Intelligent management method and system for store after-sales service reservation and storage medium

Abstract

The invention discloses a store after-sales service reservation intelligent management method and system and a storage medium, and relates to the technical field of intelligent management, and the method comprises the following steps: obtaining historical work order data and external environment data, and predicting the service demand distribution of a specified time period in the future based on an ARI MA-LSTM hybrid model; a multi-objective optimization function is constructed according to service demand distribution, resource pre-allocation is carried out based on a genetic algorithm, and optimization objectives comprise minimum customer waiting time, service personnel moving time and a resource vacancy rate; acquiring real-time position information of service personnel and current traffic road condition data, and adjusting resource pre-allocation based on a dynamic path planning algorithm; the problems of large demand prediction deviation, resource distribution rigidity and response lag caused by single prediction model, resource scheduling target solidification and insufficient real-time dynamic adaptation in traditional store after-sales service management are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management, and more specifically, to an intelligent management method, system and storage medium for store after-sales service appointment. Background Art

[0002] With the continuous deepening of the digital transformation of the retail and service industries, the intelligent management of store after-sales service has not only become a key means to improve customer satisfaction, but also the core support for optimizing operation efficiency, reducing service costs, and enhancing brand stickiness. By introducing advanced technologies such as artificial intelligence, big data analysis, Internet of Things (IoT), and customer relationship management (CRM), stores can achieve the full-process digitization of after-sales service processes, including the automation and intelligent decision-making of multiple links such as service appointment, problem diagnosis, work order dispatch, service tracking, customer feedback, and closed-loop return visit.

[0003] Traditional after-sales service appointment systems mostly rely on manual experience or static rules for resource scheduling, and have defects such as large prediction deviations, rigid resource allocation, and lag in dynamic response. Especially in scenarios with frequent demand fluctuations, the existing technologies are difficult to effectively integrate the impact of historical work order data and external environmental factors (such as weather, holidays, regional events) on service demand, resulting in insufficient capture ability of the prediction model for non-linear time series characteristics. In addition, the resource scheduling mostly adopts a single-objective optimization strategy, which cannot balance the contradictions among customer waiting time, personnel movement cost, and resource idle rate, and lacks dynamic adaptation to real-time traffic conditions and service personnel locations, prone to task assignment conflicts and inefficient path planning problems.

[0004] In view of the above problems, the present invention proposes a solution. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an intelligent management method, system and storage medium for store after-sales service appointment, by constructing an intelligent management method for store after-sales service appointment, to solve the problems of large demand prediction deviation, rigid resource allocation and response lag caused by a single prediction model, fixed resource scheduling target and insufficient real-time dynamic adaptation in traditional store after-sales service management.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An intelligent management method for store after-sales service appointment, comprising the following steps: obtaining historical work order data and external environment data, and predicting the service demand distribution in a future preset period based on an ARIMA-LSTM hybrid model; constructing a multi-objective optimization function according to the service demand distribution, and performing resource pre-allocation based on a genetic algorithm, wherein the optimization objectives include minimizing customer waiting time, service staff movement time, and resource idle rate; obtaining the real-time location information of service staff and current traffic condition data, and adjusting the resource pre-allocation based on a dynamic path planning algorithm.

[0008] In a preferred embodiment, the steps of obtaining historical work order data and external environment data, and predicting the service demand distribution in a future preset period based on an ARIMA-LSTM hybrid model are as follows: obtaining historical work order data, and aggregating the historical work order data through a time window to form a service request time series; synchronously accessing external environment data, aligning the external environment data with the work order data with the timestamp as the primary key, and constructing a multi-dimensional time feature input matrix; performing trend modeling and preliminary fitting on the historical time series data through an ARIMA model, capturing the linear change trend, and extracting the ARIMA model residual sequence as the non-linear feature input, wherein the historical time series data includes the service request time series and the multi-dimensional time feature input matrix; training the historical time series data and the ARIMA residuals in an LSTM model to construct a service request prediction model; based on the service request prediction model, outputting future demand prediction values, and dynamically correcting the output results through rolling window training and optimizing the loss function to generate a demand distribution map; performing resource pre-allocation according to the demand distribution map.

