User pain point service event grading processing method and system

Abstract

The invention discloses a user pain point service event grading processing method and system, relates to the technical field of power supply business management, and solves the problems that user emotion is difficult to perceive and response is not timely in the prior art. According to the invention, by analyzing and calculating the emotional intensity value of the complaint work order and the user score indicated by the electrical load data, the urgency and importance of the user appeal can be accurately described; and the direction of the disposal scheme is indicated according to the equipment fault probability calculated by combining the equipment operation parameters, and the three forms an event feature vector to match similar historical cases and generate the disposal scheme, so that the disposal scheme corresponding to the urgent degree represented by the emotion of the user can be obtained, the response speed is improved, and the user satisfaction is improved.

Description

Technical Field

[0001] This invention relates to the field of power supply service management technology, and in particular to a method and system for graded handling of user pain point service events. Background Technology

[0002] Against the backdrop of smart grid construction, power supply companies face significant challenges in the refined management of user service needs. According to the 2023 Public Service Quality Monitoring Report of the National Energy Administration, 72.3% of user complaints in the power supply sector are concentrated on three major pain points: delayed response (average processing time greater than 48 hours), repeated handling (more than 3 complaints about the same issue), and inaccurate handling plans.

[0003] Existing methods rely on simple classification based on structured fields (such as complaint type codes), failing to perceive the intensity of user emotions in text / voice messages. Highly sensitive events may be downgraded due to incorrect labeling, leading to response timeouts.

[0004] The industry has attempted to introduce a basic work order management system, but its rule engine only supports classification based on preset conditions and cannot perceive users' emotions and specific demands. As users become more aware of their rights, there is an urgent need to build an intelligent handling system that integrates emotion recognition, experience reuse, and full process control.

[0005] Therefore, a hierarchical handling method and system for user pain point service incidents is needed. Summary of the Invention

[0006] To address the shortcomings of existing technologies in sensing user emotions and responding slowly, this invention provides a method and system for tiered handling of user pain point service incidents, which can better sense user emotions and improve response speed and user satisfaction. The specific technical solution is as follows:

[0007] In a first aspect, embodiments of this application provide a method for graded handling of user pain point service incidents, including:

[0008] Obtain the electricity load data and complaint work order of the complaining user, as well as the equipment operation parameters of the equipment corresponding to the complaint work order; calculate the user score of the complaining user based on the electricity load data; calculate the equipment failure probability of the equipment based on the equipment operation parameters; calculate the emotional intensity value of the complaint work order based on the work order text and / or work order voice; generate an event feature vector based on the emotional intensity value, the user score, and the equipment failure probability; match similar historical cases in a preset case library according to the event feature vector; and generate a handling plan based on the similar historical cases.

[0009] Preferably, the electricity load data includes electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity bill amount; the calculation of the user score for the complaining user based on the electricity load data includes: weighted summation of the electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity bill amount to obtain the user score; the formula for calculating the user score includes:

[0010] Value = w R ·R+w F ·F+w M ·M

[0011] w R =0.5-0.1×P gdp ;

[0012] w M =0.2 + 0.15 × I s

[0013] Where R is the electricity freshness, F is the average monthly electricity usage frequency, M is the average monthly electricity cost, and w R w is the electricity freshness coefficient. F w is the frequency coefficient for electricity consumption. M P is the electricity billing factor. gdp I represents the national percentile of the GDP of the user's district / county. s Industry sensitivity coefficient; w F The electricity frequency coefficient is a preset value.

[0014] Preferably, the calculation of the device failure probability based on the device's operating parameters includes: extracting the spatiotemporal features of the device's operating parameters through a preset Long Short-Term Memory (LSTM) network model; and performing a weighted calculation based on the spatiotemporal features using an attention mechanism to obtain the device failure probability.

