Broadband installation and maintenance work order processing method and device, equipment and storage medium

Abstract

The invention discloses a broadband installation and maintenance work order processing method and device, equipment and a storage medium, relates to the technical field of household broadband operation and maintenance, and can improve the management efficiency of broadband installation and maintenance work orders. According to the specific scheme, the method comprises the steps of obtaining a training sample set, and performing multiple times of training on a preset initial decision model by using the training sample set; a splitting mechanism in the initial decision model is determined according to information gain and business key indexes, the initial decision model comprises a self-adaptive weight adjustment mechanism, the self-adaptive weight adjustment mechanism is used for adjusting the weight of sample information, the initial decision model further comprises a self-adaptive comprehensive adjustment mechanism, and the self-adaptive comprehensive adjustment mechanism is used for adjusting the weight of the sample information. The self-adaptive comprehensive adjustment mechanism is used for adjusting a learning strategy of the decision model; and utilizing the target decision model to predict the processing time and the fault level of the currently received broadband installation and maintenance work order in real time, obtaining the predicted processing time and the predicted fault level, and generating a corresponding processing strategy according to the predicted processing time and the predicted fault level.

Description

Technical Field

[0001] The present application relates to the field of home broadband operation and maintenance technology, and in particular to a method, device, equipment and storage medium for processing broadband installation and maintenance work orders. Background Art

[0002] With the rapid development of the communications industry, the number of installation and maintenance work orders of broadband operators has increased dramatically. Traditional work order management methods can no longer meet the needs of efficient and accurate services. The existing work order management system lacks a real-time evaluation and early warning mechanism for the work efficiency of installation and maintenance personnel, resulting in uneven resource allocation and poor response timeliness. Summary of the invention

[0003] The present application provides a method, device, equipment and storage medium for processing broadband installation and maintenance work orders, which can improve the management efficiency of broadband installation and maintenance work orders.

[0004] In order to achieve the above objectives, this application adopts the following technical solutions:

[0005] In a first aspect of an embodiment of the present application, a method for processing a broadband installation and maintenance work order is provided, the method comprising:

[0006] Acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of the sample information includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes an actual processing time and a fault level of the broadband installation and maintenance work order;

[0007] The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0008] The initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined according to information gain and business criticality indicators, the business criticality indicators are determined according to the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined according to the learning rate of the model, the prediction error and the business criticality indicators, the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model;

[0009] The target decision model is used to predict in real time the processing time and fault level of the currently received broadband installation and maintenance work order, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

[0010] As a possible implementation manner, before obtaining the training sample set, the method further includes:

[0011] Acquire multiple important factors that affect the processing time and fault level of the work order, wherein the important factors include the work order type, the customer level, and the fault type;

[0012] Obtaining a weight coefficient of each of the important factors, and constructing the business key indicator according to the important factors and the weight coefficients;

[0013] The key business indicators are:

[0014] K i =w1·Type i +w2·Level i +w3·Fault i

[0015] Among them, K i Indicates business key indicators, Type i Is the work order type, Level i It is the customer level, Fault i is the fault type, w1, w2, w3 are the corresponding weight coefficients.

[0016] As a possible implementation manner, after constructing the business key indicators, the method further includes:

[0017] Constructing the decision model splitting mechanism based on the key business indicators and information gain;

[0018] The splitting mechanism is:

[0019] SC=α·Gain(S,F)-β·K i

[0020] Among them, SC represents the splitting mechanism, S is the sample set in the node, F is the feature considered for splitting, Gain is the information gain, which indicates the improvement of data purity after splitting, and K i is a key business indicator, and α and β are adjustment parameters in the model.

[0021] As a possible implementation manner, after constructing the business key indicators, the method further includes:

[0022] Constructing the adaptive weight adjustment mechanism based on the learning rate, prediction error and business key indicators of the model;

[0023] The adaptive weight adjustment mechanism is:

[0024] w new =wold +γ·Adjust(S,Error,K i )

[0025] Among them, w new is the sample adjusted weight, w old is the weight of the sample before adjustment, γ is the learning rate, and Adjust is the adjustment value;

[0026] The calculation formula of the adjustment value is:

[0027]

[0028] Among them, S represents the sample set, |S| represents the size of the sample set, and Error i represents the prediction error of the ith sample, Ki represents the business critical indicator of the ith sample, and max(Error) represents the maximum prediction error in the sample set.

