Intelligent work order management method, system, device and medium

Abstract

The present invention is applicable to the field of work order management, and discloses an intelligent work order management method, system, equipment and medium. The method first classifies work orders according to the differences in work order types to obtain work orders of different categories, and vectorizes the characteristics of each type of work order from two dimensions: work order processing complexity and required resources, to construct a work order feature vector; based on the work order feature vector, a work order processing strategy selection model is established; in the process of work order processing, resource consumption data of each link is collected to construct a feature matrix of work order resource consumption and value contribution; a work order resource allocation recommendation model is established using a collaborative filtering algorithm; after the work order processing is completed, a work order cost accounting model is established based on the actual resource consumption and work order characteristics; in addition, in the entire process of work order processing, data of each link is continuously collected, and each model is updated and optimized in real time to achieve a dynamic balance between processing efficiency and cost sharing.

Description

Technical Field

[0001] The present invention relates to the field of work order management, and in particular to an intelligent work order management method, system, equipment and medium. Background Art

[0002] Currently, work order management systems are widely used in various fields, such as IT service management, customer support, equipment maintenance, etc. However, traditional work order management methods face many challenges in terms of processing complexity and resource allocation. Different types of work orders have significant differences in processing complexity and required resources, which requires the work order processing system to be able to flexibly adapt to the characteristics of work orders and dynamically adjust processing strategies and resource allocation.

[0003] Traditional work order classification methods often rely on manual judgment or simple rule matching, which is difficult to accurately reflect the actual characteristics and needs of work orders. This not only leads to low processing efficiency, but may also cause unreasonable allocation of resources. At the same time, in the process of work order processing, the evaluation of resource consumption and value contribution lacks scientific basis, resulting in problems such as uneven cost sharing and unreasonable pricing. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes an intelligent work order management method, system, device and medium.

[0005] In a first aspect, the present application proposes an intelligent work order management method, comprising: classifying work orders according to differences in work order types, obtaining work orders of different categories, and constructing a work order feature vector based on the characteristics of each type of work order, combined with the complexity of work order processing and required resources;

[0006] Based on the work order feature vector, a work order processing strategy selection model is established;

[0007] In the process of work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed;

[0008] Constructing a work order resource allocation recommendation model according to a feature matrix of the work order resource consumption and the work order value contribution;

[0009] After the work order is processed, a work order cost accounting model is constructed based on the resource consumption data and work order feature data of the work order;

[0010] In the entire process of work order processing, data from each link is continuously collected, and each model is updated and optimized to balance the work order processing efficiency and processing cost.

[0011] More specifically, in the above technical solution, work orders are classified according to the differences in work order types to obtain work orders of different categories. Based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources, a work order feature vector is constructed, including:

[0012] Acquire a historical work order data set, wherein the historical work order data set includes a plurality of historical work order data;

[0013] Clustering the historical work order data set to obtain multiple clustering results;

[0014] Calculate the silhouette coefficient corresponding to each clustering result, determine the clustering result with the largest silhouette coefficient as the optimal clustering result, and obtain the number of clusters corresponding to the optimal clustering result;

[0015] Dividing the historical work order data set into a plurality of work order categories according to the number of clusters;

[0016] For each of the work order categories, extract the features of each historical work order data in the work order category, the features of the historical work order data include the average time required for work order processing and the One-Hot encoding of the required resources, and construct a work order feature vector based on the average time and the One-Hot encoding;

[0017] Inputting the work order feature vector corresponding to each work order category into a support vector machine for training to obtain a work order classification model;

[0018] Acquire a newly received work order, extract features of the newly received work order, input the features of the newly received work order into the work order classification model, and predict the category of the newly received work order;

[0019] Obtaining a preset processing flow and required resource type corresponding to the category of the newly received work order;

[0020] For each processing link of the preset processing flow, a time rotation method is used to perform task scheduling, and a resource instance is selected from an available resource pool for allocation according to the required resource type, so as to perform automated processing of the newly received work order;

[0021] Acquire the actual processing time and actually used resources of the newly received work order, and update the work order feature vector of the work order category corresponding to the newly received work order;

[0022] The updated work order feature vector is input into the support vector machine for retraining to obtain an updated work order classification model.

[0023] More specifically, in the above technical solution, based on the work order feature vector, a work order processing strategy selection model is established, including:

[0024] The work order feature vector is used as an input feature and the processing strategy is used as an output label, and a processing strategy selection model is trained by using a CART decision tree algorithm.

[0025] More specifically, in the above technical solution, during the work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed, including:

[0026] Obtain resource consumption data for each link in the work order processing process and obtain work order priority data;

[0027] Associating the resource consumption data with the priority data of the work order to obtain associated work order resource data;

[0028] Constructing a work order feature matrix including work order priority and resource consumption data according to the associated work order resource data;

[0029] Performing cluster analysis on the work order feature matrix, dividing the work orders into different categories, and obtaining the central feature vector of each category of work orders;

[0030] Calculate the mean and variance of resource consumption and the average value contribution of each type of work order, and use the t-test to determine whether there is a significant difference in resource allocation during the work order processing;

[0031] If there are significant differences in resource allocation, recommendations for optimizing resource allocation are generated based on the differences.

[0032] More specifically, in the above technical solution, a work order resource allocation recommendation model is constructed according to the feature matrix of the work order resource consumption and the work order value contribution, including:

[0033] The project-based collaborative filtering in the collaborative filtering algorithm is used to process the feature matrix of the work order resource consumption and the work order value contribution, and the similarity between the work orders is calculated to obtain a work order similarity matrix, wherein the cosine similarity is used to measure the similarity between the work orders;

[0034] Generate a recommendation list for each work order according to the work order similarity matrix, wherein the recommendation list includes several work orders with the highest cosine similarity to the current work order;

[0035] Obtaining resource allocation status of the work orders in the recommendation list to form a reference plan for resource allocation;

[0036] Matching the reference plan for resource allocation with the actual requirements of the current work order, and selecting the resource allocation strategy that best matches the requirements of the current work order;

[0037] A heuristic algorithm is used to optimize the resource allocation strategy to minimize overall resource consumption while meeting work order requirements;

[0038] Apply the optimized resource allocation strategy to the work order processing process.