[0015] Preferably, the complaint work order includes the work order text and the work order audio; calculating the emotional intensity value of the complaint work order based on the work order text and / or work order audio includes: extracting the semantic feature vector of the work order text using a preset BERT model; the attention weights set for preset emotional keywords in the BERT model are higher than the attention weights for non-emotional keywords; performing Mel-spectrum transformation on the work order audio to obtain a Mel-spectrum image; performing convolution calculation on the Mel-spectrum image to obtain the acoustic feature vector of the work order audio; the weights assigned to acoustic features representing negative emotions in the convolution calculation are higher than the weights assigned to acoustic features representing positive emotions; and fusing the semantic feature vector and the acoustic feature vector to obtain the emotional intensity value.

[0016] Preferably, matching similar historical cases in a preset case library based on the event feature vector includes: comparing the elements of the event feature vector with preset threshold elements to obtain a comparison result; determining the handling level of the complaint work order based on the comparison result; and matching similar historical cases in the case library with response times corresponding to the handling level based on the handling level.

[0017] Preferably, after generating a disposal plan based on the similar historical case, the method further includes: writing the disposal plan and the execution record of the disposal plan into a blockchain for evidence storage.

[0018] Preferably, the operating parameters of the equipment include the equipment's physical parameters, environmental parameters, and topology-related parameters; the physical parameters include temperature, oil level, and current harmonic distortion rate; the environmental parameters include air temperature, humidity, and wind speed; and the topology-related parameters include upstream load rate and the number of downstream user complaints.

[0019] Secondly, embodiments of this application provide a user pain point service incident tiered handling system, applied to the method described in the first aspect, the system comprising:

[0020] The acquisition module is used to acquire the electricity load data and complaint work order of the complaining user, as well as the equipment operation parameters of the equipment corresponding to the complaint work order;

[0021] The calculation module is used to calculate the user score of the complaining user based on the electricity load data;

[0022] This calculation module is also used to calculate the probability of equipment failure based on the equipment's operating parameters;

[0023] The calculation module is also used to calculate the emotional intensity value of the complaint work order based on the work order text and / or work order voice.

[0024] The module is used to generate an event feature vector based on the emotion intensity value, the user rating, and the device failure probability;

[0025] The matching module is used to match similar historical cases in a preset case library based on the event feature vector;

[0026] The generation module is used to generate a treatment plan based on the similar historical case.

[0027] Thirdly, embodiments of this application provide a computing device, including: a memory for storing a program; and a processor for loading the program to execute the method as described in the first aspect.

[0028] Fourthly, embodiments of this application provide a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the method described in the first aspect.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows: by analyzing and calculating the emotional intensity value of the complaint work order and the user rating indicated by the power load data, the urgency and importance of the user's demands can be accurately described; and by combining the equipment failure probability calculated by the equipment operating parameters to indicate the direction of the handling plan, the three factors constitute an event feature vector to match similar historical cases and generate a handling plan, which can obtain a handling plan corresponding to the urgency of the user's emotional expression, improve the response speed, and thus improve user satisfaction. Attached Figure Description

[0030] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0031] Figure 1 A flowchart illustrating a method for tiered handling of user pain point service incidents provided in this application embodiment;

[0032] Figure 2 A schematic diagram of a user pain point service incident classification and handling system provided in this application embodiment;

[0033] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0036] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0037] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0038] To address the issues of difficulty in perceiving user emotions and delayed responses in traditional methods, this invention provides a hierarchical handling method and system for user pain point service events, which can better perceive user emotions and improve response speed and user satisfaction.

[0039] Please see Figure 1 , Figure 1 This application provides a flowchart illustrating a method for tiered handling of user pain point service incidents, applicable to computing devices. Figure 1 As shown, the method includes:

[0040] Step 101: The computing device obtains the electricity load data and complaint work order of the complaining user, as well as the equipment operation parameters of the equipment corresponding to the complaint work order.