[0009] In a preferred embodiment, the step of dynamically correcting the output results through rolling window training and optimizing the loss function is as follows: extracting a set of input samples from each time step of the historical time series data according to the time series; performing sequence prediction on each set of input samples through the trained service request prediction model to output the service request prediction values within the future time steps; if the service request prediction value exceeds the actual service request volume, comparing the actual service request volume with the service request prediction value to output error data, and continuously calculating the error data in each sliding window and summarizing it to form an error trend sequence; correcting the future demand prediction value according to the error trend sequence.

[0010] In a preferred embodiment, the correction of the future demand prediction value according to the error trend sequence is as follows: the error trend sequence is divided into several sequence blocks of equal length; the local peaks and valleys in each sequence block are obtained, and the error points whose absolute values exceed the set threshold are screened out, and the error burst frequency in each sequence block is counted; based on the error burst frequency and the mean absolute error, the error sensitivity is output; according to the error sensitivity and the error direction corresponding to each prediction time period, the future demand prediction value is corrected to form an updated service request prediction value.

[0011] In a preferred embodiment, the multi-objective optimization function is constructed according to the service demand distribution, and resource pre-allocation is performed based on the genetic algorithm. The objective optimization includes minimizing the customer waiting time, the service staff movement time, and the resource idle rate, which is specifically as follows: the future service demand distribution is obtained, and the service demand quantity is matched with the actual available resources as the constraint condition for resource scheduling. The service demand distribution represents the service request quantity in each time period; according to the actual demand and resource constraints, a multi-objective optimization function is constructed. The multi-objective optimization function includes minimizing the customer waiting time, minimizing the service staff movement time, and minimizing the resource idle rate; the multi-objective optimization function is solved based on the genetic algorithm, and resource pre-allocation is performed.

[0012] In a preferred embodiment, the solution of the multi-objective optimization function based on the genetic algorithm and resource pre-allocation are as follows: several initial solution populations are randomly generated. Each solution population represents a resource scheduling scheme, and each scheme includes the allocation of different service staff, service time, and work station allocation; the fitness evaluation results of each solution population are output according to the multi-objective optimization function; the optimal individuals are selected according to the fitness evaluation results, and crossover and mutation operations are performed to generate a new generation of solution populations; the fitness of each generation of solution populations is evaluated until the fitness change fluctuation is lower than the preset value, and the optimal resource scheduling scheme is output.

[0013] In a preferred embodiment, the real-time location information of the service staff and the current traffic condition data are obtained, and the resource pre-allocation is adjusted based on the dynamic path planning algorithm as follows: obtain the first data of the service staff, where the first data includes the real-time location information and the traffic condition data between the service staff and the target service location, and the traffic condition data includes road passing speed, congestion level, predicted time consumption, and traffic abnormal events, and the location information includes service staff identification, geographical coordinates, and timestamp; construct a task path map of the service staff according to the first data, and establish a path relationship map structure between the current task of the service staff and the subsequent assigned tasks; call the dynamic path planning algorithm based on the path map to generate an optimized path, where the path planning algorithm takes the shortest time consumption, least congestion, or minimum task delay as the optimization goal; dynamically adjust the original resource pre-allocation plan according to the path planning result, and the adjustment includes task re-assignment, adjustment of task execution order, and change of service time period; synchronize the updated resource scheduling plan to the service staff terminal and the client system.

[0014] Technical effects and advantages of an intelligent management method, system, and storage medium for store after-sales service reservation of the present invention:

[0015] The present invention realizes the intelligent and efficient management of store after-sales service reservation by integrating ARIMA-LSTM hybrid model prediction, error trend correction, multi-objective optimization, and dynamic path planning. By combining historical work orders and external environment data, the accuracy of service demand prediction is improved, and the prediction results are dynamically corrected using a rolling window and error sensitivity mechanism to effectively cope with non-linear fluctuations and sudden demand changes. In terms of resource scheduling, a multi-objective optimization function with customer waiting time, service staff movement time, and resource idle rate as the objectives is constructed, and the genetic algorithm is used to obtain the optimal resource pre-allocation plan to improve the overall scheduling efficiency and service quality. Further, real-time location information and traffic conditions are introduced, and the scheduling plan is intelligently adjusted in combination with the dynamic path planning algorithm to ensure the optimal path of the service staff, timely response, and smooth task execution of the service staff, significantly improving resource utilization and customer satisfaction. Brief Description of the Drawings

[0016] Figure 1 It is a flowchart of an intelligent management method for store after-sales service reservation of the present invention.