[0039] More specifically, in the above technical solution, after the work order processing is completed, a work order cost accounting model is constructed according to the resource consumption data and work order feature data of the work order, including:

[0040] Obtain resource consumption data and work order feature data of completed work orders, wherein the resource consumption data includes CPU time, memory usage, and network traffic, and the work order feature data includes work order type, priority, complexity, and processing time;

[0041] Performing Z-Score standardization processing on the resource consumption data;

[0042] One-hot encoding is used to process the work order type;

[0043] Convert the priority of the work order from P1 to P4 to a value from 4 to 1, and multiply it by the processing time to form a first new feature;

[0044] Convert the complexity of the work order from high, medium, and low to values ​​of 3, 2, and 1, and multiply them by the processing time to form a second new feature;

[0045] The preprocessed data is divided into training set data and test set data in a ratio of 7:3;

[0046] Using the training set data to train a ridge regression model to obtain a first prediction model;

[0047] The test set data is used to evaluate the first prediction model to obtain the mean square error (MSE) and R-square value;

[0048] If the MSE is less than 0.5 and the R-square value is greater than 0.95, the first prediction model is determined as the work order cost prediction model;

[0049] If the MSE is greater than or equal to 0.5 or the R-square value is less than or equal to 0.95, the support vector regression model is trained using the training set data to obtain a second prediction model, and the second prediction model is determined as the work order cost prediction model;

[0050] Acquire new completed work order data, and process it using the same data preprocessing method as the training set data to obtain new work order data after preprocessing;

[0051] The preprocessed new work order data is input into the work order cost prediction model to obtain the cost prediction value of the new work order.

[0052] More specifically, in the above technical solution, in the whole process of work order processing, data of each link is continuously collected, and each model is updated and optimized to balance the work order processing efficiency and processing cost, including:

[0053] Obtain data on each link of the entire work order processing process, including the work order submission time, processor, processing time, processing results and customer satisfaction;

[0054] Obtain a pre-trained work order classification model and the data of each link, and when the amount of new work order data reaches a preset threshold, use an incremental learning algorithm to update the work order classification model to obtain a first updated model;

[0055] Using a cross-validation method to perform performance evaluation on the first updated model to obtain a first evaluation result;

[0056] If the first evaluation result meets the preset performance requirement, deploying the first updated model to the online service to obtain a first deployed model;

[0057] If the first evaluation result does not meet the preset performance requirement, adjusting the parameters of the work order classification model and re-executing the incremental learning algorithm;

[0058] Acquire newly submitted work order data, call the first deployment model to classify the newly submitted work order data, assign a handler according to the classification result and generate a reply content, and feed back the processing result to the customer;

[0059] Obtain the labor cost data and work order processing results of each link, and calculate the average processing cost and cost sharing ratio of each link;

[0060] Obtain the work order backlog and processing efficiency of each link. When the backlog of any link exceeds the preset backlog threshold or the processing efficiency is lower than the preset efficiency threshold, adjust the manpower input of this link.

[0061] In a second aspect, the present application also proposes an intelligent work order management system, comprising:

[0062] The first processing module: classifies the work orders according to the differences in work order types, obtains work orders of different categories, and constructs a work order feature vector based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources;

[0063] The second processing module: establishing a work order processing strategy selection model based on the work order feature vector;

[0064] The third processing module: during the work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed;

[0065] Fourth processing module: constructing a work order resource allocation recommendation model according to the feature matrix of the work order resource consumption and the work order value contribution;

[0066] The fifth processing module: after the work order processing is completed, a work order cost accounting model is constructed according to the resource consumption data and work order feature data of the work order;

[0067] The sixth processing module: in the whole process of work order processing, continuously collect data from each link, and update and optimize each model to balance the work order processing efficiency and processing cost.

[0068] In a third aspect, the present application further proposes a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the methods described above when executing the computer program.

[0069] In a fourth aspect, the present application further proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method as described in any one of the above items is implemented.

[0070] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0071] The present invention uses a clustering algorithm to intelligently classify work orders, and constructs work order feature vectors based on the complexity of work order processing and the required resources, so that the system can more accurately understand the characteristics and requirements of various types of work orders. Based on these feature vectors, the system can establish an effective work order processing strategy selection model, automatically assign the most appropriate processing path and resources to different types of work orders, thereby avoiding the problem of low processing efficiency caused by human misjudgment or lack of experience in traditional methods. This intelligent processing method can significantly improve the processing speed and quality of work orders.

[0072] During the work order processing, the present invention not only focuses on processing efficiency, but also takes into account the rationality of cost allocation. By collecting resource consumption data of each link in real time and combining the priority and value contribution of the work order, the system can construct a feature matrix of work order resource consumption and value contribution, and then establish a work order resource allocation recommendation model and a cost accounting model. These models can help enterprises more accurately evaluate the cost of each link, achieve a dynamic balance of cost allocation, and avoid resource waste and cost overruns. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0074] Figure 1 It is a flowchart of an intelligent work order management method provided by one embodiment of the present invention;

[0075] Figure 2 It is a structural diagram of an intelligent work order management system provided by one embodiment of the present invention;

[0076] Figure 3 It is a schematic diagram of the structure of the terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0077] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0078] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0079] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0080] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0081] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0082] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0083] See also Figure 1 The present application proposes an intelligent work order management method, including: S101 classifying work orders according to the differences in work order types to obtain work orders of different categories, constructing a work order feature vector based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources, S102 establishing a work order processing strategy selection model based on the work order feature vector, S103 in the process of work order processing, by collecting resource consumption data of each link, combined with the priority of the work order, constructing a feature matrix of work order resource consumption and work order value contribution, S104 constructing a work order resource allocation recommendation model based on the feature matrix of work order resource consumption and work order value contribution, S105 after the work order processing is completed, constructing a work order cost accounting model based on the resource consumption data and work order feature data of the work order; S106 in the whole process of work order processing, continuously collecting data from each link, and updating and optimizing each model to balance the work order processing efficiency and processing cost.

