Method and device for determining position of monitoring equipment, equipment and storage medium

Abstract

The invention discloses a method and device for determining the position of monitoring equipment, equipment and a storage medium, and belongs to the technical field of intelligent traffic. The method comprises the steps of obtaining traffic flow data collected by at least one existing monitoring device of a target road section, wherein the target road section is a road section area where the monitoring device is to be arranged; inputting the traffic flow data into a traffic congestion state prediction model, and determining the traffic congestion state of each area corresponding to the existing monitoring equipment; inputting the arrangement information of at least one existing monitoring device, the traffic flow data and the traffic congestion state to a target monitoring device position prediction model to obtain the position information of the monitoring device to be arranged; the target monitoring equipment position prediction model is obtained by training the to-be-trained monitoring equipment position prediction model based on the state space, the action space and the reward function. By using the technical scheme provided by the invention, the effectiveness and the accuracy of traffic flow data acquisition can be improved, and the scientificity of emergency lane temporary starting decision making is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and particularly to a method, device, equipment and storage medium for determining the position of a monitoring device. Background Art

[0002] With the acceleration of the urbanization process, the problem of highway traffic congestion has become increasingly serious, affecting the road traffic capacity. Emergency lanes are provided on highways for emergency vehicles, and the reasonable use of emergency lanes can avoid serious congestion. Therefore, in order to alleviate the traffic congestion problem, the reasonable activation of emergency lanes has become one of the key measures. However, the traditional decision-making for activating emergency lanes often relies on empirical judgment, lacking scientificity and systematicness. Therefore, the highway management department decides whether to allow the temporary use of emergency lanes based on experience, lacking a theoretical basis, which is prone to disputes and difficult to evaluate the effect of alleviating congestion.

[0003] In order to improve the scientificity of the decision-making for temporarily activating emergency lanes, highway monitoring video data can be provided for the problem of temporarily activating emergency lanes. How to arrange video monitoring points on the highway is an urgent problem to be solved.

[0004] Therefore, more reliable solutions are needed. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method, device, equipment and storage medium for determining the position of a monitoring device, which can reasonably arrange the monitoring device to improve the effectiveness and accuracy of traffic flow data collection, thereby improving the scientificity of the decision-making for temporarily activating emergency lanes, alleviating traffic congestion, and improving traffic safety.

[0006] To achieve the above purpose, the present invention is implemented by adopting the following technical solutions:

[0007] On the one hand, the present invention provides a method for determining the position of a monitoring device, the method comprising:

[0008] Obtaining traffic flow data collected by at least one existing monitoring device on a target section, where the target section is a section area where a monitoring device is to be arranged;

[0009] Inputting the traffic flow data into a traffic congestion state prediction model to determine the traffic congestion state of the area corresponding to each existing monitoring device;

[0010] Inputting the arrangement information of the at least one existing monitoring device, the traffic flow data, and the traffic congestion state into a target monitoring device position prediction model to obtain the position information of the monitoring device to be arranged;

[0011] The target monitoring device location prediction model is obtained by training the monitoring device location prediction model to be trained based on a state space, an action space, and a reward function; the state space includes a plurality of state data, and the plurality of state data includes the arrangement information of a plurality of deployed monitoring devices on the sample road section where the monitoring device to be arranged is located, a plurality of sample traffic flow data, and a plurality of sample traffic congestion states; the plurality of sample traffic flow data is obtained based on the arrangement information of the plurality of deployed monitoring devices, and the plurality of sample traffic congestion states is obtained based on the plurality of sample traffic flow data and the traffic congestion state prediction model; the action space includes a plurality of action data, and the plurality of action data includes a plurality of action sequences corresponding to the adjusted monitoring device arrangement information; the reward function is used to evaluate the quality of the action data.

[0012] In some possible implementation manners, the target monitoring device location prediction model is constructed in the following manner:

[0013] Input the current state data into the monitoring device location prediction model to be trained to generate current action data; the current state data includes the arrangement information of the currently deployed monitoring devices on the sample road section where the monitoring device to be arranged is located, the current sample traffic flow data, and the current sample traffic congestion state; the current sample traffic flow data is obtained based on the arrangement information of the currently deployed monitoring devices, and the current sample traffic congestion state is obtained based on the sample traffic flow data and the traffic congestion state prediction model; the currently deployed monitoring devices are determined based on a plurality of preset monitoring device arrangement information or the current action data; the current action data includes an action sequence corresponding to the adjusted monitoring device arrangement information; the adjusted monitoring device arrangement information includes enabling any monitoring device and disabling any monitoring device;

[0014] Determine the successor state data according to the current action data;

[0015] Determine the current reward according to the successor state data and the current state data in combination with the reward function;

[0016] Use the successor state data as the current state data, and input the current state data, the current action data, and the current reward data into the monitoring device location prediction model to be trained to obtain successor action data;

[0017] Use the successor action data as the current action data, and jump to Determine the successor state data according to the current action data until a preset convergence condition is met.

[0018] In some possible embodiments, the to-be-trained monitoring device location prediction model includes a preliminary feature extraction module, a feature location marking module, a deep feature extraction module, and a monitoring device location prediction module. Inputting the current state data, the current action data, and the current reward data into the to-be-trained monitoring device location prediction model to obtain subsequent action data includes:

[0019] Inputting the current state data, the current action data, and the current reward data into the preliminary feature extraction module for preliminary feature extraction processing to obtain preliminary combined features;

[0020] Inputting the preliminary combined features into the feature location marking module for feature location marking processing to obtain ordered combined features;

[0021] Inputting the ordered combined features into the deep feature extraction module for deep feature extraction processing to obtain high-level combined features;

[0022] Inputting the high-level combined features into the monitoring device location prediction module for monitoring device location prediction processing to obtain subsequent action data.

[0023] In some possible embodiments, the method includes:

[0024] Setting the objective function Z of the to-be-trained monitoring device location prediction model as shown in the following formula:

[0025] ;

[0026] Wherein, represents the influence of deploying a monitoring device at the i-th monitoring device location on the accuracy of the traffic congestion state prediction model; represents the cost of deploying a monitoring device at the i-th monitoring device location; Whether to deploy a monitoring device at the i-th monitoring device location; represents a weight parameter for measuring the relationship between the emergency lane enabling decision support effect and the cost; n represents the number of monitoring device locations to be deployed;

[0027] The constraint conditions of the objective function are:

[0028] , ;

[0029] ;

[0030] , ;

[0031] Wherein, represents a set of positions of the monitoring devices to be deployed for the target area covering the sample road section where the monitoring devices are to be deployed; m represents the number of target areas of the sample road section where the monitoring devices are to be deployed; B represents the preset cost of deploying the monitoring devices.