[0017] Figure 2 It is a system structure diagram of an intelligent management method for store after-sales service reservation of the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] It should be noted in advance that the acquisition and processing of all information or data in the present invention are carried out on the premise of complying with the corresponding national data protection regulations and policies and obtaining the authorization of the authorized administrator.

[0020] Embodiment 1, Figure 1 A smart management method for after-sales service appointment in a store according to the present invention is given, including the following steps:

[0021] S1. Obtain historical work order data and external environment data, and predict the service demand distribution in a future preset time period based on the ARIMA-LSTM hybrid model;

[0022] In this example, obtaining historical work order data and external environment data, and predicting the service demand distribution in a future preset time period based on the ARIMA-LSTM hybrid model are specifically as follows:

[0023] Obtain historical work order data, and aggregate the historical work order data through a time window to form a service request time series;

[0024] Synchronously access external environment data, align the external environment data with the work order data using the time stamp as the primary key, and construct a multi-dimensional time feature input matrix;

[0025] Use the ARIMA model to perform trend modeling and preliminary fitting on the historical time series data, capture the linear change trend, and extract the ARIMA model residual sequence as the non-linear feature input. The historical time series data includes the service request time series and the multi-dimensional time feature input matrix;

[0026] Input the historical time series data and the ARIMA residuals into the LSTM model for training to construct a service request prediction model;

[0027] Based on the service request prediction model, output the future demand prediction value, and dynamically correct the output result through rolling window training and optimizing the loss function to generate a demand distribution map;

[0028] Pre-allocate resources according to the demand distribution map.

[0029] It should be noted that historical work order data refers to the work order records accumulated during the after-sales service process of the store, including the service request initiation time and completion time, service request type, the number of service requests per unit time (such as per hour, per day), the actual service time consumed for a single work order, the staff ID handling the work order, skill level, affiliated area, geographical coordinates or area code where the customer is located, etc.

[0030] The external environment refers to the external influencing factors related to the store's service demand, including weather data, holiday information, traffic conditions, etc.

[0031] In this example, the specific steps for predicting using the ARIMA-LSTM hybrid model are as follows:

[0032] Aggregate the historical work order data into a service request time series according to a time window, and align the external environment data (temperature, holiday flag) to generate a multi-dimensional input matrix;

[0033] The ARIMA model fits the linear trend and outputs the residuals. The LSTM takes the residuals as input to learn the non-linear features, and optimizes the model parameters through the mean square error loss function;

[0034] Adopt a rolling window training strategy, and dynamically correct the future prediction values according to the prediction error trend series (such as error outbreak frequency, mean absolute error) to improve the reliability of the demand distribution map.

[0035] The specific model expression for the ARIMA model to fit the linear trend is as follows:

[0036]

[0037] where, y t is the service request volume of the time series at time t, L is the lag operator, φ i is the coefficient of the autoregressive term, θ j is the coefficient of the moving average term, ε t is the white noise error term, and d is the number of differences.

[0038] Furthermore, by integrating the advantages of ARIMA and LSTM models, accurate prediction of service demand is achieved. The ARIMA model is good at capturing linear trends in historical data and can effectively model the basic trend of service requests. LSTM, on the other hand, has the ability to handle non - linear and long - term dependence relationships and can deeply explore complex dynamic changes in residuals. Aggregating work order data through time windows and introducing multi - dimensional external environment features improve the expressive ability and spatio - temporal correlation of input data, enhancing the generalization ability and practical adaptability of the model. At the same time, adopting a rolling window training mechanism and a dynamic loss optimization strategy can continuously monitor and correct prediction errors, ensuring the stability and real - time nature of prediction results. The finally generated service demand distribution map provides a scientific basis for subsequent resource pre - allocation, making resource allocation more forward - looking and flexible, and significantly improving the intelligent level and response efficiency of after - sales service appointments in stores.