[0084] Specifically, work order classification is the first step in optimizing the work order processing process. For example, IT support work orders can be divided into network problems, software problems, hardware problems, account problems, etc. Customer service work orders of e-commerce platforms can be divided into product consultation, order problems, after-sales service, complaints and suggestions, etc. The purpose of this classification is to better understand the characteristics of different types of work orders, so as to formulate more effective processing strategies. Next, it is necessary to construct a work order feature vector. Taking the "network problem" of IT support work orders as an example, its feature vector can include: the number of users affected, network delay time, fault duration, frequency of problem occurrence, whether it is a core business system, etc. Each feature is assigned a different weight, for example, the weight of the number of users affected is the highest, followed by whether it is a core business system. The feature vector of software problems can be: software version, operating system, error message, reproduction steps, scope of impact, etc. Based on the work order feature vector, a work order processing strategy selection model can be established. For example, network problems that affect a large number of users and involve core business systems should be handled first and assigned to experienced engineers; software problems with a small impact range and non-core business can be assigned to junior engineers or handled using automated tools. This can effectively improve the efficiency of work order processing and reduce resource consumption. During the work order processing, it is necessary to collect resource consumption data of each link. For example, the resource consumption of processing a network problem may include: engineer's working hours, network diagnostic tools used, required server resources, etc. Combined with the priority of the work order, a feature matrix of work order resource consumption and work order value contribution can be constructed. Assume that the work order value contribution of a high priority network problem is 10, and the work order value contribution of a low priority software problem is 1. Based on the feature matrix of work order resource consumption and work order value contribution, a work order resource allocation recommendation model can be constructed. The model can recommend suitable engineers, tools and resources based on the value contribution and required resources of the work order. For example, for a network problem with high value contribution but high resource consumption, the model may recommend experienced engineers and advanced network diagnostic tools. After the work order is processed, it is necessary to construct a work order cost accounting model based on the resource consumption data and work order feature data of the work order. For example, the cost of a network problem work order may include the salary of the engineer, the cost of using the network diagnostic tool, the cost of server resources, etc. This model can help better understand the cost of work order processing and provide data support for subsequent cost optimization. Finally, in the entire process of work order processing, it is necessary to continuously collect data from each link and update and optimize each model. For example, if it is found that certain types of work order processing efficiency is low and costly, its feature vector can be analyzed, the processing strategy can be adjusted, or resource allocation can be optimized. In this way, the efficiency and cost of work order processing can be constantly balanced to achieve continuous improvement. For example, the customer service department of an e-commerce platform receives a work order about order delays. The feature vector of the work order may include: order amount, user level, delay time, user sentiment, etc.According to the feature vector, the system determines that the work order has a higher priority and assigns it to an experienced customer service staff. The customer service staff inquires about the logistics information and finds that the order is delayed due to weather reasons. The customer service staff explains the situation to the user and provides a corresponding compensation plan. Finally, the system records the processing time of the work order, the customer service staff's working hours, the cost of the compensation plan and other data for subsequent model optimization. In another example, the user reports that an App function cannot be used. The feature vector may include: App version, operating system, network status, user description, etc. Based on these features, the system determines that it may be caused by a software bug and assigns it to the developer for investigation. After the developer locates the problem, fixes the bug and releases a new version. The system then records the processing time of this work order, the developer's working hours, the severity of the bug and other data. These two examples reflect the entire process from work order classification, feature extraction to resource allocation and cost accounting, and also show how to formulate different processing strategies according to the characteristics of different types of work orders.

[0085] In some embodiments, work orders are classified according to the differences in work order types to obtain work orders of different categories. Based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources, a work order feature vector is constructed, including:

[0086] Acquire a historical work order data set, wherein the historical work order data set includes a plurality of historical work order data;

[0087] Clustering the historical work order data set to obtain multiple clustering results;

[0088] Calculate the silhouette coefficient corresponding to each clustering result, determine the clustering result with the largest silhouette coefficient as the optimal clustering result, and obtain the number of clusters corresponding to the optimal clustering result;

[0089] Dividing the historical work order data set into a plurality of work order categories according to the number of clusters;

[0090] For each of the work order categories, extract the features of each historical work order data in the work order category, the features of the historical work order data include the average time required for work order processing and the One-Hot encoding of the required resources, and construct a work order feature vector based on the average time and the One-Hot encoding;

[0091] Inputting the work order feature vector corresponding to each work order category into a support vector machine for training to obtain a work order classification model;

[0092] Acquire a newly received work order, extract features of the newly received work order, input the features of the newly received work order into the work order classification model, and predict the category of the newly received work order;

[0093] Obtaining a preset processing flow and required resource type corresponding to the category of the newly received work order;

[0094] For each processing link of the preset processing flow, a time rotation method is used to perform task scheduling, and a resource instance is selected from an available resource pool for allocation according to the required resource type, so as to perform automated processing of the newly received work order;

[0095] Acquire the actual processing time and actually used resources of the newly received work order, and update the work order feature vector of the work order category corresponding to the newly received work order;

[0096] The updated work order feature vector is input into the support vector machine for retraining to obtain an updated work order classification model.

[0097] Specifically, historical work order data contains rich business rules. Cluster analysis can reveal different types of work orders. For example, an e-commerce platform has accumulated a large amount of after-sales work order data, including returns, exchanges, repairs, complaints, etc. By clustering these historical work order data, the work orders can be roughly divided into categories such as "product quality problems", "logistics problems", "seven-day unconditional returns and exchanges", and "malicious bad reviews". The silhouette coefficient can evaluate the quality of the clustering results. The value of the silhouette coefficient is between -1 and 1, and the larger the value, the better the clustering effect. Assuming that in the example of e-commerce after-sales work orders, try to use different numbers of clusters (for example, 3, 4, and 5 categories) for clustering, and calculate the silhouette coefficients respectively. If the silhouette coefficient is the largest when clustering into 4 categories, the optimal number of clusters is determined to be 4, and the historical work order data is divided into 4 work order categories. For each work order category, work order features need to be extracted. Still taking the e-commerce after-sales work order as an example, the work order features of the "product quality problem" category may include: the average time required for work order processing (for example, 3 days), the one-hot code of the required resources (for example, customer service personnel, quality inspection personnel, and warehouse personnel are required). By combining the average time and the one-hot code, the feature vector of the work order category can be constructed, for example: [3,1,1,1], where 3 means the average processing time is 3 days, and the three 1s respectively indicate that customer service, quality inspection, and warehouse personnel are required. By inputting the feature vector corresponding to each work order category into the support vector machine for training, a work order classification model can be obtained. This model can predict the category to which the new work order belongs based on its features. For example, the feature vector of a new work order is [2,1,0,1]. When it is input into the trained model, the model predicts that the work order belongs to the "seven-day unconditional return and exchange" category. Each work order category corresponds to a preset processing flow and the type of resources required. The preset processing flow of the "seven-day unconditional return and exchange" category may be: customer service confirms the return and exchange information → warehouse receives the return → finance refunds. The types of resources required may be: customer service staff, warehouse staff, and financial staff. The time rotation method can schedule tasks for each link in the processing flow. For example, customer service staff need to process multiple return and exchange work orders. The system can assign customer service staff to each work order in turn according to the time rotation method to ensure that each work order can be processed in a timely manner. According to the required resource type, the system can select resource instances from the available resource pool for allocation. For example, the system can select one from the idle warehouse staff to handle the warehousing operation of a return work order. After the new work order is processed, the actual processing time and the actual resources used need to be recorded. For example, the actual processing time of a "seven-day unconditional return and exchange" work order is 2 days, and customer service A, warehouse B, and financial C are actually used. These data can be used to update the feature vector corresponding to the work order category, for example, updating the average processing time to 5 days.By inputting the updated feature vector into the support vector machine for retraining, the work order classification model can be continuously optimized, and the classification accuracy and resource allocation efficiency can be improved. The benefit of this is that the model can continuously adapt to new work order data and business changes, thereby better predicting work order categories and optimizing resource allocation. Through the above steps, the automated processing of work orders, the optimal allocation of resources, and the effective control of costs can be achieved. Continuously collecting data and updating the model can continuously improve system performance and ultimately achieve a balance between work order processing efficiency and costs.