[0032] In some possible implementation manners, the reward function is constructed in the following manner:

[0033] ;

[0034] wherein represents the change in the sample traffic flow data between the successor state data and the current state data, represents the change in the sample traffic congestion state between the successor state data and the current state data, represents the change in the accuracy of the traffic congestion state prediction model between the successor state data and the current state data, represents the change in the preset cost between the successor state data and the current state data, , , and respectively represent the first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function.

[0035] In some possible implementation manners, the traffic congestion state prediction model is constructed in the following manner:

[0036] Obtain the traffic flow sample data collected by the monitoring devices of the current sample road section, and the preset traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample road section;

[0037] Input the traffic flow sample data into the traffic congestion state prediction model to be trained, and obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample road section;

[0038] Based on the preset traffic congestion state and the predicted traffic congestion state, train the traffic congestion state prediction model to be trained to obtain the traffic congestion state prediction model.

[0039] In some possible implementation manners, the traffic congestion state prediction model to be trained includes: a clustering module, a gating module, and a plurality of prediction modules;

[0040] The step of inputting the traffic flow sample data into the traffic congestion state prediction model to be trained, and obtaining the predicted traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample road section includes:

[0041] Input the traffic flow sample data into the clustering module for partitioning processing to obtain a data partitioning result;

[0042] Input the data partitioning result into the gating module to determine the weight of each prediction module;

[0043] Input the data partitioning result into the corresponding prediction module according to the weight of each prediction module to obtain the traffic congestion status prediction result of each prediction module;

[0044] Based on the weights of each prediction module, perform a weighted summation process on the traffic congestion status prediction results of each prediction module to obtain the predicted traffic congestion status.

[0045] On the other hand, a device for determining the location of a monitoring device is provided. The device includes:

[0046] A data acquisition module for acquiring traffic flow data collected by at least one existing monitoring device on a target road section, where the target road section is a road section area where a monitoring device is to be arranged;

[0047] A status determination module for inputting the traffic flow data into a traffic congestion status prediction model to determine the traffic congestion status of the area corresponding to each existing monitoring device;

[0048] A location determination module for inputting the arrangement information of the at least one existing monitoring device, the traffic flow data, and the traffic congestion status into a target monitoring device location prediction model to obtain the location information of the monitoring device to be arranged; the target monitoring device location prediction model is obtained by training a monitoring device location prediction model to be trained based on a state space, an action space, and a reward function; the state space includes a plurality of state data, and the plurality of state data includes the arrangement information of a plurality of existing monitoring devices on a sample road section of the monitoring device to be arranged, a plurality of sample traffic flow data, and a plurality of sample traffic congestion statuses; the plurality of sample traffic flow data is obtained based on the arrangement information of the plurality of existing monitoring devices, and the plurality of sample traffic congestion statuses is obtained based on the plurality of sample traffic flow data and the traffic congestion status prediction model; the action space includes a plurality of action data, and the plurality of action data includes a plurality of action sequences corresponding to adjusted monitoring device arrangement information; the reward function is used to evaluate the quality of the action data.

[0049] On the other hand, an electronic device is provided. The device includes a processor and a memory. At least one instruction and at least one program segment are stored in the memory, and the at least one instruction and the at least one program segment are loaded and executed by the processor to implement the method for determining the location of the monitoring device as described above.

[0050] On the other hand, a computer-readable storage medium is provided. At least one instruction and at least one program segment are stored in the computer storage medium. The at least one instruction and the at least one program segment are loaded and executed by a processor to implement the method for determining the position of the monitoring device as described above.

[0051] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0052] In the present invention, traffic flow data collected by at least one existing monitoring device on the target road section where the monitoring device to be arranged is located is obtained; the traffic flow data is input into a traffic congestion state prediction model to determine the traffic congestion state of the area corresponding to each existing monitoring device; then, the arrangement information, traffic flow data, and traffic congestion state of at least one existing monitoring device are input into a target monitoring device position prediction model to obtain the position information of the monitoring device to be arranged. The position of the monitoring device can be dynamically adjusted according to the current traffic environment state. By using the target monitoring device position prediction model to determine a reasonable monitoring device arrangement plan, the effectiveness and accuracy of traffic flow data collection on this road section can be improved. Furthermore, corresponding data can be provided for the decision-making of temporarily enabling the emergency lane, thereby improving the scientificity of the decision-making for temporarily enabling the emergency lane, alleviating traffic congestion, and improving the smoothness and traffic safety of road traffic. And the target monitoring device position prediction model is trained based on the state space, action space, and reward function for the monitoring device position prediction model to be trained, which can improve the effectiveness, accuracy, and robustness of the model, and further improve the effectiveness and accuracy of determining the position information of the monitoring device to be arranged. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 is a flowchart showing a method for determining the position of a monitoring device provided by an embodiment of the present invention;

[0055] Figure 2 is a flowchart showing a method for constructing a traffic congestion state prediction model provided by an embodiment of the present invention;

[0056] Figure 3 is a flowchart showing a method for constructing a target monitoring device position prediction model provided by an embodiment of the present invention;

[0057] Figure 4It is a schematic flowchart of a process for inputting current state data, current action data, and current reward data into a monitoring device position prediction model to be trained to obtain subsequent action data according to an embodiment of the present invention;

[0058] Figure 5 It is a schematic structural diagram of a device for determining the position of a monitoring device according to an embodiment of the present invention. Specific embodiments

[0059] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, 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.

[0060] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0061] In the embodiments of the present invention, the term "module" or "unit" refers to a computer program with a predetermined function or a part of a computer program, which works together with other related parts to achieve a predetermined goal, and can be fully or partially implemented by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of the overall module or unit that includes the functions of the module or unit.

[0062] The various exemplary embodiments, features, and aspects of the present invention will be described in detail below with reference to the accompanying drawings. The same reference numerals in the drawings denote elements with the same or similar functions. Although various aspects of the embodiments are shown in the drawings, unless otherwise specified, the drawings do not have to be drawn to scale.

[0063] As used herein, the term "exemplary" means "serving as an example, embodiment, or illustration". Any embodiment described herein as "exemplary" is not necessarily to be construed as superior or better than other embodiments.

[0064] As used herein, the term "and / or" is merely a description of the relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" as used herein means any one of a plurality or any combination of at least two of a plurality. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0065] In addition, to better illustrate the present invention, numerous specific details are given in the following detailed description. Those skilled in the art should understand that the present invention can also be implemented without certain specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art are not described in detail in order to highlight the gist of the present invention.

