Path determination method and device, equipment, medium and product

Abstract

The invention discloses a path determination method and device, equipment, a medium and a product, and is applied to the field of information technology support. The method comprises the following steps: firstly, generating all possible path sets by adopting a genetic algorithm, and updating paths in the path sets based on a preset genetic rule to obtain a target path set; and then determining a target offline path with the maximum length weight sum by adopting an ant colony algorithm. According to the method, the time spent on completing each target task and the value of the client and the task are considered, and more reasonable path recommendation is provided for the target object. And in the path determination process, the genetic algorithm and the ant colony algorithm are linked through the neural network. Firstly, based on a genetic algorithm, initial pheromone distribution is generated for an ant colony algorithm by utilizing the characteristics of high early-stage global search performance and high convergence speed of the genetic algorithm; and path finding and path pheromone updating are carried out by using an ant colony algorithm, and finally a target offline path is determined.

Description

Technical Field

[0001] The present application belongs to the field of information technology support, and in particular, relates to a path determination method, device, equipment, medium and product. Background Art

[0002] In actual scenarios, staff need to perform some assigned target tasks offline, such as group visits, store inspections, marketing stalls, and building sweeps, etc. In order to improve work efficiency, staff (target objects) need to plan the execution order of target tasks in advance and complete path planning.

[0003] At present, the path planning for the target object to perform the target task is to connect the positions of each task in the order in which the target task is generated, thereby forming a path planning. However, this path planning is based on a single piece of information and does not consider other influencing factors in the actual situation. Unreasonable path planning leads to a poor experience for the target object. Summary of the invention

[0004] The embodiments of the present application provide a path determination method, device, equipment, medium and product, which can provide more reasonable path planning.

[0005] On the one hand, an embodiment of the present application provides a path determination method, including:

[0006] Acquire each target task to be performed in the target time period and target information required to perform the target task; the target information at least includes: task information and / or road network information;

[0007] Based on the target information, determining the length weight between two target positions; the target positions include the task position and the initial position of the target object; the length weight at least satisfies any one of the following or any combination: negatively correlated with the travel time between the two target positions, positively correlated with the value of the customer, and positively correlated with the value of the target task;

[0008] A genetic algorithm is used to generate a set of all possible paths according to the target task and the road network information; the paths in the path set correspond to the chromosomes of the genetic algorithm;

[0009] Based on a preset genetic rule, the paths in the path set are updated to obtain a target path set; the preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path;

[0010] Based on the target path set, an ant colony algorithm is used to determine a maximum length weight and a corresponding target offline path.

[0011] On the other hand, updating the paths in the path set based on the preset genetic rule to obtain the target path set includes:

[0012] Based on the preset genetic rule, the paths in the path set are iteratively updated multiple times according to the following operations until the first iteration stop condition is met to obtain the target path set; each update includes at least any one of the following items or any combination:

[0013] Based on the path with the highest score in the path set last time, the path with the lowest score in the path set last time is replaced; before each update, the fitness function is used to score each path in the path set to obtain a score value; the score value is positively correlated with the length weight sum of the path;

[0014] Based on the intersection of the task location, combining multiple paths in the path set from last time;

[0015] A preset number of paths in the previous path set are adjusted with a preset probability.

[0016] On the other hand, the first iteration stopping condition includes any one or any combination of the following:

[0017] The number of iterations reaches the maximum preset number of iterations;

[0018] The number of iterations is greater than the minimum preset number of iterations, and the number of consecutive iteration stagnations reaches the maximum preset number of stagnations; when the iteration stagnation occurs, it indicates that the iterative evolution rate is less than the minimum evolution rate.

[0019] On the other hand, the method of determining the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm includes:

[0020] Determining a reference pheromone corresponding to the ant colony algorithm based on each path in the target path set;

[0021] Based on the reference pheromone, determining the target pheromone of the ant colony algorithm for each path;

[0022] Based on the pheromone update rule, updating the target pheromone until the second iteration stop condition is met;

[0023] Based on the updated target pheromone, the maximum length weight and the corresponding target offline path are determined.

[0024] On the other hand, after determining the length weights between each pair of target positions based on the target information, the method further includes:

[0025] Based on each of the length weights and each of the target positions, a directed graph is constructed to obtain a full path environment map; the nodes in the full path environment map are the target positions, and the lines between each two of the nodes are assigned corresponding length weights.

[0026] On the other hand, after the maximum length weight and the corresponding target offline path are determined based on the target path set by using an ant colony algorithm, the method further includes:

[0027] Obtain the task completion rate under the corresponding weather and / or time period;

[0028] When the task completion rate is lower than a threshold, determining that the target task is not suitable for execution;

[0029] Output a prompt message that the target task is not suitable for execution.

[0030] On the other hand, after the maximum length weight and the corresponding target offline path are determined based on the target path set by using an ant colony algorithm, the method further includes:

[0031] In the case where the target task changes within the target time period, determining the task cost value corresponding to each of the task locations that the target object has not visited based on the corresponding target information; the task cost value is determined based at least on the target travel time required to reach the task location and / or the corresponding task importance;

[0032] Based on each of the task cost values, a target real-time path of the task location that the target object has not visited is re-determined.

[0033] On the other hand, the task cost is positively correlated with the target travel time and negatively correlated with the task importance;

[0034] The determining, based on the corresponding target information, the task cost value corresponding to each of the task locations that the target object has not visited, and the re-determining, based on each of the task cost values, the target real-time path of the task location that the target object has not visited, includes:

[0035] Determine each first task cost value between the current location of the target object and each task location that has not been visited;

[0036] Determine the task position corresponding to the minimum first task cost value as the next target task position in the target real-time path;

[0037] For each target task position, the following operations are performed until all the target task positions of the target real-time path are determined:

[0038] Determine each second task cost value between the current target task position and each remaining task position;

[0039] The task position corresponding to the smallest second task cost value is determined as the next target task position in the target real-time path.

[0040] On the other hand, the task information includes at least any one or any combination of the following: target object information, task type, human traffic situation, reservation situation of the target task, customer value information, value information of the target task, and estimated execution time of the target task to be completed;

[0041] The road network information includes at least any one or any combination of the following: task location, task destination, transportation mode, travel time and road congestion level.

[0042] On the other hand, the target information at least includes task information, and the task information at least includes: customer value information, value information of the target task, flow of people, and execution time of the target task;

[0043] Acquiring the target information required to perform the target task includes:

[0044] Using the hierarchical analysis method, the weight of each customer indicator data is calculated, and the customer value score is determined by weighted calculation to obtain the customer's value information; the customer indicator data includes at least any one of the following or any combination: customer account opening attribute data, customer historical consumption data, customer historical order data, and customer potential customer portrait data;

[0045] Using the hierarchical analysis method, the weight of each task indicator data is calculated, and the task value score is determined by weighted calculation to obtain the value information of the target task; the task indicator data includes at least any one of the following or any combination: expected task benefit, task type, urgency and customer relationship;

[0046] Based on the execution status of completed tasks, the k-means clustering algorithm is used to determine the flow of people; the execution status includes at least any one or any combination of the following: execution time, execution weather, whether it is a holiday, whether it is a weekend, whether it is a social event, and execution flow of people;

[0047] Based on the types and execution times of completed tasks, a k-means clustering algorithm is used to determine the estimated execution time of the target task to be completed.

