Installation and maintenance personnel monitoring method and device, storage medium and product

Abstract

The invention relates to an installation and maintenance personnel monitoring method and device, a storage medium and a product. According to the method, multi-dimensional portrait data, based on service completion conditions and customer evaluation, of installation and maintenance personnel in a plurality of recent continuous statistical periods are obtained, a first sub-model of a target model is used for predicting portrait features of a next period, and an abnormal type label is judged through a second sub-model of the target model.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of intelligent data processing, and in particular to a field maintenance personnel monitoring method and device, a storage medium and a product. BACKGROUND

[0002] Field maintenance personnel are an important force in communication network construction, opening and maintenance. The basic information, basic ability, construction standardization, service timeliness, and the like of field maintenance personnel are key factors to ensure the efficiency and quality of business opening. However, the service level of field maintenance personnel varies greatly. How to efficiently and accurately identify low-quality field maintenance personnel becomes a technical problem to be solved. SUMMARY

[0003] To solve the above problems, the present disclosure provides a field maintenance personnel monitoring method, device, storage medium and product.

[0004] According to a first aspect of the present disclosure, a field maintenance personnel monitoring method is provided, the method comprising: obtaining multi-dimensional portrait indicator data of a target field maintenance personnel in a preset number of continuous historical statistical periods before a current time, the multi-dimensional portrait indicator data of each statistical period being determined based on service order completion and customer evaluation of the target field maintenance personnel in the corresponding statistical period; generating a target time sequence sample in order based on the multi-dimensional portrait indicator data corresponding to the preset number of historical statistical periods; inputting the target time sequence sample into a target model, to predict a multi-dimensional portrait feature corresponding to a next statistical period through a first sub-model in the target model, and to predict an abnormal class label corresponding to the next statistical period through a second sub-model in the target model, wherein the first sub-model of the target model is trained based on a plurality of time sequence samples, the time sequence sample is determined based on the multi-dimensional portrait indicator data corresponding to a plurality of historical statistical periods greater than the preset number in succession, and the second sub-model of the target model is trained based on the multi-dimensional portrait indicator data corresponding to a plurality of historical statistical periods and the corresponding abnormal class label; grading and warning the target field maintenance personnel according to the abnormal class label.

[0005] In an implementation manner, the multi-dimensional portrait indicator data comprises at least one of the following indicators: a pre-visit on-time rate, a device installation on-time rate, an average opening time, a dressing standard rate, a quality inspection pass rate, a service satisfaction degree, a revisit satisfaction degree, and a customer complaint frequency.

[0006] In an implementation manner, the generating a target time sequence sample in order based on the multi-dimensional portrait indicator data corresponding to the preset number of historical statistical periods comprises: The multi-dimensional portrait indicator data corresponding to the preset number of historical statistical periods are subjected to missing value processing and outlier processing. The data after the missing value processing and the outlier processing are subjected to feature coding and data standardization. The data after the data standardization are subjected to feature selection, and target time sequence samples are generated in sequence.

[0007] In an embodiment, the first sub-model comprises a multivariate time series prediction model based on a long short-term memory network, and the second sub-model comprises a multi-classification model; and the training method of the target model comprises: Multi-dimensional portrait indicator data of the installation and maintenance personnel in a plurality of historical statistical periods and corresponding abnormal category labels are obtained, and the multi-dimensional portrait indicator data of each statistical period is determined based on service order completion and customer evaluation of the installation and maintenance personnel in the corresponding statistical period. The first sub-model in the target model is trained based on multi-dimensional portrait indicator data corresponding to a preset number of historical statistical periods in time sequence; and The second sub-model is trained based on multi-dimensional portrait indicator data corresponding to each historical statistical period and corresponding abnormal category labels. The input features of the first sub-model are used as input features of the second sub-model, and the output result of the second sub-model is the output result of the target model.

[0008] In an embodiment, the method further comprises: A prediction accuracy of the target model is obtained, and the prediction accuracy represents a ratio of samples that are truly positive in samples that are predicted to be positive in a test set of the target model. A recall rate of the target model is obtained, and the recall rate represents a prediction accuracy of positive samples in the test set. An F value of the target model is obtained according to the prediction accuracy and the recall rate, and the F value represents a model quality of the target model.

[0009] In an embodiment, the grading early warning of the target installation and maintenance personnel according to the abnormal category labels comprises real-time early warning and / or periodic early warning. The real-time early warning comprises sending early warning notifications to the installation and maintenance personnel who are identified as abnormal on the same day after the target model is called every day. The periodic early warning comprises calling the target model at a fixed time every week, and sending early warning notifications to the installation and maintenance personnel who are identified as abnormal in the same week and / or their direct managers.

[0010] In an implementation, the early warning notification is sent in at least one of a PC message, an APP message, or a short message.