[0066] Figure 1 is a schematic flowchart of a method for determining the position of a monitoring device provided by an embodiment of the present invention. This specification provides method operation steps such as in the embodiment or flowchart, but based on routine or non-creative labor, there can be more or fewer operation steps. The order of steps listed in the embodiment is only one way among the execution orders of numerous steps and does not represent the only execution order. When the actual system or server product executes, it can be executed in the order of the method shown in the embodiment or the drawings or executed in parallel (for example, in an environment of parallel processors or multi-threaded processing). Specifically, as Figure 1 shown, the above method may include:

[0067] S101: Obtain traffic flow data collected by at least one existing monitoring device on a target road section;

[0068] In a specific embodiment, the target road section can be the road section area where monitoring devices are to be arranged. Optionally, the target road section can be a road section area with high traffic flow, or frequent speed changes, or prone to traffic congestion, or with frequent traffic accidents. The target road section can be the road section area on the highway where monitoring devices are to be arranged, or the road section area near schools or residential communities where monitoring devices are to be arranged; specifically, the target road section can be determined in combination with the actual application scenario. At least one existing monitoring device can be the monitoring device that already exists on the target road section. Optionally, the monitoring device can be a device for monitoring the traffic conditions of the road section, such as a camera. Traffic flow data can be the data collected by at least one monitoring device for monitoring the corresponding area; traffic flow data can be the data representing the operating state of the traffic flow, and traffic flow data can include traffic flow volume data, traffic speed data, and traffic density data.

[0069] S102: Input the traffic flow data into the traffic congestion state prediction model to determine the traffic congestion state of the area corresponding to each existing monitoring device;

[0070] In a specific embodiment, the traffic congestion state prediction model can be a model for determining the traffic congestion state. Optionally, using the traffic congestion state prediction model to determine the traffic congestion state can improve the accuracy of determining the traffic congestion state.

[0071] In an alternative embodiment, Figure 2 is a schematic flowchart of a method for constructing a traffic congestion state prediction model provided by an embodiment of the present invention; as Figure 2 shown, the above traffic congestion state prediction model can be constructed in the following manner:

[0072] S201: Obtain the traffic flow sample data collected by the monitoring devices of the current sample road section, and the preset traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample road section;

[0073] S202: Input the traffic flow sample data into the traffic congestion state prediction model to be trained, and obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample road section;

[0074] S203: Based on the preset traffic congestion state and the predicted traffic congestion state, train the traffic congestion state prediction model to be trained to obtain the traffic congestion state prediction model.

[0075] In an alternative embodiment, the above traffic congestion state prediction model to be trained includes: a clustering module, a gating module, and a plurality of prediction modules;

[0076] Inputting the traffic flow sample data into the traffic congestion state prediction model to be trained to obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring device of the current sample section may include:

[0077] Input the traffic flow sample data into the clustering module for partitioning to obtain a data partitioning result;

[0078] Input the data partitioning result into the gating module to determine the weight of each prediction module;

[0079] Input the data partitioning result into the corresponding prediction module according to the weight of each prediction module to obtain the traffic congestion state prediction result of each prediction module;

[0080] Based on the weight of each prediction module, perform a weighted sum processing on the traffic congestion state prediction results of each prediction module to obtain the predicted traffic congestion state.

[0081] In a specific embodiment, the current sample section may include multiple sections in the drivable area, and the current sample section is used to obtain the traffic flow sample data used in the process of training the traffic congestion state prediction model; the monitoring device of the current sample section may be the monitoring device that can be used in the current sample section. The traffic flow sample data may include the data collected by the monitoring devices of multiple sections, and the traffic flow sample data may be the data used to train the traffic congestion state prediction model.

[0082] The preset traffic congestion state may be information characterizing the traffic congestion state of the monitoring area corresponding to the monitoring device of the current sample section; specifically, the traffic congestion state can be divided in combination with the actual application, such as unobstructed, slow-moving, slightly congested, moderately congested, severely congested, etc. Optionally, assuming that the total number of divisions of the road section traffic congestion state is N (N is a positive integer), the preset traffic congestion state may be a 1*N vector, and each element in the vector corresponds to a traffic congestion state. Specifically, each element may represent the probability that the traffic congestion state of the monitoring area corresponding to the monitoring device belongs to the corresponding traffic congestion state (if the traffic congestion state of the monitoring area corresponding to a certain monitoring device belongs to the corresponding traffic congestion state, correspondingly, the corresponding probability is 1; otherwise, if the traffic congestion state of the monitoring area corresponding to a certain monitoring device does not belong to the corresponding traffic congestion state, correspondingly, the corresponding probability is 0).

[0083] In a specific embodiment, the clustering module can be used to divide the traffic flow sample data into several subsets. Optionally, the clustering module can use K-means clustering or Gaussian mixture clustering to divide the traffic flow sample data. The data division result can include several subsets into which the traffic flow sample data is divided. The gating module can be used to dynamically allocate the weights of each prediction module according to the data division result to determine the importance of each prediction module for the traffic flow sample data. Specifically, if K-means clustering is used to divide the traffic flow sample data, each cluster center can correspond to a prediction module, and the weight of each prediction module can be determined according to the distance between the traffic flow sample data and each cluster center; if Gaussian mixture clustering is used to divide the traffic flow sample data, each cluster can correspond to a prediction module, and the weight of each prediction module can be determined according to the probability that the traffic flow sample data belongs to each cluster. The prediction module can predict the traffic congestion state for the data division result. Optionally, the prediction module can use the fuzzy neural network regression method to predict the traffic congestion state for the data division result. Specifically, the prediction module can include a fuzzy neural network model, and a fuzzy neural network model can be trained for each subset based on several subsets in the data division result.

[0084] In a specific embodiment, predicting the traffic congestion state can be information representing the traffic congestion state of the monitoring area corresponding to the monitoring device of the current sample road section identified by the traffic congestion state prediction model to be trained.

[0085] In a specific embodiment, training the traffic congestion state prediction model to be trained based on the preset traffic congestion state and the predicted traffic congestion state to obtain the traffic congestion state prediction model can include: determining the congestion state recognition loss of the traffic congestion state prediction model to be trained according to the preset traffic congestion state and the predicted traffic congestion state; training the traffic congestion state prediction model to be trained based on the congestion state recognition loss to obtain the traffic congestion state prediction model. Optionally, the congestion state recognition loss can be calculated in combination with a first preset loss function; optionally, the first preset loss function can be set according to the actual application. The above congestion state recognition loss can represent the accuracy of the traffic congestion state recognition of the current traffic congestion state prediction model to be trained.