[0029] As a possible implementation manner, after constructing the business key indicators, the method further includes:

[0030] Construct the adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is:

[0031] IAF(x)=δ(K i )·γ(H(x))

[0032] δ(K i )=1+ρ·tanh(K i )

[0033]

[0034] Among them, IAF(x) represents the comprehensive adjustment mechanism, δ(K i ) is the business criticality index adjustment factor, γ(H(x)) is the historical performance adjustment factor of the decision module, ρ is the adjustment factor preset in the decision model, which is used to control the influence of the business criticality index on the learning rate of the decision model. E(x) is the number of errors of sample x, μ and σ are the mean and standard deviation of the number of errors, respectively, and θ is the adjustment factor;

[0035] For each sample, the learning rate used is η(x) = η0·IAF(x), where η0 is the preset initial learning rate;

[0036] For each erroneous sample, the weight in the loss function of the decision model is adjusted to w(x)=IAF(x).

[0037] As a possible implementation, the training sample set is used to train a preset initial decision model until the prediction error of the decision model meets a preset condition to obtain a target decision model, including:

[0038] The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0039] The current training process in the multiple trainings is as follows:

[0040] The prediction error, learning rate and business key indicators of each sample in the previous training process, as well as the maximum prediction error in the training sample set in the previous training process, are input into the adaptive weight adjustment mechanism to obtain the current weight of each sample in the current training process;

[0041] Input the adjustment factor of the decision module in the previous training process, the number of errors of each sample, the preset initial learning rate of each sample, the average and standard deviation of the number of sample errors in the training sample set into the adaptive comprehensive adjustment mechanism to obtain the current learning rate of each sample in the current training process and the current weight of each error sample in the loss function of the decision model;

[0042] Obtaining a current sample set according to the current weight of each sample in the current training process, the current learning rate, and the current weight of each erroneous sample in the loss function of the decision model;

[0043] Perform decision splitting on the current sample set based on the decision model splitting mechanism to obtain a prediction error of the current training process.

[0044] As a possible implementation manner, generating a corresponding processing strategy according to the predicted processing time and the predicted fault level includes:

[0045] Generate a corresponding work order emergency level according to the predicted processing time and the predicted fault level;

[0046] Corresponding installation and maintenance personnel and processing time are assigned according to the urgency level of the work order.

[0047] According to a second aspect of an embodiment of the present application, a device for processing a broadband installation and maintenance work order is provided, the device comprising:

[0048] An acquisition module is used to acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of which includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes the actual processing time and fault level of the broadband installation and maintenance work order;

[0049] A training module, used to train the preset initial decision model multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0050] The initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined according to information gain and business criticality indicators, the business criticality indicators are determined according to the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined according to the learning rate of the model, the prediction error and the business criticality indicators, the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model;

[0051] The prediction module is used to use the target decision model to predict the processing time and fault level of the currently received broadband installation and maintenance work order in real time, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

[0052] According to a third aspect of an embodiment of the present application, there is provided an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for processing broadband installation and maintenance work orders according to the first aspect of an embodiment of the present application is implemented.

[0053] According to a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for processing broadband installation and maintenance work orders according to the first aspect of the embodiments of the present application is implemented.

[0054] The beneficial effects brought by the technical solution provided by the embodiment of the present application include at least:

[0055] The method for processing broadband installation and maintenance work orders provided in the embodiment of the present application obtains a training sample set, wherein the training sample set includes multiple sample information, each of which includes work order feature information of the broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, and the label information includes the actual processing time and fault level of the broadband installation and maintenance work order; the preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0056] The initial decision model is a decision model based on a gradient boosting tree. The splitting mechanism in the initial decision model is determined based on information gain and business criticality indicators. The business criticality indicators are determined based on work order type, customer level and fault type. The initial decision model includes an adaptive weight adjustment mechanism. The adaptive weight adjustment mechanism is used to adjust the weight of sample information. The adaptive weight adjustment mechanism is determined based on the learning rate, prediction error and business criticality indicators of the model. The initial decision model also includes an adaptive comprehensive adjustment mechanism. The adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model. The target decision model is used to predict the processing time and fault level of the currently received broadband installation and maintenance work order in real time to obtain the predicted processing time and predicted fault level, and a corresponding processing strategy is generated based on the predicted processing time and predicted fault level. This can improve the processing efficiency of broadband installation and maintenance work orders, thereby improving user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A flowchart of a method for processing a broadband installation and maintenance work order provided in an embodiment of the present application;

[0058] Figure 2 A structural diagram of a broadband installation and maintenance work order processing device provided in an embodiment of the present application;

[0059] Figure 3 A schematic diagram of the internal structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions 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 are within the scope of protection of this application.

[0061] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present disclosure, unless otherwise specified, "plurality" means two or more.

[0062] Additionally, the use of “based on” or “according to” is meant to be open and inclusive, as a process, step, calculation, or other action “based on” or “according to” one or more conditions or values ​​may, in practice, be based on additional conditions or beyond values.

[0063] With the rapid development of the communications industry, the number of installation and maintenance work orders of broadband operators has increased dramatically. Traditional work order management methods can no longer meet the needs of efficient and accurate services. The existing work order management system lacks a real-time evaluation and early warning mechanism for the work efficiency of installation and maintenance personnel, resulting in uneven resource allocation and poor response timeliness.