[0011] According to a second aspect of the embodiments of the present disclosure, a device for monitoring installation and maintenance personnel is provided, and the device comprises: an acquisition module configured to acquire multi-dimensional portrait index data of a target installation and maintenance personnel in a preset number of historical statistical periods before a current time, the multi-dimensional portrait index data of each statistical period being determined based on service order completion and customer evaluation of the target installation and maintenance personnel in the corresponding statistical period; a generation module configured to generate a target time sequence sample in sequence based on the multi-dimensional portrait index data of the preset number of historical statistical periods; a prediction module configured to input the target time sequence sample into a target model, to predict a multi-dimensional portrait feature corresponding to a next statistical period by a first sub-model in the target model, and to predict an abnormal category label corresponding to the next statistical period by a second sub-model in the target model, wherein the first sub-model of the target model is trained based on a plurality of time sequence samples, the time sequence samples being determined based on the multi-dimensional portrait index data corresponding to a plurality of historical statistical periods greater than the preset number in sequence, and the second sub-model of the target model is trained based on the multi-dimensional portrait index data corresponding to a plurality of historical statistical periods and corresponding abnormal category labels; an early warning module configured to perform hierarchical early warning on the target installation and maintenance personnel according to the abnormal category label.

[0012] According to a third aspect of the embodiments of the present disclosure, an electronic device is provided, and the electronic device comprises: a memory having a computer program stored thereon; a processor configured to execute the computer program in the memory to implement the steps of the method according to any one of the first aspect.

[0013] According to a fourth aspect of the embodiments of the present disclosure, a non-transitory computer-readable storage medium is provided, and the medium has a computer program stored thereon, the program being executed by a processor to implement the steps of the method according to any one of the first aspect.

[0014] According to a fifth aspect of the embodiments of the present disclosure, a computer program product is provided, and the computer program product comprises a computer program, the computer program being executed by a processor to implement the steps of the method according to any one of the first aspect.

[0015] The embodiments of the present disclosure can achieve the following beneficial effects: by acquiring multi-dimensional portrait data of the service completion and customer evaluation of the installation and maintenance personnel in recent continuous multiple statistical periods, using the first sub-model of the target model to predict the portrait features of the next period, and using the second sub-model of the target model to judge the abnormal type label, the installation and maintenance personnel can be graded and warned according to the predicted abnormal label, which can realize accurate evaluation of the service quality of the installation and maintenance personnel, eliminate low-quality installation and maintenance personnel, and further ensure the efficiency and quality of related business opening.

[0016] Other features and advantages of the present disclosure will be described in detail in the following detailed description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of this application and help to explain the illustrative embodiments of the present application and their descriptions. In the drawings: Figure 1 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0018] Figure 2 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0019] Figure 3 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0020] Figure 4 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0021] Figure 5 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0022] Figure 6 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure.

[0023] Figure 7 FIG. 1 is a flowchart of an installation and maintenance personnel monitoring method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0024] The specific embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present disclosure, and are not used to limit the present disclosure.

[0025] It should be understood that the term "includes" and its variations, as used herein, are meant to be open-ended, that is, "includes but is not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment." The term "another embodiment" means "at least one additional embodiment." The term "some embodiments" means "at least some embodiments." Related definitions are given throughout the detailed description.

[0026] It should be noted that the terms "first", "second", and the like used in the present disclosure are only used to distinguish different devices, modules or units, and are not intended to limit the order or interdependence of the functions performed by these devices, modules or units. The modification of "one" or "multiple" mentioned in the present disclosure is illustrative and not limiting, and those skilled in the art should understand that "one" or "multiple" should be understood as "one or more" unless otherwise explicitly stated in the context. In the description of the present disclosure, "multiple" means two or more than two, and other quantifiers are similar; "at least one", "one or more" or similar expressions mean any combination of these items, including any combination of single or multiple items.

[0027] Although the operations or steps are described in a specific order in the accompanying drawings in the embodiments of the present disclosure, it should not be understood as requiring the operations or steps to be performed in the specific order or serial order shown, or requiring all the operations or steps to be performed to obtain the desired results. In the embodiments of the present disclosure, the operations or steps can be performed in series; the operations or steps can be performed in parallel; or a part of the operations or steps can be performed.

[0028] The names of the messages or information exchanged between the multiple devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of the messages or information. It can be understood that before using the technical solutions disclosed in the embodiments of the present disclosure, the type of personal information involved in the present disclosure, the scope of use, the use scenario, etc. should be informed to the user and the authorization of the user should be obtained according to relevant laws and regulations.

[0029] Figure 1 is a flowchart of a method for monitoring installation and maintenance personnel provided by an embodiment of the present disclosure. As shown in Figure 1 The method for monitoring installation and maintenance personnel provided by the embodiment of the present disclosure can include the following steps: In step S10, the multi-dimensional portrait index data of the target installation and maintenance personnel in a preset number of historical statistical periods before the current time is obtained, and the multi-dimensional portrait index data of each statistical period is determined based on the service order completion and customer evaluation of the target installation and maintenance personnel in the corresponding statistical period.

