A method and system for planning a distribution operation plan for a food distribution vehicle
By constructing anomaly tree model and prediction model, evaluating the abnormal operation mode of the food delivery vehicle and optimizing the distribution plan, the problem that traditional technology cannot effectively deal with dynamic abnormal situations is solved, and the accuracy and economic benefits of distribution are improved.
Patent Information
- Application Number
- CN202411267844.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-11
AI Technical Summary
The dispatch and operational optimization of existing food delivery vehicles relies on traditional data analysis and empirical judgments, and cannot effectively deal with complex and dynamic abnormal situations such as weather changes, traffic jams or vehicle failures, resulting in delivery delays, increased costs and customer satisfaction. reduce.
By obtaining and standardizing the historical data of food delivery vehicles, applying reliability engineering and machine learning algorithms, building exception tree models and prediction models, evaluating the cost of abnormal operation modes, and formulating an optimized delivery plan.
It realizes accurate prediction of abnormal operations and optimization of distribution plans, improves the accuracy, efficiency and economic benefits of distribution, enhances the ability to adapt to dynamic environmental changes, and reduces operating costs.
Smart Images

Figure CN118798762B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent logistics technology, and in particular to a method and system for planning a delivery operation plan for a food delivery vehicle. Background Art
[0002] In today's rapidly developing logistics and distribution industry, the scheduling and operation optimization of food delivery vehicles are crucial. Current technologies mainly rely on traditional data analysis and empirical judgment to plan delivery operations. However, these methods have some limitations, such as strong dependence on historical data, insufficient prediction of abnormal operations, and slow response to real-time environmental changes. Traditional technologies usually respond to potential delivery problems by manually adjusting delivery routes and times, but this approach often cannot effectively handle complex and dynamic abnormal situations such as weather changes, traffic jams or vehicle failures. These problems lead to delivery delays, increased costs and reduced customer satisfaction, affecting overall delivery efficiency and economic benefits.
[0003] Therefore, there is a need for a method and system that can effectively predict abnormal operations that may occur in the future, evaluate their impact on distribution, and formulate optimized distribution plans, thereby improving distribution efficiency, reducing costs, and enhancing the ability to adapt to abnormal situations. Summary of the invention
[0004] The purpose of the present invention is to provide a method and system for planning a delivery operation plan for a food delivery vehicle to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0005] In a first aspect, the present application provides a method for planning a delivery operation plan for a food delivery vehicle, comprising:
[0006] Obtain historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records, and delivery operation environment data of food delivery vehicles, and standardize and normalize the data to obtain cleaned and standardized data;
[0007] Performing statistical analysis using reliability engineering methods based on the standardized data, identifying major abnormal operation modes of historical distribution operations, and establishing an abnormal tree model to obtain abnormal operation mode data;
[0008] According to the abnormal operation mode data, the impact of different abnormal operation modes on the delivery operation is evaluated, and the cost data set of the abnormal operation mode is obtained by quantifying the direct and indirect costs of various abnormal operation modes;
[0009] Based on the cost data set of the abnormal operation mode, a machine learning algorithm is used to perform model building processing, and the model is verified and evaluated to obtain an abnormal operation mode prediction model;
[0010] According to the abnormal operation mode prediction model, a preventive distribution operation plan is formulated, and the preventive distribution operation plan is optimized by evaluating the cost-effectiveness of different distribution operation plans to obtain an optimized distribution operation plan.
[0011] In a second aspect, the present application also provides a delivery operation plan planning system for a food delivery vehicle, comprising:
[0012] An acquisition unit is used to acquire historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records and delivery operation environment data of food delivery vehicles, and standardize and normalize the data to obtain cleaned standardized data;
[0013] An analysis unit, configured to perform statistical analysis based on the standardized data using a reliability engineering method, identify the main abnormal operation modes of historical delivery operations, and establish an abnormal tree model to obtain abnormal operation mode data;
[0014] An evaluation unit, configured to evaluate the impact of different abnormal operation modes on the delivery operation according to the abnormal operation mode data, and obtain a cost data set of the abnormal operation mode by quantifying the direct and indirect costs of various abnormal operation modes;
[0015] A construction unit is used to perform model construction processing using a machine learning algorithm according to the cost data set of the abnormal operation mode, and to perform model verification and evaluation to obtain an abnormal operation mode prediction model;
[0016] The optimization unit is used to formulate a preventive distribution operation plan according to the abnormal operation mode prediction model, optimize the preventive distribution operation plan by evaluating the cost-effectiveness of different distribution operation plans, and obtain an optimized distribution operation plan.
[0017] The beneficial effects of the present invention are:
[0018] The present invention significantly improves the accuracy, efficiency and economic benefits of distribution operations by accurately predicting abnormal operations, optimizing distribution plans, improving the scientificity and efficiency of decision-making, and enhancing the ability to adapt to dynamic environmental changes. Using advanced data analysis and machine learning technology, this method can accurately identify potential problems, formulate optimal distribution strategies, reduce operating costs, and flexibly respond to environmental changes, thereby improving the stability and performance of the overall distribution system.
[0019] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0021] Figure 1 A schematic diagram of a method for planning a delivery operation plan for a food delivery vehicle according to an embodiment of the present invention;
[0022] Figure 2 It is a schematic diagram of the structure of a delivery operation plan planning system for food delivery vehicles described in an embodiment of the present invention.
[0023] In the figure: 701, acquisition unit; 702, analysis unit; 703, evaluation unit; 704, construction unit; 705, optimization unit; 7021, first calculation subunit; 7022, first processing subunit; 7023, first analysis subunit; 7024, second analysis subunit; 7025, second processing subunit; 7031, second calculation subunit; 7032, third calculation subunit; 7033, third processing subunit; 7034, fourth processing subunit; 7041, first construction subunit; 7042, second construction subunit; 7043, third construction subunit; 7044, fourth construction subunit; 7045, fifth construction subunit; 7051, fifth processing subunit; 7052, fourth calculation subunit; 7053, sixth processing subunit; 7054, seventh processing subunit. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0025] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0026] Embodiment 1:
[0027] This embodiment provides a method for planning a delivery operation plan for a food delivery vehicle.