[0048] On the other hand, an embodiment of the present application provides a path determination device, the device comprising:

[0049] An acquisition module, used to acquire each target task to be performed in a target time period and target information required to perform the target task; the target information at least includes: task information and / or road network information;

[0050] A determination module is used to determine the length weight between two target positions based on the target information; the target positions include the task position and the initial position of the target object; the length weight satisfies at least any one of the following or any combination: negatively correlated with the travel time between the two target positions, positively correlated with the value of the customer, and positively correlated with the value of the target task;

[0051] A generation module, used to generate a set of all possible paths according to the target task and the road network information by using a genetic algorithm; the paths in the path set correspond to the chromosomes of the genetic algorithm;

[0052] An updating module, used for updating the paths in the path set based on a preset genetic rule to obtain a target path set; the preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path;

[0053] The determination module is further used to determine the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm;

[0054] The judging module is used to judge whether the target offline path meets the preset requirements; if the target offline path does not meet the preset requirements, trigger the updating module until the target offline path meets the preset requirements.

[0055] In yet another aspect, an embodiment of the present application provides a path determination device, the device comprising: a processor and a memory storing computer program instructions;

[0056] When the processor executes the computer program instructions, the path determination method as described above is implemented.

[0057] On the other hand, an embodiment of the present application provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the path determination method described above is implemented.

[0058] On the other hand, an embodiment of the present application provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the path determination method as described above.

[0059] A path determination method in an embodiment of the present application obtains each target task to be executed and the required target information, and determines the length weight between each target position based on the target information. Then, a genetic algorithm is used to generate all possible path sets according to the target task and the road network information, and the paths in the path set are updated based on the preset genetic rules to obtain the target path set. Then, based on the target path set, an ant colony algorithm is used to determine the maximum length weight and the corresponding target offline path, and determine whether the target offline path meets the preset requirements; if not, the target path set is reconstructed until the target offline path meets the preset requirements. The embodiment of the present application plans the path, taking into account the time spent on completing each target task and the value of the customer and the task, so as to determine the optimal target offline path and provide a more reasonable path recommendation for the target object. And in the process of path determination, the genetic algorithm and the ant colony algorithm are finally linked through a neural network. First, based on the genetic algorithm, the initial pheromone distribution is generated for the ant colony algorithm by utilizing the characteristics of the genetic algorithm's strong global search and fast convergence speed in the early stage; then the ant colony algorithm is used to find the path and update the path pheromone, and finally the target offline path is determined. It can be seen that this scheme can efficiently and accurately determine the appropriate target offline path. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0061] Figure 1 A schematic diagram of a process flow of a path determination method provided by an embodiment of the present application is shown;

[0062] Figure 2 An overall flow chart of offline path planning provided in an embodiment of the present application;

[0063] Figure 3 An overall flow chart of real-time path planning provided by an embodiment of the present application;

[0064] Figure 4 This is a schematic diagram of the personal customer value assessment system;

[0065] Figure 5 This is a schematic diagram of the value assessment system for government and enterprise customers;

[0066] Figure 6 This is a schematic diagram of the peak flow recognition model;

[0067] Figure 7 A specific environmental modeling rendering provided in an embodiment of the present application;

[0068] Figure 8 A schematic diagram of the entire process of an offline path planning model provided in an embodiment of the present application;

[0069] Fig. 9 A schematic diagram of the structure of a path determination device provided in an embodiment of the present application is shown;

[0070] Fig.10 A schematic diagram of the hardware structure of a path determination device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0071] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0072] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "include..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0073] In order to solve the problems of the traditional solution, the embodiments of the present application provide a path determination method, device, equipment, medium and product. The path determination method provided by the embodiments of the present application is first introduced below. Figure 1 FIG. 1 is a flow chart of a path determination method provided by an embodiment of the present application; Figure 1 As shown, the method includes the following steps: S101 to S105.

[0074] S101: Acquire target tasks to be performed in a target time period and target information required to perform the target tasks.

[0075] Before the target time period, the target offline path of each target task to be executed in the target time period needs to be planned in advance, so as to recommend the optimal path for the target object. The specific range of the target time period is not limited and is determined according to actual needs. Generally, the target offline path is output in days.

[0076] Before planning a route, you need to first determine the target tasks that need to be performed within the target time period, and obtain the target information required to perform the target tasks, and then you can plan the route for the corresponding target tasks. The specific content of the target information is obtained according to actual needs, and at least includes: task information and / or road network information. Task information is related to the target task to be performed, including target object information, task type, traffic flow, target task reservation status, customer value information, target task value information, and the estimated execution time of the target task to be completed, etc. The road network information is related to the actual task location, task end point, transportation method, travel time, and road congestion.

[0077] S102: Determine the length weight between each pair of target positions based on the target information.

[0078] After obtaining the target information, it is necessary to determine the length weight between the target positions based on the target information. The target position includes the task position and the initial position of the target object.

[0079] The closer the target task is, the higher the priority of execution should be; the higher the value of the customer and the value of the target task are, the higher the priority of processing should be. Therefore, the length weight satisfies at least one of the following or any combination: negatively correlated with the travel time between the two target locations, positively correlated with the value of the customer, and positively correlated with the value of the target task.

[0080] Through the road network information in the target information obtained above, the travel time between any two target locations can be calculated; through the customer's value information and the target task's value information in the task information, the corresponding customer's value and the corresponding target task's value can be obtained.

[0081] S103: Genetic algorithm is used to generate a set of all possible paths according to the target task and road network information.

[0082] S104: Based on the preset genetic rule, the paths in the path set are updated to obtain a target path set.

[0083] The paths in the path set correspond to the chromosomes of the genetic algorithm. By using the genetic algorithm, the paths are represented as chromosomes, which can imitate the globality, randomness, and robustness of the natural process. And a suitable fitness function can be selected to evaluate the superiority of the chromosome (i.e., the path), that is, to determine the length weight and the path, so as to replace the path in the path set with a more suitable path.

[0084] Specifically, the paths in the path set are iteratively updated by a preset genetic rule to obtain a target path set. The preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path.

[0085] S105: Based on the target path set, an ant colony algorithm is used to determine a maximum length weight and a corresponding target offline path.

[0086] S106: Determine whether the target offline path meets the preset requirements; if not, return to S104 until the target offline path meets the preset requirements.

[0087] After determining the target path set, the maximum length weight and the corresponding target offline path can be quickly determined by further using the ant colony algorithm. The ant colony algorithm can quickly find the shortest path by updating pheromones. Because the ant colony algorithm lacks pheromones in the early stage, the convergence speed is slow. Therefore, in the embodiment of the present application, the target pheromone of each path can be determined by the target path set, skipping the pheromone shortage stage of the ant colony algorithm in the early stage, and directly entering the high convergence speed stage. Through the ant colony algorithm, the optimal path of the final task, that is, the target offline path, can be quickly locked from the optimal path set.

[0088] If the obtained target offline path does not meet the preset requirement, the process returns to S104 until the target offline path meets the preset requirement. The preset requirement can be set according to actual needs, and it is necessary to ensure that the final target offline path meets the actual needs.

[0089] The following is a specific implementation method of offline path planning: Figure 2 An overall flow chart of an offline path planning provided in an embodiment of the present application; Figure 2 As shown, the following steps are included:

[0090] S201: Determine whether it is suitable for execution based on the task execution weather, task appointment status and task time.

[0091] S202: Abstract the task location and task path into a weighted directed graph to obtain a global road network map.

[0092] S203: Obtain a set of target paths with the greatest travel value through a genetic algorithm.

[0093] S204: Using an ant colony algorithm, find a target offline path with the greatest travel value in the target path set.

[0094] S205: Determine whether the target offline path meets the end condition; if not, return to S203; if so, enter S206.

[0095] S206: Output the current target offline path.

[0096] It can be seen that through the genetic algorithm, a set of target paths with the largest pass value is obtained, and then the target offline path with the largest pass value is found in the target path set through the ant colony algorithm. The pass value can be determined based on the length weight and. At this time, if the obtained target offline path does not meet the preset requirements, it returns to the genetic algorithm stage, reconstructs the target path set, and then re-determines a new target offline path. If the target offline path meets the preset requirements, it can be directly recommended to the target object.