[0030] In this step, multi-dimensional profile indicator data for a preset number of consecutive historical statistical periods prior to the current moment of the target installation and maintenance personnel is obtained. The multi-dimensional profile indicator data for each statistical period is determined based on the service order completion status and customer feedback of the target installation and maintenance personnel within the corresponding statistical period. For example, multi-dimensional profile indicator data for a preset number of consecutive historical statistical periods prior to the current moment of the target installation and maintenance personnel can be obtained. This data is derived from the personnel's service order completion status and customer feedback within the corresponding period, specifically covering indicators such as on-time appointment rate and on-time installation rate. The on-time appointment rate is calculated as the ratio of on-time work orders to the total number of work orders, and the number of customer complaints is calculated according to a rule of deducting 15 points per complaint. The statistical period can be hourly, daily, weekly, monthly, or other time periods; this disclosure does not impose any restrictions.

[0031] In step S20, target time-series samples are generated sequentially based on the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods.

[0032] In this step, target time-series samples are generated sequentially based on a preset number of historical statistical periods corresponding to multi-dimensional profile indicator data. For example, missing value processing and outlier processing can be performed on the preset number of historical statistical periods corresponding to multi-dimensional profile indicator data first, then feature encoding and data standardization can be performed on the data after missing value processing and outlier processing, then feature selection can be performed on the standardized data, and target time-series samples can be generated sequentially.

[0033] In step S30, the target time series sample is input into the target model so that the multi-dimensional profile features corresponding to the next statistical period can be predicted by the first sub-model in the target model, and the anomaly category label corresponding to the next statistical period can be predicted by the second sub-model in the target model.

[0034] The first sub-model of the target model is trained based on multiple sets of time-series samples, which are determined based on multi-dimensional profile indicator data corresponding to a continuous historical statistical period greater than a preset number. The second sub-model of the target model is trained based on multi-dimensional profile indicator data and corresponding anomaly category labels corresponding to multiple historical statistical periods.

[0035] In this step, the target time-series samples are input into the target model to predict the multi-dimensional profile features corresponding to the next statistical period through the first sub-model of the target model, and to predict the anomaly category labels corresponding to the next statistical period through the second sub-model of the target model. The first sub-model of the target model is trained based on multiple sets of time-series samples, which are determined based on multi-dimensional profile indicator data corresponding to a continuous historical statistical period greater than a preset number. The second sub-model of the target model is trained based on multi-dimensional profile indicator data and corresponding anomaly category labels corresponding to multiple historical statistical periods. For example, multi-dimensional profile indicator data and corresponding anomaly category labels for installation and maintenance personnel in multiple historical statistical periods can be obtained first. The multi-dimensional profile indicator data for each statistical period is determined based on the service order completion status and customer evaluations of installation and maintenance personnel within the corresponding statistical period. Then, the first sub-model of the target model is trained based on multi-dimensional profile indicator data corresponding to a preset number of continuous historical statistical periods. Finally, the second sub-model is trained based on the multi-dimensional profile indicator data and corresponding anomaly classification labels corresponding to each historical statistical period. The input features of the first sub-model are used as the input features of the second sub-model, and the output of the second sub-model is the output of the target model.

[0036] In step S40, the target installation and maintenance personnel are given a graded warning based on the anomaly category label.

[0037] In this step, tiered warnings are issued to target installation and maintenance personnel based on anomaly category labels. For example, tiered warnings may include real-time warnings and / or periodic warnings; real-time warnings include sending a warning notification to the installation and maintenance personnel identified as abnormal that day after the target model is invoked daily; periodic warnings include invoking the target model at a fixed time each week and sending a warning notification to the installation and maintenance personnel identified as abnormal that week, and / or their direct managers.

[0038] By acquiring multi-dimensional profile data of installation and maintenance personnel based on service completion status and customer evaluation over several consecutive statistical periods, the first sub-model of the target model is used to predict the profile characteristics for the next period, and the second sub-model of the target model is used to determine the anomaly type label. Based on the predicted anomaly label, the installation and maintenance personnel are given graded early warnings, which can accurately assess the service quality of installation and maintenance personnel, eliminate low-quality installation and maintenance personnel, and thus ensure the efficiency and quality of related business activation.

[0039] In one possible implementation, the multi-dimensional profile indicator data includes at least one of the following indicators: on-time appointment rate, on-time installation rate, average activation duration, dress code compliance rate, quality inspection pass rate, service satisfaction, follow-up satisfaction, and number of customer complaints.

[0040] For example, the on-time appointment rate = number of on-time appointment work orders / total number of work orders * 100; on-time appointment work orders are work orders whose activation check-in time is earlier than the scheduled appointment time.