[0028] See also Figure 1 , the figure shows that the method includes step S1, step S2, step S3, step S4 and step S5.
[0029] Step S1, obtaining historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records and delivery operation environment data of food delivery vehicles, and standardizing and normalizing the data to obtain cleaned standardized data;
[0030] It should be noted that the historical delivery operation data, abnormal delivery operation records, modification records of delivery operation plans and delivery operation environment data of food delivery vehicles are derived from the information accumulated during the long-term operation of food delivery vehicles, including the operation time, transportation times, transportation operation plans and abnormal transportation operation of food delivery vehicles, as well as key parameters of the delivery environment of food delivery vehicles such as the plant area of the delivery operation, the road conditions of the delivery operation, the temperature and humidity of the delivery operation process, etc. In practice, the demand side needs to ensure that the data obtained from the delivery side is complete and accurate. These data are not only used to evaluate the performance of past delivery operation plans, but also provide a basis for future predictions and decisions. By standardizing and normalizing the data, the differences between different data sources can be eliminated to ensure the consistency and comparability of the data. For example, for the data of food delivery vehicles of different time periods or different models, standardization can unify their units and data formats so that they can be effectively processed by the same set of analysis tools.
[0031] Step S2: applying reliability engineering methods to perform statistical analysis based on the standardized data, identifying the main abnormal operation modes of historical delivery operations, and establishing an abnormal tree model to obtain abnormal operation mode data;
[0032] It should be noted that reliability engineering refers to the analysis and improvement of the reliability of products, systems or processes through a systematic approach to ensure their stability and durability under predetermined conditions. Its main goal is to identify and prevent potential failures, extend the life of the system, and improve its operating efficiency in various environments. Reliability engineering usually involves multiple aspects such as failure mode identification, reliability prediction, preventive maintenance and risk management, aiming to reduce the occurrence of unexpected failures and optimize maintenance strategies and resource allocation. In this embodiment, the reliability engineering method is applied to the logistics distribution system to improve the stability and efficiency of the distribution process.
[0033] Furthermore, in this step, the standardized data is used to construct an abnormal tree model, in which each node represents a basic event that may cause an abnormal operation, and the logic gates connecting these nodes represent the causal relationship between the events. In this step, step S2 includes step S21, step S22, step S23, step S24 and step S25.
[0034] Step S21, calculating the occurrence frequency and distribution of historical abnormal operations according to the standardized data to obtain statistical characteristic data of historical abnormal operations;
[0035] It is understandable that the frequency of abnormal operations refers to the number of abnormal operations in the delivery process within a certain time range. This data can be obtained by statistically analyzing historical abnormal operation records. The distribution of abnormal operations describes the distribution characteristics of different abnormal operations in time and space, and reflects the occurrence rules and trends of abnormal operations. By conducting in-depth analysis of these data, the abnormal operation patterns under different conditions in the delivery process can be revealed, thereby providing a key basis for subsequent model construction. The calculation formula is as follows:
[0036] ;
[0037] ;
[0038] in, An index indicating an abnormal operation mode; Represents the index of the delivery operation record; Indicates the total number of delivery operations; Indicates The frequency of abnormal operation. Indicates The number of occurrences of class abnormal operations; Indicates The time of the delivery operation; Indicates environmental parameters; Indicates operation parameters; Indicates Complexity coefficient of class exception operations; Indicates the influencing factors of environmental factors; Indicates Probability distribution of abnormal operations; Represents the total time of the delivery operation.
[0039] Step S22, performing clustering processing according to the statistical feature data of the historical abnormal operations, identifying the main abnormal modes of the historical abnormal operations by combining the distribution operation environment data to obtain mode classification data of the abnormal operations;
[0040] It can be understood that this step can effectively extract abnormal operation modes with similar characteristics from a large amount of historical abnormal data through cluster analysis, reducing the subjectivity and errors in the data analysis process. A systematic understanding of historical abnormal operation modes has been established, providing a basis for the establishment of subsequent abnormal tree models. Combining the analysis of operating environment data with abnormal operation modes can more comprehensively evaluate the impact and probability of abnormal operations under different environmental conditions, and provide support for the construction of subsequent cost evaluation and prediction models. Among them, the distribution operation environment data includes environmental factors such as weather conditions, traffic conditions, and distribution routes, which can affect the occurrence and impact of abnormal operations. Suppose we have a set of historical abnormal data of food delivery vehicles, including weather conditions and traffic congestion. By clustering these data, we may find pattern classification data of abnormal distribution operations such as delayed distribution in hot weather and route blockage caused by traffic accidents.
[0041] Step S23, performing association analysis according to the pattern classification data of the abnormal operation, analyzing the association relationship between different abnormal operations in the historical delivery operations and extracting abnormal operation combinations based on a preset dynamic threshold mechanism, to obtain pattern association data of the abnormal operation;
[0042] It is understandable that this step can reveal the intrinsic relationship between different abnormal operation modes through association analysis, which helps to understand abnormal behaviors in complex distribution operation environments. According to the results of association rule mining, abnormal operation combinations that often occur simultaneously or have mutual influence are extracted. The final pattern association data includes the frequency, association strength, conditions for occurrence and impact of each abnormal operation combination. The extraction of abnormal operation combinations based on association analysis can improve the accuracy of subsequent anomaly detection and prediction models. Understanding the association relationship of abnormal operation modes can help to formulate more effective preventive measures and reduce the probability and impact of abnormal occurrence.
