the event in the queue and make a authenticity judgment.

[0040] S53. If the new event is a stationary event and the stationary time exceeds the threshold t4 (t4 is equal to 6 minutes), remove the event in the queue and make a authenticity judgment; this step can effectively handle the situation of short pauses during the loading and unloading process.

[0041] S54. If the action type of the new event is opposite to that of the old event, the event in the queue is removed and a true or false judgment is made, and the new event is put into the queue.

[0042] S6. Authenticity judgment of truck action events: Action events removed from the action event queue may have a certain probability of being false events due to data interference.

[0043] For the truck action event in this step, since it has gone through the merging process in step S5, if it is slow loading and unloading, the slope of the truck action event is smaller but the time must be longer, while fast loading and unloading generally has a larger slope and a shorter time. In addition, whether it is fast loading and unloading or slow loading and unloading, there must be a certain difference in the starting and ending load values. After comprehensive analysis, if the absolute value of the difference between the starting and ending load values ​​of the truck action event is greater than the threshold value m, the product of the duration of the event can be judged as a true event, otherwise it is judged as a false event and ignored; the threshold value m is equal to 9000*60 (i.e. 540000).

[0044] S7. Use spectrum analysis to further identify false events: For some low-precision load-bearing sensors, their data contain large interference components, and the aforementioned method alone may still contain some false events. The original load-bearing data corresponding to the start and end time of the action event can be used for fast Fourier transform to obtain the spectrum data of the event window. The high-frequency components of real events are relatively small, while the high-frequency components of false events are relatively large.

[0045] Take the ratio of the maximum amplitude in the low frequency band (excluding zero frequency) to the maximum amplitude in the high frequency band. If the ratio is greater than the threshold value n, it can be judged as a real event; if the ratio is less than the threshold value n, a larger threshold value is needed to perform a stricter screening of the event in step S6; this step can further improve the accuracy of loading and unloading event recognition. The threshold value n is equal to 2.

[0046] Other aspects of the invention that are not detailed are all conventional techniques known to those skilled in the art.

[0047] It should be noted that the terms "comprises", "includes" or any other variations are intended to cover non-exclusive inclusion, so that a process, method, article or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or apparatus.

[0048] The protection scope of the invention is not limited to the technical solutions disclosed in the specific implementation methods. Any modifications, equivalent substitutions, improvements, etc. made to the above embodiments based on the technical essence of the invention shall fall within the protection scope of the invention.

Claims

1. A method for automatically determining the accurate start and end times of loading and unloading goods of a freight truck, characterized in that, It includes the following steps: S1. Construct an original data queue and set an original data sliding time window: The original data reported by the vehicle load sensor each time forms an original data event. All the original data events within a fixed duration t are sorted according to the data timestamp to form a rectangular original data queue; the original data in each original data event includes a timestamp, a load value, a speed value, and longitude and latitude; Set an original data sliding time window on the original data queue. The horizontal length of the original data sliding time window is a fixed duration t1; the sliding step of the original data sliding time window is T, where T is the time interval for the vehicle load sensor to report data; S2. Construct a mean data queue and set a mean data sliding time window: Every time the original data sliding time window slides one step, use the original data within the original data sliding time window at this time to calculate the mean timestamp, mean load, mean speed, and mean longitude and latitude. The above mean data forms a new mean data event. All the mean data events within a fixed duration t are sorted in timestamp order to form a mean data queue; Set a mean data sliding time window on the mean data queue. The horizontal length of the mean data sliding time window is a fixed duration t2; the sliding step of the mean data sliding time window is one mean data event; S3. Construct a slope data queue and set a slope data sliding time window: Every time the mean data sliding time window slides one step, use the mean data within the mean data sliding time window at this time to perform a linear regression on the mean load. The slope of the regression line is used as the slope of the load curve of this mean data sliding time window. At the same time, calculate the quadratic speed mean of the mean speed within the mean data sliding time window at this time, and record the start and end timestamps of this mean data sliding time window. The above data forms a new slope data event. All the slope data events within a fixed duration t are sorted in time order to form a slope data queue; Set a slope data sliding time window on the slope data queue. The horizontal length of the slope data sliding time window is a fixed duration t3; the sliding step of the mean data sliding time window is one slope data event; S4. Construct an action event queue: Every time the slope data sliding time window slides one step, use the slope of the load curve within the slope data sliding time window at this time to calculate the mean slope, and further smooth the mean slope. At the same time, calculate the cubic speed mean of the quadratic speed mean within the slope data sliding time window again, and record the start and end timestamps, start and end mean loads, and start longitude and latitude of this slope data sliding time window. Use the above data to generate a new truck action event. All the truck action events within a fixed duration t are sorted in timestamp order to form the action event queue of the truck; the types of the truck action events include running events, loading events, unloading events, and stationary events; S5. Processing of the action event queue: Whenever a new truck action event occurs, first check the action event queue. If the action event queue is empty and the new event is a loading event or an unloading event, then put the new event into the action event queue; otherwise, if there is an old event in the action event queue, handle it according to the following steps: S51. If the action types of the old event and the new event are the same, it means that two adjacent events are in the same continuous loading or unloading process. Only extend the end timestamp of the old event to the end timestamp of the new event, and update the end load value of the old event with the end load value of the new event; S52. If the new event is a running event, remove the event from the queue and make a authenticity judgment; S53. If the new event is a stationary event and the stationary time exceeds the threshold t4, then remove the event from the queue and make a authenticity judgment; S54. If the action type of the new event is opposite to that of the old event, then remove the event from the queue and make a authenticity judgment, and put the new event into the queue; S6. Authenticity judgment of truck action events; Make an authenticity judgment on the truck action events removed from the action event queue in steps S52 - S54.

2. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 1, wherein : The t is equal to 0.5 hours to 8 hours; T is equal to 30 seconds to 60 seconds; t1 is equal to 3 minutes; t2 is equal to 5 - 8 minutes; t3 is equal to 3 minutes; t4 is equal to 6 minutes.

3. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 1, characterized in that : In step S4, the recognition steps of the truck action event are as follows: S41. If the average speed exceeds the threshold s, it is judged as a running event, otherwise execute step S42 or step S43; S42. If the slope of the load curve is positive and greater than the threshold k1, then it is judged as a loading event; S43. If the slope of the load curve is negative and less than the threshold k2, then it is judged as an unloading event; S44. If none of the above situations exist, then it is judged as a stationary event.

4. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 3, wherein : The s is equal to 5 km / h; the k1 is equal to 15; the k2 is equal to -15.

5. The method for automatically determining the accurate start and end times of truck loading and unloading according to any one of claims 1-4, characterized in that : In step S6, the authenticity judgment steps of the truck action event are: The product of the absolute value of the difference between the start and end load values of the truck action event and the event duration. If it is greater than the threshold m, it can be judged as a true event, otherwise it is judged as a false event and ignored.

6. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 5, wherein : The threshold m is equal to 9000 * 60.

7. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 6, characterized in that It also includes S7. Further distinguish false events using spectrum analysis, and the steps are: Take the ratio of the maximum amplitude in the low frequency band to the maximum amplitude in the high frequency band. If the ratio is greater than the threshold n, it can be judged as a true event; if the ratio is less than the threshold n, then a larger threshold is needed to make a more strict distinction for the event in step S6.

8. The method for automatically determining the accurate start and end times of truck loading and unloading according to claim 7, characterized in that: The threshold n is equal to 2.

Images