[0097] In addition, in practical applications, there may be a situation where the task is changed. Therefore, this embodiment provides a target real-time path planning method when the task is changed; Figure 3 An overall flow chart of real-time path planning provided by an embodiment of the present application; Figure 3 As shown, the following steps are included:

[0098] S301: Setting the valuation function.

[0099] S302: Obtain an evaluation function between the initial position and all remaining task positions, select the point with the minimum cost, and obtain the next target task position.

[0100] S303: Obtain an evaluation function between the current target task position and all remaining task positions, select the point with the minimum cost, and obtain the next target task position until the target real-time path is obtained.

[0101] Here, the next target task location is selected once through the evaluation function, so as to obtain the complete target real-time path. The calculation process is simple and efficient, which is suitable for temporary emergency processing.

[0102] A path determination method in an embodiment of the present application obtains each target task to be executed and the required target information, and determines the length weight between each target position based on the target information. Then, a genetic algorithm is used to generate all possible path sets according to the target task and the road network information, and the paths in the path set are updated based on the preset genetic rules to obtain the target path set. Then, based on the target path set, an ant colony algorithm is used to determine the maximum length weight and the corresponding target offline path, and determine whether the target offline path meets the preset requirements; if not, the target path set is reconstructed until the target offline path meets the preset requirements. The embodiment of the present application plans the path, taking into account the time spent on completing each target task and the value of the customer and the task, so as to determine the optimal target offline path and provide a more reasonable path recommendation for the target object. And in the process of path determination, the genetic algorithm and the ant colony algorithm are finally linked through a neural network. First, based on the genetic algorithm, the initial pheromone distribution is generated for the ant colony algorithm by utilizing the characteristics of the genetic algorithm's strong global search and fast convergence speed in the early stage; then the ant colony algorithm is used to find the path and update the path pheromone, and finally the target offline path is determined. It can be seen that this scheme can efficiently and accurately determine the appropriate target offline path.

[0103] Taking a specific implementation as an example, the scheme of the embodiment of the present application is divided into four parts according to the time of task issuance and the needs of front-line personnel: the first part is the construction of the pre-model, the main purpose of which is to build a pre-model and provide input data for subsequent model construction. The second part is an offline path planning scheme, which is based on the tasks pre-planned to be completed before the task day, and uses a genetic and ant colony fusion model based on a neural network for offline path planning for this part of the task. The third part is a real-time path planning scheme, which can solve the path planning scenarios in the case of temporary additions, changes and cancellations on the task day, supports front-line personnel to independently select target tasks, and uses the A-STAR model for real-time path planning. The fourth part is the output of the scheme. According to the output of the offline path planning in the second part or the output of the real-time path planning in the third part, the task sequence and route are marked on the map, and the front-line personnel can get the task execution sequence and path suggestions on the work page.

[0104] First, the first part of the above text is explained. As mentioned above, the target information includes task information and / or road network information. The task information includes at least any one of the following or any combination: target object information, task type, traffic flow, target task reservation, customer value information, target task value information, and the estimated execution time of the target task to be completed. Road network information includes at least any one of the following or any combination: task location, task destination, transportation mode, travel time, and road congestion. With this information, the target path can be determined faster and more accurately.

[0105] Most of the above-mentioned task information and road network information can be directly obtained, but the customer's value information, the value information of the target task, the flow of people, and the execution time of the target task need to be obtained by building a pre-model. The specific process of obtaining the four types of information is as follows:

[0106] First, obtain the customer's value information. Use the hierarchical analysis method to calculate the weight of each customer indicator data, and determine the customer value score through weighted calculation to obtain the customer's value information; customer indicator data includes at least any one of the following or any combination: customer account opening attribute data, customer historical consumption data, customer historical ordering data, and customer potential customer portrait data.

[0107] Specifically, in order to support the fitness function calculation of the genetic algorithm in the offline path planning algorithm and the pheromone concentration calculation in the ant colony algorithm, a customer value model is constructed. When inputting the model, customers are divided into individual customers and government and enterprise customers. Figure 4 It is a schematic diagram of the individual customer value assessment system; Figure 4 As shown, it includes a first target layer 401 , a first criterion layer 402 and a first solution layer 403 . Figure 5 This is a schematic diagram of the government and enterprise customer value assessment system; Figure 5 As shown, it includes a second target layer 501, a second criterion layer 502 and a second solution layer 503. Figure 4 as well as Figure 5 The customer value is evaluated from several aspects. The indicator data of four dimensions, namely, customer account opening attribute data in the operator's database, customer historical consumption data, customer historical ordering data, and potential customer portrait data based on the operator's understanding of the customer, are used. The hierarchical analysis method is used to calculate the weight of each indicator data, and the customer value score is calculated in a weighted manner to evaluate the value of the task target customer.

[0108] The algorithm of the customer value model is: calculate the weight of each indicator according to the hierarchical analysis method, multiply the weight by the weight, and the final score is the customer value score. The model output is: customer identification, customer type and value score.

[0109] Second, obtain the value information of the target task. Use the analytic hierarchy process to calculate the weight of each task indicator data, and determine the task value score through weighted calculation to obtain the value information of the target task; the task indicator data includes at least any one of the following or any combination: expected task benefits, task type, urgency, and customer relationship.

[0110] Specifically, in order to support the fitness function calculation of the genetic algorithm in the offline path planning algorithm and the pheromone concentration calculation in the ant colony algorithm, a task value model is constructed to evaluate the value of the task itself. The model will evaluate the task value from the perspectives of expected task benefits (based on the operator precipitation model estimation), task type, urgency (task deadline), and customer relationship.

[0111] The algorithm of the task value model calculates the weight of each indicator according to the hierarchical analysis method, and multiplies the weighted scores of all indicators by the weights. The final score is the task value score. Model output: Output task identification, task type, and task value score.

[0112] Third, obtain the flow of people. Based on the execution of completed tasks, the k-means clustering algorithm is used to determine the flow of people; the execution status includes at least any one of the following or any combination: execution time, execution weather, whether it is a holiday, whether it is a weekend, whether it is a social event, and execution flow of people.

[0113] Specifically, in order to support the screening of the best time rules for the path planning model, a peak passenger flow recognition model is constructed to identify the objective conditions corresponding to the peak passenger flow, provide a reference for the time arrangement of path planning, and associate the offline path planning model tasks as the output fields of the path planning model.

[0114] Figure 6 This is a schematic diagram of the peak flow recognition model; Figure 6 As shown in the figure, the peak flow recognition model is fed with the execution status of completed tasks: execution time, execution weather, whether it is a holiday, whether it is a weekend, whether it is a social activity, and the number of people executed. Introduction to the algorithm of the peak flow recognition model: The k-means clustering algorithm is used. The idea is to classify data according to different latitudes of the data. In this model, it is used to calculate the concentration of other data corresponding to the peak flow. Model output: task location, task type, time, weather, whether it is a weekend, whether it is a holiday, whether it is a social activity, and the mean of the flow of people.

[0115] Fourth, the execution time of the target task is obtained. Based on the type and execution time of the completed tasks, the k-means clustering algorithm is used to determine the estimated execution time of the target task to be completed.

[0116] In order to support the screening of the best time rules for the path planning model, a task duration recognition model is built to identify the average execution time of different task types, which serves as a reference for the time arrangement of path planning. The task type of the offline path planning model is associated and used as the output field of the path planning model.

[0117] The task duration recognition model inputs the execution status of completed tasks: task type and task duration. Introduction to the algorithm of the task duration recognition model: The k-means clustering algorithm is used. The idea is to classify data according to different dimensions. In this model, it is used to calculate the average task duration corresponding to each task type. Model output: task type and average execution time.

[0118] The embodiments of the present application can efficiently determine the corresponding target information through the above algorithms. The customer value information can be obtained through the customer value model, the value information of the target task can be obtained through the task value model, the peak traffic flow situation can be obtained through the peak traffic flow recognition model, and the execution time of the target task can be calculated through the task duration recognition model.