[0098] In some embodiments, based on the work order feature vector, a work order processing strategy selection model is established, including:

[0099] The work order feature vector is used as an input feature and the processing strategy is used as an output label, and a processing strategy selection model is trained by using a CART decision tree algorithm.

[0100] Specifically, the work order feature vector can be understood as a digital abstraction of the work order. It uses some key indicators to describe the characteristics of the work order so that the computer can understand and process it. For example, a work order describing a network failure can have a feature vector that includes the duration of the failure, the number of affected users, the type of network equipment, etc. These indicators together form a vector, such as [10 minutes, 100 people, router], which represents the characteristics of this work order. One-Hot encoding is a method of converting categorical data into numerical data. For example, the resource type required for a work order can be hardware, software, or network. Using One-Hot encoding, hardware can be represented as [1,0,0], software can be represented as [0,1,0], and network can be represented as [0,0,1]. In this way, the original text description can be converted into a digital form that can be processed by the computer. Support vector machine (SVM) is a commonly used machine learning algorithm that can be used for classification and regression. In this scenario, it is used to train a work order classification model. The input of the model is the work order feature vector, and the output is the work order category. Through training, SVM can learn the boundaries between different work order categories, so as to accurately classify new work orders. For example, a work order about database problems can be classified into the category of "database failure" after its feature vector is processed by the trained SVM model. The time rotation method is a commonly used task scheduling algorithm. It arranges all tasks into a circular queue and allocates processor resources to each task in turn according to the time slice, so as to ensure that each task can get a fair processing opportunity; for example, there are three work orders that need to be processed, which take 10 minutes, 5 minutes and 2 minutes respectively. Using the time rotation method, the time slice can be set to 2 minutes, then the first work order needs 5 time slices, the second work order needs 3 time slices (rounded up), and the third work order needs 1 time slice. The available resource pool refers to a collection of resources that can be dynamically allocated, such as servers, database connections, network bandwidth, etc. When a work order needs to be processed, the appropriate resource instance can be selected from the resource pool for allocation to meet the processing requirements of the work order. For example, a work order that needs to access a database can obtain an available connection from the database connection pool to perform database operations. CART decision tree is a machine learning algorithm for classification and regression. It recursively divides the data set into smaller subsets to eventually form a tree structure. Each node represents a feature, each branch represents a feature value, and each leaf node represents an output label.In this scenario, the CART decision tree can select the appropriate processing strategy based on the work order feature vector. For example, if the work order feature vector contains "fault duration" and "number of affected users", the CART decision tree can classify the work order into different processing strategies based on these two features. For example, if the fault duration exceeds 1 hour and the number of affected users exceeds 1,000, the "emergency processing" strategy is selected; otherwise, the "regular processing" strategy is selected. Assume that there are three types of work orders: network failure, software failure, and hardware failure. The average processing time of the work order and the required resource type are extracted as features. For example, the average processing time of network failure is 30 minutes, and the required resource is the network device; the average processing time of software failure is 60 minutes, and the required resource is the server; the average processing time of hardware failure is 90 minutes, and the required resource is the hardware device. One-Hot encoding is used to represent the resource type, network devices are [1,0,0], servers are [0,1,0], and hardware devices are [0,0,1]. The feature vector of network failure is [30, [1, 0, 0]], the feature vector of software failure is [60, [0, 1, 0]], and the feature vector of hardware failure is [90, [0, 0, 1]]. These feature vectors are input into the CART decision tree for training. The CART decision tree can be divided according to the features in the feature vector. For example, if the average processing time is less than 60 minutes, it is divided into one category; otherwise, it is divided into another category. Different processing strategies can be defined on each leaf node. For example, for work orders with an average processing time of less than 60 minutes, a fast processing strategy is adopted; for work orders with an average processing time of greater than or equal to 60 minutes, a standard processing strategy is adopted. When a new work order arrives, its feature vector is extracted and input into the trained CART decision tree. The decision tree will divide the work order into the corresponding leaf nodes according to the feature vector and select the corresponding processing strategy. For example, a new work order with an average processing time of 45 minutes and a required resource of network equipment has a feature vector of [45, [1, 0, 0]]. When it is input into the decision tree, it will be divided into the leaf node of the quick processing strategy, so that the quick processing strategy can be selected for processing. By using the CART decision tree, the appropriate processing strategy can be automatically selected according to the feature vector of the work order, thereby improving the efficiency of work order processing.

[0101] In some embodiments, during the work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed, including:

[0102] Obtain resource consumption data for each link in the work order processing process and obtain work order priority data;

[0103] Associating the resource consumption data with the priority data of the work order to obtain associated work order resource data;

[0104] Constructing a work order feature matrix including work order priority and resource consumption data according to the associated work order resource data;

[0105] Performing cluster analysis on the work order feature matrix, dividing the work orders into different categories, and obtaining the central feature vector of each category of work orders;

[0106] Calculate the mean and variance of resource consumption and the average value contribution of each type of work order, and use the t-test to determine whether there is a significant difference in resource allocation during the work order processing;

[0107] If there are significant differences in resource allocation, recommendations for optimizing resource allocation are generated based on the differences.

