Method, system, medium and electronic device for determining a cargo switching behavior

Abstract

The application discloses a stolen and replaced goods behavior judgment method, system, medium and electronic equipment. The method comprises the following steps: receiving the track point information reported by the vehicle terminal of each truck; judging whether the parking behavior or the deviation behavior of each truck occurs according to the track point information; if so, determining a plurality of characteristic values related to the parking behavior or the deviation behavior; inputting the plurality of characteristic values into a pre-trained stolen and replaced goods behavior judgment model; the pre-trained stolen and replaced goods behavior judgment model is generated based on a model training sample and a neural network model, and the model training sample is simulated by a sample generator based on a Monte Carlo simulation algorithm; outputting the stolen and replaced goods risk coefficient corresponding to each truck; and judging whether each truck has the stolen and replaced goods behavior based on the stolen and replaced goods risk coefficient. Therefore, by adopting the embodiment of the application, the hardware cost can be reduced, all vehicles can be comprehensively monitored to determine whether there is a stolen and replaced goods risk, and the comprehensiveness of the stolen and replaced goods behavior judgment is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, system, medium and electronic device for determining the act of swapping goods. Background Technology

[0002] With the rapid development of the logistics industry, more and more management problems have emerged. For example, during transportation, there have been incidents of high-value goods being swapped during full-truckload shipments, which can cause significant losses to businesses. Therefore, ensuring the safety of goods during logistics transportation and protecting the interests of enterprises is of paramount importance.

[0003] In related technologies, surveillance videos are installed on trucks for manual or automated analysis based on video monitoring. However, this method requires the installation of camera and video recording hardware on each truck, which is costly and has a small coverage area, thus reducing the comprehensiveness of determining the theft of goods. Summary of the Invention

[0004] This application provides a method, system, medium, and electronic device for determining the act of swapping goods. To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its sole purpose is to present some concepts in a simple form as a prelude to the detailed description that follows.

[0005] In a first aspect, embodiments of this application provide a method for determining the act of swapping goods, the method comprising:

[0006] Receive trajectory point information reported by the on-board terminal of each truck in real time;

[0007] Based on the trajectory point information, determine whether each truck has stopped or deviated from its course;

[0008] In each case of a truck stopping or veerging behavior, multiple feature values ​​associated with the stopping or veerging behavior are determined;

[0009] Multiple feature values ​​are input into a pre-trained model for detecting theft and substitution of goods. The pre-trained model for detecting theft and substitution of goods is generated based on model training samples and a neural network model. The model training samples are simulated by a sample generator built based on a Monte Carlo simulation algorithm.

[0010] Output the risk coefficient for cargo theft / swapping for each truck;

[0011] Based on the risk coefficient of cargo theft and substitution, it is determined whether each truck has engaged in cargo theft and substitution.

[0012] Optionally, a pre-trained model for detecting theft or substitution of goods is generated by following these steps:

[0013] Based on the historical trajectory information of the target vehicle, create multiple types of parking features and multiple types of non-parking features;

[0014] For each type of docking feature and non-docking feature, multiple sub-class features are set;

[0015] Obtain a pre-set trigger probability and a risk coefficient threshold for each sub-category feature when the sub-category feature is triggered;

[0016] Based on each subclass feature and the corresponding trigger probability and risk coefficient threshold, a sample generator based on the Monte Carlo simulation algorithm is used to generate model training samples.

[0017] Create a neural network model;

[0018] The neural network model is trained based on the model training samples to generate a pre-trained model for judging the act of switching goods.

[0019] Optional features include various stop characteristics such as stop road type, stop distance, stop route percentage, stop start time, and stop duration; and various non-stop characteristics such as yaw distance, vehicle delay status, cargo value, and number of times the driver cooperates.

[0020] For each type of docking feature and non-docking feature, multiple sub-class features are set, including:

[0021] The types of roads for stopping are divided into three sub-types: high-grade roads, medium-grade roads, and low-grade roads, resulting in the sub-class characteristics of road types for stopping.

[0022] The docking distance is divided into three sub-types: long distance, medium distance, and short distance, resulting in sub-type features of docking distance;

[0023] The percentage of stops is divided into three sub-types: two ends, slightly two ends, and middle, resulting in sub-class features of the percentage of stops.

[0024] The docking start time period is divided into three sub-types: daytime, early morning and evening, and late night, to obtain the sub-type features of the docking start time period;

[0025] The docking duration is divided into three sub-types: very short, relatively short, and long, to obtain the sub-class features of docking duration.

[0026] The yaw distance is divided into 5 different subclasses according to multiple preset distance ranges to obtain the subclass features of yaw distance;

[0027] The vehicle's on-time status is divided into two sub-types: on-time and not on-time, thus obtaining the sub-class features of whether a vehicle is on-time.

[0028] The value of goods is divided into three subcategories: high, medium, and low, resulting in the sub-category characteristics of the value of goods.

[0029] The number of times drivers cooperated was divided into three sub-types: no cooperation, cooperation, and multiple cooperations, thus obtaining the sub-class features of the number of times drivers cooperated.