[0119] Then, the second part in the above text is explained.

[0120] In practical applications, the target offline path is generally output in days. The offline path planning scheme is based on the tasks planned and completed before the task day. It uses relevant information of the planned tasks in the operator's data system, historical data of completed tasks, historical data of task target customers, road network data, weather data and other indicators. According to the travel time, task value and customer value, the task itinerary is planned offline in advance to generate the optimal path.

[0121] First, based on the output of the pre-model and the information in the operator system data, a task information input table is formed, and the target object information, task execution information (type, location), weather, reservation, value information (task, customer) and recommended execution status (duration, time) are input. Based on the road network information, a road network information input table is formed, and the task start and end points, transportation methods, travel time and road congestion are input. The above two input tables are used as input data for this part.

[0122] When executing a task, it is necessary to ensure that the task is in the best execution condition. Therefore, after determining the maximum length weight and the corresponding target offline path based on the target path set and using the ant colony algorithm, the method further includes: obtaining the task completion rate under the corresponding weather and / or corresponding time period, and when the task completion rate is lower than the threshold, determining that the target task is not suitable for execution, and outputting a prompt message that the target task is not suitable for execution.

[0123] Determining whether the task is in the best execution condition, that is, rule filtering business logic, means formulating rules based on business logic to put the task in the best execution condition, which includes two conditions: best weather and best time.

[0124] Best weather rule: According to front-line execution experience and historical data, the transaction rate threshold is set, and the transaction rate of different task types and execution weather combinations is calculated. If it is lower than the threshold, it is considered unsuitable for execution. The hourly weather forecast is introduced from China Weather Network. After the rule is triggered, it will be recommended that the task is not suitable for execution now, and the corresponding prompt will be output.

[0125] Best time rules:

[0126] 1. Determine whether there are any customers with appointments on that day: If there are customers with appointments, the day is divided into N+1 segments (N: the number of tasks in the appointment time) according to the appointment time, and the model is converted into N path planning models with determined starting and ending points plus 1 path planning model with a determined starting point; if there are no customers with appointments, the model is a path planning model with an agreed starting point.

[0127] 2. Based on the front-line execution experience and historical data, analyze the transaction rates of different customer groups and different task methods in different time periods to form time rules. After the time rule is triggered, it will be recommended that the task is more suitable for execution now, and the corresponding prompt will be output. The time rules are formed as follows: Enterprise customer task time rule analysis: Cluster analysis of enterprise type, enterprise scale, task type, target business, visit time (weekday morning, weekday afternoon, weekday evening, non-working day) and transaction rate to find the best time rule. Personal user task time rule analysis: Cluster analysis of customer market type, whether it is a high-value customer, customer age (whether retired, no fixed job, etc.), task type, target business, visit time (weekday morning, weekday afternoon, weekday evening, non-working day) and transaction rate to find the best time rule. Group marketing task time rule analysis: Peak traffic time identified by the peak traffic identification model.

[0128] The embodiment of the present application determines whether the current time is suitable for executing the task through the task completion rate corresponding to the weather and time period, thereby avoiding waste of resources.

[0129] In practical applications, after determining the length weights between each target position based on the target information, a directed graph can be constructed based on each length weight and each target position to obtain a full path environment map; the nodes in the full path environment map are the target positions, and the lines between each node are assigned corresponding length weights.

[0130] The starting point (permanent office location), target task location, etc. in the road network information are abstracted as nodes, and the roads between the nodes are abstracted as lines. Based on the road network information and business understanding, this embodiment proposes a length weight function based on the travel time (the initial time v of the vehicle plus the weather impact factor c plus the congestion degree h), the task value, and the customer value:

[0131]

[0132] Among them, F time is the path travel time function, which is obtained by the initial time v of the vehicle, the weather impact factor c, and the congestion level h. value1 is the value of the current target task, F value2 is the value of the current task to the customer. α and β represent weight factors, α+β=1.

[0133] Figure 7 A specific environmental modeling effect diagram provided in the embodiment of the present application; Figure 7 As shown, the starting point is the starting position 701 of the target object. Three virtual cells are used as task positions, including the first cell 702, the second cell 703 and the third cell 704. The data between each point is the corresponding length weight; a is 15, b is 13, c is 25, d is 19, e is 33, and f is 9.

[0134] For a given weighted directed graph G(V,{E}), where V is a vertex set, E is a line set, and c(v,w) is the weight of line (v,w). st =(v0=v s v1=...,v n =v t ) is V by v s to v t The optimal path problem can be expressed as the following mathematical model:

[0135]

[0136] The embodiment of the present application improves the comprehensibility of path optimization by constructing a full path environment map.

[0137] In the above-mentioned embodiment, it is mentioned that the paths in the path set need to be updated based on the preset genetic rules to obtain the target path set. Because the present application adopts a genetic algorithm, it is necessary to take advantage of its strong search performance and fast convergence speed. Therefore, a specific iterative scheme is provided here:

[0138] First, based on the preset genetic rules, the paths in the path set are updated multiple times according to the following operations until the first iteration stop condition is met to obtain the target path set; each update includes at least any one of the following items or any combination: 1. Based on the path with the highest score in the previous path set, the path with the lowest score in the previous path set is replaced (selection operator); before each update, the fitness function is used to score each path in the path set to obtain a score value; the score value is positively correlated with the sum of the length weights of the path. 2. Based on the intersection of the task location, multiple paths in the previous path set are combined (crossover operator). 3. A preset number of paths in the previous path set are adjusted with a preset probability (mutation operator).

[0139] Specifically, through genetic algorithms, the paths are represented as chromosomes, imitating the globality, randomness and robustness of natural processes, and a suitable fitness function can be selected to evaluate the superiority of the chromosomes (i.e., roads).

[0140] The fitness function of this scheme is optimized based on business experience and is defined as a composite function of task value, customer value, and travel time function.

[0141] The genetic algorithm has a higher convergence speed to the optimal solution in the early stage (fewer chromosome divisions). Based on this characteristic, the genetic algorithm is used to perform early random global search, selection, crossover, and mutation to generate a set of solutions, that is, to generate a set of optimal solutions for a set of paths.

[0142] First, the initial population of the genetic algorithm is generated, and based on the above environmental modeling diagram, the algorithm is globally searched, that is, all possible paths are generated. Due to business rules, the number of tasks per day is generally less than or equal to 5. Therefore, the global search is not only less likely to fall into the local optimal solution, but also the number will not be too large to affect the running speed.

[0143] Then, a fitness function is designed. The fitness function represents the difficulty of an evolution and has two conditions: being accessible within the path range and keeping the length weight of the path as short as possible. In this embodiment, the total length of the path map lines is the shortest: the fitness function is the length of the path map lines.

[0144] Next, design genetic operators, including selection operator, crossover operator and mutation operator. Selection operator: select chromosomes with high fitness. This embodiment selects the elite strategy. The idea of ​​this strategy is that when the fitness value of the new generation is less than that of the previous generation, the best individuals of the previous generation replace the worst individuals of the new generation with the same number to ensure the continuity of the good chromosomes. Crossover operator: select chromosomes with high fitness with probability P. crossThe process of randomly combining two individuals of the previous generation to produce a new individual. This patent selects a single crossover point of random crossover points and randomly exchanges any task position. Mutation operator: The process of introducing new genes. This patent adopts the substitution mutation method to replace P in any task position except the current task position. variation Probability of random variation.

[0145] The embodiment of the present application updates the paths in the path set based on the updating rules of chromosomes by the genetic algorithm, so that the algorithm has the advantages of strong searchability and fast convergence speed.