[0043] Step S24, performing model construction processing according to the pattern association data of the abnormal operation, determining the logical relationship between each basic event and the top event of the historical distribution operation by using a fault tree analysis algorithm, and constructing an abnormal tree model;
[0044] It should be noted that the fault tree is a tool for systematic analysis of abnormal operation mechanisms. It describes various possible basic events that lead to abnormal operations and their relationships through a logical tree structure. The use of fault trees can provide a comprehensive understanding and quantitative analysis of abnormal operation modes. Its advantages are: first, the fault tree can clearly show the logical relationship and causal chain between different abnormal operation events, helping to identify and understand various abnormal modes that may occur in distribution operations. This structured analysis method makes it easier to determine the core factors that dominate abnormal operations in complex systems; second, fault tree analysis allows for the construction of models based on historical data, taking into account multiple possible abnormal operation situations and their probability distribution, and more accurately predicting the probability of occurrence of different abnormal operation modes, thereby effectively formulating preventive distribution operation plans and optimizing cost management.
[0045] It can be understood that this step helps identify and focus on the key factors that lead to top abnormal events through a deep understanding of the causal relationship between abnormal operation events through fault tree analysis. The abnormal tree model can be used to assess the risk level and potential impact of different abnormal operation modes, which helps to optimize distribution operation plans and resource allocation. In addition, fault tree analysis provides data-driven decision support, enabling managers to make more informed decisions when facing complex distribution environments and abnormal situations.
[0046] Step S25: performing decomposition processing according to the abnormal tree model, generating abnormal operation mode data by identifying and extracting abnormal probabilities and influencing factors of key paths and key events.
[0047] It can be understood that this step can focus on the factors that have the greatest impact on distribution operations by decomposing and identifying key paths and key events, and improve the pertinence and effectiveness of abnormal operation management. Among them, abnormal operation mode data provides a detailed understanding of key risks, which helps to optimize risk management strategies and reduce the probability and impact of abnormal events.
[0048] Step S3: Based on the abnormal operation mode data, evaluate the impact of different abnormal operation modes on the delivery operation, and obtain a cost data set of the abnormal operation mode by quantifying the direct and indirect costs of various abnormal operation modes;
[0049] It is understandable that this step helps managers better understand and manage the risks of various abnormal operation modes by quantifying cost data, and take targeted measures to reduce the risk level. This step also provides economic data support to help decision makers make wise decisions when formulating distribution strategies and optimizing resource allocation. This step performs cost-benefit analysis based on the cost data set to find the abnormal operation mode with the highest cost, and give priority to the abnormal operation mode with the greatest impact on the cost-benefit ratio. In this step, step S3 includes step S31, step S32, step S33 and step S34.
[0050] Step S31, inputting the abnormal operation mode data into a preset fuzzy comprehensive evaluation model, constructing a fuzzy matrix and determining a membership function to perform fuzzy evaluation on the influence of all abnormal operation modes under different distribution operation environments, and calculating the risk value of each abnormal operation mode to obtain abnormal operation mode risk value data;
[0051] It should be noted that the fuzzy matrix is used to describe the fuzzy relationship and weights between different factors. These factors include the probability of abnormal operation, influencing factors (such as environmental conditions, etc.), and the severity of the consequences of abnormal operation. The membership functions and mutual correlations of these factors are determined through expert experience or based on historical data. Constructing a fuzzy matrix is the basis of the evaluation process. The membership function is used to quantify the degree of fuzziness of each factor. For example, environmental conditions are divided into different membership levels (such as high, medium, and low) so that the model can handle uncertainty and complexity. The fuzzy judgment and calculation process considers not only the possibility of abnormal operation itself, but also the possible environmental impact and consequences when it occurs. Finally, through fuzzy logic operations and weighted calculations, the risk value data of each abnormal operation mode is obtained. These data directly reflect the potential impact of each abnormal operation on the overall distribution. The membership function is as follows:
[0052] ;
[0053] in, Indicates abnormal operation mode In delivery conditions The membership value under ; An index representing the category of abnormal operation mode; An index representing the delivery conditions; represents the index of the Fourier term; represents the total number of Fourier terms; and They are the parameters for adjusting the membership shape and offset respectively; , and are the amplitude, frequency and phase, respectively, terms used in the Fourier transform.
[0054] Step S32: According to the abnormal operation mode risk value data, the activity-based cost method is used to quantify the direct cost of each abnormal operation mode, and the cost of each operation is calculated by identifying each delivery operation and the resources consumed to obtain direct cost data;
[0055] It can be understood that this step determines the specific distribution activities related to the abnormal operation mode. For example, loading goods, transportation, delivery, waiting, etc. Then analyze the resources consumed by each activity, including manpower, materials, equipment, time, etc. Among them, the activity-based costing method improves the accuracy of cost allocation and can more accurately quantify the direct costs of various abnormal operation modes. It provides detailed cost data, reveals the cost structure of various distribution operations and resource consumption, and improves cost transparency. It provides an accurate data basis for subsequent cost-benefit analysis and optimization, helping managers to formulate more effective prevention and optimization measures.
[0056] Step S33: construct an abnormal operation impact hierarchy model according to the direct cost data, and evaluate the impact of various abnormal operations on shutdown losses, transportation losses and additional costs based on the abnormal operation impact hierarchy model to obtain indirect cost data;
[0057] Specifically, when constructing the abnormal operation impact hierarchy model, it is necessary to first conduct a detailed analysis of the actual scenarios in the distribution operation process to identify various types of abnormal operations that may occur. These abnormal operations include mechanical failures of distribution vehicles, sudden traffic jams encountered during transportation, temporary changes in distribution routes, and inventory errors or item damage in warehouse management. The top layer of the model is set as the overall target layer, and its main goal is to evaluate the impact of abnormal operations on the overall distribution cost, especially the impact on downtime losses, transportation losses, and additional costs. The middle layer is refined into the influencing factor layer, which specifically includes downtime losses (such as repair time and spare vehicle scheduling time caused by vehicle failure), transportation losses (such as increased fuel consumption and delay time caused by traffic congestion or route changes), and additional costs (such as additional fuel consumption caused by inventory management errors and labor costs for rescheduling). The bottom layer lists specific abnormal operations, which have different degrees of impact on each factor in the middle layer, such as vehicle failure, traffic congestion, and inventory management errors. The construction process includes data collection and preprocessing, extracting information such as the frequency, duration, and direct impact of various abnormal operations from historical operation data and real-time monitoring systems. By analyzing the direct consequences of these operations, we determine how they affect downtime losses, transportation losses, and additional costs, and then use the hierarchical analysis method to assign weights. Finally, we use methods such as weighted sum to calculate the impact of each abnormal operation on the total cost, and quantify these impacts as indirect cost data. This hierarchical structure model not only helps companies make more accurate decisions when formulating risk management strategies, but also optimizes resource allocation by identifying high-risk operations.