[0064] In addition, the processing of installation and maintenance work orders in the communications industry involves many factors, including work order type, customer level, fault type, etc. These factors are complex and changeable, making it challenging to estimate processing time and fault level. Traditional algorithms usually ignore these business characteristics. Existing gradient boosting trees perform well in predicting tasks on static data sets, but are stretched when dealing with real-time changing data streams and complex business requirements. The present invention significantly improves the accuracy of the model's estimated processing time and fault level for work order predictions by introducing business criticality indicators, a dynamic node splitting mechanism, and adaptive weight adjustment.

[0065] The present application embodiment provides a method for processing a broadband installation and maintenance work order, such as Figure 1 As shown, the method comprises the following steps:

[0066] Step 101: Acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of which includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes the actual processing time and fault level of the broadband installation and maintenance work order;

[0067] The work order characteristic information includes the following characteristic information:

[0068] Work order types include: installation, maintenance, upgrade, inspection, etc. Customer levels include: VIP customers, ordinary customers, corporate customers, etc., which may affect the processing priority. Fault severity: minor, moderate, severe, urgent. Customer feedback information: customer description and feedback on the fault. Historical maintenance record: past maintenance history of the customer or equipment. Geographic location information: customer location may affect the arrival time of maintenance personnel. Time information: the date and time the work order was created, which may affect the processing speed (difference between holidays and weekdays). Accessory inventory status: whether the required accessories are in stock. Weather conditions: extreme weather may affect maintenance work. Equipment type and age: different equipment may have different failure rates and maintenance difficulties.

[0069] The process of obtaining the broadband installation and maintenance work order information can be as follows:

[0070] Data source interface definition: Choose backend technologies such as Node.js or Java for API development. Write code to call the API of the communication service system, which handles real-time requests and receives work order data, and returns response data in JSON format. Security and authentication: Configure the server and client to support HTTPS. This protocol is a secure transmission protocol based on SSL / TLS, which can ensure the confidentiality, integrity and credibility of network communications, and is used to protect user privacy and prevent network attacks. Implement the OAuth 2.0 authentication process, which authorizes third-party applications to access protected resources without leaking user passwords, improves security and user experience, and ensures that each data request is authorized. Real-time data monitoring and acquisition: Implement the connection and maintenance logic of WebSocket on the server side. This full-duplex communication protocol based on the TCP protocol can establish a persistent connection between the client and the server, realizing real-time data transmission and the development of interactive applications. Set up data reception callbacks to process pushed work order data in real time. Data preprocessing: Implement data validity verification to ensure the quality of collected data; and convert the collected data into the required format. This invention uses JSON as an embodiment. JSON is a lightweight data exchange format based on text, easy to read and write, and widely used for data transmission and storage in Web applications. Data storage and synchronization: Deploy and configure a real-time database. This invention uses Kafka as an embodiment. Kafka is a high-throughput, distributed message queue system for processing large-scale data streams and real-time data pipelines, supporting horizontal expansion and fault tolerance; and realizes instant data writing and backup to ensure data integrity and consistency.

[0071] The above-mentioned process of acquiring broadband installation and maintenance work order information can ensure the real-time, accuracy and security of data collection in the broadband installation and maintenance work order intelligent management system, and provide solid data support for the efficient operation of the system.

[0072] Step 102: Use the training sample set to train the preset initial decision model multiple times until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model.

[0073] Among them, the initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined based on information gain and business criticality indicators, the business criticality indicators are determined based on the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined based on the learning rate, prediction error and business criticality indicators of the model, and the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model.

[0074] Step 103: Use the target decision model to predict in real time the processing time and fault level of the currently received broadband installation and maintenance work order, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

[0075] The method for processing broadband installation and maintenance work orders provided in the embodiment of the present application obtains a training sample set, wherein the training sample set includes multiple sample information, each of which includes work order feature information of the broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, and the label information includes the actual processing time and fault level of the broadband installation and maintenance work order; the preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0076] The initial decision model is a decision model based on a gradient boosting tree. The splitting mechanism in the initial decision model is determined based on information gain and business criticality indicators. The business criticality indicators are determined based on work order type, customer level and fault type. The initial decision model includes an adaptive weight adjustment mechanism. The adaptive weight adjustment mechanism is used to adjust the weight of sample information. The adaptive weight adjustment mechanism is determined based on the learning rate, prediction error and business criticality indicators of the model. The initial decision model also includes an adaptive comprehensive adjustment mechanism. The adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model. The target decision model is used to predict the processing time and fault level of the currently received broadband installation and maintenance work order in real time to obtain the predicted processing time and predicted fault level, and a corresponding processing strategy is generated based on the predicted processing time and predicted fault level. This can improve the processing efficiency of broadband installation and maintenance work orders, thereby improving user satisfaction.