[0146] The iterative process of the above path set will continuously optimize each path in the set, but a suitable iterative stop condition is required to ensure the validity of the path set while avoiding too long iteration time. Therefore, the first iterative stop condition includes any one or any combination of the following: 1. The number of iterations reaches the maximum preset number of iterations. 2. The number of iterations is greater than the minimum preset number of iterations, and the number of consecutive iterations that stagnate reaches the maximum preset number of stagnations; when the iteration stagnation indicates that the iterative evolution rate is less than the minimum evolution rate.

[0147] Specifically, in order to ensure the fusion point of the genetic algorithm and the ant colony algorithm, this embodiment uses a method of presetting the number of iterations, setting the maximum number of iterations N max , minimum number of iterations V min , the maximum number of stagnation times N stop , minimum evolution rate N rate The end conditions are divided into the following two cases: 1. The number of evolutions reaches the maximum number of iterations N. max 2. The number of evolutions is greater than the minimum number of iterations N min , and the number of consecutive stagnations reaches the maximum number of stagnations N stop , each stagnant evolution rate is less than the minimum evolution rate N rate If any of the above two conditions are met, the genetic algorithm stage will stagnate and enter the ant colony algorithm stage.

[0148] The embodiment of the present application provides specific content of a first iteration stop condition. The first iteration stop condition serves as a fusion point of a genetic algorithm and an ant colony algorithm, retaining the advantages of the genetic algorithm's strong early search ability and fast convergence speed, allowing the ant colony algorithm to skip the early pheromone scarcity stage and quickly determine the target offline path.

[0149] The above embodiment mentioned that after obtaining the target path set, it is necessary to use the ant colony algorithm based on the target path set to determine the maximum length weight and the corresponding target offline path. The embodiment of the present application provides a specific solution, first based on each path in the target path set, determine the reference pheromone corresponding to the ant colony algorithm. Then, based on the reference pheromone, determine the target pheromone of the ant colony algorithm for each path. Then, based on the pheromone update rule, update the target pheromone until the second iteration stop condition is met. Finally, based on the updated target pheromone, determine the maximum length weight and the corresponding target offline path.

[0150] Specifically, this step uses an ant colony algorithm to obtain the final optimal target offline path based on the optimal path set generated by the genetic algorithm.

[0151] The ant colony algorithm imitates the behavior of ants leaving pheromones on the paths they walk during the foraging process, connects the ants (task locations) in series, and uses pheromones (i.e. the length calculation formula of the path map line, which is also the fitness function of the genetic algorithm, a composite function of the task value, customer value and travel time function) for feedback, ultimately allowing the ant colony (task location) to approach the optimal solution.

[0152] The convergence speed of the ant colony algorithm in the later stage increases rapidly due to the early accumulation of pheromones. Based on this characteristic, the optimal solution set generated by the genetic algorithm in the early stage is used as the initial information point of the ant colony algorithm, so that the ant colony algorithm skips the early stage of pheromone scarcity and enters the high convergence speed stage. Through the ant colony algorithm, the optimal path for the final task is locked from the optimal path set. The steps of the ant colony algorithm are as follows:

[0153] Initial pheromone settings: Based on the ant colony algorithm, the pheromone settings are adjusted to facilitate the transition between the ant colony algorithm and the genetic algorithm. The initial pheromone settings are set as:

[0154] τ=τ0+τ G ;

[0155] Where τ0 is the initial pheromone of the ant colony algorithm, T G is a constant representing the pheromone converted from the result set of the genetic algorithm.

[0156] Movement probability calculation model: Initially, the pheromone concentration on the path is equal to a constant τ. All ants start from the starting point of the path, and the transfer probability of ant k moving from task position i to task position j at time t is:

[0157]

[0158] Among them A k represents the nodes that ant k can pass through; α is the information heuristic factor, β is the expected heuristic factor; η ijIndicates the expected degree of transfer from node i to node j, usually Among them, d ij Generally, it represents the distance from node i to node j, and here it represents the length weight between the two nodes.

[0159] The pheromone update model adopts a global iteration method, that is, only the ants that have experienced the shortest path in this iteration will update their pheromones:

[0160]

[0161] Among them: 0<ρ<1 is the pheromone volatility coefficient, Δτ ij (t) represents the increment of pheromone concentration after this cycle, Δτ ij k (t) represents the pheromone concentration left by the kth ant after this cycle, L k It represents the length of the path taken by the kth ant in this cycle, and Q is the amount of pheromone.

[0162] The embodiment of the present application obtains a point target path set through a genetic algorithm as a reference factor for the pheromone of the ant colony algorithm, thereby enabling the ant colony algorithm to skip the early pheromone scarcity stage and directly enter the high convergence speed stage through the determined target pheromone, thereby quickly determining the target offline path.

[0163] Using the adaptive fusion algorithm, the optimal path generated by the ant colony algorithm is operated in the following steps to achieve effective fusion with the genetic algorithm. The steps are as follows:

[0164] Step 1: Use the genetic algorithm to initialize a set of paths, that is, the optimal path set (parameter set of the neural network) generated in step 4.

[0165] Step 2: Feed the optimal path set into the neural network for evaluation and calculate its performance on the training set.

[0166] Step 3: Take all paths as parameters of the neural network, train it, and calculate the performance of the neural network on the validation set.

[0167] Step 4: Use the genetic algorithm to select the path, and based on its performance on the validation set, form the next generation population. Use the ant colony algorithm to select the path, and adjust the parameters of the neural network based on its performance on the validation set.

[0168] Step 5: Use genetic algorithm to perform crossover and mutation operations on the selected path to generate new paths, use ant colony algorithm to transfer information on the selected path, and adjust the parameters of the neural network.

[0169] Step 6: Add the new paths generated by genetic algorithm and ant colony algorithm to the path set.

[0170] Step 7: Repeat steps 3-6 until the second iteration stop condition is reached.

[0171] Figure 8 A schematic diagram of the entire process of an offline path planning model provided in an embodiment of the present application; Figure 8 As shown, the following steps are included:

[0172] S801: Environment construction.

[0173] This step is the preprocessing process.

[0174] S802: Genetic algorithm solution.

[0175] S803: Generate an initial optimal target path set.

[0176] S802 and S803 are the genetic algorithm stages.

[0177] S804: Ant colony algorithm solution.

[0178] S805: Neural network fusion.

[0179] S804 and S805 are the ant colony algorithm stages.

[0180] S806: Output the optimal target offline path.

[0181] The above step algorithm draws on the advantages of the two algorithms, increases the convergence speed, and makes the algorithm run faster. Linking the two independent algorithms through a neural network not only optimizes the time efficiency, but also improves the shortcomings of the algorithm being too random and having poor globality. That is, the neural network can optimize the path selection and provide a more comprehensive set of optimal paths.

[0182] Finally, the model results are output. By integrating the calculation of the fusion model with the results of the peak flow recognition model and the task duration recognition model, the target object information, task information, task sequence, recommended execution time, and recommended execution duration are output and updated in days.

[0183] The result table provides information about the target object and the tasks to be performed, and matches the execution order, recommended execution time, and execution duration of the task on the day. The output information of the model result table will be intelligently empowered to the target object workbench, provide a work schedule for the target object, and provide support for the visualization of the task location and intelligent planning map of the itinerary path output by the solution.

[0184] In the above part, the offline precise itinerary planning based on the adaptive neural network algorithm of genetics and ant colonies is carried out. The genetic algorithm has strong global search ability and fast convergence speed in the early stage to generate the initial pheromone distribution for the ant colony algorithm. The ant colony algorithm is used to find the path and update the path pheromone by combining the output results of the pre-data model such as task value, customer value, peak passenger flow, and high-precision data sources such as road network information and weather information. Finally, the two algorithms are linked through a neural network, which not only optimizes the time efficiency, but also improves the shortcomings of the algorithm being too random and having poor globality. That is, the neural network can optimize the path selection and give the optimal path recommendation.