[0041] On-time installation rate is calculated as the percentage of work orders whose completion date and scheduled on-site arrival date are both earlier than the required completion date; On-time installation rate = Number of on-time installation work orders / Total work orders * 100.

[0042] The average activation time requires statistical analysis of all activated work orders within the specified period. The activation time for completed work orders is calculated as the completion time minus the dispatch time, and the activation time for incomplete work orders is calculated as the current time minus the dispatch time. The average value is then taken, and scores are allocated based on a baseline value and a challenge value. The calculation method for the average activation time score is as follows: x < baseline: 60 points * (x / baseline); baseline <= x < challenge: 60 points * (x / baseline); x >= challenge: 90 points * (x / challenge), maximum score 100 points.

[0043] The dress code compliance rate = (Number of dress code compliance work orders / Total number of work orders) * 100. Among these, dress code compliance work orders are those whose safety helmets passed quality inspection among all work orders dispatched by the installation and maintenance APP within the statistical period.

[0044] Quality inspection pass rate = (Number of quality inspection pass-through work orders / Total number of work orders) * 100. Among them, the quality inspection pass-through work orders are those whose image quality inspection (PON device model, service tag / service card recognition) is qualified among all activation work orders dispatched by the installation and maintenance APP within the statistical period.

[0045] Service satisfaction = Number of satisfactory service orders / Total number of service orders * 100. Among these, satisfactory service orders are those whose dispatch time / arrival time at the installation and maintenance APP is within the statistical period and for which service satisfaction data is available.

[0046] Satisfaction rate during follow-up visits = (Number of work orders with satisfactory follow-up visits / Total number of work orders) * 100. Work orders with satisfactory follow-up visits are those whose dispatch time / arrival time at the installation / maintenance APP falls within the statistical period and for which satisfaction rate data is available.

[0047] The number of customer complaints is directly calculated based on the number of work orders with complaints within the specified period. This indicator uses a point deduction system: 15 points are deducted for each work order with a customer complaint, until all points are deducted.

[0048] By selecting these indicators, we can comprehensively cover the core aspects of installation and maintenance personnel's services, such as timeliness, standardization, efficiency, and customer feedback. This avoids the one-sidedness of evaluating a single indicator and provides comprehensive and accurate data support for subsequent model training and anomaly identification, thereby improving the accuracy of evaluating the service quality of installation and maintenance personnel.

[0049] Figure 2This is a flowchart illustrating a monitoring method for installation and maintenance personnel according to an embodiment of this disclosure. Figure 2 As shown, generating target time-series samples sequentially from multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods may include the following steps: In step S201, missing value processing and outlier processing are performed on the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods.

[0050] In this step, missing value processing and outlier processing are performed on the multi-dimensional profile indicator data corresponding to a preset number of historical statistical periods. For example, missing value processing may include the following: Remove features with missing values: If a variable has a high missing rate (greater than 80%), low coverage, and low importance, the variable can be directly deleted.

[0051] Mean imputation: Data attributes are divided into interval and non-interval types. If the missing values ​​are interval types, the missing values ​​are imputed using the average of the existing values ​​of that attribute; if the missing values ​​are non-interval types, the missing values ​​are imputed using the mode of that attribute (i.e., the value that occurs most frequently) according to the mode principle in statistics.

[0052] Maximum Likelihood (ML): Under the condition of random missing data, assuming that the model is correct for the complete sample, the unknown parameters can be estimated by the marginal distribution of the observed data.

[0053] Interpolation methods include random interpolation, multiple interpolation, Lagrange interpolation, and Newton interpolation.

[0054] For example, outlier processing may include the following: The absolute value difference median method: First, find the median Xmedian of all factors; second, obtain the absolute deviation Xi of each factor from the median. Xmedian; the third step is to obtain the median absolute deviation (MAD); finally, determine the parameter n to determine a reasonable range [Xmedian]. [n*MAD, Xmedian+n*MAD], and make adjustments for factor values ​​that exceed the reasonable range.

[0055] The 3σ standard deviation method: The standard deviation itself reflects the dispersion of a factor and is based on the factor's mean, Xmean. In outlier processing, Xmean ± nσ can be used to measure the distance between a factor and the mean.

[0056] Percentile method: Sort the factor values ​​in ascending order and discard factor values ​​that are ranked higher than 97.5% or lower than 2.5%.

[0057] Clustering: Using clustering algorithms, small clusters that are far from other clusters are discarded.

[0058] In step S202, the data after missing value processing and outlier processing are subjected to feature encoding and data standardization.

[0059] In this step, the data after missing value and outlier processing are subjected to feature encoding and data standardization. For example, feature encoding may include the following methods: Feature binarization / diversification: The process of feature binarization is to convert numerical attributes into Boolean attributes and set a threshold as the dividing point between attribute values ​​of 0 and 1.

