Task scheduling method and device, nonvolatile storage medium and electronic equipment

Abstract

The invention discloses a task scheduling method and device, a nonvolatile storage medium and electronic equipment. The method comprises the steps that influence factors related to a target scheduling task and scheduling factors corresponding to the influence factors are determined according to historical work order data; a target weight set is determined, the target weight set comprises target weights corresponding to all the scheduling factors, the target weight set is a weight set enabling the change rate of a cost function of the task scheduling model to be within a preset change rate range, and the cost function is composed of a loss function and a normalized item; the normalization item is positively correlated with the weight of each scheduling factor, and the value of the loss function is correlated with the weight of each scheduling factor; and processing the scheduling factor and the target weight set through a task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task. The technical problem of low accuracy of a scheduling method caused by insufficient response of a traditional scheduling method to dynamic influence factors is solved.

Description

Technical Field

[0001] The present application relates to the field of task scheduling, and in particular, to a task scheduling method, device, non-volatile storage medium and electronic device. Background Art

[0002] Traditional task scheduling methods suffer from poor training effects and low scheduling decision accuracy when faced with dynamic environments and complex scheduling scenarios, which limits their application effects and customer experience in actual services.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a task scheduling method, device, non-volatile storage medium and electronic device to at least solve the technical problem of low accuracy of scheduling methods caused by insufficient response of traditional scheduling methods to dynamic influencing factors.

[0005] According to one aspect of an embodiment of the present application, a task scheduling method is provided, including: determining the influencing factors involved in a target scheduling task and the scheduling factors corresponding to each type of influencing factor based on historical work order data; determining a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of a cost function of a task scheduling model within a preset rate of change range, the cost function consists of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; processing the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0006] Optionally, the cost function is expressed as follows: Among them, C represents the cost function, C0 represents the loss function, is the normalization term, n is the number of samples in the training set, represents the weight, It represents the square accumulation of the weights corresponding to each scheduling factor, and λ is a preset normalization parameter used to adjust the influence of the normalization term on the value of the cost function.

[0007] Optionally, a positive correlation is preset between the normalization parameter and the sample size.

[0008] Optionally, the loss function comprises a weighted combination of various scheduling factors.

[0009] Optionally, determining the target weight set includes: determining the initial weights of each scheduling factor based on historical work order data; in the process of training the task scheduling model, adjusting the initial weights by back propagation and gradient descent until the rate of change of the cost function is within a preset rate of change, thereby obtaining the target weights of each scheduling factor.

[0010] Optionally, determining the influencing factors involved in the target scheduling task based on historical work order data, and the scheduling factors corresponding to each type of influencing factor include: determining the task type of the target scheduling task; determining the feature type of feature data involved in the task type from the historical work order data based on the task type, wherein the feature type is the influencing factor; and cleaning and preprocessing the feature data to obtain the scheduling factor.

[0011] Optionally, the scheduling factors include at least one of the following: a distance factor, a personnel skill factor, a work order appointment time factor, and a weather factor, wherein the personnel skill factor includes a skill vector for measuring the skill level of each operator and a skill difference vector for the skill level differences between different operators, and the weather factor includes a climate vector for reflecting the degree of impact of different climates.

[0012] Optionally, processing the scheduling factors and target weight sets through the task scheduling model includes: the task scheduling model determines the weights corresponding to the scheduling factors corresponding to the influencing factors and the scheduling factors affected by the affected factors based on the influencing factors involved in the target scheduling task; and multiplying the weights corresponding to the scheduling factors corresponding to the influencing factors by the scheduling factors affected by the affected factors.

[0013] According to another aspect of an embodiment of the present application, a task scheduling device is also provided, including: a data acquisition module, used to determine the influencing factors involved in the target scheduling task based on historical work order data, and the scheduling factors corresponding to each type of influencing factors; a processing module, used to determine a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of the cost function of the task scheduling model within a preset rate of change range, the cost function is composed of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; an execution module, used to process the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0014] According to another aspect of an embodiment of the present application, a non-volatile storage medium is provided, in which a program is stored, wherein when the program is running, a device where the non-volatile storage medium is located is controlled to execute a task scheduling method.

[0015] According to another aspect of an embodiment of the present application, there is further provided an electronic device, including: a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the task scheduling method is executed when the program is running.

