Task allocation method and system based on greening data processing

Abstract

The invention relates to the technical field of electricity data processing, in particular to a task allocation method and system based on greening data processing. The method comprises the following steps: respectively acquiring green plant data and people flow density data of each road section in a preset environmental sanitation sub-road section set; the green plant data comprises plant species, number and tree age of green plants in each road section; determining the types of greening wastes and the quantity of various greening wastes of each road section in a future preset time period based on the greening plant data; taking the greening waste type, the quantity of various greening wastes and the people flow density data as input of an environmental sanitation work load prediction model to obtain environmental sanitation work load values, influenced by greening waste state conversion, of each road section in a future preset time period; and based on the environmental sanitation operation load value, environmental sanitation tasks are allocated to each environmental sanitation group through a preset combination optimization algorithm. According to the method provided by the invention, environmental sanitation work can be distributed more reasonably.

Description

Technical Field

[0001] The present application relates to the technical field of electrical data processing, and in particular to a task allocation method based on greening data processing. Background Art

[0002] Existing sanitation task allocation methods generally use resource allocation strategies based on static indicators. This involves obtaining and evaluating the workload of cleaning tasks based on directly measurable physical quantities such as the weight and volume of green waste, and scheduling personnel and equipment accordingly.

[0003] However, this allocation method still has some problems, which leads to the unreasonable allocation of sanitation work by the current sanitation task allocation method. Summary of the Invention

[0004] In order to solve the above technical problems or at least partially solve the above technical problems, the present application provides a task allocation method based on greening data processing, which can more reasonably allocate sanitation work tasks.

[0005] In a first aspect, the present application proposes a task allocation method based on greening data processing, the task allocation method based on greening data processing comprising the following steps: Obtain green plant data and pedestrian density data for each road section in the preset sanitation sub-section set; The green plant data includes the plant species, quantity and age of green plants in each road section; Determining the types of greenery waste and the amounts of each type of greenery waste in a preset time period in the future for each road section based on the greenery plant data; The types of green waste, the amount of each type of green waste, and the pedestrian density data are used as inputs to the sanitation operation load prediction model to obtain the sanitation operation load value of each road section affected by the green waste state transformation in a preset time period in the future; Based on the sanitation operation load value, the sanitation tasks are allocated to each sanitation team through a preset combination optimization algorithm.

[0006] Optionally, the task allocation method based on greening data processing further includes: Obtain microenvironmental characteristic data of each road section; When predicting the sanitation operation load value of each road section: The microenvironment characteristic data, together with the green waste type, the quantity of each type of green waste and the pedestrian density data, are used as the input of the sanitation operation load prediction model to obtain the sanitation operation load value of each road section affected by the green waste state transformation under the influence of pedestrian interaction and microenvironment in a preset time period in the future.

[0007] Optionally, the microenvironment characteristic data include ventilation index, humidity index, average light duration and average light intensity.

[0008] Optionally, the sanitation operation load prediction model includes: The feature input layer receives the green waste type code, green waste quantity, pedestrian density data and microenvironmental feature data of each section in the sanitation sub-section set; The green waste type identification layer is connected to the feature input layer to obtain the green waste type code and green waste quantity of each road section, which is used to identify the dominant green waste type of each road section; The cross-modal attention layer is connected to the green waste type recognition layer and the feature input layer. Based on the output of the green waste type recognition layer, it dynamically generates an attention weight matrix for the microenvironmental feature data of each road section in the feature input layer. The specific rules are as follows: When it is identified as fruit-type greening waste, a higher weight is given to the humidity index and ventilation index in the microenvironmental characteristic data; When petal-type green waste is identified, higher weights are assigned to the ventilation index, average light duration, and average light intensity in the microenvironmental characteristic data; The feature fusion layer is connected to the cross-modal attention layer and the feature input layer. It combines the attention-weighted microenvironmental features of each road section with the amount of green waste and pedestrian density data to form a fused feature vector and output it externally. The hidden layer is connected to the feature fusion layer to obtain the fused feature vector output by the feature fusion layer, and the fused feature vector is nonlinearly transformed and then outputted externally; The output layer is connected to the hidden layer to obtain the output of the hidden layer for processing and then output the sanitation operation load value of each road section.

[0009] Optionally, crowd density data is obtained by calling thermal data provided by a map service provider.