[0086] In a specific embodiment, training the traffic congestion state prediction model to be trained based on the congestion state recognition loss to obtain a traffic congestion state prediction model may include: updating the model parameters of the traffic congestion state prediction model to be trained based on the congestion state recognition loss, and based on the updated traffic congestion state prediction model to be trained, repeatedly inputting traffic flow sample data into the traffic congestion state prediction model to be trained, obtaining the predicted traffic congestion state to the first training iteration step of updating the model parameters of the traffic congestion state prediction model to be trained based on the congestion state recognition loss until the first preset convergence condition is satisfied. The above-mentioned satisfaction of the first preset convergence condition may be that the congestion state recognition loss is less than or equal to the first preset loss threshold, or the number of times of the first training iteration step reaches the first preset number, etc. The specific first preset loss threshold and the first preset number may be set in combination with the model accuracy and training speed requirements in actual applications.

[0087] In the above embodiment, using the traffic congestion state prediction model to determine the traffic congestion state can improve the efficiency and accuracy of determining the traffic congestion state, and further improve the effectiveness and accuracy of determining the position information of the monitoring devices to be arranged.

[0088] S103: Input the arrangement information of at least one existing monitoring device, traffic flow data, and traffic congestion state into the target monitoring device position prediction model to obtain the position information of the monitoring device to be arranged;

[0089] In a specific embodiment, the arrangement information of at least one existing monitoring device may represent the position information of the existing monitoring devices on the target road section and the relationship between the existing monitoring devices. The position information of the monitoring device to be arranged may be the position information of the monitoring device to be arranged on the target road section. Optionally, based on the position information of the monitoring device to be arranged, monitoring devices can be reasonably arranged on the target road section to improve the accuracy of monitoring the target road section, timely understand the traffic conditions of the target road section, and further improve the smoothness and safety of the target road section. Specifically, when the target road section is the road section area where monitoring devices are to be arranged on the highway, based on the position information of the monitoring device to be arranged, monitoring devices are reasonably arranged in this road section area, and further improve the effectiveness and accuracy of traffic flow data collection, which can provide corresponding data for the decision-making of temporarily enabling the emergency lane, and further improve the scientificity of the decision-making of temporarily enabling the emergency lane, thereby alleviating traffic congestion and improving the smoothness of road traffic and traffic safety; when the target road section is the road section area where monitoring devices are to be arranged near schools or residential areas, based on the position information of the monitoring device to be arranged, monitoring devices are reasonably arranged in this road section area, which can improve the effectiveness and accuracy of traffic flow data collection, and further timely understand the traffic conditions of this road section area, facilitating the staff to take corresponding measures for this road section area to improve the smoothness and safety of road traffic.

[0090] In a specific embodiment, the target monitoring device location prediction model is obtained by training the monitoring device location prediction model to be trained based on a state space, an action space, and a reward function; the state space includes multiple state data, and the multiple state data includes the layout information of multiple deployed monitoring devices on the sample road section where the monitoring device to be deployed is located, multiple sample traffic flow data, and multiple sample traffic congestion states; the multiple sample traffic flow data is obtained based on the layout information of the multiple deployed monitoring devices, and the multiple sample traffic congestion states are obtained based on the multiple sample traffic flow data and a traffic congestion state prediction model; the action space includes multiple action data, and the multiple action data includes action sequences corresponding to multiple adjusted monitoring device layout information; the reward function is used to evaluate the quality of the action data.

[0091] In a specific embodiment, the sample road section where the monitoring device to be deployed is located may include road sections of multiple monitoring devices to be deployed in the drivable area. The multiple deployed monitoring devices may include the monitoring devices deployed on the road sections of the multiple monitoring devices to be deployed; the multiple deployed monitoring devices may be determined based on preset monitoring device layout data or multiple action data; the layout information of the multiple deployed monitoring devices may represent the position information of the deployed monitoring devices on the sample road section and the relationship between the deployed monitoring devices. The multiple sample traffic flow data may include data collected by the deployed monitoring devices on multiple sample road sections. The multiple sample traffic congestion states may be information characterizing the traffic congestion states of the monitoring areas corresponding to the deployed monitoring devices on the sample road sections. The multiple sample traffic flow data and the multiple sample traffic congestion states may be data used for training the target monitoring device location prediction model.

[0092] In a specific embodiment, the above-mentioned multiple sample traffic flow data may be determined in the following manner: according to the layout information of the multiple deployed monitoring devices, determine the position information of each deployed monitoring device; according to the position information of each deployed monitoring device, collect sample traffic flow data of the area corresponding to the position information of each deployed monitoring device through each deployed monitoring device. The above-mentioned multiple sample traffic congestion states may be determined in the following manner: input the multiple sample traffic flow data into a traffic congestion state prediction model, and output multiple sample traffic congestion states.

[0093] In an alternative embodiment, Figure 3 is a schematic flowchart of a method for constructing a target monitoring device location prediction model provided by an embodiment of the present invention; as Figure 3 shown, the target monitoring device location prediction model is constructed in the following manner:

[0094] S301: Input the current state data into the monitoring device location prediction model to be trained, and generate current action data;

[0095] S302: Determine the successor state data according to the current action data;

[0096] S303: Determine the current reward according to the successor state data and the current state data in combination with the reward function;

[0097] S304: Use the successor state data as the current state data, and input the current state data, the current action data, and the current reward data into the to-be-trained monitoring device location prediction model to obtain the successor action data;

[0098] S305: Use the successor action data as the current action data, and jump to determine the successor state data according to the current action data until the preset convergence condition is met.

[0099] In a specific embodiment, the current state data includes the layout information of the currently deployed monitoring devices on the sample road section where the monitoring devices are to be arranged, the current sample traffic flow data, and the current sample traffic congestion state; the current sample traffic flow data is obtained based on the layout information of the currently deployed monitoring devices, and the current sample traffic congestion state is obtained based on the sample traffic flow data and the traffic congestion state prediction model; the currently deployed monitoring devices are determined based on multiple preset monitoring device layout information or the current action data; the current action data includes the action sequence corresponding to the adjusted monitoring device layout information; adjusting the monitoring device layout information includes enabling any monitoring device and disabling any monitoring device.

[0100] In a specific embodiment, the multiple preset monitoring device layout information can be set in combination with the actual application. Specifically, the multiple preset monitoring device layout information can include arranging a monitoring device every 3 meters, arranging a monitoring device every 5 meters, etc. The current action data is generated based on the model. The action sequence corresponding to the adjusted monitoring device layout information includes multiple actions, that is, the corresponding actions during the process of adjusting the monitoring device layout information. Specifically, the current action data can include enabling the monitoring device at every 3 meters, enabling the monitoring devices at 3 meters and 9 meters in this road section, and disabling the monitoring device at 6 meters in this road section, etc.