[0039] In this example, the output result is dynamically corrected through rolling window training and optimizing the loss function, as follows:

[0040] According to the time series, a set of input samples are extracted from each time step in historical time - series data;

[0041] Each set of input samples is used for sequence prediction through a trained service request prediction model, and the predicted values of service requests within future time steps are output;

[0042] If the predicted value of the service request exceeds the actual service request volume, the actual service request volume is compared with the predicted value of the service request, and error data is output. The error data is continuously calculated in each sliding window and aggregated to form an error trend sequence;

[0043] The predicted value of future demand is corrected according to the error trend sequence.

[0044] It should be noted that the dynamic correction mechanism can continuously correct and enhance the prediction results of service requests through rolling window training and loss function optimization. By extracting input samples at each time step and performing real - time prediction, the system can quickly capture the changing trend of recent service demand. Combining the error between the actual request volume and the predicted value to construct an error trend sequence enables the model to not only have prediction ability during the training stage but also achieve online learning and adaptive adjustment during the operation stage. Continuously tracking the error trend helps to discover prediction biases of the model for periodic, sudden, or long - term offsets, enhancing the robustness and prediction accuracy of the model. This mechanism is especially suitable for scenarios with frequent demand fluctuations in the retail and service industries, providing a more stable and reliable decision - making basis for the allocation of after - sales service resources in stores.

[0045] In this example, the predicted value of future demand is corrected according to the error trend sequence, as follows:

[0046] Divide the error trend sequence into several sequence blocks of equal length;

[0047] Obtain the local peaks and valleys in each sequence block, filter out the error points whose absolute values exceed the set threshold, and count the error burst frequency in each sequence block;

[0048] Output the error sensitivity based on the error burst frequency and the mean absolute error;

[0049] According to the error sensitivity and error direction corresponding to each prediction time period, correct the future demand prediction value to form an updated service request prediction value.

[0050] Among them, the calculation formula of the error sensitivity is as follows:

[0051]

[0052] Among them, X k is the error sensitivity, f k is the number of error points exceeding the set threshold in the k-th sequence block, L2 is the length of each sequence block, e i is the value of each error point, and α and β are preset weight coefficients respectively.

[0053] It should be noted that through the block analysis of the error trend sequence, the local abnormal fluctuation characteristics are accurately captured, thereby improving the model's response ability to sudden deviations. By extracting the error peaks and valleys within each sequence block, and combining the error burst frequency and the mean absolute error, an error sensitivity index is constructed to effectively measure the intensity and frequency of the prediction deviation. This dynamic correction strategy based on error sensitivity can, on the basis of considering the error direction, make targeted adjustments to the prediction values of future time periods, avoiding overall prediction distortion caused by local fluctuations. Compared with static prediction methods, this mechanism is more flexible and adaptable, especially suitable for business scenarios with complex demand fluctuations and frequent environmental changes, further enhancing the stability and accuracy of the service request prediction model, and providing more forward-looking support for resource scheduling and pre-allocation.

[0054] S2. Construct a multi-objective optimization function according to the service demand distribution, and perform resource pre-allocation based on the genetic algorithm. The optimization objectives include minimizing the customer waiting time, the service staff movement time, and the resource idle rate;

[0055] In this example, a multi-objective optimization function is constructed according to the service demand distribution, and resource pre-allocation is performed based on the genetic algorithm. The optimization objectives include minimizing the customer waiting time, the service staff movement time, and the resource idle rate, specifically as follows:

[0056] Obtain the future service demand distribution, match the service demand quantity with the actual available resources, and use it as a constraint condition for resource scheduling. The service demand distribution represents the service request quantity in each time period;

[0057] Construct a multi-objective optimization function according to the actual demand and resource constraints. The multi-objective optimization function includes minimizing the customer waiting time, minimizing the service staff movement time, and minimizing the resource idle rate;

[0058] Solve the multi-objective optimization function based on the genetic algorithm and perform resource pre-allocation.