[0185] Next, the third part in the above text is explained.

[0186] In the daily execution of tasks, there may be temporary additions, changes, and cancellations of tasks. Based on the above problems, the model supports the target object to autonomously select any task, and performs real-time path planning for the selected task based on the travel distance to generate the optimal path.

[0187] Therefore, this embodiment uses an ant colony algorithm based on the target path set to determine the maximum length weight and the corresponding target offline path. Then, when the target task changes within the target time period, the task cost value corresponding to each task location that the target object has not visited is determined based on the corresponding target information; the task cost value is determined based on at least the target travel time required to reach the task location and / or the corresponding task importance; then, based on each task cost value, the target real-time path of the task location that the target object has not visited is re-determined.

[0188] The embodiment of the present application re-determines the target real-time path of the task location that has not been visited through the task cost value, and efficiently provides a new recommended path to the target object when the target task changes within the target time period, thereby avoiding the situation where the target offline path is unavailable due to task changes, which in turn causes the target object to be unable to obtain a suitable recommended path.

[0189] A specific real-time target path determination solution is also provided here. In this embodiment, the task cost is positively correlated with the target travel time and negatively correlated with the task importance.

[0190] Based on the corresponding target information, determining the task cost value corresponding to each task location that the target object has not visited, and based on each task cost value, re-determining the target real-time path of the task location that the target object has not visited, including the following process:

[0191] First, the first task cost values ​​between the current position of the target object and the task positions that have not been visited are determined, and then the task position corresponding to the smallest first task cost value is determined as the next target task position in the target real-time path.

[0192] Next, for each target task position, the following operations are performed until all target task positions of the target real-time path are determined: the second task cost values ​​between the current target task position and the remaining task positions are determined. The task position corresponding to the smallest second task cost value is determined as the next target task position in the target real-time path.

[0193] The following is an explanation of a specific implementation method. First, corresponding target information is also required for real-time path planning. The real-time path planning model supports planning paths according to the real-time situation of the task. Based on the information in the operator system data, a task information input table is formed to input the target object information and the execution task information (type, location). Based on the road network information, a road network information input table is formed to input the starting and ending points of the task, the mode of transportation, and the travel time. The above two input tables are used as the input data for the third part.

[0194] Then, the cost function (ie, evaluation function) of the A-STAR algorithm is calculated. In this embodiment, the A-STAR algorithm is used for real-time path planning. The algorithm has simple steps and fast operation speed, and is suitable for real-time planning while quickly outputting results.

[0195] A-STAR is a combination of a heuristic algorithm (an algorithm that only considers the shortest path from the current point to the next point) and a depth-first search algorithm (which only searches for the global optimal path and may have no solution). It uses a path quality evaluation formula (i.e., cost function) to search for the next point with the highest cost.

[0196] The innovation of this solution is that based on the understanding of mobile business, the cost function is set as a composite function of task importance and travel time function. According to business understanding and actual road network information, the cost function is divided into two parts: one is the embodiment of task importance, which comes from the simple value ranking of task types; the other is the travel time. Cost function:

[0197]

[0198] Where: a is the current task location, b is the next task location, time(a,b) is the time between the two task locations: the time between any two points can be obtained based on the road network information (input data). value(b) is the importance of the next task location: the calculation method is the average value of the income / task duration of different task types. k is an adjustable constant.

[0199] The steps for generating the optimal target real-time path are as follows: Starting from the initial point (the current position of the target object), calculate the cost function between the initial point and all task positions. Then select the task position with the minimum cost function as the next target task position. Calculate the cost function between the current target task position and all remaining task positions; select the task position with the minimum cost function as the next target task position. Repeat the above steps until all task positions are traversed and path planning is completed.

[0200] Real-time fast itinerary planning based on the A-STAR algorithm is mainly used when there is a temporary change in the task during the execution of the task and the itinerary needs to be changed. In this way, the task information, road network information, transportation tools and other factors are used to construct a suitable valuation function, and the value maximization of "current value plus future valuation" is given priority for expansion, providing the optimal itinerary planning in real time.

[0201] The embodiment of the present application determines the target task position in sequence, and each time, uses the task position corresponding to the minimum task cost value as the next target task position, thereby determining the target real-time path in sequence. The recommended target real-time path has a certain reliability and a fast calculation speed.

[0202] Finally, according to the A-STAR algorithm, the target object information, execution task information, and target task ranking are output and updated on a daily basis.

[0203] The result table provides information about the tasks selected by the target object in real time, and outputs a recommended order for the selected tasks to be executed. The output information of the model result table is intelligently enabled to the target object workbench, which updates the target object's work schedule in real time and provides support for real-time task location visualization and intelligent planning of travel paths on the workbench.

[0204] Finally, the fourth part in the above text is explained.

[0205] The application based on the solution proposed in this embodiment accesses the task to-do list data of the target object, performs offline and real-time analysis on the work task information, and calculates the work sequence. The application fills the gap in task path planning and realizes the overall improvement of the target object's work efficiency.

[0206] Through the above technical solution, the recommended output is obtained for the business usage scenario of the target object. The front-line executives use the itinerary planning method through the page to obtain the following usage recommendations: the order of the work plan, that is, the order of execution of various tasks on the current working day, combined with the map display, to show the order of execution of each itinerary point to form a recommended path. Itinerary suggestions, showing itinerary recommendation information: including recommended itinerary execution time, task optimal execution time information and recommended execution time.

[0207] The scheme of this embodiment obtains the optimal solution for the parameter combination of each module of the algorithm through a large amount of experimental data in different path-finding environments. Finally, the superiority of the algorithm of this embodiment is proved through simulation comparison with ant colony algorithm, genetic algorithm and A-STAR algorithm under the same parameters.

[0208] Development and operating environment settings: The development environment is Python 3.7.16, and the operating environment is Mac M1 chip operating system, core CPU 8 and 16G memory. The performance comparison is shown in the following table:

[0209] Table 1 Parameter setting table

[0210]

[0211] The effect comparison is shown in the following table:

[0212] Table 2 Environment 1: Experimental data test results for 5 task locations and 10 routes

[0213]

[0214]

[0215] Table 3 Environment 2: Test results of experimental data for 10 routes at 3 task locations

[0216]

[0217] Table 4 Environment 3: Experimental data test results for 8 task locations and 10 routes

[0218]

[0219] The A-STAR algorithm is an algorithm with high computational efficiency in the current static environment and is widely used in the field of computer path planning. The improved A-STAR algorithm is still the algorithm with the best running time, but the effect is lower in the optimal planning of path length and task value. However, its superior computational efficiency can bear the real-time path planning with low data volume and simple manual adjustment. The fusion algorithm shows good results in the optimal combination of path length and task value, with a 10% improvement in path length effect, and a certain efficiency improvement, and a 5% reduction in running time.

[0220] This embodiment successfully solves the current situation of target objects running blindly, running according to experience, and low performance when performing target tasks through offline and real-time planning, improves the execution efficiency of front-line staff, and improves marketing conversion rate. The comparison of the results before and after the application of the technical solution shows that the average time for the target objects to perform tasks is greatly shortened after the application of the technical solution.

[0221] An offline itinerary planning scheme based on genetic and ant colony adaptive neural network algorithms; first, collect task information and road network information and integrate them into a unified data set, then build a neural network model that takes the task information and road network information as input and outputs the optimal itinerary planning scheme, then initialize and iterate the genetic algorithm and ant colony algorithm, integrate the genetic algorithm and ant colony algorithm into the neural network model to form an adaptive neural network algorithm (the genetic algorithm is responsible for global search, the ant colony algorithm is responsible for local search, and the two cooperate with each other), jointly optimize the weights of the neural network model, and finally input the task information and road network information into the adaptive neural network algorithm to obtain the optimal itinerary planning scheme, and output the scheme to the target object for reference.