[0058] Step S34: Integrate the direct cost data and the indirect cost data according to the mode to obtain a cost data set of the abnormal operation mode.
[0059] It can be understood that this step achieves comprehensive coverage of the costs of various abnormal operation modes by integrating direct costs and indirect costs. Through standardized data integration processing, the integrity and consistency of cost data are ensured, which is convenient for subsequent analysis and decision-making.
[0060] Step S4: Based on the cost data set of the abnormal operation mode, a machine learning algorithm is used to perform model construction processing, and the model is verified and evaluated to obtain an abnormal operation mode prediction model;
[0061] It should be noted that machine learning algorithms are suitable for processing large-scale and complex data, and can extract hidden patterns and associations from them, which are difficult to fully capture in traditional statistical methods. In this step, the application of machine learning can help formulate more accurate and effective distribution plans by analyzing historical abnormal operation data and predicting abnormal operations that may occur in the future. The construction of this prediction model not only considers the influence of a single factor, but also improves the accuracy and practicality of the prediction by comprehensively considering the complex relationship between multiple variables. Through model verification and evaluation, the algorithm parameters are further optimized to ensure the applicability and robustness of the model under different distribution environment conditions, thereby providing reliable decision support for distribution management. In this step, step S4 includes step S41, step S42, step S43, step S44 and step S45.
[0062] Step S41, performing feature selection and extraction processing on the cost data set of the abnormal operation mode to obtain an initial data set;
[0063] Furthermore, this step determines which features are most critical for evaluating and predicting the cost of abnormal operation modes through various cost data and related influencing factors in the cost data set of abnormal operation modes. Next, the feature selection algorithm is used to screen out key features, remove redundant or irrelevant features, and standardize these key feature data to form an optimized initial data set, so that it can better reflect the cost characteristics under different abnormal operation modes, which is convenient for subsequent modeling and analysis.
[0064] Step S42: Based on the initial data set, the abnormal operation mode features that have the greatest impact on the historical delivery operations of the food delivery vehicle are predicted by the LASSO regression method to obtain a feature data set;
[0065] It can be understood that the LASSO regression method (Least Absolute Shrinkage and Selection Operator) in this step is a linear regression model that performs variable selection and regularization by adding an L1 norm penalty term, which has the advantages of automatic feature selection, simplified model and prevention of overfitting. Specifically, the initial data set is input into the LASSO regression model, and the abnormal operation mode features that have the greatest impact on cost prediction are screened out by adjusting the penalty term parameters, and finally a feature data set is obtained. These features are analyzed in depth to identify the factors that have the greatest impact on cost, providing a scientific basis for formulating more reasonable maintenance strategies. Although the preliminary feature selection has screened the key features, the LASSO regression method further automatically selects these features to ensure that the model is more concise and excludes redundant features, improves the stability and accuracy of the prediction model, and thus more reasonably performs the cost accounting of abnormal operations. Through the LASSO regression method, abnormal operation mode data can be efficiently processed, key features can be extracted, and more reasonable cost accounting can be achieved, thereby improving the scientificity and efficiency of maintenance management.
[0066] Step S43, dividing the feature data set into a training data set and a test data set according to preset time points;
[0067] It should be noted that this step divides the data into training data sets and test data sets by time point segmentation to ensure the effectiveness of model training and evaluation. By adapting to the characteristics of time series data, the time sequence of training and test data sets is ensured to be reasonable, thus improving the performance of the model in practical applications.
[0068] Step S44, training and optimizing the training data set, constructing a random forest model to predict the historical delivery operation costs of food delivery vehicles, and using a grid search method to optimize model parameters, and training to obtain an initial abnormal operation prediction model;
[0069] It is understandable that this step uses the training set to build a random forest model. Random forest is an integrated learning method based on decision trees. By building multiple decision trees and combining their prediction results, the stability and accuracy of the model are enhanced. Then the grid search method is used to optimize the model parameters. The grid search finds the parameter set that makes the model perform best by exhaustively enumerating the preset parameter combinations to improve the prediction ability of the model. Specific parameter optimization includes the adjustment of parameters such as the number of trees, the maximum depth of the tree, and the minimum number of sample splits. The optimized random forest model can effectively process high-dimensional data and has strong generalization ability and robustness, so as to accurately predict the distribution operation cost and obtain a preliminary fault prediction model.
[0070] Step S45: perform model verification and evaluation based on the initial abnormal operation prediction model, verify the model by using a test data set, evaluate the prediction performance of the model, and adjust and optimize to obtain a final abnormal operation mode prediction model.
[0071] Furthermore, the purpose of model validation is to evaluate the performance of the model on unseen data, so as to determine the generalization ability and prediction performance of the model. In this process, evaluation indicators can include mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), R² value, etc., to quantify the prediction accuracy and robustness of the model. If the evaluation results show that the model's performance in some aspects is not ideal, the model parameters can be further adjusted and optimized, such as by adjusting the number of trees in the random forest model, depth, minimum number of samples for splitting nodes, etc. Through continuous verification, evaluation and adjustment, a fault prediction model with better performance is finally obtained.
[0072] Step S5: formulate a preventive distribution operation plan according to the abnormal operation mode prediction model, optimize the preventive distribution operation plan by evaluating the cost-effectiveness of different distribution operation plans, and obtain an optimized distribution operation plan.