[0077] Optionally, before obtaining the training sample set, the method further includes:

[0078] Acquire multiple important factors that affect the processing time and fault level of the work order, wherein the important factors include the work order type, the customer level, and the fault type;

[0079] Obtaining a weight coefficient of each of the important factors, and constructing the business key indicator according to the important factors and the weight coefficients;

[0080] The key business indicators are:

[0081] K i =w1·Type i +w2·Level i +w3·Fault i

[0082] Among them, K i Indicates business key indicators, Type i Is the work order type, Level i It is the customer level, Fault i is the fault type, w1, w2, w3 are the corresponding weight coefficients.

[0083] For example, the following is a database example that shows how to combine business data to obtain business key indicators. As shown in Table 1, the business key indicator of the example in the table is K i =2.7.

[0084]

[0085] Optionally, after constructing the business key indicators, the method further includes:

[0086] Constructing the decision model splitting mechanism based on the key business indicators and information gain;

[0087] The splitting mechanism is:

[0088] SC=α·Gain(S,F)-β·K i

[0089] Among them, SC represents the splitting mechanism, S is the sample set in the node, F is the feature considered for splitting, Gain is the information gain, which indicates the improvement of data purity after splitting, and K i is a key business indicator, and α and β are adjustment parameters in the model.

[0090] For example, suppose there is a decision tree node that is considering whether to split based on the fault type feature. The following is an example of how to apply the dynamic node splitting mechanism:

[0091] Sample value S in the node: contains 100 samples, of which 60 samples are network failures and 40 are hardware failures. Consider split feature F: failure type. Information gain Gain (S, F): Assumed to be 0.3. Business criticality indicator K i It can be calculated by the above formula, and the average value at this node is assumed to be 2.0. Adjustment parameters: α and β are set to 1.0 and 0.5 respectively. Then the calculation of the splitting standard SC is:

[0092] SC=α·Gain(S,F)-β·Ki=1.0·0.3-0.5·2.0=0.3-1.0=-0.7

[0093] Since SC is negative, this indicates that this split may not be the best choice based on business criticality.

[0094] Optionally, after constructing the business key indicators, the method further includes:

[0095] Constructing the adaptive weight adjustment mechanism based on the learning rate, prediction error and business key indicators of the model;

[0096] The adaptive weight adjustment mechanism is:

[0097] w new =w old +γ·Adjust(S,Error,K i )

[0098] Among them, w new is the sample adjusted weight, w old is the weight of the sample before adjustment, γ is the learning rate, and Adjust is the adjustment value;

[0099] The calculation formula of the adjustment value is:

[0100]

[0101] Among them, S represents the sample set, |S| represents the size of the sample set, and Error i represents the prediction error of the ith sample, Ki represents the business critical indicator of the ith sample, and max(Error) represents the maximum prediction error in the sample set.

[0102] It is understandable that in order to make the model more sensitive to business changes, an adaptive weight adjustment mechanism is introduced to dynamically adjust the weight coefficient according to the prediction error and business feedback.

[0103] Optionally, after constructing the business key indicators, the method further includes:

[0104] Construct the adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is:

[0105] IAF(x)=δ(K i )·γ(H(x))

[0106] δ(K i )=1+ρ·tanh(K i )

[0107]

[0108] Among them, IAF(x) represents the comprehensive adjustment mechanism, δ(K i ) is the business criticality index adjustment factor, γ(H(x)) is the historical performance adjustment factor of the decision module, ρ is the adjustment factor preset in the decision model, which is used to control the influence of the business criticality index on the learning rate of the decision model. E(x) is the number of errors of sample x, μ and σ are the mean and standard deviation of the number of errors, respectively, and θ is the adjustment factor;

[0109] For each sample, the learning rate used is η(x) = η0·IAF(x), where η0 is the preset initial learning rate;

[0110] For each erroneous sample, the weight in the loss function of the decision model is adjusted to w(x)=IAF(x).

[0111] It is understandable that broadband installation and maintenance services are faced with rapidly changing service requirements, technology upgrades, and changing customer needs and market conditions. In view of the fact that the model needs to be constantly updated to reflect the latest business status and data trends, the present invention designs an adaptive comprehensive adjustment mechanism that takes into account the business key indicators K defined in the previous text. i The historical performance of the model on similar samples directly affects the learning rate used each time the model is updated and adjusts the weights of erroneous samples, thereby improving the ability of the training process to be strongly coupled with the business.

[0112] For each sample x, the actual learning rate used is η(x) = η0·IAF(x), that is, for the business-critical sample (K i The model has a higher learning rate, which means that the model pays more attention to business key points to ensure that the model quickly adapts to important samples and accurately predicts them. For each wrong sample x, its weight in the loss function is adjusted to w(x) = IAF(x), giving greater weight to wrongly predicted business key points or difficult-to-predict samples, promoting performance improvements for these samples.