[0101] In a specific embodiment, the current state data can reflect the environmental situation where the model is currently located (i.e., the layout information of the currently deployed monitoring devices); the subsequent state data can reflect the environmental situation after the model executes the current action data, and the subsequent state data can be data determined based on the environmental changes caused by the current action data (e.g., changes in the layout information of the deployed monitoring devices); specifically, the current action data can adjust the layout information of the monitoring devices, thereby changing the layout information of the deployed monitoring devices, and thus changing the sample traffic flow data and sample traffic congestion. At this time, compared with the current state data, the state data changes, and then the subsequent state data is determined. Optionally, the subsequent state data can also be determined based on the environmental changes caused by the preset monitoring device layout information.

[0102] The current reward data can be used to evaluate the quality of the execution of the current action data, provide timely feedback to the model, and facilitate the timely adjustment of the model.

[0103] The subsequent action data can be action data determined based on the subsequent state data (the new state data relative to the current state data, i.e., the new environment relative to the current environment), where the current action data and the current reward data can be used by the monitoring device location prediction model to be trained to determine the quality of the current action data, facilitating the further training of the monitoring device location prediction model to be trained.

[0104] In a specific embodiment, determining the subsequent state data according to the current action data as described above may include: generating the layout information of the adjusted monitoring devices according to the monitoring device location prediction model to be trained executing the current action data; determining the adjusted sample traffic flow data based on the layout information of the adjusted monitoring devices; obtaining the adjusted sample traffic congestion state based on the adjusted traffic flow data and the traffic congestion state prediction model; and determining the layout information of the adjusted monitoring devices, the adjusted sample traffic flow data, and the adjusted sample traffic congestion state as the subsequent state data.

[0105] In an alternative embodiment, the above reward function can be constructed in the following manner:

[0106] ;

[0107] where represents the change in the sample traffic flow data between the subsequent state data and the current state data, represents the change in the sample traffic congestion state between the subsequent state data and the current state data, represents the change in the accuracy of the traffic congestion state prediction model between the subsequent state data and the current state data, represents the change in the preset cost between the subsequent state data and the current state data, , , and respectively represent the first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function.

[0108] In a specific embodiment, the reward function can be used to evaluate the effect of the monitoring device layout plan. The construction process of the reward function needs to consider multiple factors, including: reducing the average travel time of traffic flow, reducing traffic congestion, improving the emergency response speed, and maintaining a low cost. The accuracy of the traffic congestion state prediction model can be used as an important indicator to measure the decision-making support role of the monitoring device layout information for the emergency lane activation. The accuracy of the traffic congestion state prediction model can be used to evaluate the model performance, can affect the emergency lane activation decision, and the model accuracy can be determined using preset evaluation indicators, such as precision, recall, and F1 score, etc. The preset cost can be based on the layout information of the monitoring device determined by the current action data, and the cost required to set up the corresponding monitoring device. The first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function can represent the importance of the above various consideration factors in the reward function. Specifically, the first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function can respectively represent the importance of the change in the sample traffic flow data between the successor state data and the current state data, the importance of the change in the sample traffic congestion state between the successor state data and the current state data, the importance of the change in the accuracy of the traffic congestion state prediction model between the successor state data and the current state data, and the importance of the change in the preset cost between the successor state data and the current state data; Optionally, the first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function can be set in combination with the actual application.

[0109] In a specific embodiment, the current reward data can include positive rewards and negative punishments. The positive rewards can come from the realization of the above-mentioned reduction of the average travel time of traffic flow, reduction of traffic congestion, improvement of the emergency response speed, and reduction of cost. The negative punishments can come from the increase in the average travel time of traffic flow, aggravation of traffic congestion, slowdown of the emergency response speed, and cost exceeding the budget.

[0110] In an alternative embodiment, Figure 4 is a schematic flow diagram of inputting the current state data, the current action data, and the current reward data into the monitoring device location prediction model to be trained to obtain the successor action data; the above-mentioned monitoring device location prediction model to be trained includes a preliminary feature extraction module, a feature location marking module, a deep feature extraction module, and a monitoring device location prediction module, as Figure 4As shown, the above-mentioned current state data, current action data, and current reward data are input into the monitoring device position prediction model to be trained, and subsequent action data is obtained, including:

[0111] S401: Input the current state data, current action data, and current reward data into the preliminary feature extraction module for preliminary feature extraction processing to obtain preliminary joint features;

[0112] S402: Input the preliminary joint features into the feature position marking module for feature position marking processing to obtain ordered joint features;

[0113] S403: Input the ordered joint features into the deep feature extraction module for deep feature extraction processing to obtain high-level joint features;

[0114] S404: Input the high-level joint features into the monitoring device position prediction module for monitoring device position prediction processing to obtain subsequent action data.

[0115] In a specific embodiment, the preliminary feature extraction module is used to preliminarily extract the feature information of the current state data, current action data, and current reward data to obtain preliminary joint features; optionally, the preliminary feature extraction module can be set according to the actual application. Specifically, the preliminary feature extraction module can include a linear embedding layer. The feature position marking module is used to perform position marking processing on the preliminary joint features to generate ordered joint features; optionally, the feature position marking module performs position encoding on the preliminary joint features to retain the order information of the features. Optionally, the feature position marking module can be set according to the actual application. The deep feature extraction module is used to deeply extract the feature information of the ordered joint features to obtain high-level joint features; optionally, the deep feature extraction module can be set according to the actual application; specifically, the deep feature extraction module can include a Transformer encoder, and the Transformer encoder combines the multi-head self-attention mechanism to capture the dependencies between features. The monitoring device position prediction module is used to perform monitoring device position prediction processing on the high-level joint features to determine subsequent action data; optionally, the monitoring device position prediction module can be set according to the actual application; specifically, the monitoring device position prediction module can include a linear decoder, and the linear decoder combines the reinforcement learning strategy to generate future action sequences.

[0116] In a specific embodiment, a preset loss function can be used to optimize the model. Specifically, the layout information of the currently deployed monitoring devices determined according to the current action data and the layout information of the existing monitoring devices on the sample road section of the monitoring device to be deployed can be combined, and the action generation loss can be determined in combination with the preset loss function. Then, based on the action generation loss, the monitoring device position prediction model to be trained is trained to obtain the target monitoring device position prediction model. Optionally, the preset loss function can be set in combination with the actual application. Specifically, the preset loss function can be the mean square error loss function. The above action generation loss can characterize the accuracy of the action data generation of the currently trained monitoring device position prediction model.

[0117] In a specific embodiment, the above-mentioned training the monitoring device position prediction model to be trained based on the action generation loss to obtain the target monitoring device position prediction model may include: based on the action generation loss, updating the model parameters of the monitoring device position prediction model to be trained, and based on the updated monitoring device position prediction model to be trained, repeating the training iteration steps from determining the successor state data according to the current action data to updating the model parameters of the monitoring device position prediction model to be trained based on the action generation loss until the preset convergence condition is met. The above-mentioned meeting the preset convergence condition may be that the action generation loss is less than the preset loss threshold, or the number of training iteration steps reaches the preset number, etc. The specific preset loss threshold and preset number can be set in combination with the model accuracy and training speed requirements in the actual application.