[0060] One-hot encoding: The encoding uses an N-bit state register to encode N possible values. Each state is represented by an independent register, and only one bit is valid at any given time.

[0061] It is necessary to eliminate the influence of different attributes of the samples having different magnitudes: ① Differences in magnitude will cause attributes with larger magnitudes to dominate; ② Differences in magnitude will slow down the iterative convergence speed; ③ Algorithms that rely on sample distance are very sensitive to the magnitude of the data. After normalization, the optimization process has a smaller range, the optimization process becomes smoother, and it is easier to converge correctly to the optimal solution. For example, data standardization can include the following methods: Min-max normalization (normalization): normalizes the maximum value to 1, the minimum value to 0 / -1, and other values ​​to be distributed within this range. For each attribute, let minA and maxA be the minimum and maximum values ​​of attribute A, respectively. A raw value x of A is mapped to a value x' in the interval [0,1] through min-max normalization. The formula is: New data = (Original data - Minimum value) / (Maximum value - Minimum value).

[0062] z-score standardization (normalization): Generally, the mean is normalized to 0, and the variance is normalized to 1. It standardizes data based on the mean and standard deviation of the original data. The original value x of attribute A is standardized to x' using z-score. The z-score standardization method is suitable when the maximum and minimum values ​​of attribute A are unknown, or when there are outliers outside the range of values. New data = (Original data - Mean) / Standard deviation.

[0063] In step S203, feature selection is performed on the standardized data, and target time series samples are generated sequentially.

[0064] In this step, feature selection is performed on the standardized data, and target time-series samples are generated sequentially. For example, the process of selecting a relevant subset of features from a given feature set is called feature selection. Two main reasons for feature selection are: mitigating the curse of dimensionality and reducing the difficulty of the learning task. Feature selection must ensure that important features are not lost. Common dimensionality reduction methods include: Singular Value Decomposition (SVD), Principal Component Analysis (PCA), and Linear Discriminant Analysis (LDA).

[0065] This process can effectively improve data quality, reduce noise and redundant information interference, and ensure that the samples input into the model are accurate and efficient, laying the foundation for the accuracy of model predictions.

[0066] Figure 3 This is a flowchart illustrating a monitoring method for installation and maintenance personnel according to an embodiment of this disclosure. The first sub-model includes a multivariate time-series prediction model based on a long short-term memory network, and the second sub-model includes a multi-classification model. Figure 3 As shown, the training method for the target model may include the following steps: In step S301, the multi-dimensional profile indicator data and corresponding anomaly category labels of the installation and maintenance personnel in multiple historical statistical periods are obtained. The multi-dimensional profile indicator data of each statistical period are determined based on the service order completion status and customer evaluation of the installation and maintenance personnel in the corresponding statistical period.

[0067] In step S302, the first sub-model in the target model is trained based on multi-dimensional profile index data corresponding to a preset number of historical statistical periods with continuous time. In step S303, a second sub-model is trained based on the multi-dimensional profile indicator data and corresponding anomaly classification labels corresponding to each historical statistical period. The input features of the first sub-model are used as the input features of the second sub-model, and the output of the second sub-model is the output of the target model.

[0068] For example, the target model comprises a first sub-model (a multivariate temporal prediction model based on a long short-term memory network) and a second sub-model (a multi-classification model). The training process is as follows: First, multi-dimensional profile indicator data (from service order completion status and customer evaluations for the corresponding period) and corresponding anomaly category labels for installation and maintenance personnel are acquired over multiple historical statistical periods. Then, the first sub-model is trained: indicator data from a predetermined number of consecutive historical statistical periods are selected to construct training samples. The model learns the temporal dependence of the indicator data, enabling it to predict the multi-dimensional profile features for the next statistical period. During training, historical data samples are used, with 80% randomly ordered as training data and 20% as test data to optimize the model. Simultaneously, the second sub-model is trained: using indicator data from each historical statistical period and corresponding anomaly category labels as training data, and using the input features of the first sub-model as its own input features, a multi-classification algorithm (such as using a softmax function) is employed to train it to output the anomaly category labels for the next statistical period. Finally, the output of the second sub-model is used as the output of the target model.

[0069] The sample characteristics may include: on-time appointment rate, on-time installation rate, average activation time, dress code compliance rate, quality inspection pass rate, service satisfaction, follow-up satisfaction, and number of customer complaints. Anomaly category labels may include: untimely service, non-standard service, unsatisfactory service, unprofessional construction, and no anomalies. The hidden layers of the multivariate time-series prediction model based on Long Short-Term Memory networks can be up to three, and the statistical period can be hours, days, or other time periods; this disclosure does not impose any restrictions.