[0030] Optionally, model training samples are generated based on each subclass feature and the corresponding trigger probability and risk coefficient threshold, combined with a sample generator established based on the Monte Carlo simulation algorithm, including:

[0031] A sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate and trigger multiple types of docking features and multiple types of non-docking features in order to obtain the target decimal corresponding to the triggered features.

[0032] The subclass features to be triggered are determined based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each subclass feature.

[0033] A sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate the triggering of the subclass features to be triggered, thereby determining the triggered subclass features.

[0034] From the risk coefficient thresholds corresponding to each sub-category feature, determine the target risk coefficient threshold corresponding to the triggered sub-category feature;

[0035] Based on the target risk coefficient threshold corresponding to the triggered sub-class feature, the risk coefficient corresponding to the triggered sub-class feature is randomly generated;

[0036] The risk coefficients corresponding to the triggered sub-category features are summed to obtain the risk coefficient for theft and replacement of goods.

[0037] Based on the risk coefficient of theft and replacement of goods and the characteristics of multiple subclasses, model training samples are generated.

[0038] Optionally, the sub-class features to be triggered are determined based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each sub-class feature, including:

[0039] From each subclass feature, identify multiple first subclass features that belong to the triggered feature;

[0040] From the trigger probabilities corresponding to each subclass feature, find the trigger probability corresponding to each first subclass feature;

[0041] From the trigger probabilities corresponding to each first subclass feature, determine multiple second subclass features whose trigger probabilities correspond to the target decimals of the triggered features;

[0042] Multiple second-subclass features are identified as subclass features to be triggered.

[0043] Optionally, model training samples are generated based on the risk coefficient of theft and substitution and multiple subclass features, including:

[0044] Multiple subclass features are identified as model features;

[0045] The risk factor for theft and substitution of goods is set as the target value;

[0046] By associating the model features with the target value, a training sample is obtained;

[0047] Continue executing the steps of creating multiple types of parking features and multiple types of non-parking features based on the historical trajectory point information of the target vehicle until the number of training samples reaches the preset number, and then use all training samples as model training samples.

[0048] Optionally, based on the risk coefficient of cargo theft, determine whether each truck has engaged in cargo theft, including:

[0049] When the risk coefficient for goods swapping exceeds a preset risk coefficient threshold, each truck is determined to have engaged in goods swapping; or...

[0050] When the risk coefficient for theft or substitution is less than or equal to the preset risk coefficient threshold, it is determined that there is no theft or substitution of goods in each truck.

[0051] Secondly, embodiments of this application provide a system for determining the act of swapping goods, the system comprising:

[0052] The trajectory point information receiving module is used to receive trajectory point information reported by the on-board terminal of each truck in real time.

[0053] The behavior analysis module is used to determine whether each truck has stopped or deviated from its course based on trajectory point information.

[0054] The feature value determination module is used to determine multiple feature values ​​related to each truck stopping or veerging behavior.

[0055] The feature value input module is used to input multiple feature values ​​into the pre-trained theft and substitution behavior detection model; the pre-trained theft and substitution behavior detection model is generated based on the model training samples and the neural network model, and the model training samples are simulated by a sample generator built based on the Monte Carlo simulation algorithm;

[0056] The theft and substitution risk coefficient output module is used to output the theft and substitution risk coefficient for each truck.

[0057] The cargo swapping behavior detection module is used to determine whether each truck has engaged in cargo swapping behavior based on the cargo swapping risk coefficient.

[0058] Thirdly, embodiments of this application provide a computer storage medium storing multiple instructions adapted for loading and execution of the above-described method steps by a processor.

[0059] Fourthly, embodiments of this application provide an electronic device that may include: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed by the above-described method steps.

[0060] The technical solutions provided in this application embodiment may include the following beneficial effects:

[0061] In this embodiment, a pre-trained model for judging the behavior of swapping goods is generated based on model training samples and neural network models. This model can perform real-time evaluation of transport vehicles with potential risks of swapping goods during the transport journey, and can promptly determine whether the truck has engaged in the behavior of swapping goods. The model training samples are simulated by a sample generator based on the Monte Carlo simulation algorithm. The judgment of the behavior of swapping goods can be realized through software, which is low-cost and has a wide coverage, thereby improving the comprehensiveness of the judgment of the behavior of swapping goods.

[0062] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0063] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0064] Figure 1 This is a flowchart illustrating a method for determining the act of swapping goods provided in an embodiment of this application;

[0065] Figure 2 This is a schematic block diagram illustrating a model training process provided in an embodiment of this application;

[0066] Figure 3 This is a schematic flowchart illustrating a process for determining the act of swapping goods, as provided in an embodiment of this application.

[0067] Figure 4 This is a schematic diagram of the structure of a product substitution detection system provided in an embodiment of this application;

[0068] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0069] The following description and accompanying drawings fully illustrate specific embodiments of this application to enable those skilled in the art to practice them.

[0070] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0071] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of systems and methods consistent with some aspects of this application as detailed in the appended claims.

[0072] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.