[0118] In an alternative embodiment, the above method includes:

[0119] Set the objective function Z of the monitoring device position prediction model to be trained as shown in the following formula:

[0120] ;

[0121] Where, represents the impact of deploying a monitoring device at the i-th monitoring device position on the accuracy of the traffic congestion state prediction model; represents the cost of deploying a monitoring device at the i-th monitoring device position; represents whether to deploy a monitoring device at the i-th monitoring device position; represents the weight parameter used to measure the relationship between the emergency lane activation decision support effect and the cost; n represents the number of positions of the monitoring device to be deployed;

[0122] The constraint conditions of the objective function are:

[0123] , ;

[0124] ;

[0125] , ;

[0126] wherein, represents the set of positions of the monitoring devices to be arranged in the target area covering the sample section of the monitoring devices to be arranged; m represents the number of target areas of the sample section of the monitoring devices to be arranged; B represents the preset cost of arranging the monitoring devices.

[0127] In a specific embodiment, represents arranging a monitoring device at the i-th monitoring device position; not arranging a monitoring device at the i-th monitoring device position; the target areas of the sample section of the monitoring devices to be arranged may include areas on the sample section with large traffic flow, frequent changes in traffic speed, prone to traffic congestion or frequent occurrence of traffic accidents.

[0128] In a specific embodiment, the goals achieved by determining the objective function include: maximizing the support of the monitoring devices for the emergency lane activation decision, minimizing the total cost of arranging the monitoring devices, and ensuring the effective monitoring of the target area; combining constraint conditions such as preset cost limitations to train the monitoring device position prediction model to be trained.

[0129] In a specific embodiment, after the training of the target monitoring device position prediction model is completed, the trained model can be tested in the actual environment to evaluate its performance and robustness; through simulation testing, the effectiveness of the model can be verified to ensure that the model maintains high performance under various possible traffic conditions.

[0130] In a specific embodiment, in order to further improve the performance of the model, the ε-greedy strategy can be used, which balances exploration and exploitation. In the initial stage of training, the model can traverse all state data in the state space for more exploration, try different action combinations, so that the model can find the optimal arrangement information of the monitoring devices in the changing environment.

[0131] In the above embodiment, using the target monitoring device position prediction model to determine the position information of the monitoring devices to be arranged can improve the accuracy and effectiveness of determining the monitoring device position information, and further improve the accuracy and effectiveness of traffic flow data collection. The model can continuously adjust the arrangement of the monitoring devices, dynamically generate optimal action data through the collected traffic flow data to adapt to the changing traffic conditions. And a reasonable set of arrangement information of the monitoring devices can be selected under a limited cost budget to ensure the maximum improvement of the effectiveness and accuracy of traffic flow data collection within the target section, effectively monitor the traffic conditions and support the temporary activation decision of the emergency lane.

[0132] As can be seen from the technical solutions provided in the embodiments of this specification above, this specification obtains traffic flow data collected by at least one existing monitoring device on the target road section where the monitoring device to be arranged is located; inputs the traffic flow data into the traffic congestion state prediction model to determine the traffic congestion state of the area corresponding to each existing monitoring device; then, inputs the arrangement information, traffic flow data, and traffic congestion state of at least one existing monitoring device into the target monitoring device position prediction model to obtain the position information of the monitoring device to be arranged; can dynamically adjust and confirm the position of the monitoring device according to the current traffic environment state, use the target monitoring device position prediction model to determine a reasonable monitoring device arrangement plan, which can improve the effectiveness and accuracy of traffic flow data collection on this road section, and a reasonable monitoring device arrangement plan can provide more comprehensive traffic flow data and traffic information, and for the area where traffic flow data has not been collected on this target road section, it can improve the accuracy of predicting traffic flow data in this area based on the collected traffic flow data, and further can provide corresponding data for the decision-making of temporarily enabling the emergency lane, thereby improving the scientific nature of the decision-making of temporarily enabling the emergency lane to relieve traffic congestion and improve the smoothness and traffic safety of road traffic. And the target monitoring device position prediction model is trained based on the state space, action space, and reward function for the monitoring device position prediction model to be trained, which can improve the effectiveness, accuracy, and robustness of the model, and further can improve the effectiveness and accuracy of determining the position information of the monitoring device to be arranged.

[0133] An embodiment of the present invention further provides a device for determining the position of a monitoring device. Correspondingly, Figure 5 is a schematic structural diagram of a device for determining the position of a monitoring device provided by an embodiment of the present invention; as Figure 5 shown, the above device includes:

[0134] A data acquisition module 510, configured to acquire traffic flow data collected by at least one existing monitoring device on a target road section, where the target road section is a road section area where the monitoring device to be arranged is located;

[0135] A state determination module 520, configured to input the traffic flow data into a traffic congestion state prediction model to determine the traffic congestion state of the area corresponding to each existing monitoring device;

[0136] A location determination module 530 is configured to input the layout information of the at least one existing monitoring device, the traffic flow data, and the traffic congestion status into a target monitoring device location prediction model to obtain the location information of the monitoring device to be arranged; the target monitoring device location prediction model is obtained by training a monitoring device location prediction model to be trained based on a state space, an action space, and a reward function; the state space includes a plurality of state data, and the plurality of state data includes the layout information of a plurality of arranged monitoring devices on the sample road section of the monitoring device to be arranged, a plurality of sample traffic flow data, and a plurality of sample traffic congestion statuses; the plurality of sample traffic flow data is obtained based on the layout information of the plurality of arranged monitoring devices, and the plurality of sample traffic congestion statuses are obtained based on the plurality of sample traffic flow data and the traffic congestion status prediction model; the action space includes a plurality of action data, and the plurality of action data includes action sequences corresponding to a plurality of adjusted monitoring device layout information; the reward function is used to evaluate the quality of the action data.

[0137] In an optional embodiment, the device includes a target monitoring device location prediction model construction module for constructing a target monitoring device location prediction model;

[0138] The target monitoring device location prediction model construction module includes:

[0139] An action data generation unit is configured to input the current state data into the monitoring device location prediction model to be trained to generate current action data; the current state data includes the layout information of the currently arranged monitoring devices on the sample road section of the monitoring device to be arranged, current sample traffic flow data, and current sample traffic congestion statuses; the current sample traffic flow data is obtained based on the layout information of the currently arranged monitoring devices, and the current sample traffic congestion statuses are obtained based on the sample traffic flow data and the traffic congestion status prediction model; the currently arranged monitoring devices are determined based on a plurality of preset monitoring device layout information or the current action data; the current action data includes action sequences corresponding to adjusted monitoring device layout information; the adjusted monitoring device layout information includes enabling any monitoring device and disabling any monitoring device;

[0140] A state data determination unit is configured to determine subsequent state data according to the current action data;

[0141] A reward determination unit is configured to determine the current reward according to the subsequent state data and the current state data in combination with the reward function;

[0142] A model training unit, configured to use the subsequent state data as the current state data, and input the current state data, the current action data, and the current reward data into the monitoring device position prediction model to be trained to obtain subsequent action data; use the subsequent action data as the current action data, and jump to the unit for determining according to the state data until a preset convergence condition is met.