[0059] It should be noted that by constructing a multi-objective optimization function including customer waiting time, service staff movement time, and resource idle rate, both service efficiency and resource utilization rate are comprehensively considered. On the basis of predicting the future service demand distribution, the service request quantity is matched with the actual available resources to form a constraint condition for resource scheduling, thus ensuring the practical feasibility of the optimization process. Using the genetic algorithm to solve the multi-objective optimization function can explore the optimal or near-optimal resource allocation scheme in the global search, avoid falling into the local optimum, and improve the global optimality of the scheduling strategy. This method is particularly suitable for complex resource scheduling scenarios affected by multiple factors. By effectively reducing the customer waiting time and service staff movement cost, and at the same time reducing the waste caused by resource idleness, it realizes the dynamic balance between service quality and operation efficiency, and provides intelligent and high-efficiency resource pre-allocation decision support for enterprises.

[0060] Furthermore, minimizing the customer waiting time is achieved by reasonably arranging service staff and resources so that customers can obtain a response as soon as possible after submitting a service request. This goal can improve customer satisfaction and reduce complaints or customer churn caused by excessive waiting. During the peak service demand period, dynamically scheduling resources to shorten the response delay helps to enhance the service carrying capacity of the system and realize an efficient service system centered on customers.

[0061] The goal of minimizing the service staff movement time is to optimize the spatial distribution of service tasks, reasonably plan the paths and task sequences of service staff, and reduce unnecessary movement distances or times. This can not only save transportation costs and physical exertion, but also improve the task execution efficiency of service staff, cover more service requests within a limited time, thereby improving the overall service capacity and reducing operating costs.

[0062] Minimizing the resource idle rate focuses on improving the utilization efficiency of service resources and avoiding the situation where resources such as service staff, equipment, or venues fail to be put into use in a timely manner when there is demand. By reasonably predicting demand and scientifically allocating resources so that they can fully play their roles in each time period, it is possible to reduce resource waste, lower the operating expenses of redundant configurations, and maximize resource benefits in resource-constrained scenarios.

[0063] In this example, the multi-objective optimization function is solved based on the genetic algorithm, and resource pre-allocation is performed, specifically as follows:

[0064] Generate a number of initial solution populations randomly. Each solution population represents a resource scheduling plan, and each plan includes the allocation of different service personnel, service time, and work station allocation;

[0065] Output the fitness evaluation results of each solution population according to the multi-objective optimization function;

[0066] Select the optimal individuals according to the fitness evaluation results, perform crossover and mutation operations, and generate a new generation of solution populations;

[0067] Perform fitness evaluation on each generation of solution populations until the fitness change fluctuation is lower than the preset value, and output the optimal resource scheduling plan.

[0068] It should be noted that by introducing the genetic algorithm to solve the multi-objective optimization function, it has the advantages of strong global optimization ability, high convergence efficiency, and adaptability to complex scheduling scenarios. Through the random generation of the initial population, it is possible to search for diverse resource scheduling plans in a wide solution space and avoid falling into local optima. The fitness evaluation mechanism quantifies the trade-off results of the multi-objective optimization function into an evaluation criterion, enabling the algorithm to achieve a dynamic balance among optimizing the customer waiting time, service personnel movement time, and resource idle rate. The crossover and mutation operations enhance the breadth and depth of the solution space exploration, contributing to continuously evolving better scheduling strategies. As the iteration progresses, the algorithm determines the convergence through the fitness fluctuation and finally outputs the globally optimal resource allocation plan, which has good generalization ability and applicability, providing intelligent decision-making support for the efficient utilization and scheduling of resources in dynamic and high-concurrency service scenarios.

[0069] S3. Obtain the real-time location information of service personnel and the current traffic condition data, and adjust the resource pre-allocation based on the dynamic path planning algorithm.

[0070] In this example, a number of initial solution populations are randomly generated. Each solution population represents a resource scheduling plan, and each plan includes the allocation of different service personnel, service time, and work station allocation;

[0071] Output the fitness evaluation results of each solution population according to the multi-objective optimization function;

[0072] Select the optimal individuals according to the fitness evaluation results, perform crossover and mutation operations, and generate a new generation of solution populations;

[0073] Perform fitness evaluation on each generation of solution populations until the fitness change fluctuation is lower than the preset value, and output the optimal resource scheduling plan.