[0010] Optionally, determining the types of greenery waste and the amounts of each type of greenery waste in a preset time period in the future for each road section based on the greenery plant data includes the following steps: Classify the green plants in each road section according to their species, and calculate the number and average age of different green plant species; The number, average age, and current season information of each greening plant species are input into the corresponding greening waste quantity prediction model. The greening waste quantity prediction model uses an LSTM time series neural network to predict the type and quantity of greening waste generated by each greening plant species in a preset time period in the future; The prediction results of all green plant species in each road section are summarized, and the green waste types are classified and accumulated to obtain the green waste types and the quantity of each type of green waste in each road section in the future preset time period.

[0011] Optionally, based on the sanitation workload value, allocating sanitation tasks to the sanitation teams through a preset combined optimization algorithm includes the following steps: Each road section is regarded as a task item to be assigned, and each task item includes a road section identifier and a corresponding sanitation operation load value; Construct a road segment adjacency graph, treat each road segment as a graph node, establish undirected edge connections based on the spatial adjacency between road segments, and form an undirected graph structure that describes the continuity of the road segments; The preset working capacity limit is used as a constraint condition, so that each sanitation team corresponds to a packing container with a capacity equal to the preset working capacity limit; Based on the bin packing algorithm, a road segment continuity constraint is added. Adjacent road segments in the undirected graph are preferentially assigned to the same sanitation team. The optimization is performed with load balancing and road segment continuity as the dual objective function. This ensures that the total load value assigned to each sanitation team does not exceed the preset working capacity limit, and that the assigned road segments form a connected subgraph in the undirected graph as much as possible. After running the above packing algorithm process, the task allocation plan for each sanitation team is output, including a list of road section signs that each sanitation team is responsible for.

[0012] In the second aspect, the present application proposes a task allocation system based on greening data processing, including a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, the at least one program, the code set or instruction set is loaded and executed by the processor to implement the task allocation method based on greening data processing as described in any one of the first aspects.

[0013] The technical solution provided by this application has the following advantages compared with the existing technology: One of its beneficial effects and its working principle is: Existing technologies mainly allocate sanitation resources based on static physical parameters such as weight and volume of green waste, ignoring the impact of the state evolution process of certain green waste in the actual environment on the difficulty of cleaning.

[0014] For example, the cleaning difficulty of dry, loose fallen leaves of the same weight and rotten fruits that are tightly stuck to the ground after being trampled by pedestrians and crushed by vehicles is several times different. However, existing technology is unable to recognize and deal with this difference, resulting in unreasonable allocation of sanitation tasks.

[0015] This application obtains data on green plants and pedestrian density for each road section, determines the types of green waste based on the green plant data, and predicts the amount of each type of green waste. Furthermore, an artificial intelligence model is used to learn the impact of green waste state transformation under the interaction of pedestrian flow.

[0016] Then, the predicted green waste type, quantity and crowd density data are input into the sanitation operation load prediction model to obtain the sanitation operation load value reflecting the impact of green waste state transformation. Finally, task allocation is performed based on the load value through a combinatorial optimization algorithm.

[0017] In this way, artificial intelligence is used to learn the laws of physical state changes of green waste under different pedestrian trampling intensities, and the learned state change laws are used to calculate the actual sanitation load after considering the state transformation.

[0018] Based on this, the allocation of sanitation tasks based on dynamic load assessment is realized, enabling the sanitation management department to accurately predict the actual cleaning difficulty of each road section, avoiding the problem of improper allocation of sanitation resources due to ignoring the transformation of the status of green waste.

[0019] Therefore, the task allocation method based on greening data processing proposed in this application can more reasonably allocate sanitation work.

[0020] The second beneficial effect and its working principle are: In different areas with similar pedestrian flow, the degradation rate and cleaning difficulty of the same type of green waste are still different. This is because the microenvironment composed of different greening configurations and the environment in which each area is located will also affect the transformation process of the green waste status.

[0021] Different microenvironmental conditions, such as ventilation, humidity, and light intensity, can alter the degree of change in the physical properties of landscaping waste and the resulting difficulty of cleaning. For example, in a microenvironment with high humidity and low ventilation, fruit-like landscaping waste will decay faster and become more adherent to the ground, making it more difficult to clean. In contrast, in a microenvironment with good ventilation and ample light, petals-like landscaping waste will dry quickly and disperse, expanding the cleaning area.

[0022] Therefore, when predicting the impact of the transformation of greening waste status on the sanitation operation load, it is also necessary to consider the microenvironmental characteristics of the greening waste, so as to improve the accuracy of the prediction of the actual sanitation operation load.

[0023] This application uses an artificial intelligence model to establish a correlation model between microenvironmental characteristics, human flow interactions, and the transformation of landscaping waste. The model inputs microenvironmental characteristics, along with data on landscaping waste type, quantity, and human flow density, to more accurately calculate the impact of landscaping waste transformation on workload under specific microenvironmental conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart of a task allocation method based on greening data processing provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical solution in this application will be described below with reference to the accompanying drawings.