[0113] By using an adaptive comprehensive adjustment mechanism, the decision model can automatically consider business criticality and historical performance when updating, allowing the model to learn adaptively based on the business importance and historical performance of each sample. In practice, this means that the model will pay more attention to work orders that have a large impact on the business, such as high-priority or high-risk failures. By updating the learning rate and sample weights in real time, the model can respond quickly to business changes and improve the accuracy of predictions for urgent and important events.

[0114] Optionally, the using the training sample set to train a preset initial decision model until the prediction error of the decision model meets a preset condition to obtain a target decision model includes:

[0115] The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0116] The current training process in the multiple trainings is as follows:

[0117] The prediction error, learning rate and business key indicators of each sample in the previous training process, as well as the maximum prediction error in the training sample set in the previous training process, are input into the adaptive weight adjustment mechanism to obtain the current weight of each sample in the current training process;

[0118] Input the adjustment factor of the decision module in the previous training process, the number of errors of each sample, the preset initial learning rate of each sample, the average and standard deviation of the number of sample errors in the training sample set into the adaptive comprehensive adjustment mechanism to obtain the current learning rate of each sample in the current training process and the current weight of each error sample in the loss function of the decision model;

[0119] Obtaining a current sample set according to the current weight of each sample in the current training process, the current learning rate, and the current weight of each erroneous sample in the loss function of the decision model;

[0120] Perform decision splitting on the current sample set based on the decision model splitting mechanism to obtain a prediction error of the current training process.

[0121] In the actual implementation process, the training process of the decision model can be:

[0122] Data preparation and preprocessing: Feature processing converts features such as work order type, customer level, and fault type into numerical form, which is suitable for model processing. Target variables: are the processing time and fault level of the work order.

[0123] Dynamic node splitting mechanism: Node splitting: In each decision tree of the decision model, node splitting is not only based on traditional information gain, but also considers the business key indicator K i The splitting criterion uses the above-mentioned dynamic node classification mechanism as the splitting criterion to balance the information gain and business importance of the data.

[0124] Adaptive weight adjustment: Weight adjustment logic: According to the prediction error and K of each sample i Dynamically adjust weights so that the model pays more attention to samples that are difficult to predict or important in business during training. Specifically, for samples with large prediction errors, especially those with high K i The samples with the same value increase their relative importance in subsequent training. new =w old +γ*Adjust(S, Error, K_i), where the Adjust function is based on the sample set S, the prediction error Error and K i Calculate weight adjustment values.

[0125] The adjusted weights affect the construction of newly added trees: samples with higher weights have a greater influence on determining the split point of the tree and forming the final prediction. This step focuses on how to change the contribution of a single sample to the model training process; and directly acts on the sample level to adjust the weight of each sample during the training process.

[0126] The comprehensive adjustment mechanism adjusts the learning rate of the new decision trees added in each iteration, so that the model pays more attention to the correction of business-critical or historically difficult-to-predict samples during the learning process. Specifically, IAF considers the business criticality K of each sample i and historical performance. The learning rate is adjusted to η(x) = η0*IAF(x), which affects the model update speed. For samples with incorrect predictions, their weights in the loss function are adjusted according to the comprehensive adjustment mechanism, prompting the model to pay more attention to these samples. The application of the comprehensive adjustment mechanism affects the learning strategy of the entire model, including the learning rate of new trees and the importance of erroneous samples when the model is updated. In this step, more attention is paid to the adjustment of the entire model learning process, including the adjustment of the learning rate and the overall sample weight; and it acts on the model update level, affecting the learning process of the entire model by adjusting the learning rate and weight.

[0127] In each iteration, a new tree based on the residual of the current model is constructed using the weights and learning rates adjusted as above.

[0128] Model validation and optimization: Use the test set to verify the accuracy of the model and adjust parameters as needed.

[0129] Model deployment: Deploy the trained gradient boosting tree model to the actual work order processing system.

[0130] Optionally, generating a corresponding processing strategy according to the predicted processing time and the predicted fault level includes: generating a corresponding work order urgency level according to the predicted processing time and the predicted fault level; and allocating corresponding installation and maintenance personnel and processing time according to the work order urgency level.

[0131] In addition, considering the current number of work orders accepted by installation and maintenance personnel and the timely processing rate of historical work orders, the system combines the algorithm prediction results and dynamically adjusts the early warning strategy to achieve accurate and effective early warnings and provide graded reminders to ensure the efficient operation of the entire system.