[0074] It should be noted that by integrating the real-time location information of service personnel with traffic condition data and optimizing and adjusting resource pre-allocation based on the dynamic path planning algorithm, it has higher timeliness and response flexibility. During the optimization process, a diverse initial solution population is generated through the genetic algorithm to comprehensively explore the resource scheduling space and ensure the diversity and coverage of the scheduling scheme. The fitness evaluation combines multi-objective indicators such as customer waiting time, service personnel movement time, and resource idle rate to achieve a comprehensive evaluation of the solution population. Through the crossover and mutation operations in the iteration, the resource scheduling strategy is continuously optimized to increase the probability of obtaining the global optimal solution. Introducing real-time data to dynamically adjust the path effectively reduces the impact of traffic congestion on the scheduling efficiency and ensures faster service response and more reasonable resource utilization. The finally output scheduling scheme not only meets multi-objective optimization but also has dynamic adaptability, providing efficient and accurate technical support for intelligent scheduling in complex service systems.

[0075] Embodiment 2 Figure 2 A system for an intelligent management method for after-sales service reservation in a store according to the present invention is provided, including a service demand prediction module, a resource pre-allocation module, and a resource allocation adjustment module, and there are connections between the modules;

[0076] The service demand prediction module is used to obtain historical work order data and external environment data, and predict the service demand distribution in a future preset period based on the ARIMA-LSTM hybrid model;

[0077] The resource pre-allocation module is used to construct a multi-objective optimization function according to the service demand distribution and perform resource pre-allocation based on the genetic algorithm, and the optimization objectives include minimizing customer waiting time, service personnel movement time, and resource idle rate;

[0078] The resource allocation adjustment module is used to obtain the real-time location information of service personnel and current traffic condition data, and adjust the resource pre-allocation based on the dynamic path planning algorithm.

[0079] The present invention also provides a computer-readable storage medium, and the readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the steps in the above data processing method for intelligent construction can be implemented.

[0080] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product.

[0081] Those of ordinary skill in the art will realize that the modules and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0082] In addition, in each embodiment of this application, the various functional modules can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0083] As mentioned above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0084] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent management method for appointment of store after-sales service, characterized in that, The steps are as follows: Obtain historical work order data and external environment data, and predict the service demand distribution in a future preset period based on the ARIMA-LSTM hybrid model; Construct a multi-objective optimization function according to the service demand distribution, and perform resource pre-allocation based on the genetic algorithm. The optimization objectives include minimizing the customer waiting time, the service staff movement time, and the resource idle rate; Obtain the real-time location information of the service staff and the current traffic condition data, and adjust the resource pre-allocation based on the dynamic path planning algorithm.

2. An intelligent management method for store after-sales service reservation according to claim 1, characterized in that Obtain historical work order data and external environment data, and predict the service demand distribution in a future preset period based on the ARIMA-LSTM hybrid model, specifically as follows: Obtain historical work order data, and aggregate the historical work order data through a time window to form a service request time series; Synchronously access external environment data, align the external environment data with the work order data using the timestamp as the primary key, and construct a multi-dimensional time feature input matrix; Conduct trend modeling and preliminary fitting on the historical time series data through the ARIMA model to capture the linear change trend, and extract the ARIMA model residual sequence as the non-linear feature input. The historical time series data includes the service request time series and the multi-dimensional time feature input matrix; Input the historical time series data and the ARIMA residuals into the LSTM model for training to construct a service request prediction model; Based on the service request prediction model, output the future demand prediction value, and dynamically correct the output result through rolling window training and optimizing the loss function to generate a demand distribution map; Perform resource pre-allocation according to the demand distribution map.

3. An intelligent management method for after-sales service reservation in a store according to claim 2, characterized in that, Dynamically correct the output result through rolling window training and optimizing the loss function, specifically as follows: Extract a set of input samples from each time step in the historical time series data according to the time series; Perform sequence prediction on each set of input samples through the trained service request prediction model, and output the service request prediction value within the future time step; If the service request prediction value exceeds the actual service request volume, compare the actual service request volume with the service request prediction value, output the error data, and the error data is continuously calculated in each sliding window and aggregated to form an error trend sequence; Correct the future demand prediction value according to the error trend sequence.