[0073] This application provides a method, system, medium, and electronic device for determining the act of swapping goods, to solve the problems existing in the aforementioned related technologies. The technical solution provided in this application generates a pre-trained model for determining the act of swapping goods based on model training samples and a neural network model. This model can perform real-time assessments of transport vehicles with potential risks of swapping goods during their transport journey, and can promptly determine whether a truck has engaged in such behavior. The model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm, enabling the determination of the act of swapping goods through software. This method is low-cost, has a wide coverage, and thus improves the comprehensiveness of the determination of the act of swapping goods. Exemplary embodiments are described in detail below.

[0074] The following will be combined with the appendix Figure 1 - Appendix Figure 3 This application provides a detailed description of the method for determining theft and substitution of goods provided in its embodiments. This method can be implemented using a computer program and can run on a von Neumann architecture-based system for determining theft and substitution of goods. This computer program can be integrated into an application or run as a standalone utility application.

[0075] Please see Figure 1 This is a flowchart illustrating a method for determining the act of swapping goods, as provided in an embodiment of this application. Figure 1 As shown, the method in this application embodiment may include the following steps:

[0076] S101 receives real-time trajectory point information reported by the on-board terminal of each truck.

[0077] Each truck is a transport vehicle on its journey, and the on-board terminal is a positioning device installed on the vehicle, which can be used to generate real-time trajectory point information of the truck.

[0078] In this embodiment of the application, the on-board terminal of each truck obtains the truck's location information in real time. Based on the obtained location information, the vehicle terminal can generate trajectory point information for each truck. The on-board terminal can report the trajectory point information of each truck to the cargo swapping behavior determination system. The system receives the trajectory point information reported by the on-board terminal of each truck in real time.

[0079] S102, Based on the trajectory point information, determine whether each truck has stopped or deviated from its course;

[0080] In this embodiment of the application, after obtaining the trajectory point information, the system can determine whether the trajectory point information has changed within a preset period. If there is a change, it means that the vehicle has not stopped. If there is no change, it means that the vehicle has stopped. At the same time, it can determine whether the distance between the trajectory point information and the preset transportation route is greater than a preset value. If it is greater than the preset value, it means that the truck has deviated from its course.

[0081] S103, in the case of each truck stopping or veerging behavior, determine multiple feature values ​​related to the stopping or veerging behavior;

[0082] In this embodiment, when a truck stops or veers off course, a pre-defined set of feature values ​​related to that stop or veer off course can be identified. The pre-defined feature values ​​include multiple types of stopping features and multiple types of non-stopping features. The stopping features include stopping road type, stopping distance, stopping route percentage, stopping start time, and stopping duration. The non-stopping features include veer-off distance, whether the vehicle is late, cargo value, and number of times the driver cooperates. The stopping road type includes feature values ​​for three subtypes: high-grade roads, medium-grade roads, and low-grade roads. The stopping distance includes long distance... The system includes 28 pre-defined feature values: distance, medium distance, and short distance; stop route percentage (including two ends, slightly towards the ends, and midway); stop start time (daytime, early morning and evening, and late night); stop duration (very short, relatively short, and long); yaw distance (five different subcategories based on multiple preset distance ranges); vehicle delay status (delayed and not delayed); cargo value (high, medium, and low); and driver cooperation frequency (no cooperation, cooperated before, and cooperated multiple times).

[0083] S104, input multiple feature values ​​into the pre-trained theft and substitution behavior detection model; the pre-trained theft and substitution behavior detection model is generated based on the model training samples and the neural network model, and the model training samples are simulated based on the sample generator established by the Monte Carlo simulation algorithm;

[0084] Among them, the pre-trained model for judging the theft and substitution of goods is a mathematical model that outputs the risk coefficient of theft and substitution for each truck.

[0085] In this embodiment, a pre-trained model for determining theft of goods can be generated by following these steps: First, based on the historical trajectory information of the target vehicle, multiple types of stopping features and multiple types of non-stopping features are created; then, multiple sub-features are set for each type of stopping feature and non-stopping feature; next, a trigger probability and a risk coefficient threshold generated when triggering a sub-feature are obtained for each sub-feature; then, model training samples are generated by combining a sample generator based on a Monte Carlo simulation algorithm with each sub-feature and the corresponding trigger probability and risk coefficient threshold; next, a neural network model is created; finally, the neural network model is trained based on the model training samples to generate a pre-trained model for determining theft of goods.

[0086] Specifically, the various stop characteristics include the type of road where the vehicle stops, the distance between the stops, the percentage of the route taken where the vehicle stops, the start time of the stop, and the duration of the stop; the various non-stop characteristics include the deviation distance, whether the vehicle is late, the value of the goods, and the number of times the driver has cooperated.