[0041] The computing device can be a computing device or server within the power supply business management system. Specifically, it can be a personal computer or tablet, or other smart terminal, directly operated by the management or maintenance personnel of the power supply business management system. In the scenario of power supply business complaint management, this computing device can communicate with the electricity consumption information collection system, customer service system, and equipment detection system via wired or wireless means to obtain relevant information.

[0042] Specifically, after receiving a user's complaint information, the computing device can obtain the user's electricity load curve in real time from the electricity information collection system, retrieve the data record of the corresponding complaint work order from the API of the customer service system, and obtain the time-series data of the transformer temperature and voltage fluctuation of the corresponding area when the user is using electricity from the equipment detection system, as well as other equipment operating parameters.

[0043] Specifically, a complaint ticket includes at least one of two types of data: ticket text and / or ticket voice.

[0044] Preferably, the computing device can use a Kafka message queue as the data transmission hub to receive the aforementioned data. Specifically, the electricity information collection system pushes second-level updated electricity load data to the Kafka topic power_data in real time; the device detection system subscribes to device status change events through the Open Platform Communications Unified Architecture (OPC UA) protocol and writes time-series data such as temperature and voltage fluctuations to the Kafka topic device_status; the customer service system publishes the text description and voice recording (format: text UTF-8 encoded, voice MP3 compressed) of new work orders to the Kafka topic complaint_event in real time.

[0045] Preferably, the computing device can filter call tickets with voice durations of less than 3 seconds, eliminating misdialed calls. This can reduce invalid data and improve data availability.

[0046] Preferably, the computing device can predict the probability of failure when the real-time pushed equipment operating parameters fluctuate significantly. When the predicted failure probability exceeds a preset threshold, it directly triggers a maintenance work order in the power supply business management system, thus mitigating the crisis in advance. In this case, the computing device can filter out equipment operating parameter changes with voltage fluctuations of less than 2%, ignoring normal fluctuations.

[0047] After receiving a user's complaint, the computing device can merge data packets from different systems based on the user's unified ID (such as the electricity user number) and execute subsequent steps.

[0048] Specifically, the computing device can associate the user's electricity meter ID with the corresponding power supply transformer ID, and then associate it with the upstream line ID; or it can be based on the spatial matching of the user's power supply location GIS coordinates and the device's coverage area.

[0049] Step 102: The calculation device calculates the user score for the complaining user based on the electricity load data.

[0050] The user profiling engine deployed in the computing device can calculate a user score that indicates user value. Specifically, the user profiling engine can use the RFM model to calculate this user score. The RFM model is a customer segmentation tool used to analyze user electricity consumption behavior and value. Its core is to classify electricity users based on three dimensions, thereby providing data support for differentiated services, precision marketing, load management, etc. The three dimensions are Recency (i.e., electricity freshness), Average monthly electricity consumption frequency, and Monetary / Magnitude of electricity consumption. Therefore, the model is named the RFM model.

[0051] Preferably, the electricity load data includes electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity cost; the calculation device can perform a weighted sum of the electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity cost to obtain the user score; the formula for calculating the user score includes:

[0052]

[0053] Where R is the electricity freshness, F is the average monthly electricity usage frequency, M is the average monthly electricity cost, and w R w is the electricity freshness coefficient. F w is the frequency coefficient for electricity consumption. M P is the electricity billing factor. gdp I represents the national percentile of the GDP of the user's district / county. s Industry sensitivity coefficient; w F The electricity frequency coefficient is a preset value.

[0054] Specifically, GDP percentile refers to the region's ranking in the national economy as a percentage.

[0055] Where R = 1 / (current date - most recent electricity usage date + 1).

[0056] Step 103: Calculate the probability of equipment failure based on the operating parameters of the equipment.

[0057] The operating parameters of the equipment include the equipment's physical parameters, environmental parameters, and topology-related parameters. Preferably, the physical parameters include temperature, oil level, and current harmonic distortion rate; the environmental parameters include air temperature, humidity, and wind speed; and the topology-related parameters include upstream load rate and the number of downstream user complaints.