[0016] According to another aspect of an embodiment of the present application, a computer program product is also provided, including a computer program, and when the computer program is executed by a processor, a task scheduling method is implemented.

[0017] In an embodiment of the present application, the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each type of influencing factor are determined based on historical work order data; a target weight set is determined, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of the cost function of the task scheduling model within a preset rate of change range, and the cost function is composed of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; the scheduling factors and the target weight set are processed by the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task, and the purpose of optimizing the task scheduling model weight and improving the generalization ability is achieved by dynamically adjusting the scheduling factor weights and training based on L2-NORM normalization, thereby achieving the technical effect of improving the accuracy and efficiency of intelligent scheduling in different scenarios, and further solving the technical problem of low accuracy of scheduling methods caused by insufficient response of traditional scheduling methods to dynamic influencing factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 It is a structural schematic diagram of a computer terminal (mobile device) provided according to an embodiment of the present application;

[0020] Figure 2 It is a flowchart of a task scheduling method provided according to an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of factor clipping provided according to an embodiment of the present application;

[0022] Figure 4 It is a schematic diagram of the core process of a task scheduling method provided according to an embodiment of the present application;

[0023] Figure 5 It is a structural schematic diagram of a task scheduling method and device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] In order to better understand the embodiments of the present application, the technical terms involved in the embodiments of the present application are explained as follows:

[0027] Site: An area can be circled on a map as a physical grid. Customers with addresses in this area will receive marketing, distribution, installation and maintenance services from this physical grid. Multiple physical grids and the installation and maintenance personnel within the grids form a distribution station. The installation and maintenance personnel are the people who provide marketing, distribution, installation and maintenance services within the site.

[0028] L2-NORM (L2 norm): A mathematical technique for measuring the square root of the sum of the squares of each element in a vector or matrix, widely used in machine learning and deep learning. In neural networks, it is used to penalize large weight values ​​to avoid overfitting of the model, improve the generalization ability of the model, and ensure that the model performs more stably on unseen data. L2-NORM regularization helps control model complexity and make the learning process more robust. Especially when the number of features is large, it can effectively reduce the model's sensitivity to noise data to speed up the training process, while adapting to different scheduling scenarios.

[0029] In the field of task scheduling, traditional scheduling methods mainly rely on rule-based algorithms or simple machine learning models to achieve comprehensive scheduling of personnel and sites. These systems usually consider static scheduling factors such as customer distance, installation and maintenance personnel skills, and work order appointment time. However, they have significant limitations when dealing with dynamic environmental factors, especially the differences in scheduling factors in different time periods.

[0030] Problems with existing technologies:

[0031] 1. Inadequate handling of scheduling factor differences: Traditional scheduling methods fail to effectively adapt to the dynamic changes of scheduling factors at different times. For example, during peak hours or holidays, the density of customer requests, the availability of installation and maintenance personnel, and traffic conditions will change. However, these systems often assign static weights to each factor, resulting in reduced accuracy of scheduling decisions at specific times.

[0032] 2. Limitations of model training: During the model training process, traditional scheduling methods may fail to fully consider the complex relationships and dynamic changes between scheduling factors, resulting in poor model generalization ability. Especially when processing nonlinear or high-dimensional data, the model may overfit the training data and perform poorly on new data, that is, the training effect is poor.

[0033] 3. Limited ability to respond to dynamic influencing factors: Unpredictable environmental factors such as weather and traffic conditions have a direct impact on scheduling decisions. Traditional scheduling models may fail to respond to changes in these influencing factors in real time and accurately, resulting in an increase in the error between scheduling decisions and actual conditions, affecting service quality.

[0034] 4. Inefficient resource allocation: Traditional scheduling models are often inefficient in resource allocation due to the failure to effectively identify and utilize the optimal match between personnel skills and site coverage. For example, the opportunity to send the most suitable installation and maintenance personnel to the most urgent service requests may be missed.

[0035] In summary, the existing technologies have obvious deficiencies in processing dynamic scheduling factors, the generalization ability of model training, and resource allocation efficiency. These problems are particularly prominent when facing complex and changeable scheduling scenarios, resulting in reduced overall accuracy and efficiency of scheduling methods.

[0036] In order to solve the above problems, relevant solutions are provided in the embodiments of the present application, which are described in detail below.