[0087] In this embodiment, when setting multiple sub-categories for each type of parking feature and non-parking feature, the parking road type is divided into three sub-categories: high-level road, medium-level road, and low-level road, resulting in sub-categories for parking road type. The parking distance is then divided into three sub-categories: long distance, medium distance, and short distance, resulting in sub-categories for parking distance. The parking route percentage is further divided into three sub-categories: two-end, slightly two-end, and mid-range, resulting in sub-categories for parking route percentage. Finally, the parking start time is divided into three sub-categories: daytime, early morning and evening, and late night, resulting in sub-categories for parking start time. The system first identifies the sub-categories of stop times; then, it divides stop duration into three sub-types: very short, relatively short, and long, to obtain sub-categories of stop duration; next, it divides yaw distance into five different sub-categories according to multiple preset distance ranges, to obtain sub-categories of yaw distance; it also divides vehicle delay status into two sub-types: delayed and not delayed, to obtain sub-categories of vehicle delay status; then, it divides cargo value into three sub-categories: high, medium, and low, to obtain sub-categories of cargo value; finally, it divides driver cooperation frequency into three sub-types: no cooperation, cooperation, and multiple cooperations, to obtain sub-categories of driver cooperation frequency.

[0088] In this embodiment, when generating model training samples based on each sub-class feature and the corresponding trigger probability and risk coefficient threshold, combined with a sample generator based on the Monte Carlo simulation algorithm, the process is as follows: First, the sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate triggering multiple types of docking features and multiple types of non-docking features to obtain the target decimal corresponding to the triggered feature. Then, based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each sub-class feature, the sub-class feature to be triggered is determined. Next, the sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate triggering the sub-class feature to be triggered to determine the triggered sub-class feature. Then, from the risk coefficient threshold corresponding to each sub-class feature, the target risk coefficient threshold corresponding to the triggered sub-class feature is determined. Next, based on the target risk coefficient threshold corresponding to the triggered sub-class feature, the risk coefficient corresponding to the triggered sub-class feature is randomly generated. Finally, the risk coefficients corresponding to the triggered sub-class features are summed to obtain the theft and replacement risk coefficient. And based on the theft and replacement risk coefficient and multiple sub-class features, model training samples are generated.

[0089] For example, multiple types of docking features and multiple types of non-docking features are triggered by random probability. A trigger is activated if the value is greater than 0.5, and not activated if the value is less than 0.5. Subclasses are determined to identify which subclass is triggered according to the trigger probability under the condition that the multiple types of docking features and multiple types of non-docking features are triggered; when performing random simulation triggering on the multiple types of docking features and multiple types of non-docking features, assuming that the trigger probabilities of the three subclasses (long distance, medium distance, short distance) of the large docking distance category are 0.2, 0.3, and 0.5 respectively (the sum of all probabilities must be 1), then when this large category is randomly triggered, a decimal between 0 and 1 will be randomly generated. If the value x is less than 0.2, the long-distance sub-feature is triggered; if 0.2 < x < 0.5, the medium-distance sub-feature is triggered; if x > 0.5, the short-distance sub-feature is triggered. For example, for the delay type, the trigger probability of no delay is set to 0.9, and the risk coefficient threshold is 0; the trigger probability of delay is 0.1, and the risk threshold is 0.05-0.15.

[0090] In the embodiment of the present application, when determining the subclass features to be triggered according to the target decimal corresponding to the triggered feature and the trigger probability corresponding to each subclass feature, firstly, a plurality of first subclass features belonging to the triggered feature are determined from each subclass feature; then, the trigger probability corresponding to each first subclass feature is found from the trigger probabilities corresponding to each subclass feature; next, a plurality of second subclass features whose trigger probabilities correspond to the target decimal corresponding to the triggered feature are determined from the trigger probabilities corresponding to each first subclass feature; finally, the plurality of second subclass features are determined as the subclass features to be triggered.

[0091] In the embodiment of the present application, when generating model training samples according to the goods theft and replacement risk coefficient and a plurality of subclass features, firstly, the plurality of subclass features are determined as model features; then, the goods theft and replacement risk coefficient is determined as a target value; next, the model features and the target value are associated to obtain a training sample; finally, the step of creating multiple types of docking features and multiple types of non-docking features according to the historical track point information of the target vehicle is continuously executed until the number of training samples reaches the preset number, and all training samples are taken as model training samples.

[0092] In the embodiment of the present application, when training a neural network model according to the model training samples to generate a pre-trained goods theft and replacement behavior determination model, firstly, the model training samples are input into the neural network model, and the loss value of the model is output. When the loss value reaches the minimum, the pre-trained goods theft and replacement behavior determination model is obtained. When the loss value does not reach the minimum, the model training samples are continuously input into the neural network model until the loss value reaches the minimum.

[0093] For example Figure 2 as shown, Figure 2This is a schematic flowchart of a model training process provided in this application. First, five types of docking features and four types of non-docking features are generated. Then, multiple sub-features are set for each type of docking and non-docking feature. Next, a trigger probability and a risk coefficient threshold are set for each sub-feature. Then, a sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate triggering multiple types of docking and non-docking features. Next, for the features that appear after triggering, the corresponding sub-feature is triggered according to its trigger probability. Then, a risk coefficient is randomly generated according to the risk threshold of the sub-feature. The risk coefficients of all triggered sub-features are added together as the risk coefficient of the sample. If the coefficient is greater than 1, it is reset to 1. The sub-features are used as model features, and the risk coefficient is used as the target value. A large number of samples can be generated through this sample generator. Finally, a neural network model is designed, and the generated samples are used to train the model to obtain a pre-trained model for judging the theft of goods.