[0222] A real-time path planning solution based on the A-STAR algorithm; Solution input: form a task information input table based on the information in the operator system data, and form a road network information input table based on the road network information; Calculate the cost function of the A-STAR algorithm: The real-time path planning solution uses the A-STAR algorithm, which has simple algorithm steps and fast running speed, suitable for real-time planning while quickly outputting results. Optimal path generation: Calculate the cost function of the initial point and all task points from the initial point, select the task point with the smallest cost function and move to it, calculate the cost function of the task point and all remaining task points, select the task point with the smallest cost function and move to it, and repeat the above steps until all task points are traversed to complete the path planning.

[0223] In summary, this application builds a variety of front-end data models by utilizing basic data, task allocation and business needs, including task value assessment model, customer value assessment model, peak passenger flow identification model, etc., and embeds standardized analysis results into the intelligent itinerary planning model, and then integrates high-precision data sources such as road network information and weather information. Through rule setting, it integrates path planning algorithms such as genetic algorithm, ant colony algorithm, A-STAR algorithm, etc., to construct a path planning model based on neural network, realize offline and real-time intelligent itinerary planning, and meet the needs of different execution scenarios.

[0224] To solve the above problem, the present application also provides a path determination device. Fig. 9 A schematic diagram of the structure of a path determination device provided in an embodiment of the present application is shown; Fig. 9 As shown, the device includes the following modules:

[0225] The acquisition module 901 is used to acquire each target task to be performed in the target time period and the target information required to perform the target task; the target information at least includes: task information and / or road network information;

[0226] The determination module 902 is used to determine the length weight between two target positions based on the target information; the target position includes the task position and the initial position of the target object; the length weight satisfies at least any one of the following or any combination: negatively correlated with the travel time between the two target positions, positively correlated with the value of the customer, and positively correlated with the value of the target task;

[0227] A generation module 903 is used to generate all possible path sets according to the target task and the road network information by using a genetic algorithm; the paths in the path set correspond to chromosomes of the genetic algorithm;

[0228] An updating module 904 is used to update the paths in the path set based on a preset genetic rule to obtain a target path set; the preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path;

[0229] The determination module 902 is further used to determine the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm;

[0230] The judging module is used to judge whether the target offline path meets the preset requirements; if the target offline path does not meet the preset requirements, trigger the updating module until the target offline path meets the preset requirements.

[0231] In some embodiments, the update module 904 is specifically configured to:

[0232] Based on the preset genetic rule, the paths in the path set are iteratively updated multiple times according to the following operations until the first iteration stop condition is met to obtain the target path set; each update includes at least any one of the following items or any combination:

[0233] Based on the path with the highest score in the previous path set, the path with the lowest score in the previous path set is replaced; before each update, the fitness function is used to score each path in the path set to obtain a score value; the score value is positively correlated with the length weight of the path;

[0234] Based on the intersection of the task location, multiple paths in the previous path set are combined;

[0235] A preset number of paths in the previous path set are adjusted with a preset probability.

[0236] In some embodiments, the first iteration stop condition includes any one or any combination of the following:

[0237] The number of iterations reaches the maximum preset number of iterations;

[0238] The number of iterations is greater than the minimum preset number of iterations, and the number of consecutive iteration stagnations reaches the maximum preset number of stagnations; when the iteration stagnation occurs, it indicates that the iterative evolution rate is less than the minimum evolution rate.

[0239] In some embodiments, the determination module 902 is specifically configured to:

[0240] Based on each path in the target path set, determine the reference pheromone corresponding to the ant colony algorithm;

[0241] Based on the reference pheromone, determine the target pheromone of the ant colony algorithm for each path;

[0242] Based on the pheromone update rule, the target pheromone is updated until the second iteration stop condition is met;

[0243] Based on the updated target pheromone, the maximum length weight and the corresponding target offline path are determined.

[0244] In some embodiments, the path determination device further includes:

[0245] The construction module is used to construct a directed graph based on each length weight and each target position after determining the length weight between each target position based on the target information to obtain a full path environment map; the nodes in the full path environment map are the target positions, and the lines between each node are assigned corresponding length weights.

[0246] In some embodiments, the determination module 902 is further configured to:

[0247] Used to obtain the task completion rate under the corresponding weather and / or corresponding time period after determining the maximum length weight and the corresponding target offline path based on the target path set by using the ant colony algorithm;

[0248] The path determination device further includes:

[0249] A determination module, used to determine that the target task is not suitable for execution when the task completion rate is lower than a threshold;

[0250] The output module is used to output prompt information that the target task is not suitable for execution.

[0251] In some embodiments, the determination module 902 is further used to: after determining the maximum length weight and the corresponding target offline path based on the target path set by using the ant colony algorithm, determine the task cost value corresponding to each task location that the target object has not visited based on the corresponding target information when the target task changes within the target time period; the task cost value is determined based on at least the target travel time required to reach the task location and / or the corresponding task importance;

[0252] Based on the cost value of each task, the target real-time path of the task location that the target object has not reached is re-determined.

[0253] In some embodiments, the task cost is positively correlated with the target travel time and negatively correlated with the task importance;

[0254] The determination module 902 is specifically configured to:

[0255] Determine each first task cost value between the current location of the target object and each task location that has not been visited;

[0256] Determine the task position corresponding to the minimum first task cost value as the next target task position in the target real-time path;

[0257] For each target task location, perform the following operations until all target task locations of the target real-time path are determined:

[0258] Determine each second task cost value between the current target task position and each remaining task position;

[0259] The task position corresponding to the smallest second task cost value is determined as the next target task position in the target real-time path.

[0260] In some embodiments, the task information includes at least any one or any combination of the following: target object information, task type, traffic flow, reservation status of the target task, customer value information, target task value information, and estimated execution time of the target task to be completed;

[0261] The road network information includes at least any one or any combination of the following: task location, task destination, transportation mode, travel time and road congestion level.

[0262] In some embodiments, the target information includes at least task information, and the task information includes at least: customer value information, target task value information, traffic flow, and execution time of the target task;

[0263] The acquisition module 901 is specifically used for:

[0264] Use the analytic hierarchy process to calculate the weight of each customer indicator data, and determine the customer value score through weighted calculation to obtain the customer's value information; the customer indicator data includes at least any one of the following or any combination: customer account opening attribute data, customer historical consumption data, customer historical order data, and customer potential customer portrait data;

[0265] Use the analytic hierarchy process to calculate the weight of each task indicator data, and determine the task value score through weighted calculation to obtain the value information of the target task; the task indicator data includes at least any one of the following or any combination: expected task benefits, task type, urgency, and customer relationship;

[0266] Based on the execution status of completed tasks, the k-means clustering algorithm is used to determine the flow of people; the execution status includes at least any one or any combination of the following: execution time, execution weather, whether it is a holiday, whether it is a weekend, whether it is a social event, and execution flow of people;

[0267] Based on the type and execution time of completed tasks, the k-means clustering algorithm is used to determine the estimated execution time of the target tasks to be completed.

[0268] The device provided in the embodiment of the present application is the same as the method in the above embodiment, so both have the same embodiments and beneficial effects, which will not be repeated here.

[0269] Fig.10 FIG. 1 shows a schematic diagram of the hardware structure of a path determination device provided in an embodiment of the present application; Fig.10 As shown, the path determination device may include a processor 111 and a memory 112 storing computer program instructions.

[0270] Specifically, the processor 111 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0271] The memory 112 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 112 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 112 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 112 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 112 is a non-volatile solid-state memory.

[0272] The memory 112 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0273] The processor 111 implements any one of the path determination methods in the above embodiments by reading and executing computer program instructions stored in the memory 112 .