[0070] This training method enables the two sub-models to work together. The time-series features captured by the first sub-model can provide more accurate input for the second sub-model, improving the overall prediction and classification accuracy of the model. At the same time, relying on a large amount of historical data for training ensures that the model can adapt to the service data characteristics of different installation and maintenance personnel, enhancing the model's versatility and reliability.

[0071] Figure 4 This is a flowchart illustrating a monitoring method for installation and maintenance personnel according to an embodiment of this disclosure. Figure 4 As shown, the method may further include the following steps: In step S50, the prediction accuracy of the target model is obtained, whereby the prediction accuracy represents the ratio of samples that are actually positive among those predicted as positive in the test set of the target model.

[0072] In this step, the prediction accuracy of the target model is obtained. Prediction accuracy represents the ratio of truly positive samples among those predicted as positive in the test set by the target model. For example, the prediction accuracy of the target model can be obtained using the following formula: Prediction accuracy = True Positive Cases (TP) / (True Positive Cases (TP) + False Positive Cases (FP)) This metric is used to measure the accuracy of the model in predicting positive samples.

[0073] In step S60, the recall rate of the target model is obtained, whereby the recall rate characterizes the prediction accuracy of positive samples in the test set.

[0074] In this step, the recall rate of the target model is obtained. Recall rate represents the prediction accuracy of positive samples in the test set. For example, the recall rate of the target model can be obtained by the following formula: Recall = True Positive Instances (TP) / (True Positive Instances (TP) + False Negative Instances (FN)) This metric is used to measure the model's ability to cover positive samples.

[0075] In step S70, the F-value of the target model is obtained based on the prediction accuracy and the recall rate, wherein the F-value characterizes the model quality of the target model.

[0076] In this step, the F-score of the target model is obtained based on the prediction accuracy and recall rate. The F-score characterizes the model quality of the target model. For example, the F-score can be obtained using the following formula: F-score = (Precision × Recall × 2) / (Precision + Recall) This indicator is used to comprehensively reflect the overall quality of the model.

[0077] By evaluating model performance from different dimensions using these three indicators, we can avoid the bias of evaluation based on a single indicator, comprehensively grasp the accuracy and coverage of the model in anomaly identification, ensure the reliability of model performance, provide accurate model support for subsequent anomaly identification and early warning by installation and maintenance personnel, and reduce monitoring errors caused by model errors.

[0078] Figure 5 This is a flowchart illustrating a monitoring method for installation and maintenance personnel according to an embodiment of this disclosure. Figure 5 As shown, the tiered early warning system may include real-time early warning and / or periodic early warning. The step of issuing tiered early warnings to the target maintenance personnel based on the anomaly category label may include the following steps: In step S401, after the target model is invoked daily, a warning notification is sent to the installation and maintenance personnel identified as abnormal on that day.

[0079] In step S402, at a fixed time each week, the target model is invoked to send early warning notifications to the installation and maintenance personnel identified as abnormal during the week and / or their direct managers.

[0080] For example, tiered early warnings can be issued to target installation and maintenance personnel based on anomaly category tags. These include real-time and periodic warnings, which can be implemented individually or simultaneously. The real-time warning process involves: daily scheduled calls to the target model to analyze the daily data of installation and maintenance personnel; identifying personnel with anomalies; and immediately sending an early warning notification to them, ensuring they are aware of their issues immediately. The periodic warning process involves: weekly calls to the target model at a fixed time (e.g., Friday) to summarize and analyze the weekly data of installation and maintenance personnel; identifying personnel with anomalies; and sending early warning notifications to them and their direct managers (e.g., team leaders), allowing managers to simultaneously monitor personnel anomalies. By combining real-time and periodic warnings, timely reminders to personnel with anomalies can be issued for immediate rectification, while managers can periodically understand the team's situation. This effectively prevents installation and maintenance personnel from being blacklisted during monthly evaluations due to accumulated problems, reducing service interruptions and customer dissatisfaction caused by poor personnel management, and improving the efficiency of installation and maintenance service management and customer satisfaction.

[0081] In one possible implementation, the warning notification is sent via at least one of PC-based messages, APP-based messages, or SMS. For example, when sending warning notifications to installation and maintenance personnel and relevant management personnel, at least one of these methods is used. During the sending process, the system selects an appropriate notification method based on preset rules or user settings: if personnel frequently use PCs for office work, PC-based messages are sent first; if personnel are frequently on the field and frequently use the installation and maintenance APP, APP-based messages are sent first; to ensure no notification is missed, SMS messages can also be sent simultaneously, delivering the warning information directly to personnel's mobile phones via the operator's network. The combination of multiple notification methods can adapt to different work scenarios and usage habits of installation and maintenance personnel, greatly improving the delivery rate of warning notifications, avoiding personnel missing rectification reminders due to the failure of a single notification method, ensuring timely and effective transmission of warning information, buying time for subsequent problem rectification and service optimization, and ensuring closed-loop management of the monitoring process.