[0037] According to an embodiment of the present application, a method embodiment of a task scheduling method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a task scheduling method. Figure 1 As shown, the computer terminal 10 (or mobile device 10) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0039] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As involved in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0040] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the task scheduling method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned task scheduling method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0041] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0042] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0043] In the above operating environment, the present application embodiment provides a task scheduling method, such as Figure 2 As shown, the method comprises the following steps:

[0044] Step S202, determining the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each of the influencing factors according to the historical work order data;

[0045] Optionally, feature data such as the distance between the customer and the site, the skills of the installation and maintenance personnel, and the work order appointment time are included from the historical work order data.

[0046] In the technical solution provided in step S202, the influencing factors involved in the target scheduling task are determined based on the historical work order data, and the scheduling factors corresponding to each type of influencing factor include: determining the task type of the target scheduling task; determining the feature type of the feature data involved in the task type from the historical work order data based on the task type, wherein the feature type is the influencing factor; and cleaning and preprocessing the feature data to obtain the scheduling factor.

[0047] Optionally, historical work order data will be used as the basis for analysis and scheduling. Clean and preprocess these data to remove outliers, fill missing values, and standardize or normalize numerical features to ensure data quality and consistency. After determining the target scheduling task, first identify the type of task, such as maintenance scheduling, installation scheduling, or marketing task scheduling. Next, determine the feature type of feature data involved in the task type from the historical work order data based on the task type. The feature type actually corresponds to the influencing factors that need to be considered in the scheduling process. For example, for maintenance scheduling tasks, it may be necessary to consider the skill level of the installation and maintenance personnel, the distance to the nearest site, and weather conditions. Clean and preprocess the collected feature data to ensure the reliability and applicability of the data, so as to obtain the scheduling factors corresponding to each influencing factor.

[0048] In the technical solution provided in step S202, the scheduling factors include at least one of the following: distance factor, personnel skill factor, work order appointment time factor, and weather factor, wherein the personnel skill factor includes a skill vector for measuring the skill level of each operator and a skill difference vector for the skill level differences between different operators, and the weather factor includes a climate vector for reflecting the degree of impact of different climates.

[0049] Optionally, the distance factor is particularly critical in the scheduling problem, which is crucial for optimizing scheduling decisions, improving service efficiency and customer satisfaction. Specifically, the distance matrix uses the geographic coordinate information of the site to calculate the spatial distance between sites or between the site and the customer's order address through the site coordinates to determine the value of the distance factor, providing a quantitative indicator for the scheduling algorithm. In most cases, the shorter the distance, the faster the service response time may be and the lower the cost may be. Therefore, the distance factor plays an extremely critical role in the scheduling problem.

[0050] The distance formula can be used to calculate the L2 norm distance between two sites. If there are two sites A(a1,b1) and B(a2,b2), the distance between them can be calculated by the following formula:

[0051]

[0052] Among them, parameters a1 and b1 represent the horizontal coordinate (such as longitude) and vertical coordinate (such as latitude) of site A in the two-dimensional coordinate system; similarly, parameters a2 and b2 represent the horizontal coordinate and vertical coordinate of site B in the same coordinate system. Here, the choice of horizontal and vertical coordinates depends on the actual application scenario. For example, in geographic scheduling, longitude and latitude are usually used as coordinates to represent the location. Through this method, we can get the straight-line distance between site A and site B. This distance is used as a key decision basis in the scheduling algorithm to help the system evaluate and optimize the response time of service personnel or service sites to customer addresses, thereby improving scheduling efficiency and service quality.

[0053] Optionally, the personnel skill factor is calculated by a skill vector: each installation and maintenance personnel is represented by a skill vector, which contains the level value of each skill of the installation and maintenance personnel. For example, the skill vector of an installation and maintenance personnel P can be where s i represents the level of the i-th skill.