[0026] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application may also be implemented in other ways than those described herein. It is apparent that the embodiments described in the specification are only some of the embodiments of the present application, not all of them. It should be noted that the embodiments of the present application and the features therein may be combined with each other unless there is a conflict.

[0027] In the first aspect, this application proposes a task allocation method based on greening data processing, referring to Figure 1 As shown, the task allocation method based on greening data processing includes the following steps: S1: Obtain green plant data and pedestrian density data for each road section in the preset sanitation sub-road section set.

[0028] The green plant data includes the plant species, quantity and age of the green plants in each road section.

[0029] Specifically, first, based on the city's existing municipal sanitation planning map, the target area is divided into several sanitation sub-sections with unique numbers to form a preset sanitation sub-section set; an on-site survey is conducted on each sub-section, and a combination of manual identification and plant species image recognition technology is used to record detailed information on all green plants in the section, including the specific types of plants (such as banyan trees, kapok trees, sycamore trees, mango trees, poinciana trees and magnolia trees, etc.), the number of each plant, and the average tree age estimated by tree diameter measurement or recorded in the database after manual tree diameter measurement using a tape measure.

[0030] Finally, the collected green plant data is associated with the corresponding road section number and stored to establish a complete green plant database.

[0031] Specifically, pedestrian density data is obtained by calling the thermal data provided by map service providers. Through the API interface, the historical thermal data services of mainstream map service providers such as Baidu Maps or Amap are called to obtain historical pedestrian density information for each road section.

[0032] The average pedestrian density characteristic value of each road section is calculated to form the standardized pedestrian density data of each road section.

[0033] S2: Determine the types of greenery waste and the quantities of each type of greenery waste in a preset time period in the future for each road section based on the greenery plant data.

[0034] Specifically, the following steps are included: The green plants in each road section were classified according to their species, and the number and average age of different green plant species were counted.

[0035] The number of green plant species, average tree age, and current season information are input into a green waste quantity prediction model corresponding to each green plant species. The green waste quantity prediction model uses an LSTM time series neural network to predict the type and quantity of green waste generated by each green plant species in a preset time period in the future; Specifically, the number, average age and time coding information (current season time information) of various types of green plants are input into the corresponding green waste quantity prediction model. Each plant species has its own exclusive prediction model.

[0036] The time code information uses a 7-day unit encoding method, dividing the entire year (365 days) into 52 time units. Each time unit corresponds to a unique code value, which is used to represent the current time in the annual cycle. The green waste quantity prediction model uses an LSTM time series neural network.

[0037] Specifically, the network structure of the LSTM model includes: The input layer receives the number of corresponding plant species, average tree age, and 7-day unit time encoding as feature vectors; The hidden layer adopts a bidirectional LSTM structure with 128 hidden units, which is used to learn the temporal variation pattern of the waste amount of this plant species; The output layer contains two branches. The first branch outputs the code of the dominant waste type produced by the plant species, and the second branch outputs the predicted value of the quantity of this waste type in the future preset time period.

[0038] Specifically, the greening waste quantity prediction model for each plant species is trained through the following steps: collecting historical waste quantity data of the plant species under different time units, tree ages and quantity conditions as a training set, where each training sample contains the number of plants of this species, the average tree age, and the corresponding time code as input features, and the actual waste type and quantity data generated in the corresponding time period as labels.

[0039] The number of plants, average tree age and time code in the training set are used as inputs of the LSTM network; The corresponding waste type code and quantity are used as the supervision labels of the LSTM network, and multi-task learning is adopted for training. The cross entropy loss function is used for the classification task, and the mean square error loss function is used for the regression task. In this way, a greening waste quantity prediction model for each greening plant species described in the embodiment of the present application is obtained through training.

[0040] The prediction results of all green plant species in each road section are summarized, and the green waste types are classified and accumulated to obtain the green waste types and the quantity of each type of green waste in each road section in the future preset time period.

[0041] Specifically, in this embodiment of the present application, the preset future time period is set to 7 days. This means that the system uses a weekly forecast cycle to provide the sanitation management department with a weekly sanitation workload forecast and task allocation plan for each road section. Generally speaking, this value should be consistent with the time step parameter of the aforementioned time code, which uses 7 days as a unit.

[0042] S3: Obtain microenvironmental characteristic data of each road section.

[0043] The microenvironmental characteristic data are obtained by periodically conducting field sampling on each road section.