[0108] Specifically, obtain the resource consumption data of each link in the work order processing process, such as the processing time, labor cost, number of software tools used, etc. of each link. At the same time, obtain the priority data of the work order, such as P1, P2, P3, etc. The larger the value, the lower the priority. Associate the resource consumption data with the priority data of the work order. For example, a work order with a P1 priority consumes 2 hours in the "problem location" link, 1 hour in the "solution formulation" link, and 3 hours in the "solution implementation" link, using 3 different software tools. In this way, the associated work order resource data is obtained. Based on the associated work order resource data, construct a work order feature matrix containing work order priority and resource consumption data. Each row of this matrix represents a work order, and each column represents a feature, such as work order priority, problem location time, solution formulation time, solution implementation time, number of software tools used, etc. Suppose there are three work orders, forming the following feature matrix: Work order 1: P1, 2 hours, 1 hour, 3 hours, 3 Work order 2: P2, 1 hour, 5 hours, 2 hours, 2 Work order 3: P1, 3 hours, 5 hours, 4 hours, 4 Perform cluster analysis on the work order feature matrix. For example, using the K-Means algorithm, the work orders are divided into different categories. Suppose the above three work orders are divided into two categories, the first category contains work order 1 and work order 3, and the second category contains work order 2. Get the central feature vector of each category of work orders, for example, the central feature vector of the first category is: P1, 5 hours, 25 hours, 5 hours, 5; the central feature vector of the second category is: P2, 1 hour, 5 hours, 2 hours, 2. Calculate the mean and variance of resource consumption and the average value contribution of each category of work orders. For example, the average problem location time of the first category of work orders is 5 hours, and the variance is 25; the average value contribution (assuming that it is evaluated based on the benefits or importance of the problems solved by the work orders) is 1 million yuan. The average problem location time of the second type of work order is 1 hour, and the variance is 0; the average value contribution is 500,000 yuan. The t-test is used to determine whether there is a significant difference in resource allocation during the work order processing process. For example, comparing the difference in problem location time between the first and second types of work orders, assuming that the p value of the t-test result is less than 05, it is considered that there is a significant difference in the problem location time between the two types of work orders, which means that the time spent on the problem location link of the high-priority work order (P1) is significantly longer than that of the low-priority work order (P2). If there is a significant difference in resource allocation, a suggestion for optimizing resource allocation is generated based on the difference. For example, since the value contribution of the P1 work order is higher and the time spent on the problem location link is significantly more, it can be recommended to increase the resource investment of the P1 work order in the problem location link, such as arranging more experienced engineers to handle it, or providing more advanced location tools to locate the problem faster, thereby reducing the overall processing time, improving efficiency, and ultimately bringing greater value returns.The benefit of this is that it can solve high-priority problems more quickly, reduce the losses caused by high-priority problems, and use resources more effectively to avoid resource waste. At the same time, corresponding resource optimization suggestions can also be given based on the differences in resource consumption in other links. For example, for some work orders with low value contributions, you can consider reducing resource investment or using automated tools to process them, thereby reducing processing costs. Another example, suppose that through cluster analysis, it is found that a certain type of work order is characterized by a long processing time, but a small number of software tools are used. This may indicate that the processing flow of this type of work order is relatively complex, requiring more steps of manual intervention, which cannot be effectively supported by existing tools. In response to this situation, it can be recommended to develop new tools or improve existing tools to improve processing efficiency and reduce manual intervention. The benefits of this are that it can reduce processing time, reduce labor costs, and improve processing quality and reduce human errors.

[0109] In some embodiments, a work order resource allocation recommendation model is constructed based on the feature matrix of the work order resource consumption and the work order value contribution, including:

[0110] The project-based collaborative filtering in the collaborative filtering algorithm is used to process the feature matrix of the work order resource consumption and the work order value contribution, and the similarity between the work orders is calculated to obtain a work order similarity matrix, wherein the cosine similarity is used to measure the similarity between the work orders;

[0111] Generate a recommendation list for each work order according to the work order similarity matrix, wherein the recommendation list includes several work orders with the highest cosine similarity to the current work order;

[0112] Obtaining resource allocation status of the work orders in the recommendation list to form a reference plan for resource allocation;

[0113] Matching the reference plan for resource allocation with the actual requirements of the current work order, and selecting the resource allocation strategy that best matches the requirements of the current work order;

[0114] A heuristic algorithm is used to optimize the resource allocation strategy to minimize overall resource consumption while meeting work order requirements;

[0115] Apply the optimized resource allocation strategy to the work order processing process.

[0116] Specifically, the collaborative filtering algorithm can help find similar work orders, and refer to the resource allocation of similar work orders, so as to optimize the resource allocation of the current work order. The core idea of ​​collaborative filtering is "birds of a feather flock together", that is, similar work orders usually have similar resource requirements. This solution adopts project-based collaborative filtering, that is, to find similar work orders based on the characteristics of the work orders; the feature matrix is ​​the basis of collaborative filtering. In this solution, the feature matrix contains the resource consumption and value contribution of the work order. For example, work order A requires 2 manpower, 1 equipment, and the value contribution is 10; work order B requires 1 manpower, 2 equipment, and the value contribution is 8. These data form the basis of the feature matrix. More complex features can also include work order type, urgency, processing time, etc. Next, it is necessary to calculate the similarity between work orders. This solution uses cosine similarity to measure the similarity between work orders. The range of cosine similarity values ​​is between 0 and 1. The larger the value, the more similar the work orders are. For example, the cosine similarity between work order A and work order B is 8, and the cosine similarity between work order A and work order C is 2, which means that work order A is more similar to work order B. Based on the similarity, a recommendation list can be generated for each work order, which contains several work orders that are most similar to the current work order. For example, the recommendation list of work order A contains work order B, work order D, and work order E, and their similarities with work order A are 8, 7, and 6, respectively. Obtaining the resource allocation of the work orders in the recommendation list can form a reference plan for resource allocation. For example, the resource allocation of work order B is 1 manpower and 2 equipment; the resource allocation of work order D is 2 manpower and 1 equipment; the resource allocation of work order E is 3 manpower and 0 equipment. This information can be used as a reference for the resource allocation of work order A. By matching the reference plan with the actual needs of the current work order, the most appropriate resource allocation strategy can be screened out. For example, if the actual requirement of work order A is to reduce labor costs as much as possible, then the resource allocation scheme of work order B (1 person, 2 devices) may be more appropriate than the scheme of work order E (3 people, 0 devices). Finally, the heuristic algorithm is used to optimize the resource allocation strategy. The goal of the heuristic algorithm is to minimize the overall resource consumption while meeting the needs of the work order; for example, a genetic algorithm or a simulated annealing algorithm can be used to find the optimal resource allocation scheme. Suppose through optimization, it is found that work order A only needs 5 people and 5 devices to complete, which is more efficient than the original 2 people and 1 device. Applying the optimized resource allocation strategy to the work order processing process can improve resource utilization, reduce costs, and improve overall efficiency. For example, by applying the optimized resource allocation strategy, the processing time of work order A can be reduced, thereby responding to customer needs more quickly.For example, suppose there are two work orders, work order X and work order Y. Work order X needs to handle database failures, which requires database experts and server resources. Work order Y needs to perform software upgrades, which also requires database experts and server resources. Through collaborative filtering, it is found that work order X and work order Y are very similar. Looking at the history of work order X, it is found that it used 2 database experts and 4 servers and successfully solved the problem. Therefore, you can refer to this solution and allocate 2 database experts and 4 servers to work order Y. Of course, the final resource allocation also needs to be adjusted according to the specific situation of work order Y. For example, if the software upgrade scale of work order Y is larger, more resources may be required. In this way, historical experience can be used to allocate resources more effectively and avoid waste or shortage of resources.