[0054] When calculating the skill difference between two installation and maintenance personnel, the L2 norm can be used to measure it. Suppose there are two installation and maintenance personnel P1 and P2, whose skill vectors are P1 = [p 11 ,p 12 ,...,p 1n ] and P2 = [p 21 ,p 22 ,...,p 2n ], the skill difference between them can be calculated by the following L2 norm formula:

[0055]

[0056] Optionally, the work order appointment time factor is calculated by the time difference vector: for a series of work order task appointment times, the difference between the actual delivery time and the customer appointment time can be calculated. If the appointment time of task i is The actual delivery completion time is The time difference vector of all tasks is in The total delay of all work order tasks can be calculated using the L2 norm. The L2 norm calculation formula for the total delay is:

[0057] Optionally, combined with the above factors, the final scheduling optimization goal may be to minimize a comprehensive loss function that includes all these factors. A complete loss function includes a weighted combination of the distance between sites and sites and between sites and customers, personnel skill differences, and task time errors: TotalLoss = λ1·Distance+λ2·SkillDifference+λ3·TimeDelay. In summary, the application of L2 norm in intelligent scheduling training is mainly achieved by calculating the L2 norm value of each factor, and integrating these values ​​into a loss function for optimization training. This method helps to comprehensively consider multiple factors and find the optimal scheduling solution. When new work order data is imported, the parameters of the model will be updated based on the above factors, and the dispatch strategy will be recalculated based on the new parameters.

[0058] Optionally, you can also optimize the L2-Norm calculation by adding weight factors and adjusting the weights of different factors to change the model training effect and scheduling results. For example, weather factors:

[0059] Automatically increase dispatch factors based on weather forecasts. Extreme weather has a significant impact on actual route selection and travel time. Each type of weather can be represented by a vector that contains the level of each climate. For example, each climate vector can be where w i Represents the influence of the wth climate.

[0060] Climate difference: When calculating the difference between each climate, the L2 norm can also be used to measure it. Suppose there are two climates, w1 and w2, and their climate vectors are W1 = [w 11 ,w 12 ,...,w 1n ] and w2=[w 21 ,w 22 ,...,w 2n ], the difference in climate impact between them can be calculated by the following L2 norm formula:

[0061]

[0062] Similarly, we can trim multiple types of factors and integrate them into a loss function for optimization training. For various factors that affect scheduling decisions (such as distance, time, weather, personnel capabilities, customer evaluation, etc.), we can dynamically adjust the weights of these factors in the model by analyzing historical data and actual business needs. For example, when the model recognizes that extreme weather conditions have a significant impact on service response time, the weight of weather factors can be increased through factor trimming to make it occupy a more important position in scheduling decisions. Conversely, if some factors have less impact in a specific scenario, the model will reduce the weight of these factors accordingly to prevent them from interfering with scheduling decisions. The personnel scoring factor can be added. The quality of installation and maintenance task completion and customer evaluation are important indicators for measuring the comprehensive score of employees. By introducing this factor, employees with better performance can be given priority in scheduling. By collecting and analyzing historical work order data, each employee's skill proficiency, service attitude, customer satisfaction, etc. can be quantitatively scored to form a comprehensive personnel scoring factor. In scheduling training, integrating this factor into the loss function can prompt the model to learn the strategy of prioritizing high-scoring employees, thereby achieving a better balance between service response speed and customer experience; the distribution vehicle factor can also be added to plan appropriate service areas and routes for technicians with different modes of transportation, etc. The addition of the distribution vehicle factor is mainly to adapt to the scheduling needs under different traffic conditions. Different modes of transportation (such as cars, motorcycles, and bicycles) have significant differences in driving speed and accessibility in urban traffic, especially in scenarios with traffic congestion or complex geographical environments. By considering this factor in the model, service areas and routes can be planned more reasonably. For example, during peak traffic hours, technicians who ride or walk are given priority for dispatching, while during non-peak hours or long-distance service requests, cars may be more preferred. This flexible scheduling strategy helps reduce service delays and improve overall operational efficiency.

[0063] Step S204, determining a target weight set, wherein the target weight set includes target weights corresponding to various scheduling factors, and the target weight set is a weight set that makes the change rate of the cost function of the task scheduling model within a preset change rate range, and the cost function is composed of a loss function and a normalization term, and the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor;

[0064] In the technical solution provided in step S204, the cost function is expressed as follows: Among them, C represents the cost function, C0 represents the loss function, is the normalization term, n is the number of samples in the training set, represents the weight, It represents the square accumulation of the weights corresponding to each scheduling factor, and λ is a preset normalization parameter used to adjust the influence of the normalization term on the value of the cost function.

[0065] The first term on the right side of the above formula is the conventional loss function, and the second term is the L2-NORM normalization term. λ>0 is the normalization parameter, and n is the size of the training set. The effect of normalization is to make the network tend to learn smaller weights. Large weights are only allowed when they can give a sufficient improvement to the first term of the cost function. Normalization is a compromise between finding small weights and minimizing the original cost function. The relative importance of these two parts is controlled by the value of λ. The smaller the λ value, the more inclined to minimize the original cost function, and vice versa.