[0094] S105 outputs the risk coefficient for cargo theft / swapping for each truck.

[0095] In this embodiment of the application, after processing the feature values ​​of the input trucks, the model can output the risk coefficient of cargo theft and substitution for each truck.

[0096] S106, based on the risk coefficient of cargo theft and replacement, determines whether each truck has engaged in cargo theft and replacement.

[0097] In this embodiment of the application, when determining whether each truck has engaged in theft or substitution of goods based on the risk coefficient of theft or substitution, if the risk coefficient of theft or substitution is greater than a preset risk coefficient threshold, it is determined that each truck has engaged in theft or substitution of goods; or, if the risk coefficient of theft or substitution is less than or equal to the preset risk coefficient threshold, it is determined that each truck has not engaged in theft or substitution of goods. Other interval judgment methods can also be used for judgment, which will not be elaborated here.

[0098] Furthermore, once a truck is identified as having engaged in cargo swapping, alerts can be sent in real time to the cargo owners of high-risk vehicles, urging them to take proactive measures to prevent losses.

[0099] For example Figure 3 As shown, the system first receives trajectory points in real time, and then determines whether a stop or deviation has occurred based on the trajectory points. If so, it determines the feature values ​​related to the stop or deviation. Then, it inputs the determined feature values ​​related to the stop or deviation into a pre-trained model for judging the behavior of switching goods, and calculates the risk coefficient.

[0100] In this embodiment, the system first receives trajectory point information reported by the onboard terminal of each truck; then, based on the trajectory point information, it determines whether each truck has engaged in stopping or veergency behavior; if so, it identifies multiple feature values ​​related to the stopping or veergency behavior; next, it inputs these feature values ​​into a pre-trained cargo swapping behavior determination model; the pre-trained cargo swapping behavior determination model is generated based on model training samples and a neural network model, and the model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm; then, it outputs a cargo swapping risk coefficient corresponding to each truck; finally, based on the cargo swapping risk coefficient, it determines whether each truck has engaged in cargo swapping behavior. Since this application generates a pre-trained cargo swapping behavior determination model based on model training samples and a neural network model, it can perform real-time evaluation of transport vehicles with potential cargo swapping risks during the transport journey, and can promptly determine whether a truck has engaged in cargo swapping behavior. The model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm, and cargo swapping behavior determination can be achieved through software, which is low-cost, has a wide coverage, and thus improves the comprehensiveness of cargo swapping behavior determination.

[0101] The following are system embodiments of this application, which can be used to execute the method embodiments of this application. For details not disclosed in the system embodiments of this application, please refer to the method embodiments of this application.

[0102] Please see Figure 4 This illustration shows a schematic diagram of a system for determining theft of goods provided in an exemplary embodiment of this application. This system can be implemented as all or part of an electronic device through software, hardware, or a combination of both. The system 1 includes a trajectory point information receiving module 10, a behavior analysis module 20, a feature value determination module 30, a feature value input module 40, a theft risk coefficient output module 50, and a theft behavior judgment module 60.

[0103] The trajectory point information receiving module 10 is used to receive the trajectory point information reported by the on-board terminal of each truck in real time.

[0104] The behavior analysis module 20 is used to determine whether each truck has stopped or deviated from its course based on the trajectory point information.

[0105] The feature value determination module 30 is used to determine multiple feature values ​​related to the stopping or yaw behavior when each truck stops or yaws.

[0106] The feature value input module 40 is used to input multiple feature values ​​into the pre-trained theft and substitution behavior judgment model; the pre-trained theft and substitution behavior judgment model is generated based on the model training samples and the neural network model, and the model training samples are simulated by a sample generator established based on the Monte Carlo simulation algorithm;

[0107] The theft and substitution risk coefficient output module 50 is used to output the theft and substitution risk coefficient for each truck.

[0108] The "Goods Swapping Behavior Judgment Module 60" is used to determine whether each truck has engaged in goods swapping behavior based on the goods swapping risk coefficient.

[0109] It should be noted that the above-described embodiment of the product substitution behavior determination system is only illustrated by the division of the functional modules described above when executing the product substitution behavior determination method. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the product substitution behavior determination system and the product substitution behavior determination method embodiment provided above belong to the same concept, and their implementation process is detailed in the method embodiment, which will not be repeated here.

[0110] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0111] In this embodiment, the system first receives trajectory point information reported by the onboard terminal of each truck; then, based on the trajectory point information, it determines whether each truck has engaged in stopping or veergency behavior; if so, it identifies multiple feature values ​​related to the stopping or veergency behavior; next, it inputs these feature values ​​into a pre-trained cargo swapping behavior determination model; the pre-trained cargo swapping behavior determination model is generated based on model training samples and a neural network model, and the model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm; then, it outputs a cargo swapping risk coefficient corresponding to each truck; finally, based on the cargo swapping risk coefficient, it determines whether each truck has engaged in cargo swapping behavior. Since this application generates a pre-trained cargo swapping behavior determination model based on model training samples and a neural network model, it can perform real-time evaluation of transport vehicles with potential cargo swapping risks during the transport journey, and can promptly determine whether a truck has engaged in cargo swapping behavior. The model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm, and cargo swapping behavior determination can be achieved through software, which is low-cost, has a wide coverage, and thus improves the comprehensiveness of cargo swapping behavior determination.