[0117] In some embodiments, after the work order processing is completed, a work order cost accounting model is constructed based on the resource consumption data and work order feature data of the work order, including:

[0118] Obtain resource consumption data and work order feature data of completed work orders, wherein the resource consumption data includes CPU time, memory usage, and network traffic, and the work order feature data includes work order type, priority, complexity, and processing time;

[0119] Performing Z-Score standardization processing on the resource consumption data;

[0120] One-hot encoding is used to process the work order type;

[0121] Convert the priority of the work order from P1 to P4 to a value from 4 to 1, and multiply it by the processing time to form a first new feature;

[0122] Convert the complexity of the work order from high, medium, and low to values ​​of 3, 2, and 1, and multiply them by the processing time to form a second new feature;

[0123] The preprocessed data is divided into training set data and test set data in a ratio of 7:3;

[0124] Using the training set data to train a ridge regression model to obtain a first prediction model;

[0125] The test set data is used to evaluate the first prediction model to obtain the mean square error (MSE) and R-square value;

[0126] If the MSE is less than 0.5 and the R-square value is greater than 0.95, the first prediction model is determined as the work order cost prediction model;

[0127] If the MSE is greater than or equal to 0.5 or the R-square value is less than or equal to 0.95, the support vector regression model is trained using the training set data to obtain a second prediction model, and the second prediction model is determined as the work order cost prediction model;

[0128] Acquire new completed work order data, and process it using the same data preprocessing method as the training set data to obtain new work order data after preprocessing;

[0129] The preprocessed new work order data is input into the work order cost prediction model to obtain the cost prediction value of the new work order.

[0130] Specifically, in modern enterprise management, accurate prediction of work order resource consumption and cost is the key to improving efficiency and reducing costs. First, by collecting resource consumption data of completed work orders, such as CPU time, memory usage, and network traffic, as well as feature data of work orders, including work order type, priority, complexity, and processing time, a basis for data analysis can be provided. For example, a network service company may find that work orders for handling network failures consume abnormal CPU time and network traffic, which may be the focus of optimizing resource allocation. The Z-Score standardization of these data is to eliminate the influence of different magnitudes and distributions, so that the data can be comparable under the same standard. For example, memory usage may have a higher numerical range than CPU time. After standardization, they can be placed on the same scale for effective comparison. The one-hot encoding of work order types is to convert categorical variables into numerical variables, which is very important for computer processing and model training. Assuming that the work order types include maintenance, installation, and consultation, through one-hot encoding, the maintenance type may be encoded as [1,0,0], installation as [0,1,0], and consultation as [0,0,1], so that it can be used in the algorithm. The priority and complexity of the work order are converted into numerical values ​​and multiplied by the processing time to form new features, which helps to reveal the impact of work orders of different priorities and complexities on resource consumption. For example, a work order with high priority and high complexity may consume much more resources than a work order with low priority and simpler. By dividing the data into a training set and a test set, a model can be built on the training set and the effectiveness of the model can be verified on the test set. For example, using 70% of the data to train a ridge regression model and then testing the model's predictive ability with the remaining 30% of the data is a common practice in machine learning. If the mean square error MSE of the ridge regression model is less than 5 and the R-square value is greater than 95, it indicates that the model has good predictive performance and can be used as a work order cost prediction. If these conditions are not met, it may be necessary to try other models, such as support vector regression models, to find a better prediction method. Finally, the new work order data is processed in the same preprocessing method and input into the cost prediction model to obtain the cost prediction value of the new work order. The implementation of this process, such as a telecommunications company predicting the cost of a new network maintenance work order, can help the company better plan its budget and resource allocation. Through these steps, enterprises can not only optimize resource allocation, but also predict and prepare necessary resources in advance, thereby reducing costs and improving response speed while ensuring service quality.

[0131] In some embodiments, during the entire process of work order processing, data from each link is continuously collected, and each model is updated and optimized to balance the work order processing efficiency and processing cost, including:

[0132] Obtain data on each link of the entire work order processing process, including the work order submission time, processor, processing time, processing results and customer satisfaction;

[0133] Obtain a pre-trained work order classification model and the data of each link, and when the amount of new work order data reaches a preset threshold, use an incremental learning algorithm to update the work order classification model to obtain a first updated model;

[0134] Using a cross-validation method to perform performance evaluation on the first updated model to obtain a first evaluation result;

[0135] If the first evaluation result meets the preset performance requirement, deploying the first updated model to the online service to obtain a first deployed model;

[0136] If the first evaluation result does not meet the preset performance requirement, adjusting the parameters of the work order classification model and re-executing the incremental learning algorithm;

[0137] Acquire newly submitted work order data, call the first deployment model to classify the newly submitted work order data, assign a handler according to the classification result and generate a reply content, and feed back the processing result to the customer;

[0138] Obtain the labor cost data and work order processing results of each link, and calculate the average processing cost and cost sharing ratio of each link;

[0139] Obtain the work order backlog and processing efficiency of each link. When the backlog of any link exceeds the preset backlog threshold or the processing efficiency is lower than the preset efficiency threshold, adjust the manpower input of this link.