[0066] Conventional loss functions measure the difference between the model's predicted output and the actual target, and the goal is to minimize this difference so that the model can make accurate predictions on the training data. However, if the model is too complex, that is, has large weights, it may overfit the training data and capture noise and anomalies in the data, which will lead to poor performance on unseen data.

[0067] The L2-NORM normalization term is designed to address this problem. It introduces an additional penalty term to suppress the size of weights, prompting the model to learn smaller weights. During the normalization process, large weights are given higher penalty costs, and they are "allowed" to exist only when these large weights can significantly improve the model's fit to the training data, that is, they can significantly reduce the value of the conventional loss function. This mechanism ensures that the model will not be overly complicated while pursuing high accuracy, avoiding the trap of overfitting.

[0068] The role of the normalization parameter λ is to control this trade-off. When the λ value is small, it means that the normalization term has less impact on the overall loss function, and the model is more inclined to minimize the original loss function, allowing larger weights to improve accuracy on the training data. On the contrary, when the λ value is large, the influence of the normalization term increases, and the model will be more inclined to learn and use smaller weights, even if this means that the performance on the training data may not be optimal, but it can ensure that the model has stronger generalization ability on new data.

[0069] In other words, the size of λ determines the trade-off between the model's pursuit of training data fit and prevention of overfitting. Small λ tends to pursue better performance on training data and may allow larger weights; while large λ focuses more on the simplification and generalization ability of the model and tends to learn smaller weights. By adjusting the value of λ, we can find a suitable balance between improving the model's prediction accuracy and ensuring the model's generalization ability, thereby improving the overall performance and stability of the model.

[0070] In the technical solution provided in step S204, there is a positive correlation between the preset normalization parameter and the sample size.

[0071] Optionally, if the training data set is large, that is, n is large, then the adjustment factor If the normalization parameter is close to 1, the weight decay effect is not obvious, resulting in a decrease in the normalization effect. Therefore, we set the normalization parameter larger for large data sets and smaller for small data sets.

[0072] In the technical solution provided in step S204, the loss function includes a weighted combination of various scheduling factors.

[0073] Optionally, the loss function is calculated as follows:

[0074] TotalLoss=λ1·Distance+λ2·SkillDifference+λ3·TimeDelay

[0075] In the technical solution provided in step S204, determining the target weight set includes: determining the initial weights of each scheduling factor based on historical work order data; in the process of training the task scheduling model, adjusting the initial weights by back propagation and gradient descent until the rate of change of the cost function is within the preset rate of change range, thereby obtaining the target weights of each scheduling factor.

[0076] Optionally, when training a neural network, the goal is to use backpropagation and gradient descent to optimize the weights. After adding the regularization term, the weight optimization becomes the derivative of the cost function of the above L2-NORM regularization term with respect to the parameter w: The learning rule for the weights becomes: This is the same as the normal gradient descent learning rule, except that there is an additional factor The weight ω is readjusted. This is the weight attenuation caused by the clipping factor. This method will not reduce the weight to 0, because other items in the original cost function that cause the decrease may increase the weight.

[0077] Step S206, processing the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling solution corresponding to the target scheduling task.

[0078] In the technical solution provided in step S206, processing the scheduling factors and the target weight set through the task scheduling model includes: the task scheduling model determines the weights corresponding to the scheduling factors corresponding to the influencing factors and the scheduling factors affected by the affected factors based on the influencing factors involved in the target scheduling task; and multiplies the weights corresponding to the scheduling factors corresponding to the influencing factors by the scheduling factors affected by the affected factors.

[0079] For example, adjusting the weight of weather factors: automatically adjusting the weight of dispatch factors according to the weather forecast. Extreme weather has a significant impact on actual route selection and driving time.

[0080] Distance of impact: Different weather conditions, such as rain, snow, fog, etc., define a weight for each condition. Set the weight of foggy days to 1.5, rainy days to 1.2, clear weather to 1.0, and typhoon weather to 0.3. These weights are calculated by the model based on actual historical data and traffic data.

[0081] Modify the distance formula: Add the weather weight coefficient to the distance formula. is the weather weight, then the weighted distance formula is where x i and i It is the coordinates of two locations in a certain dimension.