[0112] This application also provides a computer-readable medium having program instructions stored thereon, which, when executed by a processor, implement the methods for determining the act of swapping goods provided in the above-described method embodiments.

[0113] This application also provides a computer program product containing instructions that, when run on a computer, causes the computer to execute the methods for determining the act of swapping goods described in the above-described method embodiments.

[0114] Please see Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 1000 may include: at least one processor 1001, at least one network interface 1004, a user interface 1003, a memory 1005, and at least one communication bus 1002.

[0115] The communication bus 1002 is used to realize the connection and communication between these components.

[0116] The user interface 1003 may include a display screen and a camera. Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0117] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0118] The processor 1001 may include one or more processing cores. The processor 1001 connects to various parts within the electronic device 1000 using various interfaces and lines. It executes various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 1005, and by calling data stored in the memory 1005. Optionally, the processor 1001 may be implemented using at least one hardware form selected from Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 1001 may integrate one or more of the following: a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip, without being integrated into the processor 1001.

[0119] The memory 1005 may include random access memory (RAM) or read-only memory. Optionally, the memory 1005 may include a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 1005 may also be at least one storage system located remotely from the aforementioned processor 1001. Figure 5 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application for determining the act of stealing or swapping goods.

[0120] exist Figure 5In the illustrated electronic device 1000, the user interface 1003 is mainly used to provide an input interface for the user and to obtain the user's input data; while the processor 1001 can be used to call the product swapping behavior determination application stored in the memory 1005, and specifically perform the following operations:

[0121] Receive trajectory point information reported by the on-board terminal of each truck in real time;

[0122] Based on the trajectory point information, determine whether each truck has stopped or deviated from its course;

[0123] In each case of a truck stopping or veerging behavior, multiple feature values ​​associated with the stopping or veerging behavior are determined;

[0124] Multiple feature values ​​are input into a pre-trained model for detecting theft and substitution of goods. The pre-trained model for detecting theft and substitution of goods is generated based on model training samples and a neural network model. The model training samples are simulated by a sample generator built based on a Monte Carlo simulation algorithm.

[0125] Output the risk coefficient for cargo theft / swapping for each truck;

[0126] Based on the risk coefficient of cargo theft and substitution, it is determined whether each truck has engaged in cargo theft and substitution.

[0127] In one embodiment, when the processor 1001 executes the pre-trained model for determining the act of switching goods, it specifically performs the following operations:

[0128] Based on the historical trajectory information of the target vehicle, create multiple types of parking features and multiple types of non-parking features;

[0129] For each type of docking feature and non-docking feature, multiple sub-class features are set;

[0130] Obtain a pre-set trigger probability and a risk coefficient threshold for each sub-category feature when the sub-category feature is triggered;

[0131] Based on each subclass feature and the corresponding trigger probability and risk coefficient threshold, a sample generator based on the Monte Carlo simulation algorithm is used to generate model training samples.

[0132] Create a neural network model;

[0133] The neural network model is trained based on the model training samples to generate a pre-trained model for judging the act of switching goods.

[0134] In one embodiment, when the processor 1001 sets multiple sub-class features for each type of docking feature and non-docking feature, it specifically performs the following operations:

[0135] The types of roads for stopping are divided into three sub-types: high-grade roads, medium-grade roads, and low-grade roads, resulting in the sub-class characteristics of road types for stopping.

[0136] The docking distance is divided into three sub-types: long distance, medium distance, and short distance, resulting in sub-type features of docking distance;

[0137] The percentage of stops is divided into three sub-types: two ends, slightly two ends, and middle, resulting in sub-class features of the percentage of stops.

[0138] The docking start time period is divided into three sub-types: daytime, early morning and evening, and late night, to obtain the sub-type features of the docking start time period;

[0139] The docking duration is divided into three sub-types: very short, relatively short, and long, to obtain the sub-class features of docking duration.

[0140] The yaw distance is divided into 5 different subclasses according to multiple preset distance ranges to obtain the subclass features of yaw distance;

[0141] The vehicle's on-time status is divided into two sub-types: on-time and not on-time, thus obtaining the sub-class features of whether a vehicle is on-time.

[0142] The value of goods is divided into three subcategories: high, medium, and low, resulting in the sub-category characteristics of the value of goods.

[0143] The number of times drivers cooperated was divided into three sub-types: no cooperation, cooperation, and multiple cooperations, thus obtaining the sub-class features of the number of times drivers cooperated.

[0144] In one embodiment, when the processor 1001 generates model training samples based on each subclass feature and the corresponding trigger probability and risk coefficient threshold, combined with a sample generator established based on the Monte Carlo simulation algorithm, it specifically performs the following operations:

[0145] A sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate and trigger multiple types of docking features and multiple types of non-docking features in order to obtain the target decimal corresponding to the triggered features.