[0058] Preferably, when the physical parameters exceed a preset threshold, the computing device can directly determine the equipment fault. For example, if the transformer temperature exceeds 85°C, the computing device can either directly generate a maintenance work order or execute the maintenance plan after generating a complete solution.

[0059] Preferably, the computing device can extract the spatiotemporal features of the device's operating parameters through a preset Long Short-Term Memory (LSTM) network model; and obtain the device's failure probability by performing weighted calculation based on these spatiotemporal features through an attention mechanism.

[0060] The computing device can directly input the device state vector composed of the device operating parameters into the LSTM-ATTention model to obtain the device failure probability output by the LSTM-ATTention model.

[0061] Specifically, the calculation formulas for extracting spatiotemporal features include:

[0062] h t =LSTM(x t ,h t-1 );

[0063] Where, x t Let h be the device state vector at time t. t-1 h represents the hidden state of the LSTM. t This refers to the spatiotemporal characteristics of equipment operating parameters. The formula for calculating the probability of equipment failure includes:

[0064] α t =Softmax(W a ,h t );

[0065] P fault =σ(Σα) t ·h t );

[0066] Among them, W a Let α be the attention weight matrix. t Here, σ() represents the attention weights, and P is the sigmoid function. fault This represents the probability of failure.

[0067] Step 104: The calculation device calculates the emotional intensity value of the complaint work order based on the work order text and / or work order voice.

[0068] Preferably, the complaint work order includes the work order text and the work order audio; the computing device can extract the semantic feature vector of the work order text through a preset BERT model; perform Mel spectrogram transformation on the work order audio to obtain a Mel spectrogram; perform convolution calculation on the Mel spectrogram to obtain the acoustic feature vector of the work order audio; and perform fusion calculation on the semantic feature vector and the acoustic feature vector to obtain the emotion intensity value.

[0069] The pre-trained BERT model is derived from manually labeled work orders in the power supply business management system. For example, the BERT model could be a BERT-base-chinese model.

[0070] The BERT model first encodes the input work order text into a text feature vector, and then performs sentiment intensity regression through a fully connected layer to obtain continuous values ​​between 0 and 1 as the semantic feature vector of the work order text.

[0071] The BERT model, through the Transformer's self-attention mechanism, establishes associations between sentiment keywords and modifiers in text. The BERT model assigns high attention weights to certain pre-defined sentiment keywords; these high-attention-weighted words directly contribute to the sentiment intensity value, thus ensuring that the output semantic feature vector contains sentiment intensity information. The attention weights assigned to pre-defined sentiment keywords in the BERT model are higher than those assigned to non-sentiment keywords.

[0072] For example, emotional keywords include “unbearable,” “angry,” “disappointed,” “outrageous,” and “frustrating.”

[0073] Mel-Spectrum Conversion is a processing technique that converts the linear spectrum of an audio signal into a Mel spectrum. Its core lies in simulating the human auditory system's perception of sound frequencies; its essence is to resample the linear spectrum based on the Mel frequency scale.

[0074] Specifically, linear spectrum refers to the original representation of an audio signal in the frequency domain, with a linear frequency scale (such as a uniform distribution of 100Hz, 200Hz, 300Hz, etc.); Mel frequency scale refers to a non-linear frequency scale derived from human hearing experiments.

[0075] After Mel spectrum conversion, the computing device can obtain a two-dimensional matrix representation of the Mel spectrum, with the horizontal axis representing the time frame (frame shift is usually 10-50ms) and the vertical axis representing the Mel frequency bins (e.g., 80-128 frequency bands). The matrix values ​​represent the energy intensity at the corresponding time-frequency point (usually logarithmic and converted to decibels).