[0082] Application scenario: In actual route planning or logistics distribution, weighted distance is used to predict travel time under different weather conditions, so as to make more reasonable arrangements. For example, if a certain area is predicted to be foggy, the weight of foggy weather should be multiplied when calculating the distance from the starting point to the destination.

[0083] In the process of intelligent scheduling and logistics distribution optimization, environmental factors such as weather conditions play a vital role, especially extreme weather, which has a profound impact on the actual path selection and travel time. To address this challenge, we introduced a weather weight adjustment technology based on factor clipping and fusion, which allows the scheduling model to dynamically adjust the weights of weather-related scheduling factors according to real-time weather forecasts, thereby more accurately predicting the travel time and cost of delivery tasks under different weather conditions, and achieving efficient and adaptable scheduling decisions.

[0084] First, we defined specific weights for various typical weather conditions. Foggy days, rainy days, snowy days, and typhoon days were assigned weight coefficients of 1.5, 1.2, 1.2 (assuming the same impact as rainy days), and 0.3, respectively. These weights are not set arbitrarily, but are based on historical data and traffic statistics analysis, reflecting the degree of impact of different weather conditions on actual driving distance and time. For example, foggy days are given a higher weight because visibility is reduced and driving speed decreases; typhoon weather may completely interrupt traffic, so the weight is set very low, almost prohibiting the scheduling of tasks under such conditions.

[0085] Next, the weather weight factor is added to the traditional distance calculation formula, and the weight is integrated to form a new weighted distance formula. This formula takes into account the direct geographical distance from the starting point to the destination, and reflects the impact of weather conditions during actual driving by multiplying it by a specific weather weight. For example, in foggy conditions, even if the straight-line distance between two locations does not change, the calculated "effective distance" will increase significantly due to the increase in foggy weights, which prompts the system to prefer paths or schedules that are less affected by foggy weather when making scheduling decisions, thereby avoiding unnecessary delays and risks.

[0086] In actual scenarios, the introduction of this factor has greatly improved the flexibility and accuracy of the intelligent dispatching system. When planning routes or delivering logistics, the system uses weighted distance to predict the travel time under different weather conditions and find the optimal solution for each delivery task. For example, when fog is forecasted in an area, the system automatically adjusts the dispatching strategy by dynamically cutting and integrating weather factors, which not only prioritizes safety factors, but also ensures the timeliness of service and customer satisfaction.

[0087] The normalized network builds a relatively simple model based on the common patterns in the training data, which helps to resist the noise interference in the training data. In order to further improve the training effect and accuracy of the model, our method introduces a dynamic clipping factor This factor allows the network to more realistically learn observed phenomena such as Figure 3 , and can use the knowledge that has been learned to better generalize, and can show higher accuracy when facing many uncontrollable factors, so as to obtain the best training plan, such as Figure 3 As shown in the figure, the element indicated by the arrow will affect the element at the end of the arrow. Therefore, when calculating the dispatch factor at the end of the arrow, the weight value of the weight factor indicated by the corresponding arrow is presented. For example, weather conditions (T) can significantly affect the actual driving distance and time and affect the "effective distance" (D) we perceive. When calculating the loss function, the weight corresponding to the current weather conditions is added to the weight of the calculated distance dispatch factor.

[0088] In some embodiments of the present application, the core process of the task scheduling method is as follows: Figure 4 As shown in the figure, when a customer places an order for the first time, each scheduling factor is initialized based on historical data, the factor is dynamically trimmed based on the document information, the weight of each factor is adjusted and the weight decay is calculated. When the error value is minimized, the optimal factor weight training plan is determined, and the optimal scheduling plan is output based on the factor weight.

[0089] Through the above steps, dynamic adjustment through factor clipping and weight decay can be achieved, which significantly improves the training effect and prediction accuracy of the intelligent scheduling model, and provides a more efficient and intelligent solution for complex scheduling problems. Specifically, this solution has the following advantages:

[0090] 1. Through the strategy of factor clipping, it supports dynamic adjustment of the weights (w) in the model to ensure that the model will not overfit to the noise in the training data, thus effectively preventing overfitting. This method allows the model to focus more on the essential laws in the data during the learning process rather than accidental outliers, greatly improving the generalization ability and prediction accuracy of the model.