[0073] It can be understood that this step effectively reduces the risk of abnormal operation and improves the stability of distribution by formulating and optimizing preventive distribution operation plans. Among them, through cost-benefit analysis and optimization selection, the overall cost of distribution operations is reduced and economic benefits are improved. In this step, step S5 includes step S51, step S52, step S53 and step S54.
[0074] Step S51, obtaining real-time delivery operation simulation data of a preset delivery operation plan, and inputting the real-time delivery operation data into an abnormal operation mode prediction model, and obtaining probability data of abnormal operation by predicting the probability of abnormal operation in a future preset time period, wherein the delivery operation plan is a historical operation plan;
[0075] It should be noted that this step simulates and predicts real-time data, using an abnormal operation mode prediction model to predict the future preset time period and calculate the probability of abnormal operation. These predictions can cover various possible abnormal situations, such as delivery delays, route blockages, etc. It can accurately assess the risks of abnormal operations that may occur in the future. It provides prediction information based on the latest data to help formulate more scientific and effective delivery operation plans.
[0076] Step S52: Based on the Bayesian solution model, the probability data of the abnormal operation is combined with the redistribution cost and the abnormal operation cost to calculate the benefits of the distribution operation solutions under different scenarios, and the distribution operation solution with the greatest benefit is selected as the preventive distribution operation solution;
[0077] Furthermore, applying the Bayesian decision model involves the following steps:
[0078] Step S521: Calculate the probability of abnormal operation in a future preset time period using Bayesian theorem based on real-time monitoring data and historical data.
[0079] It is understandable that this process uses Bayesian theorem to combine the instant information of real-time monitoring data and the statistical information of historical data to evaluate the probability of abnormal operation in the future preset time period. These probabilities serve as the basis for evaluating different scenarios, which cover different changes in the food delivery process and the diversity of environmental conditions. Bayesian theorem is as follows:
[0080] ;
[0081] in, Indicates the event name; Indicates the name of the evidence; Indicated in the evidence When the event occurs Probability of occurrence; Indicates in the event Evidence of occurrence Probability of occurrence; Indicates an event The prior probability of Show evidence The prior probability of .
[0082] Step S522: Calculate the expected benefit of each distribution operation plan based on the maintenance decision benefits under different scenarios.
[0083] Specifically, for each scenario, each delivery operation plan is evaluated by combining the expected benefits of the delivery operation plan, including the expected cost of the delivery operation plan, the reduced delivery losses caused by failures, and the savings from additional energy consumption. The calculation formula is:
[0084] ;
[0085] in, The index representing the delivery operation plan; An index representing a scenario; Indicates the total number of scenarios; Indicates the delivery operation plan Expected return; Indicates the delivery operation plan In the scenario The following income; For the scenario The probability of the following.
[0086] Step S523: Calculate the total cost of each distribution operation plan.
[0087] The calculation formula is:
[0088] ;
[0089] in, Indicates the delivery operation plan Total cost under represents the delivery cost; It represents the cost of redistribution required after an abnormal distribution operation plan; For the scenario The probability of the following.
[0090] Step S524: By comparing the expected benefits of different distribution operation plans, the distribution operation plan with the greatest benefit is selected as the preventive distribution operation plan.
[0091] Finally, by calculating the expected return of each strategy, we can select the distribution operation plan with the greatest return in a given situation, and thus formulate a preliminary distribution operation plan. The calculation formula is:
[0092] ;
[0093] in, represents the optimal maintenance strategy, Indicates the delivery operation plan The expected return, Indicates the delivery operation plan The total cost of .
[0094] Step S53: performing a cost-effectiveness evaluation according to the preventive delivery operation plan, wherein the execution effects of all preventive delivery operation plans are simulated by a Monte Carlo simulation algorithm and the cost-effectiveness of each preventive delivery operation plan is quantified to obtain cost-effectiveness data;
[0095] It is understandable that in order to evaluate the actual effect of preventive distribution operation plans, it is necessary to simulate various uncertain factors, such as abnormalities in the distribution process, changes in distribution time, changes in the distribution environment, etc. The Monte Carlo simulation algorithm simulates the performance of different maintenance strategies in the actual implementation process through a large number of random sampling and repeated calculations. Specifically, these simulations include: the possible distribution operation frequency of the equipment in different time periods, the cost of each distribution operation, the distribution operation time, and the impact of these factors on the entire distribution operation. Through these simulations, the cost-effectiveness of each distribution operation plan in different scenarios can be quantified, and the cost-effectiveness data of each distribution operation can be obtained. These data directly reflect the economy and effectiveness of different distribution operation plans in actual implementation, and provide a scientific basis for further optimization and selection of the final distribution operation plan.
[0096] Step S54, using a genetic algorithm, taking the cost-benefit data as a fitness function, performing crossover and mutation operations, optimizing the execution plan of the scheme, selecting the preventive delivery operation scheme with the highest fitness, and taking the preventive delivery operation scheme with the highest fitness as the final delivery operation scheme for the food delivery vehicle.
[0097] Specifically, the population is initialized first, in which each individual represents a preventive delivery operation plan. Then, the cost-effectiveness data is used as the fitness function to evaluate the fitness of each individual. The fitness function is defined as the fitness equal to the inverse of the cost-effectiveness. According to the fitness, individuals with higher fitness are selected for reproduction, and individuals with higher fitness have a greater probability of being selected. The selected individuals are crossover operated to generate new individuals. The crossover operation generates new offspring individuals by exchanging part of the genes of the parent individuals. The offspring individuals are mutated to randomly change part of the genes of the individuals with a certain probability to increase the diversity of the population. The new individuals generated by the crossover and mutation operations form a new population. Repeat the process of fitness evaluation, selection, crossover and mutation until the termination condition is met (the fitness is no longer significantly improved). Through the above process, the genetic algorithm continuously optimizes the preventive delivery operation plan, selects the maintenance plan with the highest fitness, and forms the final preventive delivery operation plan. In this way, the genetic algorithm can effectively explore the search space of the maintenance strategy and find the optimal preventive delivery operation plan, making the optimization result more realistic and operational. The fitness function constructed in this step is:
[0098] ;
[0099] in, represents the fitness function; , , and Respectively represent weight coefficients; represents the cost of preventive delivery operation plan; It represents the cost required for redelivery after abnormal delivery operation; Indicate the benefits of preventive delivery operation plan; Indicates the probability of abnormal delivery operation; It indicates the effective utilization of resources (such as manpower, time, materials, etc.) by preventive distribution operation plan.