[0076] Mel spectrum more closely resembles the human auditory experience of emotion, allowing for the association of acoustic features with emotions. For example, pitch is reflected in speech data as a low-frequency energy distribution, and increased pitch indicates rising anger or anxiety. Speech rate is associated with urgency and can be identified through the temporal inter-frame change rate of speech data. Low-level convolutional kernels in convolutional computation can capture short-term features at the phoneme level (such as plosives indicating anger), while deep convolutional kernels can extract prosodic features, such as interrogative or accusatory intonation.

[0077] Convolutional computation assigns higher weights to acoustic features representing negative emotions than to those representing positive emotions. For example, features such as pitch increase trends, speech rate increase trends, and amplitude jitter increase trends in speech reflect a shift in the user's emotions toward negativity, and convolutional computation can assign higher weights to these features.

[0078] Therefore, the acoustic feature vectors calculated by computing devices are essentially high-dimensional nonlinear emotional projections of acoustic features.

[0079] After convolving the Mel spectrogram to obtain the normalized acoustic feature vector of the work order's speech, the computing device can calculate the sentiment intensity value by weighting the semantic and acoustic feature vectors. Specific calculation formulas may include:

[0080] S final =0.7*S text +0.3*S audio ;

[0081] Among them, S final It is the emotional intensity value, S text It is the normalized value of the semantic feature vector, S audio These are the normalized values ​​of the acoustic eigenvectors. The coefficients of 0.7 and 0.3 were calibrated experimentally.

[0082] This application embodiment achieves quantitative assessment of user emotional intensity at the engineering level by simulating the multimodal processing mechanism of human emotional cognition.

[0083] Understandably, if a complaint work order only contains one type of data—work order text or work order voice—the computing device can refer to the above scheme to calculate the emotional intensity value of that type of data, that is, to normalize the feature vector corresponding to the data that is not included to 0.

[0084] Understandably, for steps 102 to 104, the computing device can perform the computation in parallel, reducing the time spent on feature extraction and increasing the response speed.

[0085] Step 105: The device generates an event feature vector based on the emotion intensity value, the user rating, and the device failure probability.

[0086] Among them, computing devices can construct event feature vectors [emotional intensity value, user rating, device failure probability].

[0087] Step 106: The computing device matches similar historical cases in a preset case library based on the event feature vector.

[0088] The computing device can calculate the similarity between the event feature vector and the case event vector in the case library by calculating cosine similarity, Jaccard similarity or Pearson correlation coefficient, and then match similar historical cases that are similar to the current complaint work order.

[0089] The case library is linked through a knowledge graph, which can automatically associate similar cases (such as the "summer air conditioning load → transformer overload" pattern). After a complaint is processed, the scores of the handling solutions for similar cases in the case library can be updated based on user satisfaction to ensure the consistency of the optimal handling solution for similar problems.

[0090] Preferably, when the calculated similarity is greater than 85%, the computing device can directly call the relevant data of the corresponding similar historical cases and copy the treatment plan.

[0091] Preferably, the computing device can compare the elements of the event feature vector with the preset threshold elements to obtain a comparison result; determine the handling level of the complaint work order based on the comparison result; and match similar historical cases with response times corresponding to the handling level in the case library based on the handling level.

[0092] For each element in the event feature vector, the computing device can pre-set a threshold, which serves as an element of the threshold vector. Specifically, the computing device can preset an emotion intensity threshold α, a user value threshold β, and a fault probability threshold γ. For example, α = 0.7, β = 0.8, and γ = 0.6.

[0093] Understandably, since the number of cases in the case library exceeds 1 million, real-time traversal and matching would take a long time; therefore, computing devices can use processing levels as a pre-filter to narrow down the search scope.

[0094] The computing device can determine the handling level of a complaint ticket based on elements in one or more event feature vectors. For example, when the sentiment intensity value is greater than 0.7 and the user value is greater than 0.8, the handling level is high, and the computing device only matches "30-minute response" cases; when the sentiment intensity value is between 0.5 and 0.7, the handling level is medium, and the computing device matches "4-hour response" cases; when the sentiment intensity value is less than 0.5, the handling level is low, and the computing device matches "24-hour response" cases.