4. An intelligent management method for store after-sales service reservation according to claim 3, characterized in that Correct the future demand prediction value according to the error trend sequence, specifically as follows: Divide the error trend sequence into several sequence blocks of equal length; Obtain the local peaks and valleys in each sequence block, screen out the error points whose absolute values exceed the set threshold, and count the error outbreak frequency in each sequence block; Output the error sensitivity based on the error outbreak frequency and the mean absolute error; Correct the future demand prediction value according to the error sensitivity and the error direction corresponding to each prediction time period to form an updated service request prediction value.

5. An intelligent management method for store after-sales service reservation according to claim 4, characterized in that Construct a multi-objective optimization function according to the service demand distribution, and perform resource pre-allocation based on the genetic algorithm. The objective optimization includes minimizing the customer waiting time, the service staff movement time, and the resource idle rate, specifically as follows: Obtain the future service demand distribution, match the service demand quantity with the actual available resources, and use it as a constraint condition for resource scheduling. The service demand distribution represents the service request quantity in each time period; Construct a multi-objective optimization function according to the actual demand and resource constraints. The multi-objective optimization function includes minimizing the customer waiting time, minimizing the service staff movement time, and minimizing the resource idle rate; Solve the multi-objective optimization function based on the genetic algorithm and perform resource pre-allocation.

6. An intelligent management method for store after-sales service reservation according to claim 5, characterized in that, Solve the multi-objective optimization function based on the genetic algorithm and perform resource pre-allocation, specifically as follows: Randomly generate a number of initial solution populations. Each solution population represents a resource scheduling plan, and each plan includes the allocation of different service staff, service time, and work station allocation; Output the fitness evaluation results of each solution population according to the multi-objective optimization function; Select the optimal individual according to the fitness evaluation results, perform crossover and mutation operations to generate a new generation of solution populations; Perform fitness evaluation on each generation of solution populations until the fitness change fluctuation is lower than the preset value, and output the optimal resource scheduling plan.

7. An intelligent management method for after-sales service appointment in a store according to claim 6, characterized in that Obtain the real-time location information of the service staff and the current traffic condition data, and adjust the resource pre-allocation based on the dynamic path planning algorithm, specifically as follows: Obtain the first data of the service staff. The first data includes real-time location information and traffic condition data between the service staff and the target service location. The traffic condition data includes road passing speed, congestion level, predicted time consumption, and traffic abnormal events. The location information includes service staff identification, geographical coordinates, and time stamp; Construct a service staff task path graph according to the first data, and establish a path relationship graph structure between the current task of the service staff and the subsequent assigned tasks; Call the dynamic path planning algorithm based on the path graph to generate an optimized path. The path planning algorithm takes the shortest time consumption, the least congestion, or the minimum task delay as the optimization goal; Dynamically adjust the original resource pre-allocation plan according to the path planning result. The adjustment includes task re-allocation, task execution order adjustment, and service time period change; Synchronize the updated resource scheduling plan to the service staff terminal and the client system.

8. A system using an intelligent management method for appointment of store after-sales service as described in any one of claims 1-7, characterized in that, It includes a service demand prediction module, a resource pre-allocation module, and a resource pre-allocation adjustment module, and there are connections between the modules; The service demand prediction module is used to obtain historical work order data and external environment data, and predict the service demand distribution in the future preset time period based on the ARIMA-LSTM hybrid model; The resource pre-allocation module is used to construct a multi-objective optimization function according to the service demand distribution, and perform resource pre-allocation based on the genetic algorithm. The optimization goals include minimizing the customer waiting time, the service staff movement time, and the resource idle rate; The resource allocation adjustment module is used to obtain the real-time location information of the service staff and the current traffic condition data, and adjust the resource pre-allocation based on the dynamic path planning algorithm.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a smart management method for store after-sales service reservation as described in any one of claims 1 to 7.

Images