Early warning method, device and equipment for service index of network intelligent terminal and medium

Abstract

The embodiment of the invention discloses an early warning method, device and equipment for a service index of a network intelligent terminal and a medium, and relates to the technical field of computers. The method comprises the following steps: acquiring website attribute data, intelligent terminal service data and website meteorological data; inputting the network attribute data, the intelligent terminal service data and the network meteorological data into an early warning threshold prediction model to obtain a prediction early warning threshold of a target service index; and determining an early warning result according to the predicted early warning threshold value of the target business index and the actual business data of the target business index. According to the technical scheme, the website attribute data, the intelligent terminal service data and the website meteorological data are processed through the early warning threshold prediction model, and the dynamic early warning threshold can be predicted, so that the early warning accuracy can be greatly improved when early warning is performed according to the early warning threshold.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and medium for early warning of business indicators of a network intelligent terminal. Background Art

[0002] Bank branches are often equipped with smart terminals, allowing customers to conduct self-service transactions and query information. Existing monitoring systems for smart terminals at bank branches are primarily based on fixed threshold monitoring and early warning systems. This involves presetting one or more fixed thresholds. When a monitoring indicator exceeds or falls below a set threshold, the system triggers an alert.

[0003] However, current threshold settings often rely on experience and historical data, making it difficult to ensure their rationality. If thresholds are set too loosely, they can lead to sluggish monitoring system responses and even missed alerts. If they are set too strictly, they can trigger excessive false alarms, causing unnecessary trouble for operations and maintenance personnel. Furthermore, fixed thresholds are unsuitable for complex and ever-changing production scenarios, resulting in low early warning accuracy. Summary of the Invention

[0004] The present invention provides a method, device, equipment and medium for early warning of business indicators of a network intelligent terminal, which can determine dynamic early warning thresholds adapted to various application scenarios and improve the accuracy of early warnings.

[0005] According to one aspect of the present invention, a method for early warning of service indicators of a network intelligent terminal is provided, the method comprising:

[0006] Obtain network attribute data, smart terminal business data, and network weather data;

[0007] Inputting the network point attribute data, smart terminal business data and network point weather data into the warning threshold prediction model to obtain the prediction warning threshold of the target business indicator;

[0008] The warning result is determined based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator.

[0009] According to another aspect of the present invention, there is provided an early warning device for service indicators of a network point intelligent terminal, comprising:

[0010] Data acquisition module, used to obtain network attribute data, smart terminal business data and network weather data;

[0011] A prediction and warning threshold determination module is used to input the network point attribute data, smart terminal business data and network point meteorological data into the warning threshold prediction model to obtain the prediction and warning threshold of the target business indicator;

[0012] The early warning result determination module is used to determine the early warning result based on the predicted early warning threshold of the target business indicator and the actual business data of the target business indicator.

[0013] According to another aspect of the present invention, an electronic device is provided, comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the early warning method for business indicators of the network intelligent terminal according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the early warning method for business indicators of a network intelligent terminal according to any embodiment of the present invention when executed.

[0018] The technical solution of the embodiments of the present application includes: obtaining network attribute data, smart terminal service data, and network meteorological data; inputting the network attribute data, smart terminal service data, and network meteorological data into a warning threshold prediction model to obtain a predicted warning threshold for a target business indicator; and determining a warning result based on the predicted warning threshold for the target business indicator and the actual business data of the target business indicator. This technical solution processes the network attribute data, smart terminal service data, and network meteorological data through the warning threshold prediction model to predict a dynamic warning threshold, thereby greatly improving the accuracy of warnings when issuing warnings based on the warning threshold.