[0132] Among them, the warning condition setting can be: Data threshold setting: According to historical data analysis, the threshold of key indicators is set, such as the work order processing time exceeds 150% of the average time, the current work order volume of the installation and maintenance personnel exceeds 200% of the average value, etc. Real-time data monitoring: Use real-time data stream processing technology (Apache Kafka is used as an embodiment of the present invention) to monitor the work order data and calculate key indicators in real time.

[0133] The warning trigger mechanism can be: Develop a rule engine to automatically trigger a warning when the monitored data exceeds the preset threshold. Write logic judgment code to trigger a processing delay warning if the work order processing time exceeds the threshold.

[0134] The early warning notification mechanism can be: Multi-channel notification: Develop a module for sending early warning notifications, supporting multiple methods such as application push, SMS, phone calls, etc. Notification content customization: Customize the notification content according to the warning type and level, and provide necessary information and possible operational suggestions.

[0135] Definition of graded reminders: Based on the urgency of the work order and the processing progress, three reminder levels are defined: high, medium, and low.

[0136] High level: For example, the work order processing time exceeds the estimated time by 150%, the current work order volume of the installation and maintenance personnel exceeds 200% of the average, and emergency work orders are not responded to in time.

[0137] Medium level: If the work order processing time exceeds the estimated time, the current work order volume of the maintenance personnel exceeds 150% of the average, the response time is slightly delayed, etc.

[0138] Low level: general reminders, such as daily work order processing progress reminders.

[0139] Reminder logic implementation: Set up a scheduled task (the present invention uses Cron job as an example), regularly check the work order status, and trigger reminders according to the level rules. Write judgment logic, if the current time minus the work order start time exceeds a certain threshold and the work order status is still unfinished, trigger a reminder of the corresponding level.

[0140] Reminder notification mechanism: Customize notification content based on reminder level and work order status to ensure accurate and easy-to-understand information.

[0141] The present invention implements multi-channel early warning notifications, including email, SMS and application push, by designing a flexible early warning mechanism, dynamically adjusting the early warning strategy based on real-time data and machine learning predictions. It implements graded reminders based on the urgency of the work order and the progress of the processing. It sends status updates and progress reminders regularly to improve the response efficiency of installation and maintenance personnel and the efficiency of work order processing.

[0142] like Figure 2 As shown, an embodiment of the present application provides a device for processing a broadband installation and maintenance work order, the device comprising:

[0143] An acquisition module 11 is used to acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of which includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes the actual processing time and fault level of the broadband installation and maintenance work order;

[0144] A training module 12 is used to train the preset initial decision model multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0145] The initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined according to information gain and business criticality indicators, the business criticality indicators are determined according to the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined according to the learning rate of the model, the prediction error and the business criticality indicators, the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model;

[0146] The prediction module 13 is used to use the target decision model to predict the processing time and fault level of the currently received broadband installation and maintenance work order in real time, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

[0147] In one embodiment, the apparatus further comprises a building module 14, wherein the building module 14 is configured to:

[0148] Acquire multiple important factors that affect the processing time and fault level of the work order, wherein the important factors include the work order type, the customer level, and the fault type;

[0149] Obtaining a weight coefficient of each of the important factors, and constructing the business key indicator according to the important factors and the weight coefficients;

[0150] The key business indicators are:

[0151] K i =w1·Type i +w2·Level i +w3·Fault i

[0152] Among them, K i Indicates business key indicators, Type i Is the work order type, Level i It is the customer level, Fault i is the fault type, w1, w2, w3 are the corresponding weight coefficients.

[0153] In one embodiment, the building block 14 is further used to:

[0154] Constructing the decision model splitting mechanism based on the key business indicators and information gain;

[0155] The splitting mechanism is:

[0156] SC=α·Gain(S,F)-β·K i

[0157] Among them, SC represents the splitting mechanism, S is the sample set in the node, F is the feature considered for splitting, Gain is the information gain, which indicates the improvement of data purity after splitting, and K i is a key business indicator, and α and β are adjustment parameters in the model.

[0158] In one embodiment, the building block 14 is further used to:

[0159] Constructing the adaptive weight adjustment mechanism based on the learning rate, prediction error and business key indicators of the model;

[0160] The adaptive weight adjustment mechanism is:

[0161] w new =w old +γ·Adjust(S,Error,K i )

[0162] Among them, w new is the sample adjusted weight, w old is the weight of the sample before adjustment, γ is the learning rate, and Adjust is the adjustment value;

[0163] The calculation formula of the adjustment value is:

[0164]

[0165] Among them, S represents the sample set, |S| represents the size of the sample set, and Error i represents the prediction error of the ith sample, Ki represents the business critical indicator of the ith sample, and max(Error) represents the maximum prediction error in the sample set.