[0082] Figure 6 This is a block diagram of a monitoring device for installation and maintenance personnel provided in one embodiment of this disclosure. Figure 1 As shown in the figure, this disclosure provides a monitoring device 600 for installation and maintenance personnel, which may include the following modules: The acquisition module 610 is used to acquire multi-dimensional profile indicator data of the target installation and maintenance personnel for a preset number of historical statistical periods before the current time. The multi-dimensional profile indicator data of each statistical period is determined based on the service order completion status and customer evaluation of the target installation and maintenance personnel in the corresponding statistical period.

[0083] The generation module 620 is used to generate target time-series samples in sequence based on the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods.

[0084] The prediction module 630 is used to input the target time series sample into the target model, so as to predict the multi-dimensional profile features corresponding to the next statistical period through the first sub-model in the target model, and to predict the anomaly category label corresponding to the next statistical period through the second sub-model in the target model. The first sub-model of the target model is trained based on multiple sets of time series samples, and the time series samples are determined based on multi-dimensional profile indicator data corresponding to a continuous historical statistical period greater than a preset number. The second sub-model of the target model is trained based on multi-dimensional profile indicator data and corresponding anomaly category labels corresponding to multiple historical statistical periods.

[0085] The early warning module 640 is used to provide graded early warnings to the target installation and maintenance personnel based on the anomaly category label.

[0086] In one possible implementation, the multi-dimensional profile indicator data includes at least one of the following indicators: on-time appointment rate, on-time installation rate, average activation duration, dress code compliance rate, quality inspection pass rate, service satisfaction, follow-up satisfaction, and number of customer complaints.

[0087] In one possible implementation, the generation module 620 is further configured to: Missing values ​​and outliers are processed in the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods. The data after missing value processing and outlier processing are then subjected to feature encoding and data standardization. Feature selection is performed on the standardized data, and target time series samples are generated sequentially.

[0088] In one possible implementation, the first sub-model includes a multivariate temporal prediction model based on a long short-term memory network, and the second sub-model includes a multi-classification model; the prediction module 630 is further configured to: The multi-dimensional profile indicator data and corresponding anomaly category labels of installation and maintenance personnel are obtained in multiple historical statistical periods. The multi-dimensional profile indicator data of each statistical period are determined based on the service order completion status and customer evaluation of the installation and maintenance personnel in the corresponding statistical period. The first sub-model in the target model is trained based on multi-dimensional profile indicator data corresponding to a predetermined number of historical statistical periods with continuous time; and, The second sub-model is trained based on the multi-dimensional profile indicator data and corresponding anomaly classification labels for each historical statistical period. The input features of the first sub-model are used as the input features of the second sub-model, and the output of the second sub-model is the output of the target model.

[0089] In one possible implementation, the device 600 further includes an evaluation module for: Obtain the prediction accuracy of the target model, where the prediction accuracy represents the ratio of samples that are actually positive among those predicted as positive in the test set of the target model. Obtain the recall rate of the target model, where the recall rate characterizes the prediction accuracy of positive samples in the test set; Based on the prediction accuracy and the recall rate, the F-value of the target model is obtained, and the F-value characterizes the model quality of the target model.

[0090] In one possible implementation, the tiered early warning includes: real-time early warning and / or periodic early warning; the early warning module 640 is further configured to: After the target model is invoked daily, a warning notification is sent to the installation and maintenance personnel who are identified as abnormal on that day. At a fixed time each week, the target model is invoked to send early warning notifications to the installation and maintenance personnel identified as abnormal during that week, and / or their direct managers.

[0091] In one possible implementation, the warning notification is sent via at least one of PC-based messages, APP-based messages, or SMS messages.

[0092] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0093] This embodiment of the disclosure obtains multi-dimensional profile data of installation and maintenance personnel based on service completion status and customer evaluation for multiple consecutive statistical periods in recent years. It uses the first sub-model of the target model to predict the profile features for the next period and uses the second sub-model of the target model to determine the anomaly type label. Based on the predicted anomaly label, it implements graded early warning for installation and maintenance personnel, which can accurately evaluate the service quality of installation and maintenance personnel, eliminate low-quality installation and maintenance personnel, and thus ensure the efficiency and quality of related business activation.

[0094] Figure 7 This is a block diagram illustrating an electronic device according to an exemplary embodiment. Figure 7 As shown, the electronic device 700 may include a processor 701 and a memory 702. The electronic device 700 may also include one or more of a multimedia component 703, an input / output (I / O) interface 704, and a communication component 705.