[0143] In an optional embodiment, the monitoring device position prediction model to be trained includes a preliminary feature extraction module, a feature position marking module, a deep feature extraction module, and a monitoring device position prediction module. The model training unit includes:

[0144] A preliminary feature extraction unit, configured to input the current state data, the current action data, and the current reward data into the preliminary feature extraction module for preliminary feature extraction processing to obtain preliminary combined features;

[0145] A feature position marking unit, configured to input the preliminary combined features into the feature position marking module for feature position marking processing to obtain ordered combined features;

[0146] A deep feature extraction unit, configured to input the ordered combined features into the deep feature extraction module for deep feature extraction processing to obtain high-level combined features;

[0147] A monitoring device position determination unit, configured to input the high-level combined features into the monitoring device position prediction module for monitoring device position prediction processing to obtain subsequent action data.

[0148] In an optional embodiment, the device includes a monitoring device position prediction model determination module to be trained, specifically configured to

[0149] Set the objective function Z of the monitoring device position prediction model to be trained as shown in the following formula:

[0150] ;

[0151] Wherein, represents the influence of deploying a monitoring device at the i-th monitoring device position on the accuracy of the traffic congestion state prediction model; represents the cost of deploying a monitoring device at the i-th monitoring device position; whether to deploy a monitoring device at the i-th monitoring device position; represents a weight parameter for measuring the relationship between the decision-making support effect and the cost of enabling the emergency lane; n represents the number of positions where monitoring devices are to be deployed;

[0152] The constraint conditions of the objective function are:

[0153] , ;

[0154] ;

[0155] , ;

[0156] Among them, represents the set of positions of the monitoring devices to be arranged in the target area covering the sample section of the monitoring devices to be arranged; m represents the number of target areas of the sample section of the monitoring devices to be arranged; B represents the preset cost of arranging the monitoring devices.

[0157] In an alternative embodiment, the target monitoring device position prediction model construction module includes a reward function construction unit for

[0158] ;

[0159] Among them, represents the change in the sample traffic flow data between the successor state data and the current state data, represents the change in the sample traffic congestion state between the successor state data and the current state data, represents the change in the accuracy of the traffic congestion state prediction model between the successor state data and the current state data, represents the change in the preset cost between the successor state data and the current state data, , , and respectively represent the first preset coefficient, the second preset coefficient, the third preset coefficient, and the fourth preset coefficient of the reward function.

[0160] In an alternative embodiment, the device includes a traffic congestion state prediction model construction module for constructing a traffic congestion state prediction model;

[0161] The traffic congestion state prediction model construction module includes:

[0162] A data acquisition unit for acquiring the traffic flow sample data collected by the monitoring devices of the current sample section and the preset traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample section;

[0163] A traffic congestion state prediction unit for inputting the traffic flow sample data into the traffic congestion state prediction model to be trained to obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring devices of the current sample section;

[0164] A traffic congestion state prediction model training unit is used to train the traffic congestion state prediction model to be trained based on the preset traffic congestion state and the predicted traffic congestion state, so as to obtain a traffic congestion state prediction model.

[0165] In an optional embodiment, the traffic congestion state prediction model to be trained includes: a clustering module, a gating module, and a plurality of prediction modules;

[0166] The traffic congestion state prediction unit is specifically configured to

[0167] Input the traffic flow sample data into the clustering module for partitioning processing to obtain a data partitioning result;

[0168] Input the data partitioning result into the gating module to determine the weight of each prediction module;

[0169] Input the data partitioning result into the corresponding prediction module according to the weight of each prediction module to obtain the traffic congestion state prediction result of each prediction module;

[0170] Based on the weight of each prediction module, perform weighted summation processing on the traffic congestion state prediction results of each prediction module to obtain a predicted traffic congestion state.

[0171] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment related to the method, and will not be elaborated here.

[0172] An embodiment of the present invention further provides an electronic device, which includes: a processor and a memory. At least one instruction, at least one program, a code set, or an instruction set is stored in the memory. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the method for determining the position of a monitoring device as described in any one of the method embodiments.

[0173] An embodiment of the present invention further provides a computer storage medium, which can be set in a server to store at least one instruction, at least one program, a code set, or an instruction set for implementing the method embodiment. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the method for determining the position of a monitoring device as described in any one of the method embodiments.

[0174] Optionally, in an embodiment of the present invention, the above storage medium may be located in at least one of multiple network servers of a computer network. Optionally, in an embodiment of the present invention, the above storage medium may include, but is not limited to: various media that can store program codes such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), external hard drives, magnetic disks, or optical discs.

[0175] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0176] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flowcharts and / or blocks Figure 1 one block or multiple blocks.

[0177] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flowcharts and / or blocks Figure 1 one block or multiple blocks.

[0178] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flowcharts and / or blocks Figure 1 one block or multiple blocks.

[0179] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions noted in the blocks may occur in a different order than that noted in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0180] Finally, it should be noted that the embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope of the present invention as protected by the claims. All of these are within the protection scope of the present invention.

Claims

1. A method for determining the location of a monitoring device, characterized in that: The method comprises: Acquire traffic flow data collected by at least one existing monitoring device on a target road section, wherein the target road section is a road section area where the monitoring device is to be arranged; Inputting the traffic flow data into a traffic congestion state prediction model to determine the traffic congestion state of each area corresponding to an existing monitoring device; Inputting the arrangement information of the at least one existing monitoring device, the traffic flow data and the traffic congestion status into a target monitoring device location prediction model to obtain location information of the monitoring device to be arranged; The target monitoring device position prediction model is obtained by training the monitoring device position prediction model to be trained based on the state space, action space and reward function; the state space includes multiple state data, and the multiple state data include the layout information of multiple deployed monitoring devices of the sample road section to be deployed with monitoring devices, multiple sample traffic flow data and multiple sample traffic congestion states; the multiple sample traffic flow data are obtained based on the layout information of the multiple deployed monitoring devices, and the multiple sample traffic congestion states are obtained based on the multiple sample traffic flow data and the traffic congestion state prediction model; the action space includes multiple action data, and the multiple action data include multiple action sequences corresponding to adjusting the monitoring device layout information; the reward function is used to evaluate the quality of the action data.