[0166] In one embodiment, the building block 14 is further used to:

[0167] Construct the adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is:

[0168] IAF(x)=δ(K i )·γ(H(x))

[0169] δ(K i )=1+ρ·tanh(K i )

[0170]

[0171] Among them, IAF(x) represents the comprehensive adjustment mechanism, δ(K i ) is the business criticality index adjustment factor, γ(H(x)) is the historical performance adjustment factor of the decision module, ρ is the adjustment factor preset in the decision model, which is used to control the influence of the business criticality index on the learning rate of the decision model. E(x) is the number of errors of sample x, μ and σ are the mean and standard deviation of the number of errors, respectively, and θ is the adjustment factor;

[0172] For each sample, the learning rate used is η(x)=η0·IAF(x), where η0 is the preset initial learning rate; for each erroneous sample, the weight in the loss function of the decision model is adjusted to w(x)=IAF(x).

[0173] In one embodiment, the training module 12 is specifically used for:

[0174] The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model;

[0175] The current training process in the multiple trainings is as follows:

[0176] The prediction error, learning rate and business key indicators of each sample in the previous training process, as well as the maximum prediction error in the training sample set in the previous training process, are input into the adaptive weight adjustment mechanism to obtain the current weight of each sample in the current training process;

[0177] Input the adjustment factor of the decision module in the previous training process, the number of errors of each sample, the preset initial learning rate of each sample, the average and standard deviation of the number of sample errors in the training sample set into the adaptive comprehensive adjustment mechanism to obtain the current learning rate of each sample in the current training process and the current weight of each error sample in the loss function of the decision model;

[0178] Obtaining a current sample set according to the current weight of each sample in the current training process, the current learning rate, and the current weight of each erroneous sample in the loss function of the decision model;

[0179] Perform decision splitting on the current sample set based on the decision model splitting mechanism to obtain a prediction error of the current training process.

[0180] In one embodiment, the detection module 13 is specifically used for:

[0181] Generate a corresponding work order emergency level according to the predicted processing time and the predicted fault level;

[0182] Corresponding installation and maintenance personnel and processing time are assigned according to the urgency level of the work order.

[0183] The broadband installation and maintenance work order processing device provided in this embodiment can execute the above-mentioned broadband installation and maintenance work order processing method embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0184] For the specific definition of the processing device of the broadband installation and maintenance work order, please refer to the definition of the processing method of the broadband installation and maintenance work order in the above text, which will not be repeated here. Each module in the above-mentioned processing device of the broadband installation and maintenance work order can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor of the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0185] The execution subject of the broadband installation and maintenance work order processing method provided in the embodiment of the present application can be an electronic device, and the electronic device can be a computer device, a terminal device, a server or a server cluster, and the embodiment of the present application does not make any specific limitations on this.

[0186] Figure 3 This is a schematic diagram of the internal structure of an electronic device provided in an embodiment of the present application. Figure 3 As shown, the electronic device includes a processor and a memory connected via a system bus. The processor is used to provide computing and control capabilities. The memory may include a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The computer program can be executed by the processor to implement the steps of the method for processing broadband installation and maintenance work orders provided in the above embodiments. The internal memory provides a cache operating environment for the operating system and computer program in the non-volatile storage medium.

[0187] Those skilled in the art will understand that Figure 3 The internal structure diagram of the electronic device shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0188] In another embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for processing a broadband installation and maintenance work order as in the embodiment of the present application are implemented.

[0189] In another embodiment of the present application, a computer program product is also provided. The computer program product includes computer instructions. When the computer instructions are executed on a device for processing broadband installation and maintenance work orders, the device for processing broadband installation and maintenance work orders executes each step of the method for processing broadband installation and maintenance work orders in the method flow shown in the above method embodiment.

[0190] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using a software program, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loading and executing computer execution instructions on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, data center, etc. that contains one or more servers that can be integrated with a medium. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state disk (SSD)).

[0191] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0192] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for processing a broadband installation and maintenance work order, characterized in that: The method comprises: Acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of the sample information includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes an actual processing time and a fault level of the broadband installation and maintenance work order; The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model; The initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined according to information gain and business criticality indicators, the business criticality indicators are determined according to the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined according to the learning rate of the model, the prediction error and the business criticality indicators, the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model; The target decision model is used to predict in real time the processing time and fault level of the currently received broadband installation and maintenance work order, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

2. The method according to claim 1, characterized in that: Before obtaining the training sample set, the method further includes: Acquire multiple important factors that affect the processing time and fault level of the work order, wherein the important factors include the work order type, the customer level, and the fault type; Obtaining a weight coefficient of each of the important factors, and constructing the business key indicator according to the important factors and the weight coefficients; The key business indicators are: K i =w1·Type i +w2·Level i +w3·Fault i Among them, K i Indicates business key indicators, Type i Is the work order type, Level i It is the customer level, Fault i is the fault type, w1, w2, w3 are the corresponding weight coefficients.