[0091] 2. By continuously adjusting and clipping factors, it supports precise control of the degree of weight decay, gradually reducing the weight decay value, and thus reducing the error in model prediction. This not only speeds up the training process, but also enables the model to suppress the influence of unnecessary or redundant features while retaining important features, improving the overall training efficiency and model stability.

[0092] 3. Solve the problems of poor model training effect and insufficient accuracy faced by traditional intelligent scheduling systems in comprehensive scheduling of personnel and sites, especially when dealing with differences in scheduling factors in different time periods. The method embodiment of the present application can intelligently respond to scheduling needs in different scenarios, and through optimizing model training, it realizes effective management of variable factors and ensures that the best scheduling decisions can be made under various conditions.

[0093] The present application embodiment provides a task scheduling device, Figure 5 is a schematic diagram of the structure of the device, such as Figure 5 As shown, the device includes: a data acquisition module 50, which is used to determine the influencing factors involved in the target scheduling task based on historical work order data, and the scheduling factors corresponding to each type of influencing factor; a processing module 52, which is used to determine a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of the cost function of the task scheduling model within a preset rate of change range, and the cost function is composed of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; an execution module 54, which is used to process the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0094] In some embodiments of the present application, the cost function is expressed as follows: Among them, C represents the cost function, C0 represents the loss function, is the normalization term, n is the number of samples in the training set, represents the weight, It represents the square accumulation of the weights corresponding to each scheduling factor, and λ is a preset normalization parameter used to adjust the influence of the normalization term on the value of the cost function.

[0095] In some embodiments of the present application, there is a positive correlation between the preset normalization parameter and the sample size.

[0096] In some embodiments of the present application, the loss function includes a weighted combination of various scheduling factors.

[0097] In some embodiments of the present application, determining the target weight set includes: determining the initial weights of each scheduling factor based on historical work order data; in the process of training the task scheduling model, adjusting the initial weights by back propagation and gradient descent until the rate of change of the cost function is within a preset rate of change, thereby obtaining the target weights of each scheduling factor.

[0098] In some embodiments of the present application, determining the influencing factors involved in the target scheduling task based on historical work order data, and the scheduling factors corresponding to each type of influencing factor include: determining the task type of the target scheduling task; determining the feature type of feature data involved in the task type from the historical work order data based on the task type, wherein the feature type is the influencing factor; and cleaning and preprocessing the feature data to obtain the scheduling factor.

[0099] In some embodiments of the present application, the scheduling factor includes at least one of the following: a distance factor, a personnel skill factor, a work order appointment time factor, and a weather factor, wherein the personnel skill factor includes a skill vector for measuring the skill level of each operator and a skill difference vector for the difference in skill levels between different operators, and the weather factor includes a climate vector for reflecting the degree of impact of different climates.

[0100] In some embodiments of the present application, processing scheduling factors and target weight sets through a task scheduling model includes: the task scheduling model determines the weights corresponding to the scheduling factors corresponding to the influencing factors and the scheduling factors affected by the affected factors based on the influencing factors involved in the target scheduling task; and multiplies the weights corresponding to the scheduling factors corresponding to the influencing factors by the scheduling factors affected by the affected factors.

[0101] It should be noted that the various modules in the above-mentioned task scheduling device can be program modules (for example, a set of program instructions that implement a certain specific function) or hardware modules. For the latter, it can be expressed in the following forms, but is not limited to this: the expression form of each of the above-mentioned modules is a processor, or the functions of each of the above-mentioned modules are implemented by a processor.

[0102] An embodiment of the present application provides a non-volatile storage medium, in which a program is stored, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the following task scheduling method: determining the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each type of influencing factors based on historical work order data; determining a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of the cost function of the task scheduling model within a preset rate of change range, the cost function consists of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; processing the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0103] An embodiment of the present application provides an electronic device, comprising: a memory and a processor, the processor being used to run a program stored in the memory, wherein the following task scheduling method is executed when the program is running: determining the influencing factors involved in a target scheduling task and the scheduling factors corresponding to each type of influencing factor based on historical work order data; determining a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, the target weight set is a weight set that makes the rate of change of a cost function of a task scheduling model within a preset rate of change range, the cost function being composed of a loss function and a normalization term, the normalization term being positively correlated with the weight of each scheduling factor, and the value of the loss function being related to the weight of each scheduling factor; processing the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0104] An embodiment of the present application provides a computer program product, including a computer program, which implements the following task scheduling method when executed by a processor: determining the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each type of influencing factor based on historical work order data; determining a target weight set, wherein the target weight set includes target weights corresponding to each scheduling factor, and the target weight set is a weight set that makes the rate of change of the cost function of the task scheduling model within a preset rate of change range, the cost function consists of a loss function and a normalization term, the normalization term is positively correlated with the weight of each scheduling factor, and the value of the loss function is related to the weight of each scheduling factor; processing the scheduling factors and the target weight set through the task scheduling model to obtain a scheduling scheme corresponding to the target scheduling task.