[0095] It is understandable that the handling level can also be called the handling channel, and using color to represent the channel level is more intuitive. For example, a high handling level corresponds to red data; a medium handling level corresponds to yellow; and a low handling level corresponds to blue.

[0096] Step 107: The computing device generates a disposal plan based on the similar historical case.

[0097] The computing device can directly replicate the optimal handling plan from similar historical cases, or it can adjust the plan based on the differences between the current complaint and similar historical cases. Understandably, the computing device can update the optimal handling plan for a given case based on user satisfaction ratings and the degree of reduction in device failure probability after implementing the handling plan.

[0098] After generating a disposal plan, the computing device can generate a disposal path and execute it sequentially according to the node steps of the disposal path. Specifically, the disposal plan includes the responsible department, processing time limit, and resource requirements for each stage step.

[0099] For example, the handling path could be as follows: if the emotional intensity value is greater than 0.7 and the user rating is greater than 0.8, the computing device activates the red channel: the customer service manager directly contacts the site engineer and the on-site engineer responds within 30 minutes; then, if the failure probability is greater than 0.6, the device pre-inspection work order is triggered and the maintenance team is notified simultaneously.

[0100] Preferably, after generating a disposal plan based on the similar historical case, the computing device can write the disposal plan and the execution record of the disposal plan into the blockchain for evidence storage in the form of a hash chain.

[0101] Each record includes a timestamp, work order number, and processing node. The computing device can detect the processing time of each node using the processing node and timestamp.

[0102] Preferably, the blockchain evidence storage adopts a dual-chain structure: the business chain stores the data of the working units, and the evidence chain stores the original audio / image files.

[0103] After each execution of a handling plan, the computing device can update the case library and threshold elements. Specifically, for the case library, it can establish correlations between the latest complaint causes discovered during the handling process and historical cases, and update the effectiveness weights of the handling plans. For example, it can establish a correlation between "voltage instability → transformer overload → summer air conditioning load".

[0104] The computing device can update the threshold elements on a monthly basis, based on the latest historical processing success rate. The calculation formula can be as follows:

[0105] γ new =γ old +η*(S actual -S pred );

[0106] Where η is the learning rate, S actual S represents the actual success rate. pred To predict the success rate.

[0107] In this embodiment, by analyzing and calculating the emotional intensity value of the complaint work order and the user rating indicated by the power load data, the urgency and importance of the user's demands can be accurately described. Then, by combining the equipment failure probability calculated by the equipment operating parameters, the direction of the handling plan can be indicated. These three factors constitute an event feature vector to match similar historical cases, and by referring to similar historical cases, a handling plan can be quickly generated. This can improve the response speed while sensing the user's emotions, thereby improving user satisfaction.

[0108] The method provided in the embodiments of this application has been described above. The system provided in the embodiments of this application will be described below.

[0109] Please see Figure 2 , Figure 2 This application provides a schematic diagram of the structure of a user pain point service incident hierarchical handling system, as shown in the embodiments of this application. Figure 2 As shown, the system 20 includes:

[0110] The acquisition module 201 is used to acquire the electricity load data and complaint work order of the complaining user, as well as the equipment operation parameters of the equipment corresponding to the complaint work order;

[0111] Calculation module 202 is used to calculate the user score of the complaining user based on the electricity load data;

[0112] The calculation module 202 is also used to calculate the probability of equipment failure of the equipment based on the operating parameters of the equipment;

[0113] The calculation module 202 is also used to calculate the emotional intensity value of the complaint work order based on the work order text and / or work order voice.