[0044] Specifically, the microenvironment characteristic data includes ventilation index, humidity index, average light duration and average light intensity.

[0045] Specifically, the process of collecting microenvironmental characteristic data is as follows: First, prepare portable environmental monitoring equipment, including a digital anemometer, a thermometer, and a illuminance meter.

[0046] Then, representative seasons (such as spring, summer, and autumn) and weather conditions (sunny and cloudy) were selected, and field measurements were conducted on each preset sanitation sub-section at representative time points in each season (10:00 a.m., 2:00 p.m., and 5:00 p.m.).

[0047] The specific measurement method is as follows: ventilation index is measured by using a digital anemometer to measure wind speed values at multiple measuring points on the road section; The humidity index measures the humidity value of the road section through a thermometer and hygrometer; The average sunshine duration is recorded by continuous monitoring of the effective sunshine duration of the road section every day; The average light intensity is measured using a light meter on a road section at a standard time point.

[0048] During the data processing, the values measured multiple times are statistically analyzed and the average value is calculated as the stable microenvironmental characteristic data of the road section.

[0049] S4: The microenvironmental characteristic data, together with the green waste type, the quantity of each type of green waste and the pedestrian density data, are used as inputs to the sanitation operation load prediction model to obtain the sanitation operation load value of each road section affected by the transformation of the green waste state under the influence of pedestrian interaction and microenvironment in a preset time period in the future.

[0050] The sanitation operation load prediction model includes: The feature input layer receives the green waste type code, green waste quantity, pedestrian density data and microenvironmental feature data of each section in the sanitation sub-section set; The green waste type identification layer is connected to the feature input layer to obtain the green waste type code and green waste quantity of each road section, which is used to identify the dominant green waste type of each road section.

[0051] Specifically, the green waste type identification layer adopts a single-layer fully connected network, which takes the waste type code and waste quantity of each road section as input, and the output dimension is the total number of waste categories. The activation function is softmax, which is used to output the probability distribution of the dominant waste type of each road section.

[0052] Specifically, "dominant" is defined as the type of green waste with the largest amount of green waste within a road section. The specific determination rule is: the waste type with the highest probability in the probability distribution of dominant waste types is selected as the dominant type.

[0053] The cross-modal attention layer is connected to the green waste type recognition layer and the feature input layer. Based on the output of the green waste type recognition layer, it dynamically generates an attention weight matrix for the microenvironmental feature data of each road section in the feature input layer. The specific rules are as follows: When it is identified as fruit-type greening waste, a higher weight is given to the humidity index and ventilation index in the microenvironmental characteristic data; When petal-type green waste is identified, higher weights are assigned to the ventilation index, average light duration, and average light intensity in the microenvironmental characteristic data; When leaf-type greening waste is identified, equal weights are maintained for each microenvironmental characteristic dimension, and no special weight adjustment is performed.

[0054] Specifically, in an embodiment of the present application, the cross-modal attention layer implements dynamic allocation of attention weights through a weight parameter matrix; The weight parameter matrix sets that when the waste type identification layer outputs the highest probability of fruit waste, the weight parameter of the humidity index is set to 0.8, the weight parameter of the ventilation index is set to 0.7, and the weight parameters of the average light duration and the average light intensity are set to 0.2 and 0.1 respectively.

[0055] When the probability of outputting petal-type waste is the highest, the weight parameter of the ventilation index is set to 0.8, the weight parameters of the average light duration and average light intensity are set to 0.7 and 0.6 respectively, and the weight parameter of the humidity index is set to 0.2.

[0056] When the probability of leaf waste being output is highest, the weight parameters for ventilation index, humidity index, average sunlight duration, and average light intensity are all set to 0.5 to maintain an equal distribution. Because leaf-based greening waste (such as sycamore and ginkgo leaves) typically has a relatively stable physical form due to its inherent physical properties, compared to fruit and flower petal waste, the degree of state transformation under different microenvironmental conditions is limited. Therefore, the model assigns equal weight to the microenvironmental characteristics of leaf waste, avoiding the interference of excessive focus on different microenvironmental conditions on the prediction results.

[0057] The feature fusion layer is connected to the cross-modal attention layer and the feature input layer. It combines the attention-weighted microenvironmental features of each road section with the amount of green waste and pedestrian density data to form a fused feature vector and output it externally.

[0058] The feature fusion layer uses a vector splicing operation to splice the microenvironment feature vector that has undergone attention weighting processing with the amount of waste and the density of human traffic by dimension to form a unified fusion feature vector.

[0059] The hidden layer is connected to the feature fusion layer to obtain the fused feature vector output by the feature fusion layer, and the fused feature vector is output after nonlinear transformation.