[0146] The subclass features to be triggered are determined based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each subclass feature.

[0147] A sample generator based on the Monte Carlo simulation algorithm is used to randomly simulate the triggering of the subclass features to be triggered, thereby determining the triggered subclass features.

[0148] From the risk coefficient thresholds corresponding to each sub-category feature, determine the target risk coefficient threshold corresponding to the triggered sub-category feature;

[0149] Based on the target risk coefficient threshold corresponding to the triggered sub-class feature, the risk coefficient corresponding to the triggered sub-class feature is randomly generated;

[0150] The risk coefficients corresponding to the triggered sub-category features are summed to obtain the risk coefficient for theft and replacement of goods.

[0151] Based on the risk coefficient of theft and replacement of goods and the characteristics of multiple subclasses, model training samples are generated.

[0152] In one embodiment, when the processor 1001 determines the sub-feature to be triggered based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each sub-feature, it specifically performs the following operations:

[0153] From each subclass feature, identify multiple first subclass features that belong to the triggered feature;

[0154] From the trigger probabilities corresponding to each subclass feature, find the trigger probability corresponding to each first subclass feature;

[0155] From the trigger probabilities corresponding to each first subclass feature, determine multiple second subclass features whose trigger probabilities correspond to the target decimals of the triggered features;

[0156] Multiple second-subclass features are identified as subclass features to be triggered.

[0157] In one embodiment, when processor 1001 generates model training samples based on the risk coefficient of theft and substitution and multiple subclass features, it specifically performs the following operations:

[0158] Multiple subclass features are identified as model features;

[0159] The risk factor for theft and substitution of goods is set as the target value;

[0160] By associating the model features with the target value, a training sample is obtained;

[0161] Continue executing the steps of creating multiple types of parking features and multiple types of non-parking features based on the historical trajectory point information of the target vehicle until the number of training samples reaches the preset number, and then use all training samples as model training samples.

[0162] In one embodiment, when processor 1001 determines whether each truck has engaged in cargo swapping based on a cargo swapping risk coefficient, it specifically performs the following operations:

[0163] When the risk coefficient for goods swapping exceeds a preset risk coefficient threshold, each truck is determined to have engaged in goods swapping; or...

[0164] When the risk coefficient for theft or substitution is less than or equal to the preset risk coefficient threshold, it is determined that there is no theft or substitution of goods in each truck.

[0165] In this embodiment, the system first receives trajectory point information reported by the onboard terminal of each truck; then, based on the trajectory point information, it determines whether each truck has engaged in stopping or veergency behavior; if so, it identifies multiple feature values ​​related to the stopping or veergency behavior; next, it inputs these feature values ​​into a pre-trained cargo swapping behavior determination model; the pre-trained cargo swapping behavior determination model is generated based on model training samples and a neural network model, and the model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm; then, it outputs a cargo swapping risk coefficient corresponding to each truck; finally, based on the cargo swapping risk coefficient, it determines whether each truck has engaged in cargo swapping behavior. Since this application generates a pre-trained cargo swapping behavior determination model based on model training samples and a neural network model, it can perform real-time evaluation of transport vehicles with potential cargo swapping risks during the transport journey, and can promptly determine whether a truck has engaged in cargo swapping behavior. The model training samples are simulated using a sample generator based on a Monte Carlo simulation algorithm, and cargo swapping behavior determination can be achieved through software, which is low-cost, has a wide coverage, and thus improves the comprehensiveness of cargo swapping behavior determination.

[0166] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program for determining the act of swapping goods can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the methods described above. The storage medium for the program for determining the act of swapping goods can be a magnetic disk, an optical disk, a read-only memory, or a random access memory, etc.

[0167] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A method for determining the act of swapping goods, characterized in that, The method includes: Receive trajectory point information reported by the on-board terminal of each truck in real time; Based on the trajectory point information, it is determined whether each truck has stopped or deviated from its course; In the event that each truck engages in a stopping or veer behavior, multiple feature values ​​associated with the stopping or veer behavior are determined. The multiple feature values ​​are input into a pre-trained model for detecting theft and substitution of goods; the pre-trained model for detecting theft and substitution of goods is generated based on model training samples and a neural network model, and the model training samples are simulated by a sample generator based on a Monte Carlo simulation algorithm; Generate a pre-trained model for detecting theft and substitution of goods by following these steps: Based on the historical trajectory information of the target vehicle, create multiple types of parking features and multiple types of non-parking features; For each type of docking feature and non-docking feature, multiple sub-class features are set; Obtain a pre-set trigger probability for each sub-category feature and a risk coefficient threshold generated when the sub-category feature is triggered; Based on each sub-category feature and the corresponding trigger probability and risk coefficient threshold, a model training sample is generated using a sample generator based on a Monte Carlo simulation algorithm. The process includes: using the sample generator based on a Monte Carlo simulation algorithm to randomly simulate triggering the multiple docking and non-docking features to obtain the target decimal corresponding to the triggered feature; determining the sub-category feature to be triggered based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each sub-category feature; using the sample generator based on a Monte Carlo simulation algorithm to randomly simulate triggering the sub-category feature to be triggered to determine the triggered sub-category feature; determining the target risk coefficient threshold corresponding to the triggered sub-category feature from the risk coefficient thresholds corresponding to each sub-category feature; randomly generating the risk coefficient corresponding to the triggered sub-category feature based on the target risk coefficient threshold; summing the risk coefficients corresponding to the triggered sub-category feature to obtain the theft / exchange risk coefficient; and generating model training samples based on the theft / exchange risk coefficient and the multiple sub-category features. Create a neural network model; The neural network model is trained based on the model training samples to generate a pre-trained model for judging the act of switching goods. Output the risk coefficient for theft or substitution corresponding to each truck; Based on the aforementioned risk coefficient for theft and substitution, it is determined whether each truck has engaged in theft or substitution of goods.