[0114] Module 203 is used to generate an event feature vector based on the emotion intensity value, the user rating, and the device failure probability;

[0115] The matching module 204 is used to match similar historical cases in a preset case library based on the event feature vector;

[0116] The generation module 205 is used to generate a disposal plan based on the similar historical case.

[0117] Preferably, the electricity load data includes electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity bill; the calculation module 202 is specifically used to perform a weighted summation of the electricity usage freshness, the average monthly electricity usage frequency, and the average monthly electricity bill to obtain the user rating; the formula for calculating the user rating includes:

[0118] Value = w R ·R+w F ·F+w M ·M

[0119] w R =0.5-0.1×P gdp ;

[0120] w M =0.2 + 0.15 × I s

[0121] Where R is the electricity freshness, F is the average monthly electricity usage frequency, M is the average monthly electricity cost, and w R w is the electricity freshness coefficient. F w is the frequency coefficient for electricity consumption. M P is the electricity billing factor. gdp I represents the national percentile of the GDP of the user's district / county. s Industry sensitivity coefficient; w F The electricity frequency coefficient is a preset value.

[0122] Preferably, the calculation module 202 is specifically used to extract the spatiotemporal features of the device's operating parameters through a preset long short-term memory (LSTM) network model; and to obtain the device's failure probability by performing weighted calculation based on the spatiotemporal features through an attention mechanism.

[0123] Preferably, the complaint work order includes the work order text and the work order audio; the calculation module 202 is specifically used to extract the semantic feature vector of the work order text through a preset BERT model; perform Mel spectrum transformation on the work order audio to obtain a Mel spectrogram; perform convolution calculation on the Mel spectrogram to obtain the acoustic feature vector of the work order audio; and perform fusion calculation on the semantic feature vector and the acoustic feature vector to obtain the emotion intensity value.

[0124] Preferably, the matching module 204 is specifically used to compare the elements of the event feature vector with the preset threshold elements to obtain a comparison result; determine the handling level of the complaint work order based on the comparison result; and match similar historical cases with response times corresponding to the handling level in the case library based on the handling level.

[0125] Preferably, the system also includes an evidence storage module 206, which is used to write the disposal plan and the execution record of the disposal plan into the blockchain for evidence storage.

[0126] Preferably, the operating parameters of the equipment include the equipment's physical parameters, environmental parameters, and topology-related parameters; the physical parameters include temperature, oil level, and current harmonic distortion rate; the environmental parameters include air temperature, humidity, and wind speed; and the topology-related parameters include upstream load rate and the number of downstream user complaints.

[0127] The user pain point service event classification and handling system provided in this application embodiment can be understood by referring to the relevant content in the foregoing method embodiment section, and will not be repeated here.

[0128] like Figure 3 As shown, Figure 3 This is a schematic diagram of a possible logical structure of a computing device provided in an embodiment of this application. The computing device 30 includes a processor 301, a communication interface 302, a memory 303, and a bus 304. The processor 301, the communication interface 302, and the memory 303 are interconnected via the bus 304. In an embodiment of this application, the processor 301 is used to control and manage the operation of the computing device 30. For example, the processor 301 is used to execute... Figure 1 The steps in the embodiments and / or other processes used in the techniques described herein. Communication interface 302 is used to support communication by computing device 30. Memory 303 is used to store program code and data of computing device 30.

[0129] The processor 301 can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, etc. The bus 304 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0130] In another embodiment of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the above-described... Figure 1 The method described in the embodiments.

[0131] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0132] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0133] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0134] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0135] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0136] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for tiered handling of user pain point service incidents, characterized in that, The method includes: Obtain the electricity load data and complaint work orders of the complaining users, as well as the equipment operating parameters of the equipment corresponding to the complaint work orders; Calculate the user rating of the complaining user based on the electricity load data; Calculate the probability of equipment failure based on the equipment's operating parameters; Based on the work order text and / or work order voice, calculate the emotional intensity value of the complaint work order; An event feature vector is generated based on the emotional intensity value, the user rating, and the device failure probability. Based on the event feature vector, similar historical cases are matched in a preset case library; A solution is generated based on the aforementioned similar historical cases.