[0140] Specifically, the work order processing process involves multiple links, from submission to final processing completion and feedback to customers. The data of each link is crucial to cost accounting and efficiency improvement. In order to better understand the acquisition and use of data in the entire process of work order processing, an IT support work order processing process is used as an example. First, it is necessary to obtain data from each link of the entire process of work order processing. For example, a work order from submission to closure may go through five links: "problem submission", "problem classification", "technician processing", "solution verification" and "customer satisfaction survey". Each link records the corresponding data. For example, the "problem submission" link records the work order submission time, submitter and other information; the "problem classification" link records the processor, classification results, etc.; the "technician processing" link records the processing time, processing results, etc.; the "solution verification" link records the customer confirmation time, final solution, etc.; the "customer satisfaction survey" link records customer satisfaction ratings, feedback, etc. Collecting this data can help fully understand the entire process of work order processing and provide a data basis for subsequent analysis and optimization. Next, the update and deployment of the work order classification model are discussed. Assume that there is already a pre-trained work order classification model, which is used to automatically classify newly submitted work orders into different processing groups, such as network problem group, software problem group, hardware problem group, etc. As new work order data accumulates, this model needs to be updated regularly to improve its classification accuracy. When the amount of new work order data reaches a preset threshold, such as 1,000, the incremental learning algorithm will be used to update the model. The incremental learning algorithm allows the model to learn new knowledge without losing the original knowledge, avoiding the huge overhead of retraining the model every time. The updated model is called the first updated model. Then, the cross-validation method is used to evaluate the performance of the first updated model. If the evaluation results, such as accuracy and F1 value, meet the preset performance requirements, such as accuracy reaching 90% and F1 value reaching 85, the first updated model is deployed to the online service for actual use. If the evaluation results do not meet the preset performance requirements, it is necessary to adjust the model parameters, such as learning rate or regularization coefficient, and re-execute the incremental learning algorithm until the performance requirements are met. After a new work order is submitted, the system will call the deployed work order classification model to classify the work order. For example, a work order describing a network connection problem will be automatically classified into the network problem group and assigned to the technicians in that group to handle it. The model can also automatically generate some replies based on the classification results, such as "Hello, we have received your feedback on the network problem and will arrange for a technician to handle it as soon as possible." This can improve processing efficiency and enhance customer experience. After the work order is processed, the cost of each link needs to be calculated. For example, the labor cost of the "technician processing" link can be calculated by multiplying the technician's hourly salary by the processing time. The costs of other links can also be calculated based on actual conditions. By analyzing the average processing cost and cost sharing ratio of each link, you can find out the costly links and make targeted optimizations.For example, if the cost of the "problem classification" link is found to be too high, you can consider further optimizing the classification model to improve the classification efficiency. Finally, you need to pay attention to the backlog of work orders and the processing efficiency of each link. For example, the backlog of the "technician processing" link can be obtained by counting the number of work orders to be processed, and the processing efficiency can be obtained by calculating the number of work orders processed per unit time. If the backlog of a link exceeds the preset backlog threshold, such as 100, or the processing efficiency is lower than the preset efficiency threshold, such as 5 per hour, it is necessary to adjust the manpower input of this link. For example, you can increase the number of technicians in this link, or provide skills training to technicians to improve their processing efficiency. This can effectively control the backlog of work orders and ensure service quality.

[0141] See also Figure 2 , the present application also proposes an intelligent work order management system, including:

[0142] The first processing module 201: classifies the work orders according to the differences in work order types, obtains work orders of different categories, and constructs a work order feature vector based on the characteristics of each type of work order, combined with the complexity of work order processing and required resources;

[0143] Second processing module 202: establishing a work order processing strategy selection model based on the work order feature vector;

[0144] The third processing module 203: during the work order processing process, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed;

[0145] Fourth processing module 204: constructing a work order resource allocation recommendation model according to the feature matrix of the work order resource consumption and the work order value contribution;

[0146] Fifth processing module 205: after the work order processing is completed, construct a work order cost accounting model according to the resource consumption data and work order feature data of the work order;

[0147] The sixth processing module 206: in the whole process of work order processing, continuously collect data from each link, and update and optimize each model to balance the work order processing efficiency and processing cost.

[0148] It is understandable that if Figure 1 The contents of the intelligent work order management method embodiment shown in the figure are applicable to the embodiment of the intelligent work order management system. The functions specifically implemented by the embodiment of the intelligent work order management system are similar to those in the embodiment of the figure. Figure 1 The intelligent work order management method embodiment shown is the same as that shown in FIG. Figure 1 The beneficial effects achieved by the illustrated embodiment of the intelligent work order management method are also the same.

[0149] It should be noted that the information interaction, execution process and other contents between the above-mentioned systems are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0150] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0151] See also Figure 3 The embodiment of the present invention further provides a computer device 3, comprising: a memory 302 and a processor 301 and a computer program 303 stored in the memory 302. When the computer program 303 is executed on the processor 301, an intelligent work order management method as described in any one of the above methods is implemented.

[0152] The computer device 3 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Figure 3 It is only an example of computer device 3 and does not constitute a limitation on computer device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components, for example, it may also include input and output devices, network access devices, etc.

[0153] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0154] In some embodiments, the memory 302 may be an internal storage unit of the computer device 3, such as a hard disk or memory of the computer device 3. In other embodiments, the memory 302 may also be an external storage device of the computer device 3, such as a plug-in hard disk, a smart memory card (Smart Med iaCard, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 3. Further, the memory 302 may also include both an internal storage unit of the computer device 3 and an external storage device. The memory 302 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0155] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the intelligent work order management method as described in any one of the above methods is implemented.

[0156] In this embodiment, 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 present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0157] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An intelligent work order management method, characterized in that: include: According to the differences in work order types, work orders are classified to obtain work orders of different categories. Based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources, a work order feature vector is constructed; Based on the work order feature vector, a work order processing strategy selection model is established; In the process of work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed; Constructing a work order resource allocation recommendation model according to a feature matrix of the work order resource consumption and the work order value contribution; After the work order is processed, a work order cost accounting model is constructed based on the resource consumption data and work order feature data of the work order; During the entire process of work order processing, data from each link is continuously collected, and each model is updated and optimized to balance the work order processing efficiency and processing cost.