[0019] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1This is a flow chart of a method for early warning of business indicators of a network intelligent terminal provided in accordance with the first embodiment of the present application;

[0022] Figure 2 This is a flow chart of a method for early warning of business indicators of a network intelligent terminal provided in accordance with the second embodiment of the present application;

[0023] Figure 3 This is a schematic diagram of the structure of a warning threshold prediction model provided according to Example 2 of the present application;

[0024] Figure 4 This is a structural diagram of a feature selection network provided according to Example 2 of the present application;

[0025] Figure 5 This is a structural diagram of a warning device for service indicators of a network intelligent terminal provided in accordance with the third embodiment of the present application;

[0026] Figure 6 It is a structural diagram of an electronic device for implementing a method for early warning of business indicators of a network intelligent terminal according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", "target", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Example 1

[0030] Figure 1A flowchart of a method for early warning of business indicators of a branch intelligent terminal is provided for the first embodiment of the present application. The embodiment of the present application can be applied to the case of determining the early warning threshold of business data indicators of the intelligent terminal of a bank branch. The method can be executed by an early warning device for business indicators of the branch intelligent terminal. The early warning device for business indicators of the branch intelligent terminal can be implemented in the form of hardware and / or software. The early warning device for business indicators of the branch intelligent terminal can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0031] S110, obtaining network point attribute data, smart terminal service data, and network point weather data.

[0032] A branch refers to a bank branch. Within a branch, one or more smart terminals are typically located. These terminals are used to handle customer services, including but not limited to bank card processing, balance inquiries, and mobile phone number changes. Branch attribute data is associated with the prediction and warning thresholds for target business indicators. For example, branch attribute data includes but is not limited to the branch's location, the number of smart terminals it employs, and the branch's hierarchy. The closer a branch is to a customer-heavy area, the greater the business volume (using business volume as a target business indicator), and the higher the corresponding warning threshold. Smart terminal business data includes but is not limited to device operating parameters, real-time transaction volume, and application response time. Smart terminal business data is associated with the prediction and warning thresholds for target business indicators. For example, if the smart terminal business data indicates that the number of users operating the device on a given day is 100, the business volume might be 50. As the number of users operating the device increases, business volume generally increases as well. Branch meteorological data includes but is not limited to: maximum temperature, minimum temperature, current temperature, weather, wind speed, wind direction, air quality index, etc.; similarly, branch meteorological data is associated with the prediction and warning thresholds of target business indicators. Taking business volume as an example of the target business indicator, if the branch meteorological data shows rainy days, the number of customers handling business will decrease, and accordingly, the warning threshold of business volume needs to be lowered.

[0033] Specifically, since network attribute data is usually static data, that is, the data content is usually unchanged in a short period of time, network attribute data can be obtained in advance. Since smart terminal business data and network weather data change over time, real-time smart terminal business data and network weather data can be obtained.

[0034] S120: Input the network point attribute data, smart terminal business data, and network point weather data into a warning threshold prediction model to obtain a prediction warning threshold of a target business indicator.

[0035] Among them, the target business indicator can be an indicator of the application (APP) on the smart terminal. The target business indicator can be set according to the actual situation. For example, the target business indicator is transaction volume, total transaction amount or query volume on the day, etc. The warning threshold prediction model is used to predict the warning threshold of the target business indicator. If the target business indicator is business volume, the previous technical solution usually uses a fixed threshold to make a warning judgment. For example, if the threshold is 70, then when the business volume is less than 70, an early warning will be issued to remind relevant personnel whether there are problems such as equipment failure. The embodiment of the present application predicts the warning threshold through the early warning threshold prediction model, so that a dynamic prediction early warning threshold can be obtained, that is, the prediction early warning threshold of the embodiment of the present application changes with the change of the input data of the model.

[0036] For example, the branch attribute data, smart terminal service data, and branch weather data are input into the warning threshold prediction model. If the branch weather data indicates heavy rain, the resulting transaction volume prediction warning threshold may be 30. If the branch weather data indicates sunny weather, the resulting transaction volume prediction warning threshold may be 70. If the application response time is much longer than usual, the smart terminal may experience lag, and some users may cancel transactions. In this case, the resulting transaction volume prediction warning threshold may be 60.

[0037] S130: Determine a warning result based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator.