[0060] Specifically, in the embodiment of the present application, the hidden layer includes: The first layer of fully connected network is used to obtain the fused feature vector, with an output dimension of 256 and an activation function of ReLU; The input dimension of the second-layer fully connected network is 256 dimensions, the output dimension is 128 dimensions, and the activation function is ReLU; The input dimension of the third-layer fully connected network is 128 dimensions, the output dimension is 64 dimensions, and the activation function is ReLU.

[0061] The output layer is connected to the hidden layer to obtain the output of the hidden layer for processing and then output the sanitation operation load value of each road section.

[0062] Specifically, in the embodiment of the present application, the output layer adopts a single-layer fully connected network with an input dimension of 64 dimensions and an output dimension of 1 dimension.

[0063] The training process is as follows: historical data on sanitation operation loads on multiple road sections in different time periods are collected as a training set, where each training sample contains the road section's green waste type code, green waste quantity, pedestrian density data, and microenvironmental characteristic data as input features.

[0064] The actual sanitation load value for the road section during the corresponding time period is used as a label. The sanitation load value is determined by sanitation management personnel based on the actual cleaning difficulty, time required, and manpower input. The value ranges from 1 to 10, where 1 represents a very light load (such as a small amount of dry fallen leaves), 5 represents a medium load (such as a normal amount of mixed waste), and 10 represents an extremely heavy load (such as a large amount of rotten, trampled, and clinging fruit waste).

[0065] Each training sample contains the road section's green waste type code, green waste quantity, pedestrian density data, and microenvironmental characteristics as input features, and the actual sanitation workload value of the road section during the corresponding time period as a label. The input features in the training set are used as input to the sanitation workload prediction model. The corresponding sanitation operation load value is used as the supervision label of the model, the mean square error loss function is used for training, and the Adam optimizer is used to update the network parameters, thereby training the sanitation operation load prediction model described in the embodiment of the present application.

[0066] Its working principle and beneficial effects lie in that this application proposes a sanitation operation load prediction model based on a cross-modal attention mechanism, which can dynamically allocate attention weights to microenvironmental characteristics according to different waste types, so that the model pays attention to the difficulty of green waste conversion in different environments.

[0067] For example, for fruit waste, the model will focus on humidity and ventilation conditions because these factors dominate the decay rate and adhesion of the fruit.

[0068] For petal waste, the model focuses on lighting conditions because lighting affects the drying and dispersion process of petals.

[0069] This mechanism enables the model to simulate the analytical thinking of experts and learn the state transformation laws of waste under specific microenvironmental conditions, so as to accurately calculate the impact of waste state transformation on the actual cleaning difficulty.

[0070] Therefore, compared with the direct prediction process using the ternary matrix of green waste, human flow density data and microenvironment data, the sanitation operation load prediction model provided in the embodiment of the present application can more accurately make the model pay attention to the difficulty of green waste conversion under different environments, thereby more accurately controlling and predicting the state transformation process of waste under the dual effects of human flow trampling and microenvironment catalysis, thereby improving the accuracy of sanitation operation load prediction.

[0071] S5: Based on the sanitation operation load value, the sanitation tasks are allocated to each sanitation team through a preset combination optimization algorithm.

[0072] Specifically, based on the sanitation operation load value, sanitation tasks are allocated to each sanitation team through a preset combination optimization algorithm, including the following steps: Each road section is considered as a task item to be assigned. Each task item contains a road section ID and a corresponding sanitation workload value, forming a set of tasks to be assigned. The road section ID is a unique number, and the sanitation workload value is a value in the range of 1-10 predicted by the aforementioned model.

[0073] Construct a road segment adjacency graph, treat each road segment as a graph node, establish undirected edge connections based on the spatial adjacency between road segments, and form an undirected graph structure that describes the continuity of the road segments; The preset working capacity limit is used as a constraint condition, so that each sanitation team corresponds to a container with a capacity equal to the preset working capacity limit. In the embodiment of the present application, the preset working capacity limit is set to 50 load units, that is, the maximum total load value that each sanitation team can bear in one working cycle is 50.

[0074] Based on the packing algorithm, a road section continuity constraint is added, and adjacent road sections in the undirected graph are preferentially assigned to the same sanitation team. The optimization is performed with load balancing and road section continuity as the dual objective function to ensure that the total load value assigned to each sanitation team does not exceed the preset working capacity limit, and the assigned road sections form a connected subgraph in the undirected graph as much as possible.