2. The method for determining the location of a monitoring device according to claim 1, characterized in that: The target monitoring device location prediction model is constructed in the following way: Input the current state data into the monitoring equipment position prediction model to be trained to generate current action data; the current state data includes the layout information of the currently deployed monitoring equipment on the sample road section to be deployed with the monitoring equipment, the current sample traffic flow data and the current sample traffic congestion status; The current sample traffic flow data is obtained based on the arrangement information of the currently arranged monitoring equipment, and the current sample traffic congestion state is obtained based on the sample traffic flow data and the traffic congestion state prediction model; the currently arranged monitoring equipment is determined based on a plurality of preset monitoring equipment arrangement information or the current action data; the current action data includes an action sequence corresponding to the adjustment of the monitoring equipment arrangement information; the adjustment of the monitoring equipment arrangement information includes enabling any monitoring equipment and disabling any monitoring equipment; Determining subsequent state data according to the current action data; Determine a current reward based on the subsequent state data and the current state data in combination with a reward function; Using the subsequent state data as the current state data, inputting the current state data, the current action data and the current reward data into the monitoring device position prediction model to be trained, to obtain subsequent action data; The subsequent action data is used as the current action data, and the process jumps to determining the subsequent state data according to the current action data until a preset convergence condition is met.

3. The method for determining the location of a monitoring device according to claim 2, characterized in that: The monitoring device position prediction model to be trained includes a preliminary feature extraction module, a feature position marking module, a deep feature extraction module and a monitoring device position prediction module. The current state data, the current action data and the current reward data are input into the monitoring device position prediction model to be trained to obtain subsequent action data, including: Inputting the current state data, the current action data and the current reward data into the preliminary feature extraction module for preliminary feature extraction processing to obtain preliminary joint features; Inputting the preliminary joint features into the feature position marking module for feature position marking processing to obtain ordered joint features; Inputting the ordered joint features into the deep feature extraction module for deep feature extraction processing to obtain high-level joint features; The high-level joint features are input into the monitoring device position prediction module to perform monitoring device position prediction processing to obtain subsequent action data.

4. The method for determining the location of a monitoring device according to claim 2, characterized in that: The method comprises: The objective function Z of the monitoring device position prediction model to be trained is set as shown in the following formula: ； in, represents the influence of arranging the monitoring device at the i-th monitoring device position on the accuracy of the traffic congestion state prediction model; represents the cost of deploying monitoring equipment at the i-th monitoring equipment location; Whether to arrange monitoring equipment at the i-th monitoring equipment position; represents the weight parameter used to measure the relationship between the emergency lane activation decision support role and cost; n represents the number of locations where monitoring equipment is to be arranged; The constraints of the objective function are: ， ； ； ， ； in, represents the set of locations of monitoring equipment to be deployed covering the target area of ​​the sample road section where monitoring equipment is to be deployed; m represents the number of target areas of the sample road section where monitoring equipment is to be deployed; B represents the preset cost of deploying monitoring equipment.

5. The method for determining the location of a monitoring device according to claim 2, characterized in that: The reward function is constructed in the following way: ； Among them, represents the change of the sample traffic flow data between the subsequent state data and the current state data, represents a change in the sample traffic congestion state between the subsequent state data and the current state data, represents the change in the accuracy of the traffic congestion state prediction model between the subsequent state data and the current state data, represents a change in a preset cost between the subsequent state data and the current state data, , , and They respectively represent the first preset coefficient, the second preset coefficient, the third preset coefficient and the fourth preset coefficient of the reward function.

6. The method for determining the location of a monitoring device according to claim 1, characterized in that: The traffic congestion state prediction model is constructed in the following way: Obtaining traffic flow sample data collected by the monitoring equipment of the current sample road section, and a preset traffic congestion state of the monitoring area corresponding to the monitoring equipment of the current sample road section; Inputting the traffic flow sample data into the traffic congestion state prediction model to be trained to obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring equipment of the current sample road section; Based on the preset traffic congestion state and the predicted traffic congestion state, the traffic congestion state prediction model to be trained is trained to obtain a traffic congestion state prediction model.

7. The method for determining the location of a monitoring device according to claim 6, characterized in that: The traffic congestion state prediction model to be trained includes: a clustering module, a gating module and multiple prediction modules; The step of inputting the traffic flow sample data into the traffic congestion state prediction model to be trained to obtain the predicted traffic congestion state of the monitoring area corresponding to the monitoring equipment of the current sample road section includes: Inputting the traffic flow sample data into the clustering module for partitioning processing to obtain a data partitioning result; According to the data partitioning result, the data is input into the gating module to determine the weight of each prediction module; Inputting the data division result into the corresponding prediction module according to the weight of each prediction module to obtain the traffic congestion state prediction result of each prediction module; Based on the weight of each prediction module, the traffic congestion state prediction results of each prediction module are weighted and summed to obtain the predicted traffic congestion state.

8. A device for determining the position of a monitoring device, characterized in that: The device comprises: A data acquisition module, used to acquire traffic flow data collected by at least one existing monitoring device on a target road section, wherein the target road section is a road section area where a monitoring device is to be arranged; A state determination module, used for inputting the traffic flow data into a traffic congestion state prediction model to determine the traffic congestion state of each area corresponding to an existing monitoring device; A position determination module is used to input the layout information of at least one existing monitoring device, the traffic flow data and the traffic congestion status into a target monitoring device position prediction model to obtain the position information of the monitoring device to be deployed; the target monitoring device position prediction model is based on the state space, action space and reward function, and is obtained by training the monitoring device position prediction model to be trained; the state space includes a plurality of state data, and the plurality of state data include the layout information of a plurality of deployed monitoring devices of a sample road section where the monitoring device is to be deployed, a plurality of sample traffic flow data and a plurality of sample traffic congestion status; the plurality of sample traffic flow data are obtained based on the layout information of the plurality of deployed monitoring devices, and the plurality of sample traffic congestion status are obtained based on the plurality of sample traffic flow data and the traffic congestion status prediction model; the action space includes a plurality of action data, and the plurality of action data include a plurality of action sequences corresponding to adjusting the layout information of the monitoring device; the reward function is used to evaluate the quality of the action data.

9. An electronic device, comprising a processor and a memory, wherein the memory stores at least one instruction and at least one program, and the at least one instruction and the at least one program are loaded and executed by the processor to implement the method for determining the location of a monitoring device as described in any one of claims 1 to 7.

10. A computer storage medium, wherein at least one instruction and at least one program are stored in the computer storage medium, wherein the at least one instruction and the at least one program are loaded and executed by a processor to implement the method for determining the position of a monitoring device as described in any one of claims 1 to 7.

Images