[0274] In one example, the path determination device may further include a communication interface 113 and a bus 114. The processor 111, the memory 112, and the communication interface 113 are connected via the bus 114 and communicate with each other.

[0275] The communication interface 113 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0276] Bus 114 includes hardware, software or both, and the parts of the path determination device are coupled to each other. For example, but not limitation, the bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 114 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the present application considers any suitable bus or interconnection.

[0277] In addition, in combination with the path determination method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any one of the path determination methods in the above embodiment is implemented.

[0278] An embodiment of the present application also provides a computer program product, including a computer program, which implements any one of the path determination methods in the above embodiments when the computer program is processed and executed.

[0279] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0280] The functional blocks shown in the block diagram above can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0281] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0282] The above reference is to the flow chart and / or block diagram of the path determination method, device, equipment, medium and product according to the embodiment of the present disclosure, and describes various aspects of the present disclosure. It should be understood that each box in the flow chart and / or block diagram and the combination of each box in the flow chart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flow chart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flow chart and the combination of boxes in the block diagram and / or flow chart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0283] The above contents are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application.

Claims

1. A path determination method, characterized in that: include: Obtaining each target task to be performed in the target time period and the target information required to perform the target task; The target information at least includes: mission information and road network information; Based on the target information, determining the length weight between two target positions; the target positions include the task position and the initial position of the target object; the length weight at least satisfies any one of the following or any combination: negatively correlated with the travel time between the two target positions, positively correlated with the value of the customer, and positively correlated with the value of the target task; A genetic algorithm is used to generate a set of all possible paths according to the target task and the road network information; the paths in the path set correspond to the chromosomes of the genetic algorithm; Based on a preset genetic rule, the paths in the path set are updated to obtain a target path set; the preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path; Based on the target path set, an ant colony algorithm is used to determine a maximum length weight and a corresponding target offline path; In the case that the target offline path does not meet the preset requirement, the step of returning to the step of updating the paths in the path set based on the preset genetic rule to obtain the target path set is performed until the target offline path meets the preset requirement.

2. The path determination method according to claim 1, characterized in that: The updating of the paths in the path set based on the preset genetic rule to obtain the target path set includes: Based on the preset genetic rule, the paths in the path set are iteratively updated multiple times according to the following operations until the first iteration stop condition is met to obtain the target path set; each update includes at least any one of the following items or any combination: Based on the path with the highest score in the path set last time, the path with the lowest score in the path set last time is replaced; before each update, the fitness function is used to score each path in the path set to obtain a score value; the score value is positively correlated with the length weight sum of the path; Based on the intersection of the task location, combining multiple paths in the path set from last time; A preset number of paths in the previous path set are adjusted with a preset probability.

3. The path determination method according to claim 2, characterized in that: The first iteration stopping condition includes any one or any combination of the following: The number of iterations reaches the maximum preset number of iterations; The number of iterations is greater than the minimum preset number of iterations, and the number of consecutive iteration stagnations reaches the maximum preset number of stagnations; when the iteration stagnation occurs, it indicates that the iterative evolution rate is less than the minimum evolution rate.

4. The path determination method according to claim 3, characterized in that: The method of determining the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm includes: Determining a reference pheromone corresponding to the ant colony algorithm based on each path in the target path set; Based on the reference pheromone, determining the target pheromone of the ant colony algorithm for each path; Based on the pheromone update rule, updating the target pheromone until the second iteration stop condition is met; Based on the updated target pheromone, the maximum length weight and the corresponding target offline path are determined.

5. The path determination method according to claim 1, characterized in that: After determining the length weights between each pair of target positions based on the target information, the method further includes: Based on each of the length weights and each of the target positions, a directed graph is constructed to obtain a full path environment map; the nodes in the full path environment map are the target positions, and the lines between each two of the nodes are assigned corresponding length weights.

6. The path determination method according to claim 1, characterized in that: After determining the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm, the method further includes: Obtain the task completion rate under the corresponding weather and / or time period; When the task completion rate is lower than a threshold, determining that the target task is not suitable for execution; Output a prompt message that the target task is not suitable for execution.

7. The path determination method according to any one of claims 1 to 6, characterized in that: After determining the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm, the method further includes: In the case where the target task changes within the target time period, determining the task cost value corresponding to each of the task locations that the target object has not visited based on the corresponding target information; the task cost value is determined based at least on the target travel time required to reach the task location and / or the corresponding task importance; Based on each of the task cost values, a target real-time path of the task location that the target object has not visited is re-determined.

8. The path determination method according to claim 7, characterized in that: The task cost is positively correlated with the target travel time and negatively correlated with the task importance; The determining, based on the corresponding target information, the task cost value corresponding to each of the task locations that the target object has not visited, and the re-determining, based on each of the task cost values, the target real-time path of the task location that the target object has not visited, includes: Determine each first task cost value between the current location of the target object and each task location that has not been visited; Determine the task position corresponding to the minimum first task cost value as the next target task position in the target real-time path; For each target task position, the following operations are performed until all the target task positions of the target real-time path are determined: Determine each second task cost value between the current target task position and each remaining task position; The task position corresponding to the smallest second task cost value is determined as the next target task position in the target real-time path.

9. The path determination method according to claim 1, characterized in that: The task information includes at least any one or any combination of the following: target object information, task type, human traffic situation, reservation situation of the target task, customer value information, value information of the target task, and estimated execution time of the target task to be completed; The road network information includes at least any one or any combination of the following: task location, task destination, transportation mode, travel time and road congestion level.

10. The path determination method according to claim 9, characterized in that: The target information at least includes task information, and the task information at least includes: customer value information, value information of the target task, flow of people, and execution time of the target task; Acquiring the target information required to perform the target task includes: Using the hierarchical analysis method, the weight of each customer indicator data is calculated, and the customer value score is determined by weighted calculation to obtain the customer's value information; the customer indicator data includes at least any one of the following or any combination: customer account opening attribute data, customer historical consumption data, customer historical order data, and customer potential customer portrait data; Using the hierarchical analysis method, the weight of each task indicator data is calculated, and the task value score is determined by weighted calculation to obtain the value information of the target task; the task indicator data includes at least any one of the following or any combination: expected task benefit, task type, urgency and customer relationship; Based on the execution status of completed tasks, the k-means clustering algorithm is used to determine the flow of people; the execution status includes at least any one or any combination of the following: execution time, execution weather, whether it is a holiday, whether it is a weekend, whether it is a social event, and execution flow of people; Based on the types and execution times of completed tasks, a k-means clustering algorithm is used to determine the estimated execution time of the target task to be completed.

11. A path determination device, characterized in that: The device comprises: An acquisition module, used to acquire each target task to be performed in a target time period and target information required to perform the target task; the target information at least includes: task information and / or road network information; A determination module is used to determine the length weight between two target positions based on the target information; the target positions include the task position and the initial position of the target object; the length weight satisfies at least any one of the following or any combination: negatively correlated with the travel time between the two target positions, positively correlated with the value of the customer, and positively correlated with the value of the target task; A generation module, used to generate a set of all possible paths according to the target task and the road network information by using a genetic algorithm; the paths in the path set correspond to the chromosomes of the genetic algorithm; An updating module, used for updating the paths in the path set based on a preset genetic rule to obtain a target path set; the preset genetic rule is used to increase the sum of the length weights of each path in the path set; the sum of the length weights is the sum of the length weights of all two connected target positions in a path; The determination module is further used to determine the maximum length weight and the corresponding target offline path based on the target path set by using an ant colony algorithm; The judging module is used to judge whether the target offline path meets the preset requirements; if the target offline path does not meet the preset requirements, trigger the updating module until the target offline path meets the preset requirements.

12. A path determination device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the path determination method according to any one of claims 1 to 10 is implemented.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the path determination method according to any one of claims 1 to 10 is implemented.

14. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the path determination method as described in any one of claims 1 to 10.

Images