[0075] Among them, the specific optimization objectives of the dual objective function are: First, ensure that the total load value assigned to each sanitation team does not exceed the preset working capacity limit of 50; Second, maximize the balance of load distribution among sanitation groups, that is, minimize the variance of load values of each group; Third, the connectivity of the road sections assigned to the same sanitation team in the undirected graph is maximized, and connected subgraphs are formed first to reduce the mobility cost of the sanitation team.

[0076] Specifically, the embodiment of the present application also proposes an improved packing algorithm adapted to the scenario of the present application.

[0077] Specifically, the improved packing algorithm used in this application has the following workflow: Preprocessing stage: Precompute the shortest path distance matrix between road segments for rapid evaluation of connectivity gains; sort all road segments in descending order of sanitation workload (range 1-10); initialize each sanitation group to an empty set; Initial solution construction: A greedy strategy is used to generate an initial feasible solution, which sequentially allocates road sections to the sanitation team with the smallest current load and the capacity to accommodate the road section, ensuring that the total load value of each team does not exceed the preset working capacity limit of 50 load units; Taboo search optimization: Iterative optimization is performed within the range of the maximum number of iterations T = 200. Each iteration performs the following operations: Define the hybrid scoring function: Score = α × connectivity score + β × load balancing score; Among them, α = 0.6, β = 0.4; Score is a mixed score.

[0078] The values of α and β are based on the priority considerations of actual sanitation operations. Because in the embodiment of the present application, considering that the impact of connectivity on the efficiency of sanitation operations is more significant, when the road sections that the sanitation team is responsible for are continuous in space, it can greatly reduce the moving time and transportation costs and improve the overall operation efficiency; and in actual operations, moderate load differences are acceptable because the actual work capabilities and experience levels of different sanitation teams are different. Therefore, in the embodiment of the present application, α>β is set to give priority to ensuring the spatial continuity of road section allocation. At the same time, in order to take into account the relative balance of load distribution, the weight distribution of α=0.6 and β=0.4 not only reflects the priority of connectivity, but also ensures that the importance of load balancing is not ignored, achieving a reasonable balance between the two goals.

[0079] Therefore, in other embodiments, those skilled in the art may adaptively adjust the above parameters based on practical considerations.

[0080] Connectivity score = (total number of road segments in the group - number of connected components + 1) / total number of road segments in the group; Calculate the mean μ and variance σ² of the current load values of all groups; Load balancing score = 1 / (1 + σ² / μ²).

[0081] Generate candidate move operations: including relocate operations (i.e., moving a segment from one group to another) and 2-opt swap operations (i.e., swapping a segment pair between two groups); Calculate the hybrid score gain for each candidate operation that meets the capacity constraint and select the optimal non-tabu move to execute; Specifically, the hybrid score gain = Score value after executing the operation - Score value before executing the operation. The larger the gain value, the better the improvement effect of the operation on the overall solution. The criteria for determining the optimal non-tabu move are: first, exclude the operations recorded in the taboo table (that is, the move operations that have been executed in the past 7 iterations); then, among all non-tabu candidate operations, select the operation with the largest hybrid score gain to execute; when there are multiple operations with the same maximum gain value, the operation that can increase the connectivity score is preferred.

[0082] If there is still a tie, one of the operations is randomly selected to execute to avoid the algorithm falling into an infinite loop.

[0083] Update the tabu table (taboo step = 7) to record the executed move operations to prevent repetition in the short term.

[0084] Connectivity repair: For each sanitation team, perform BFS traversal in the road segment adjacency graph to identify disconnected island road segments; prioritize relocating island road segments and assign them to the neighboring team with the closest geographical distance and sufficient remaining capacity.

[0085] Termination condition: When there is no improvement in the mixed score after 20 consecutive iterations or the maximum number of iterations T=200 is reached, the search is stopped and the current optimal solution is output.

[0086] Among them, the judgment criteria for the current optimal solution are: record the mixed score value after each iteration during the entire search process, and select the road segment allocation scheme corresponding to the historical highest mixed score value as the optimal solution; when the algorithm terminates, if the mixed score value of the current scheme is equal to the historical highest value, then directly output the current scheme; if the mixed score value of the current scheme is lower than the historical highest value, then backtrack to the allocation scheme corresponding to the historical highest score and output the scheme, ensuring that the output solution is the optimal road segment allocation scheme found during the search process.

[0087] Those skilled in the art should know that the specific values of the number of consecutive iterations and the maximum number of iterations can be adjusted according to actual conditions. The purpose is to find a balance between a limited number of calculations and finding the global optimal value. This is a routine practice for those skilled in the art.

[0088] After running the improved packing algorithm, the task allocation plan for each sanitation team is output, including a list of road section signs that each sanitation team is responsible for.