2. The intelligent work order management method according to claim 1, characterized in that: According to the differences in work order types, work orders are classified to obtain work orders of different categories. Based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources, a work order feature vector is constructed, including: Acquire a historical work order data set, wherein the historical work order data set includes a plurality of historical work order data; Clustering the historical work order data set to obtain multiple clustering results; Calculating the silhouette coefficient corresponding to each clustering result, determining the clustering result with the largest silhouette coefficient as the optimal clustering result, and obtaining the number of clusters corresponding to the optimal clustering result; Dividing the historical work order data set into a plurality of work order categories according to the number of clusters; For each of the work order categories, extract the features of each historical work order data in the work order category, the features of the historical work order data include the average time required for work order processing and the One-Hot encoding of the required resources, and construct a work order feature vector based on the average time and the One-Hot encoding; Inputting the work order feature vector corresponding to each work order category into a support vector machine for training to obtain a work order classification model; Acquire a newly received work order, extract features of the newly received work order, input the features of the newly received work order into the work order classification model, and predict the category of the newly received work order; Obtaining a preset processing flow and required resource type corresponding to the category of the newly received work order; For each processing link of the preset processing flow, a time rotation method is used to perform task scheduling, and a resource instance is selected from an available resource pool for allocation according to the required resource type, so as to perform automated processing of the newly received work order; Acquire the actual processing time and actually used resources of the newly received work order, and update the work order feature vector of the work order category corresponding to the newly received work order; The updated work order feature vector is input into the support vector machine for retraining to obtain an updated work order classification model.

3. The intelligent work order management method according to claim 1, characterized in that: Based on the work order feature vector, a work order processing strategy selection model is established, including: The work order feature vector is used as an input feature and the processing strategy is used as an output label, and a processing strategy selection model is trained by using a CART decision tree algorithm.

4. The intelligent work order management method according to claim 1, characterized in that: In the process of work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed, including: Obtain resource consumption data for each link in the work order processing process and obtain work order priority data; Associating the resource consumption data with the priority data of the work order to obtain associated work order resource data; Constructing a work order feature matrix including work order priority and resource consumption data according to the associated work order resource data; Performing cluster analysis on the work order feature matrix, dividing the work orders into different categories, and obtaining the central feature vector of each category of work orders; Calculate the mean and variance of resource consumption and the average value contribution of each type of work order, and use the t-test to determine whether there is a significant difference in resource allocation during the work order processing; If there are significant differences in resource allocation, recommendations for optimizing resource allocation are generated based on the differences.

5. The intelligent work order management method according to claim 1, characterized in that: According to the feature matrix of the work order resource consumption and the work order value contribution, a work order resource allocation recommendation model is constructed, including: The project-based collaborative filtering in the collaborative filtering algorithm is used to process the feature matrix of the work order resource consumption and the work order value contribution, and the similarity between the work orders is calculated to obtain a work order similarity matrix, wherein the cosine similarity is used to measure the similarity between the work orders; Generate a recommendation list for each work order according to the work order similarity matrix, wherein the recommendation list includes several work orders with the highest cosine similarity to the current work order; Obtaining resource allocation status of the work orders in the recommendation list to form a reference plan for resource allocation; Matching the reference plan for resource allocation with the actual requirements of the current work order, and selecting the resource allocation strategy that best matches the requirements of the current work order; A heuristic algorithm is used to optimize the resource allocation strategy to minimize overall resource consumption while meeting work order requirements; Apply the optimized resource allocation strategy to the work order processing process.

6. The intelligent work order management method according to claim 1, characterized in that: After the work order is processed, a work order cost accounting model is constructed based on the resource consumption data and work order feature data of the work order, including: Obtain resource consumption data and work order feature data of completed work orders, wherein the resource consumption data includes CPU time, memory usage, and network traffic, and the work order feature data includes work order type, priority, complexity, and processing time; Performing Z-Score standardization processing on the resource consumption data; One-hot encoding is used to process the work order type; Convert the priority of the work order from P1 to P4 to a value from 4 to 1, and multiply it by the processing time to form a first new feature; Convert the complexity of the work order from high, medium, and low to values ​​of 3, 2, and 1, and multiply them by the processing time to form a second new feature; The preprocessed data is divided into training set data and test set data in a ratio of 7:3; Using the training set data to train a ridge regression model to obtain a first prediction model; The test set data is used to evaluate the first prediction model to obtain the mean square error (MSE) and R-square value; If the MSE is less than 0.5 and the R-square value is greater than 0.95, the first prediction model is determined as the work order cost prediction model; If the MSE is greater than or equal to 0.5 or the R-square value is less than or equal to 0.95, the support vector regression model is trained using the training set data to obtain a second prediction model, and the second prediction model is determined as the work order cost prediction model; Acquire new completed work order data, and process it using the same data preprocessing method as the training set data to obtain new work order data after preprocessing; The preprocessed new work order data is input into the work order cost prediction model to obtain the cost prediction value of the new work order.

7. The intelligent work order management method according to claim 1, characterized in that: In the whole process of work order processing, data from each link is continuously collected, and each model is updated and optimized to balance the work order processing efficiency and processing cost, including: Obtain data on each link of the entire work order processing process, including the work order submission time, processor, processing time, processing results and customer satisfaction; Obtain a pre-trained work order classification model and the data of each link, and when the amount of new work order data reaches a preset threshold, use an incremental learning algorithm to update the work order classification model to obtain a first updated model; Using a cross-validation method to perform performance evaluation on the first updated model to obtain a first evaluation result; If the first evaluation result meets the preset performance requirement, deploying the first updated model to the online service to obtain a first deployed model; If the first evaluation result does not meet the preset performance requirement, adjusting the parameters of the work order classification model and re-executing the incremental learning algorithm; Acquire newly submitted work order data, call the first deployment model to classify the newly submitted work order data, assign a handler according to the classification result and generate a reply content, and feed back the processing result to the customer; Obtain the labor cost data and work order processing results of each link, and calculate the average processing cost and cost sharing ratio of each link; Obtain the work order backlog and processing efficiency of each link. When the backlog of any link exceeds the preset backlog threshold or the processing efficiency is lower than the preset efficiency threshold, adjust the manpower input of this link.

8. An intelligent work order management system, characterized in that: include: The first processing module: classifies the work orders according to the differences in work order types, obtains work orders of different categories, and constructs a work order feature vector based on the characteristics of each type of work order, combined with the complexity of work order processing and the required resources; The second processing module: establishing a work order processing strategy selection model based on the work order feature vector; The third processing module: during the work order processing, by collecting resource consumption data of each link and combining the priority of the work order, a feature matrix of work order resource consumption and work order value contribution is constructed; Fourth processing module: constructing a work order resource allocation recommendation model according to the feature matrix of the work order resource consumption and the work order value contribution; The fifth processing module: after the work order processing is completed, a work order cost accounting model is constructed according to the resource consumption data and work order feature data of the work order; The sixth processing module: in the whole process of work order processing, continuously collect data from each link, and update and optimize each model to balance the work order processing efficiency and processing cost.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images