[0100] Embodiment 2:
[0101] like Figure 2 As shown, this embodiment provides a distribution operation plan planning system for food distribution vehicles, see Figure 2 The system includes an acquisition unit 701 , an analysis unit 702 , an evaluation unit 703 , a construction unit 704 and an optimization unit 705 .
[0102] The acquisition unit 701 is used to acquire the historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records and delivery operation environment data of the food delivery vehicle, and standardize and normalize the data to obtain cleaned standardized data;
[0103] An analysis unit 702 is used to perform statistical analysis based on the standardized data using a reliability engineering method, identify the main abnormal operation modes of historical delivery operations, and establish an abnormal tree model to obtain abnormal operation mode data;
[0104] The analysis unit 702 includes a first calculation subunit 7021 , a first processing subunit 7022 , a first analysis subunit 7023 , a second analysis subunit 7024 and a second processing subunit 7025 .
[0105] The first calculation subunit 7021 is used to calculate the occurrence frequency and distribution of historical abnormal operations according to the standardized data to obtain statistical feature data of the historical abnormal operations;
[0106] The first processing subunit 7022 is used to perform clustering processing according to the statistical feature data of the historical abnormal operations, and obtain pattern classification data of the abnormal operations by identifying the main abnormal patterns of the historical abnormal operations in combination with the distribution operation environment data;
[0107] The first analysis subunit 7023 is used to perform association analysis based on the pattern classification data of the abnormal operation, by analyzing the association relationship between different abnormal operations in the historical delivery operation and extracting abnormal operation combinations based on a preset dynamic threshold mechanism, to obtain the pattern association data of the abnormal operation;
[0108] The second analysis subunit 7024 is used to perform model building processing according to the pattern association data of the abnormal operation, and to construct an abnormal tree model by using a fault tree analysis algorithm to determine the logical relationship between each basic event and the top event of the historical distribution operation;
[0109] The second processing subunit 7025 is used to perform decomposition processing according to the abnormal tree model, and generate abnormal operation mode data by identifying and extracting abnormal probabilities and influencing factors of key paths and key events.
[0110] An evaluation unit 703 is used to evaluate the impact of different abnormal operation modes on the delivery operation according to the abnormal operation mode data, and obtain a cost data set of the abnormal operation mode by quantifying the direct and indirect costs of various abnormal operation modes;
[0111] The evaluation unit 703 includes a second calculation subunit 7031 , a third calculation subunit 7032 , a third processing subunit 7033 and a fourth processing subunit 7034 .
[0112] The second calculation subunit 7031 is used to input the abnormal operation mode data into a preset fuzzy comprehensive evaluation model, to perform fuzzy evaluation on the influence of all abnormal operation modes under different distribution operation environments by constructing a fuzzy matrix and determining a membership function, and to calculate the risk value of each abnormal operation mode to obtain the abnormal operation mode risk value data;
[0113] The third calculation subunit 7032 is used to quantify the direct cost of each abnormal operation mode by using the activity-based cost method according to the abnormal operation mode risk value data, calculate the cost of each operation by identifying each delivery operation and the resources consumed, and obtain direct cost data;
[0114] The third processing subunit 7033 is used to construct an abnormal operation impact hierarchy model according to the direct cost data, and evaluate the impact of various abnormal operations on shutdown losses, transportation losses and additional costs based on the abnormal operation impact hierarchy model to obtain indirect cost data;
[0115] The fourth processing sub-unit 7034 is used to integrate the direct cost data and the indirect cost data according to the mode to obtain a cost data set of the abnormal operation mode.
[0116] A construction unit 704 is used to perform model construction processing using a machine learning algorithm according to the cost data set of the abnormal operation mode, and to perform model verification and evaluation to obtain an abnormal operation mode prediction model;
[0117] The construction unit 704 includes a first construction subunit 7041 , a second construction subunit 7042 , a third construction subunit 7043 , a fourth construction subunit 7044 and a fifth construction subunit 7045 .
[0118] The first construction subunit 7041 is used to perform feature selection and extraction processing on the cost data set of the abnormal operation mode to obtain an initial data set;
[0119] The second construction subunit 7042 is used to predict the abnormal operation mode features that have the greatest impact on the historical delivery operations of the food delivery vehicles based on the initial data set by using the LASSO regression method to obtain a feature data set;
[0120] The third construction subunit 7043 is used to divide the feature data set into a training data set and a test data set according to preset time points;
[0121] The fourth construction subunit 7044 is used to perform model training and optimization on the training data set, perform prediction training on the historical delivery operation cost of food delivery vehicles by constructing a random forest model, and use a grid search method to optimize model parameters, and train to obtain an initial abnormal operation prediction model;
[0122] The fifth construction subunit 7045 is used to perform model verification and evaluation based on the initial abnormal operation prediction model, verify the model by using a test data set, evaluate the prediction performance of the model, and adjust and optimize to obtain the final abnormal operation prediction model.
[0123] The optimization unit 705 is used to formulate a preventive delivery operation plan according to the abnormal operation mode prediction model, optimize the preventive delivery operation plan by evaluating the cost-effectiveness of different delivery operation plans, and obtain an optimized delivery operation plan.
[0124] The optimization unit 705 includes a fifth processing subunit 7051 , a fourth calculation subunit 7052 , a sixth processing subunit 7053 and a seventh processing subunit 7054 .