[0089] After running the improved packing algorithm, a complete packing result can be obtained, that is, all road sections have been assigned to the corresponding sanitation teams (boxes), and the task allocation plan for each sanitation team has been obtained; the packing result of each sanitation team (box) can be converted into a road section identification list through the following steps: check each sanitation team (box), extract all the road section identifications contained in the box, that is, the list of road section identifications that the team is responsible for.

[0090] In summary, the technical solution provided by the embodiments of the present application has the following advantages compared with the prior art: One of its beneficial effects and its working principle is: Existing technologies mainly allocate sanitation resources based on static physical parameters such as weight and volume of green waste, ignoring the impact of the state evolution process of certain green waste in the actual environment on the difficulty of cleaning.

[0091] For example, the cleaning difficulty of dry, loose fallen leaves of the same weight and rotten fruits that are tightly stuck to the ground after being trampled by pedestrians and crushed by vehicles is several times different. However, existing technology is unable to recognize and deal with this difference, resulting in unreasonable allocation of sanitation tasks.

[0092] This application obtains data on green plants and pedestrian density for each road section, determines the types of green waste based on the green plant data, and predicts the amount of each type of green waste. Furthermore, an artificial intelligence model is used to learn the impact of green waste state transformation under the interaction of pedestrian flow.

[0093] Then, the predicted green waste type, quantity and crowd density data are input into the sanitation operation load prediction model to obtain the sanitation operation load value reflecting the impact of green waste state transformation. Finally, task allocation is performed based on the load value through a combinatorial optimization algorithm.

[0094] In this way, artificial intelligence is used to learn the laws of physical state changes of green waste under different pedestrian trampling intensities, and the learned state change laws are used to calculate the actual sanitation load after considering the state transformation.

[0095] Based on this, the allocation of sanitation tasks based on dynamic load assessment is realized, enabling the sanitation management department to accurately predict the actual cleaning difficulty of each road section, avoiding the problem of improper allocation of sanitation resources due to ignoring the transformation of the status of green waste.

[0096] Therefore, the task allocation method based on greening data processing proposed in this application can more reasonably allocate sanitation work.

[0097] The second beneficial effect and its working principle are: In different areas with similar pedestrian flow, the degradation rate and cleaning difficulty of the same type of green waste are still different. This is because the microenvironment of each area, which is composed of different greening configurations and the environment in which it is located, will also affect the transformation process of the green waste status.

[0098] Different microenvironmental conditions, such as ventilation, humidity, and light intensity, can alter the degree of change in the physical properties of landscaping waste and the resulting difficulty of cleaning. For example, in a microenvironment with high humidity and low ventilation, fruit-like landscaping waste will decay faster and become more adherent to the ground, making it more difficult to clean. In contrast, in a microenvironment with good ventilation and ample light, petals-like landscaping waste will dry quickly and disperse, expanding the cleaning area.

[0099] Therefore, when predicting the impact of the transformation of greening waste status on the sanitation operation load, it is also necessary to consider the microenvironmental characteristics of the greening waste, so as to improve the accuracy of the prediction of the actual sanitation operation load.

[0100] This application uses an artificial intelligence model to establish a correlation model between microenvironmental characteristics, human flow interactions, and the transformation of landscaping waste. The model inputs microenvironmental characteristics, along with data on landscaping waste type, quantity, and human flow density, to more accurately calculate the impact of landscaping waste transformation on workload under specific microenvironmental conditions.

[0101] In the second aspect, the present application proposes a task allocation system based on greening data processing, including a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, the at least one program, the code set or instruction set is loaded and executed by the processor to implement the task allocation method based on greening data processing as described in any of the above embodiments.

[0102] It should be noted that, in this document, relational terms such as "first" and "second" are used solely to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations. Furthermore, the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, article, or device. Without further limitation, an element defined by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article, or device comprising the element. Furthermore, in the description of the embodiments of this application, unless otherwise specified, " / " represents or. For example, A / B can represent either A or B. "And / or" herein is merely a description of an associative relationship between associated objects, indicating that three relationships can exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, in the description of the embodiments of the present application, “plurality” refers to two or more than two.