2. The method according to claim 1, characterized in that, The electricity load data includes electricity usage freshness, average monthly electricity usage frequency, and average monthly electricity bill; the calculation of the user score for the complaining user based on the electricity load data includes: The user score is obtained by weighted summing of the electricity usage freshness, the average monthly electricity usage frequency, and the average monthly electricity cost; the formula for calculating the user score includes: Value=w R ·R+w F ·F+w M ·M In R =0.5-0.1×P gdp ; In M =0.2+0.15×I s Where R is the electricity freshness, F is the average monthly electricity usage frequency, M is the average monthly electricity cost, and w R w is the electricity freshness coefficient. F w is the frequency coefficient for electricity consumption. M P is the electricity billing factor. gdp I represents the national percentile of the GDP of the user's district / county. s Industry sensitivity coefficient; w F The electricity usage frequency coefficient is a preset value.

3. The method according to claim 1, characterized in that, The calculation of the equipment failure probability based on the equipment operating parameters includes: The spatiotemporal features of the device's operating parameters are extracted using a pre-defined Long Short-Term Memory (LSTM) network model. The device failure probability is obtained by weighted calculation based on the spatiotemporal features using an attention mechanism.

4. The method according to claim 1, characterized in that, The complaint ticket includes the ticket text and the ticket voice; calculating the emotional intensity value of the complaint ticket based on the ticket text and / or ticket voice includes: The semantic feature vector of the work order text is extracted using a preset BERT model; the attention weight set for preset sentiment keywords in the BERT model is higher than the attention weight for non-sentiment keywords. The work order speech is subjected to Mel spectrogram conversion to obtain a Mel spectrogram. The Mel spectrogram is convolved to obtain the acoustic feature vector of the work order speech; in the weight allocation of the convolution calculation, the acoustic features representing negative emotions are assigned a higher weight than the acoustic features representing positive emotions. The semantic feature vector and the acoustic feature vector are fused and calculated to obtain the emotion intensity value.

5. The method according to any one of claims 1-4, characterized in that, The step of matching similar historical cases in a preset case library based on the event feature vector includes: The elements of the event feature vector are compared with the preset threshold elements to obtain the comparison result; The handling level of the complaint order is determined based on the comparison results; Based on the handling level, similar historical cases with response times corresponding to the handling level are matched in the case library.

6. The method according to any one of claims 1-4, characterized in that, After generating a treatment plan based on the similar historical cases, the method further includes: The proposed solution and its execution record will be stored on a blockchain for evidence preservation.

7. The method according to claim 3, characterized in that, The equipment operating parameters include the equipment's physical parameters, environmental parameters, and topology-related parameters; the physical parameters include temperature, oil level, and current harmonic distortion rate; the environmental parameters include air temperature, humidity, and wind speed; and the topology-related parameters include upstream load rate and the number of downstream user complaints.

8. A user pain point service incident classification and handling system, characterized in that, The system, applied to the method of any one of claims 1-7, comprises: The acquisition module is used to acquire the electricity load data and complaint work orders of the complaining users, as well as the equipment operation parameters of the equipment corresponding to the complaint work orders; The calculation module is used to calculate the user score of the complaining user based on the electricity load data; The calculation module is also used to calculate the probability of equipment failure of the equipment based on the equipment operating parameters; The calculation module is also used to calculate the emotional intensity value of the complaint work order based on the work order text and / or work order voice. The module is used to generate an event feature vector based on the emotion intensity value, the user rating, and the device failure probability; The matching module is used to match similar historical cases in a preset case library based on the event feature vector; The generation module is used to generate a treatment plan based on the similar historical cases.

9. A computing device, characterized in that, include: Memory, used to store programs; A processor for loading the program to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method of any one of claims 1-7.

Images