[0095] The processor 701 controls the overall operation of the electronic device 700 to complete all or part of the steps in the aforementioned monitoring method for installation and maintenance personnel. The memory 702 stores various types of data to support the operation of the electronic device 700. This data may include, for example, instructions for any application or method operating on the electronic device 700, and application-related data such as contact data, sent and received messages, pictures, audio, video, etc. The memory 702 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 703 may include a screen and audio components. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in memory 702 or transmitted via communication component 705. The audio component also includes at least one speaker for outputting audio signals. I / O interface 704 provides an interface between processor 701 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual or physical buttons. Communication component 705 is used for wired or wireless communication between the electronic device 700 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IoT, eMTC, or other 5G technologies, or combinations thereof, is not limited here. Therefore, the corresponding communication component 705 may include: a Wi-Fi module, a Bluetooth module, an NFC module, etc.

[0096] In an exemplary embodiment, the electronic device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described monitoring method for installation and maintenance personnel.

[0097] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the above-described installation and maintenance personnel monitoring method. For example, the computer-readable storage medium may be the memory 702 including program instructions, which may be executed by the processor 701 of the electronic device 700 to complete the above-described installation and maintenance personnel monitoring method.

[0098] In another exemplary embodiment, a computer program product is also provided, which includes a computer program executable by a programmable device, the computer program having a code portion for performing the above-described installation and maintenance personnel monitoring method when executed by the programmable device.

[0099] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.

[0100] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0101] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.

Claims

1. A method for monitoring installation and maintenance personnel, characterized in that, The method includes: Obtain multi-dimensional profile indicator data of the target installation and maintenance personnel for a preset number of consecutive historical statistical periods before the current moment. The multi-dimensional profile indicator data of each statistical period is determined based on the service order completion status and customer evaluation of the target installation and maintenance personnel in the corresponding statistical period. Target time-series samples are generated sequentially based on the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods. The target time series sample is input into the target model to predict the multi-dimensional profile features corresponding to the next statistical period through the first sub-model of the target model, and to predict the anomaly category label corresponding to the next statistical period through the second sub-model of the target model. The first sub-model of the target model is trained based on multiple sets of time series samples, and the time series samples are determined based on multi-dimensional profile indicator data corresponding to a continuous historical statistical period greater than a preset number. The second sub-model of the target model is trained based on multi-dimensional profile indicator data and corresponding anomaly category labels corresponding to multiple historical statistical periods. The target installation and maintenance personnel are given tiered warnings based on the anomaly category labels.

2. The method according to claim 1, characterized in that, The multi-dimensional profile indicator data includes at least one of the following indicators: on-time appointment rate, on-time installation rate, average activation time, dress code compliance rate, quality inspection pass rate, service satisfaction, follow-up satisfaction, and number of customer complaints.

3. The method according to claim 1, characterized in that, The process of generating target time-series samples based on the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods includes: Missing values ​​and outliers are processed in the multi-dimensional profile indicator data corresponding to the preset number of historical statistical periods. The data after missing value processing and outlier processing are then subjected to feature encoding and data standardization. Feature selection is performed on the standardized data, and target time series samples are generated sequentially.

4. The method according to claim 1, characterized in that, The first sub-model includes a multivariate temporal prediction model based on a long short-term memory network, and the second sub-model includes a multi-classification model; the training method for the target model includes: The multi-dimensional profile indicator data and corresponding anomaly category labels of installation and maintenance personnel are obtained in multiple historical statistical periods. The multi-dimensional profile indicator data of each statistical period are determined based on the service order completion status and customer evaluation of the installation and maintenance personnel in the corresponding statistical period. The first sub-model in the target model is trained based on multi-dimensional profile indicator data corresponding to a predetermined number of historical statistical periods with continuous time; and, The second sub-model is trained based on the multi-dimensional profile indicator data and corresponding anomaly classification labels for each historical statistical period. The input features of the first sub-model are used as the input features of the second sub-model, and the output of the second sub-model is the output of the target model.

5. The method according to claim 4, characterized in that, The method further includes: Obtain the prediction accuracy of the target model, where the prediction accuracy represents the ratio of samples that are actually positive among those predicted as positive in the test set of the target model. Obtain the recall rate of the target model, where the recall rate characterizes the prediction accuracy of positive samples in the test set; Based on the prediction accuracy and the recall rate, the F-value of the target model is obtained, and the F-value characterizes the model quality of the target model.

6. The method according to claim 1, characterized in that, The step of classifying and issuing early warnings to the target installation and maintenance personnel based on the anomaly category label includes: real-time early warning and / or periodic early warning; The real-time early warning includes: after calling the target model daily, sending an early warning notification to the installation and maintenance personnel identified as abnormal on that day; The periodic early warning includes: at a fixed time each week, calling the target model to send early warning notifications to the installation and maintenance personnel identified as abnormal in that week and / or their direct managers.

7. The method according to claim 6, characterized in that, The warning notification is sent via at least one of the following methods: PC message, APP message, or SMS.

8. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the method according to any one of claims 1-7.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

Images