[0103] The foregoing description is intended only to provide specific embodiments of the present application, which will enable those skilled in the art to understand and implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments described herein, but is intended to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. The task allocation method based on greening data processing is characterized by: The task allocation method based on greening data processing includes the following steps: Obtain green plant data and pedestrian density data for each road section in the preset sanitation sub-section set; The green plant data includes the plant species, quantity and age of green plants in each road section; Determining the types of greenery waste and the amounts of each type of greenery waste in a preset time period in the future for each road section based on the greenery plant data; The types of green waste, the amount of each type of green waste, and the pedestrian density data are used as inputs to the sanitation operation load prediction model to obtain the sanitation operation load value of each road section affected by the green waste state transformation in a preset time period in the future; Based on the sanitation operation load value, the sanitation tasks are allocated to each sanitation team through a preset combination optimization algorithm.

2. The task allocation method based on greening data processing according to claim 1 is characterized in that: The task allocation method based on greening data processing also includes: Obtain microenvironmental characteristic data of each road section; When predicting the sanitation operation load value of each road section: The microenvironment characteristic data, together with the green waste type, the quantity of each type of green waste and the pedestrian density data, are used as the input of the sanitation operation load prediction model to obtain the sanitation operation load value of each road section affected by the green waste state transformation under the influence of pedestrian interaction and microenvironment in a preset time period in the future.

3. The task allocation method based on greening data processing according to claim 2 is characterized in that: The microenvironment characteristic data include ventilation index, humidity index, average light duration and average light intensity.

4. The task allocation method based on greening data processing according to claim 3 is characterized in that: The sanitation operation load prediction model includes: The feature input layer receives the green waste type code, green waste quantity, pedestrian density data and microenvironmental feature data of each section in the sanitation sub-section set; The green waste type identification layer is connected to the feature input layer to obtain the green waste type code and green waste quantity of each road section, which is used to identify the dominant green waste type of each road section; The cross-modal attention layer is connected to the green waste type recognition layer and the feature input layer. Based on the output of the green waste type recognition layer, it dynamically generates an attention weight matrix for the microenvironmental feature data of each road section in the feature input layer. The specific rules are as follows: When it is identified as fruit-type greening waste, a higher weight is given to the humidity index and ventilation index in the microenvironmental characteristic data; When petal-type green waste is identified, higher weights are assigned to the ventilation index, average light duration, and average light intensity in the microenvironmental characteristic data; The feature fusion layer is connected to the cross-modal attention layer and the feature input layer. It combines the attention-weighted microenvironmental features of each road section with the amount of green waste and pedestrian density data to form a fused feature vector and output it externally. The hidden layer is connected to the feature fusion layer to obtain the fused feature vector output by the feature fusion layer, and the fused feature vector is nonlinearly transformed and then outputted externally; The output layer is connected to the hidden layer to obtain the output of the hidden layer for processing and then output the sanitation operation load value of each road section.

5. The task allocation method based on greening data processing according to claim 1 is characterized in that: The crowd density data is obtained by calling the thermal data provided by the map service provider.

6. The task allocation method based on greening data processing according to claim 1 is characterized in that: Determining the types of greenery waste and the amounts of each type of greenery waste in a preset time period in the future for each road section based on the greenery plant data includes the following steps: Classify the green plants in each road section according to their species, and calculate the number and average age of different green plant species; The number, average age, and current season information of each greening plant species are input into the corresponding greening waste quantity prediction model. The greening waste quantity prediction model uses an LSTM time series neural network to predict the type and quantity of greening waste generated by each greening plant species in a preset time period in the future; The prediction results of all green plant species in each road section are summarized, and the green waste types are classified and accumulated to obtain the green waste types and the quantity of each type of green waste in each road section in the future preset time period.

7. The task allocation method based on greening data processing according to claim 1 is characterized in that: Based on the sanitation workload value, sanitation tasks are allocated to each sanitation team through a preset combined optimization algorithm, including the following steps: Each road section is regarded as a task item to be assigned, and each task item includes a road section identifier and a corresponding sanitation operation load value; Construct a road segment adjacency graph, treat each road segment as a graph node, establish undirected edge connections based on the spatial adjacency between road segments, and form an undirected graph structure that describes the continuity of the road segments; The preset working capacity limit is used as a constraint condition, so that each sanitation team corresponds to a packing container with a capacity equal to the preset working capacity limit; Based on the bin packing algorithm, a road segment continuity constraint is added. Adjacent road segments in the undirected graph are preferentially assigned to the same sanitation team. The optimization is performed with load balancing and road segment continuity as the dual objective function. This ensures that the total load value assigned to each sanitation team does not exceed the preset working capacity limit, and that the assigned road segments form a connected subgraph in the undirected graph as much as possible. After running the above packing algorithm process, the task allocation plan for each sanitation team is output, including a list of road section signs that each sanitation team is responsible for.

8. The task allocation system based on greening data processing is characterized by: It includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the task allocation method based on greening data processing as described in any one of claims 1-7.

Images