2. The method according to claim 1, characterized in that, The various types of stopping characteristics include stopping road type, stopping distance, stopping route percentage, stopping start time, and stopping duration; the various types of non-stopping characteristics include deviation distance, whether the vehicle is late, cargo value, and number of times the driver cooperates. The method of setting multiple sub-class features for each type of docking feature and non-docking feature includes: The types of roads for stopping are divided into three subtypes: high-grade roads, medium-grade roads, and low-grade roads, thus obtaining the subclass features of the road types for stopping; The parking distance is divided into three sub-types: long distance, medium distance, and short distance, to obtain the sub-class features of the parking distance; The percentage of stops is divided into three subtypes: two ends, slightly two ends, and middle, to obtain the sub-class features of the percentage of stops. The docking start time period is divided into three sub-types: daytime, early morning and evening, and late night, to obtain the sub-type features of the docking start time period; The docking duration is divided into three sub-types: very short, relatively short, and long, to obtain the sub-class features of the docking duration. The yaw distance is divided into 5 different subclasses according to multiple preset distance ranges to obtain the subclass features of the yaw distance; The vehicle's delay status is divided into two subtypes: delayed and not delayed, thus obtaining the vehicle's delay status subclass feature. The value of the goods is divided into three subclasses: high, medium, and low, to obtain the subclass characteristics of the value of the goods. The number of times the drivers cooperated was divided into three subtypes: no cooperation, cooperation, and multiple cooperations, thus obtaining the subclass features of the number of times the drivers cooperated.

3. The method according to claim 1, characterized in that, The step of determining the subclass feature to be triggered based on the target decimal corresponding to the triggered feature and the trigger probability corresponding to each subclass feature includes: From each of the sub-class features, determine a plurality of first sub-class features belonging to the triggered feature; From the trigger probabilities corresponding to each sub-class feature, find the trigger probability corresponding to each first sub-class feature; From the trigger probabilities corresponding to each first sub-class feature, determine multiple second sub-class features whose trigger probabilities correspond to the target decimals of the triggered features; The plurality of second subclass features are identified as subclass features to be triggered.

4. The method according to claim 1, characterized in that, The step of generating model training samples based on the risk coefficient of theft and the multiple sub-class features includes: The multiple subclass features are determined as model features; The risk factor for theft and substitution is set as a target value; The model features and the target value are associated to obtain a training sample; Continue executing the steps of creating multiple types of parking features and multiple types of non-parking features based on the historical trajectory point information of the target vehicle until the number of training samples reaches a preset number, and then use all training samples as model training samples.

5. The method according to claim 1, characterized in that, The determination of whether each truck has engaged in cargo swapping based on the cargo swapping risk coefficient includes: When the risk coefficient for goods swapping exceeds a preset risk coefficient threshold, it is determined that each truck has engaged in goods swapping; or... When the risk coefficient for theft or substitution is less than or equal to a preset risk coefficient threshold, it is determined that there is no theft or substitution of goods in each truck.

6. A system for determining the act of switching goods, implemented using the method described in any one of claims 1-5, characterized in that, The system includes: The trajectory point information receiving module is used to receive trajectory point information reported by the on-board terminal of each truck in real time. The behavior analysis module is used to determine whether each truck has stopped or deviated from its course based on the trajectory point information. The feature value determination module is used to determine multiple feature values ​​related to the stopping behavior or yaw behavior when each truck engages in stopping behavior or yaw behavior. The feature value input module is used to input the multiple feature values ​​into a pre-trained theft and substitution behavior determination model; the pre-trained theft and substitution behavior determination model is generated based on model training samples and a neural network model, and the model training samples are simulated based on a sample generator established by a Monte Carlo simulation algorithm; The anti-theft and substitution risk coefficient output module is used to output the anti-theft and substitution risk coefficient for each truck. The goods swapping behavior judgment module is used to determine whether each truck has engaged in goods swapping behavior based on the goods swapping risk coefficient.

7. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions adapted for loading by a processor and executing the method as described in any one of claims 1-5.

8. An electronic device, characterized in that, include: A processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed as described in any one of claims 1-5.

Images