[0105] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0107] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0108] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0109] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0110] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A task scheduling method, characterized in that: include: Determine the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each influencing factor based on historical work order data; Determine a target weight set, wherein the target weight set includes target weights corresponding to each of the scheduling factors, and the target weight set is a weight set that makes the change rate of the cost function of the task scheduling model within a preset change rate range, the cost function is composed of a loss function and a normalization term, the normalization term is positively correlated with the weight of each of the scheduling factors, and the value of the loss function is related to the weight of each of the scheduling factors; The scheduling factor and the target weight set are processed by the task scheduling model to obtain a scheduling solution corresponding to the target scheduling task.

2. The task scheduling method according to claim 1, characterized in that: The expression of the cost function is as follows: Wherein, C represents the cost function, C0 represents the loss function, is the normalization term, n is the number of samples in the training set, represents the weight, represents the square accumulation of the weights corresponding to the various scheduling factors, and λ is a preset normalization parameter used to adjust the influence of the normalization term on the value of the cost function.

3. The task scheduling method according to claim 2, characterized in that: There is a positive correlation between the preset normalization parameter and the sample size.

4. The task scheduling method according to claim 1, characterized in that: The loss function includes a weighted combination of each of the scheduling factors.

5. The task scheduling method according to claim 1, characterized in that: Determining the target weight set includes: Determining the initial weight of each of the scheduling factors according to the historical work order data; In the process of training the task scheduling model, the initial weights are adjusted by back propagation and gradient descent until the rate of change of the cost function is within a preset rate of change, thereby obtaining the target weights of the various scheduling factors.

6. The task scheduling method according to claim 1, characterized in that: Determine the influencing factors involved in the target scheduling task based on historical work order data, and the scheduling factors corresponding to each type of influencing factor include: Determine the task type of the target scheduling task; Determining, according to the task type, from the historical work order data, a feature type of feature data related to the task type, wherein the feature type is the influencing factor; The characteristic data is cleaned and preprocessed to obtain the scheduling factor.

7. The task scheduling method according to claim 1, characterized in that: The scheduling factors include at least one of the following: distance factor, personnel skill factor, work order appointment time factor, and weather factor, wherein the personnel skill factor includes a skill vector for measuring the skill level of each operator and a skill difference vector for the skill level differences between different operators, and the weather factor includes a climate vector for reflecting the degree of impact of different climates.

8. The task scheduling method according to claim 1, characterized in that: Processing the scheduling factor and the target weight set by the task scheduling model includes: The task scheduling model determines the weights corresponding to the scheduling factors corresponding to the influencing factors and the scheduling factors affected by the influencing factors according to the influencing factors involved in the target scheduling task; The weight corresponding to the scheduling factor corresponding to the influencing factor is multiplied by the scheduling factor affected by the influencing factor.

9. A task scheduling device, characterized in that: include: The data collection module is used to determine the influencing factors involved in the target scheduling task and the scheduling factors corresponding to each influencing factor based on the historical work order data; A processing module, configured to determine a target weight set, wherein the target weight set includes target weights corresponding to each of the scheduling factors, and the target weight set is a weight set that makes the change rate of the cost function of the task scheduling model within a preset change rate range, the cost function is composed of a loss function and a normalization term, the normalization term is positively correlated with the weight of each of the scheduling factors, and the value of the loss function is related to the weight of each of the scheduling factors; An execution module is used to process the scheduling factor and the target weight set through the task scheduling model to obtain a scheduling solution corresponding to the target scheduling task.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores a program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the task scheduling method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program executes the task scheduling method described in any one of claims 1 to 8 when running.

12. A computer program product, characterized in that The method comprises a computer program, which implements the task scheduling method according to any one of claims 1 to 8 when executed by a processor.

Images