3. The method according to claim 2, characterized in that After constructing the business key indicators, the method further includes: Constructing the decision model splitting mechanism based on the key business indicators and information gain; The splitting mechanism is: SC=α·Gain(S,F)-β·K i Among them, SC represents the splitting mechanism, S is the sample set in the node, F is the feature considered for splitting, Gain is the information gain, which indicates the improvement of data purity after splitting, and K i is a key business indicator, and α and β are adjustment parameters in the model.

4. The method according to claim 3, characterized in that After constructing the business key indicators, the method further includes: Constructing the adaptive weight adjustment mechanism based on the learning rate, prediction error and business key indicators of the model; The adaptive weight adjustment mechanism is: w new =w old +γ·Adjust(S,Error,K i ) Among them, w new is the sample adjusted weight, w old is the weight of the sample before adjustment, γ is the learning rate, and Adjust is the adjustment value; The calculation formula of the adjustment value is: Among them, S represents the sample set, |S| represents the size of the sample set, and Error i represents the prediction error of the i-th sample, K i It represents the business key indicator of the i-th sample, and max(Error) represents the maximum prediction error in the sample set.

5. The method according to claim 4, characterized in that After constructing the business key indicators, the method further includes: Construct the adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is: IAF(x)=δ(K i )·γ(H(x)) δ(K i )=1+ρ·tanh(K i ) Among them, IAF(x) represents the comprehensive adjustment mechanism, δ(K i ) is the business criticality index adjustment factor, γ(H(x)) is the historical performance adjustment factor of the decision module, ρ is the adjustment factor preset in the decision model, which is used to control the influence of the business criticality index on the learning rate of the decision model, E(x) is the number of errors of sample x, μ and σ are the mean and standard deviation of the number of errors respectively, and θ is the adjustment factor; For each sample, the learning rate used is η(x) = η0·IAF(x), where η0 is the preset initial learning rate; For each erroneous sample, the weight in the loss function of the decision model is adjusted to w(x)=IAF(x).

6. The method according to claim 5, characterized in that The method of training a preset initial decision model by using the training sample set until the prediction error of the decision model meets a preset condition to obtain a target decision model includes: The preset initial decision model is trained multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model; The current training process in the multiple trainings is as follows: The prediction error, learning rate and business key indicators of each sample in the previous training process, as well as the maximum prediction error in the training sample set in the previous training process, are input into the adaptive weight adjustment mechanism to obtain the current weight of each sample in the current training process; Input the adjustment factor of the decision module in the previous training process, the number of errors of each sample, the preset initial learning rate of each sample, the average and standard deviation of the number of sample errors in the training sample set into the adaptive comprehensive adjustment mechanism to obtain the current learning rate of each sample in the current training process and the current weight of each error sample in the loss function of the decision model; Obtaining a current sample set according to the current weight of each sample in the current training process, the current learning rate, and the current weight of each erroneous sample in the loss function of the decision model; Perform decision splitting on the current sample set based on the decision model splitting mechanism to obtain a prediction error of the current training process.

7. The method according to claim 1, characterized in that Generating a corresponding processing strategy according to the predicted processing time and predicted fault level includes: Generate a corresponding work order emergency level according to the predicted processing time and the predicted fault level; Corresponding installation and maintenance personnel and processing time are assigned according to the urgency level of the work order.

8. A device for processing broadband installation and maintenance work orders, characterized in that: The device comprises: An acquisition module is used to acquire a training sample set, wherein the training sample set includes a plurality of sample information, each of which includes work order feature information of a broadband installation and maintenance work order and label information of the broadband installation and maintenance work order, wherein the label information includes the actual processing time and fault level of the broadband installation and maintenance work order; A training module, used to train the preset initial decision model multiple times using the training sample set until the prediction error of the decision model meets the preset conditions, thereby obtaining a target decision model; The initial decision model is a decision model based on a gradient boosting tree, the splitting mechanism in the initial decision model is determined according to information gain and business criticality indicators, the business criticality indicators are determined according to the work order type, customer level and fault type, the initial decision model includes an adaptive weight adjustment mechanism, the adaptive weight adjustment mechanism is used to adjust the weight of sample information, the adaptive weight adjustment mechanism is determined according to the learning rate of the model, the prediction error and the business criticality indicators, the initial decision model also includes an adaptive comprehensive adjustment mechanism, the adaptive comprehensive adjustment mechanism is used to adjust the learning strategy of the decision model; The prediction module is used to use the target decision model to predict the processing time and fault level of the currently received broadband installation and maintenance work order in real time, obtain the predicted processing time and predicted fault level, and generate a corresponding processing strategy based on the predicted processing time and predicted fault level.

9. An electronic device, characterized in that: It includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, it implements the method for processing the broadband installation and maintenance work order according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for processing a broadband installation and maintenance work order as described in any one of claims 1 to 7 is implemented.

Images