[0125] The fifth processing sub-unit 7051 is used to obtain real-time delivery operation simulation data of a preset delivery operation plan, and input the real-time delivery operation data into the abnormal operation mode prediction model to obtain the probability data of abnormal operation by predicting the probability of abnormal operation in a future preset time period, wherein the delivery operation plan is a historical operation plan;
[0126] The fourth calculation subunit 7052 is used to combine the probability data of the abnormal operation with the redistribution cost and the abnormal operation cost based on the Bayesian solution model, calculate the benefits of the delivery operation solutions under different scenarios, and select the delivery operation solution with the maximum benefit as the preventive delivery operation solution;
[0127] The sixth processing sub-unit 7053 is used to perform cost-effectiveness evaluation according to the preventive delivery operation plan, wherein the execution effects of all preventive delivery operation plans are simulated by a Monte Carlo simulation algorithm and the cost-effectiveness of each preventive delivery operation plan is quantified to obtain cost-effectiveness data;
[0128] The seventh processing sub-unit 7054 is used to utilize a genetic algorithm, take the cost-benefit data as a fitness function, perform crossover and mutation operations, optimize the execution plan of the scheme, select the preventive delivery operation scheme with the highest fitness, and use the preventive delivery operation scheme with the highest fitness as the final delivery operation scheme for the food delivery vehicle.
[0129] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.
[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
[0131] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A method for planning a distribution operation plan for a food distribution vehicle, characterized in that: include: Obtain historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records, and delivery operation environment data of food delivery vehicles, and standardize and normalize the data to obtain cleaned and standardized data; Applying reliability engineering methods to perform statistical analysis based on the standardized data, identifying abnormal operation patterns of historical delivery operations, and establishing an abnormal tree model to obtain abnormal operation pattern data; According to the abnormal operation mode data, the impact of different abnormal operation modes on the delivery operation is evaluated, and the cost data set of the abnormal operation mode is obtained by quantifying the direct and indirect costs of various abnormal operation modes; Based on the cost data set of the abnormal operation mode, a machine learning algorithm is used to perform model building processing, and the model is verified and evaluated to obtain an abnormal operation mode prediction model; Formulate a preventive delivery operation plan according to the abnormal operation mode prediction model, optimize the preventive delivery operation plan by evaluating the cost-effectiveness of different delivery operation plans, and obtain an optimized delivery operation plan; According to the standardized data, a reliability engineering method is applied to perform statistical analysis, and abnormal operation mode data of historical distribution operations are obtained by identifying abnormal operation modes and establishing an abnormal tree model, including: Calculating the occurrence frequency and distribution of historical abnormal operations according to the standardized data to obtain statistical characteristic data of the historical abnormal operations; Performing clustering processing according to the statistical feature data of the historical abnormal operations, identifying the abnormal patterns of the historical abnormal operations by combining the distribution operation environment data to obtain pattern classification data of the abnormal operations; Performing association analysis based on the pattern classification data of the abnormal operation, by analyzing the association relationship between different abnormal operations in the historical delivery operation and extracting abnormal operation combinations based on a preset dynamic threshold mechanism, to obtain pattern association data of the abnormal operation; Performing model building processing according to the pattern association data of the abnormal operation, determining the logical relationship between each basic event and the top event of the historical distribution operation by using a fault tree analysis algorithm, and constructing an abnormal tree model; Decomposition processing is performed according to the abnormal tree model, and abnormal operation mode data is generated by identifying and extracting abnormal probabilities and influencing factors of key paths and key events; According to the abnormal operation mode data, the impact of different abnormal operation modes on the distribution operation is evaluated, and the cost data set of the abnormal operation mode is obtained by quantifying the direct and indirect costs of various abnormal operation modes, including: The abnormal operation mode data is input into a preset fuzzy comprehensive evaluation model, and the influence of all abnormal operation modes under different distribution operation environments is fuzzily judged by constructing a fuzzy matrix and determining a membership function, and the risk value of each abnormal operation mode is calculated to obtain the abnormal operation mode risk value data; According to the abnormal operation mode risk value data, the activity-based cost method is used to quantify the direct cost of each abnormal operation mode, and the cost of each operation is calculated by identifying each distribution operation and the resources consumed to obtain direct cost data; Constructing an abnormal operation impact hierarchy model according to the direct cost data, and evaluating the impact of various abnormal operations on shutdown losses, transportation losses and additional costs based on the abnormal operation impact hierarchy model to obtain indirect cost data; Integrate the direct cost data and the indirect cost data according to the mode to obtain a cost data set of the abnormal operation mode; According to the cost data set of the abnormal operation mode, a model is constructed using a machine learning algorithm, and the model is verified and evaluated to obtain an abnormal operation mode prediction model, including: Performing feature selection and extraction processing on the cost data set of the abnormal operation mode to obtain an initial data set; Based on the initial data set, the abnormal operation mode features that have the greatest impact on the historical delivery operations of the food delivery vehicle are predicted by the LASSO regression method to obtain a feature data set; Dividing the feature data set into a training data set and a test data set according to preset time points; The training data set is used for model training and optimization, a random forest model is constructed to predict the historical delivery operation costs of food delivery vehicles, and a grid search method is used to optimize model parameters to obtain an initial abnormal operation mode prediction model; Performing model verification and evaluation based on the initial abnormal operation mode prediction model, verifying the model by using a test data set, evaluating the prediction performance of the model, and adjusting and optimizing to obtain a final abnormal operation mode prediction model; According to the abnormal operation mode prediction model, a preventive distribution operation plan is formulated, and the preventive distribution operation plan is optimized by evaluating the cost-effectiveness of different distribution operation plans to obtain an optimized distribution operation plan, including: Acquire real-time delivery operation simulation data of a preset delivery operation plan, and input the real-time delivery operation data into an abnormal operation mode prediction model, and obtain abnormal operation probability data by predicting the probability of abnormal operation in a future preset time period, wherein the delivery operation plan is a historical operation plan; Based on the Bayesian solution model, the probability data of the abnormal operation is combined with the redistribution cost and the abnormal operation cost to calculate the benefits of the distribution operation plan under different scenarios, and the distribution operation plan with the largest benefit is selected as the preventive distribution operation plan; Performing a cost-effectiveness evaluation according to the preventive delivery operation plan, wherein the execution effects of all preventive delivery operation plans are simulated by a Monte Carlo simulation algorithm and the cost-effectiveness of each preventive delivery operation plan is quantified to obtain cost-effectiveness data; By using a genetic algorithm, the cost-benefit data is used as a fitness function to perform crossover and mutation operations, optimize the execution plan of the scheme, select the preventive delivery operation scheme with the highest fitness, and use the preventive delivery operation scheme with the highest fitness as the final delivery operation scheme for the food delivery vehicle.
2. A distribution operation plan planning system for food distribution vehicles, characterized in that: include: An acquisition unit is used to acquire historical delivery operation data, delivery operation abnormality records, delivery operation plan modification records and delivery operation environment data of food delivery vehicles, and standardize and normalize the data to obtain cleaned standardized data; An analysis unit, configured to perform statistical analysis based on the standardized data using a reliability engineering method, identify abnormal operation patterns of historical delivery operations, and establish an abnormal tree model to obtain abnormal operation pattern data; An evaluation unit, configured to evaluate the impact of different abnormal operation modes on the delivery operation according to the abnormal operation mode data, and obtain a cost data set of the abnormal operation mode by quantifying the direct and indirect costs of various abnormal operation modes; A construction unit is used to perform model construction processing using a machine learning algorithm according to the cost data set of the abnormal operation mode, and to perform model verification and evaluation to obtain an abnormal operation mode prediction model; An optimization unit, configured to formulate a preventive delivery operation plan according to the abnormal operation mode prediction model, optimize the preventive delivery operation plan by evaluating the cost-effectiveness of different delivery operation plans, and obtain an optimized delivery operation plan; The analysis unit comprises: A first calculation subunit, configured to calculate the occurrence frequency and distribution of historical abnormal operations according to the standardized data, and obtain statistical characteristic data of the historical abnormal operations; A first processing subunit is used to perform clustering processing according to the statistical feature data of the historical abnormal operations, and obtain pattern classification data of the abnormal operations by identifying the abnormal patterns of the historical abnormal operations in combination with the distribution operation environment data; A first analysis subunit is used to perform association analysis according to the pattern classification data of the abnormal operation, by analyzing the association relationship between different abnormal operations in the historical delivery operation and extracting abnormal operation combinations based on a preset dynamic threshold mechanism, to obtain the pattern association data of the abnormal operation; The second analysis subunit is used to perform model building processing according to the pattern association data of the abnormal operation, and to construct an abnormal tree model by using a fault tree analysis algorithm to determine the logical relationship between each basic event and the top event of the historical distribution operation; A second processing subunit is used to perform decomposition processing according to the abnormal tree model, and generate abnormal operation mode data by identifying and extracting abnormal probabilities and influencing factors of key paths and key events; The evaluation unit comprises: The second calculation subunit is used to input the abnormal operation mode data into a preset fuzzy comprehensive evaluation model, to perform fuzzy evaluation on the influence of all abnormal operation modes under different distribution operation environments by constructing a fuzzy matrix and determining a membership function, and to calculate the risk value of each abnormal operation mode to obtain the abnormal operation mode risk value data; A third calculation subunit is used to quantify the direct cost of each abnormal operation mode by using the activity-based cost method according to the abnormal operation mode risk value data, calculate the cost of each operation by identifying each distribution operation and the resources consumed, and obtain direct cost data; A third processing subunit is used to construct an abnormal operation impact hierarchy model according to the direct cost data, and evaluate the impact of various abnormal operations on shutdown losses, transportation losses and additional costs based on the abnormal operation impact hierarchy model to obtain indirect cost data; A fourth processing subunit is used to integrate the direct cost data and the indirect cost data according to the mode to obtain a cost data set of the abnormal operation mode; The construction unit comprises: A first construction subunit is used to perform feature selection and extraction processing on the cost data set of the abnormal operation mode to obtain an initial data set; The second construction subunit is used to predict the abnormal operation mode features that have the greatest impact on the historical delivery operations of the food delivery vehicles based on the initial data set by using the LASSO regression method to obtain a feature data set; A third construction subunit is used to divide the feature data set into a training data set and a test data set according to preset time points; The fourth construction subunit is used to perform model training and optimization on the training data set, predict the historical delivery operation cost of the food delivery vehicle by constructing a random forest model, and use a grid search method to optimize the model parameters, and train to obtain an initial abnormal operation mode prediction model; A fifth construction subunit is used to perform model verification and evaluation according to the initial abnormal operation mode prediction model, verify the model by using a test data set, evaluate the prediction performance of the model, and adjust and optimize to obtain a final abnormal operation mode prediction model; The optimization unit comprises: a fifth processing subunit, configured to obtain real-time delivery operation simulation data of a preset delivery operation scheme, and input the real-time delivery operation data into an abnormal operation mode prediction model, and obtain abnormal operation probability data by predicting the probability of abnormal operation in a future preset time period, wherein the delivery operation scheme is a historical operation scheme; A fourth calculation subunit is used to combine the probability data of the abnormal operation with the redistribution cost and the abnormal operation cost based on the Bayesian solution model, calculate the benefits of the distribution operation solutions under different scenarios, and select the distribution operation solution with the maximum benefit as the preventive distribution operation solution; a sixth processing subunit, configured to perform a cost-effectiveness evaluation according to the preventive delivery operation plan, wherein the execution effects of all preventive delivery operation plans are simulated by a Monte Carlo simulation algorithm and the cost-effectiveness of each preventive delivery operation plan is quantified to obtain cost-effectiveness data; The seventh processing subunit is used to utilize a genetic algorithm, take the cost-benefit data as a fitness function, perform crossover and mutation operations, optimize the execution plan of the scheme, select the preventive delivery operation scheme with the highest fitness, and use the preventive delivery operation scheme with the highest fitness as the final delivery operation scheme for the food delivery vehicle.