[0038] Specifically, after obtaining the prediction warning threshold, the warning threshold of the target business indicator can be set as the prediction warning threshold at the time corresponding to the prediction warning threshold, thereby determining the warning result based on the actual business data and the prediction warning threshold.

[0039] For example, if the predicted warning threshold is 50 transactions on January 1, then on January 1, actual business data is obtained. If the actual business data shows a transaction volume of 55, no warning is issued. If the actual business data shows a transaction volume of 45, it is determined that there may be a problem with the transaction volume on that day, the warning result is determined to be necessary, and the relevant data is sent to the warning processing party.

[0040] The technical solution of the embodiments of the present application includes: obtaining network attribute data, smart terminal service data, and network meteorological data; inputting the network attribute data, smart terminal service data, and network meteorological data into a warning threshold prediction model to obtain a predicted warning threshold for a target business indicator; and determining a warning result based on the predicted warning threshold for the target business indicator and the actual business data of the target business indicator. This technical solution processes the network attribute data, smart terminal service data, and network meteorological data through the warning threshold prediction model to predict a dynamic warning threshold, thereby greatly improving the accuracy of warnings when issuing warnings based on the warning threshold.

[0041] Example 2

[0042] Figure 2 This is a flow chart of a method for early warning of business indicators of a network intelligent terminal provided in the second embodiment of the present application. The embodiment of the present application is optimized based on the above embodiment.

[0043] like Figure 2 As shown, the method of the embodiment of the present application specifically includes the following steps:

[0044] S210, obtaining network point attribute data, smart terminal business data, and network point weather data.

[0045] After obtaining the network attribute data, smart terminal business data, and network weather data, outliers are removed and normalized to make the data at the same level, eliminating the impact of different data dimensions on subsequent model training and avoiding unnecessary numerical problems. The transformation function for data normalization is:

[0046]

[0047] Where, X max Indicates the maximum value of the dimension data; X min Indicates the minimum value of the dimension data; X i 、X i ′ represents the value of the i-th sample of the dimension data before and after normalization.

[0048] After data preprocessing, the time dependency and multi-dimensional correlation in the data are extracted through the warning threshold prediction model based on the deep learning model to achieve dynamic prediction of the warning threshold. The warning threshold prediction model is a time series prediction model, and its overall structure is as follows: Figure 3 As shown, in Figure 3 In the proposed method, the warning threshold prediction model performs feature enhancement on the input data, processes the enhanced features through a multi-layer encoder, and finally outputs the prediction results through a multi-layer decoder.

[0049] The following steps describe how the early warning threshold prediction model processes input data to obtain output:

[0050] S220 , performing feature enhancement processing on the network point attribute data, smart terminal service data, network point weather data, location code of model input data, and timestamp code of model input data to obtain enhanced features.

[0051] The positional encoding of the model input data reflects the order of the input data. Each piece of input data is a sample. For example, a sample includes network attribute data, smart terminal business data, and network weather data, and these data correspond to a certain time, such as May 1st. In this case, different samples correspond to different times, for example, May 1st corresponds to a sample, and May 2nd corresponds to a sample. The positional encoding of the model input data can mark the temporal order of the samples. The timestamp encoding of the model input data can reflect the specific data acquisition time of the input data.

[0052] Specifically, the network point attribute data, smart terminal business data, network point meteorological data, location code of model input data, and timestamp code of model input data are processed to obtain enhanced features, and then the enhanced features are input into the encoder of the warning threshold prediction model for processing. This solution is set up in such a way that the features input into the encoder of the warning threshold prediction model are enhanced, and not only include the smart terminal business data of the network point, but also include network point attribute data and network point meteorological data. The model can more accurately predict the threshold from dimensions such as weather, and the enhanced features include the location code of the model input data and the timestamp code of the model input data. These features clearly reflect the data acquisition time of the input data, as well as the order between the input data, so that the model can extract more information from it, thereby improving the rationality of the prediction and warning threshold output by the model.

[0053] In an embodiment of the present application, optionally, the warning threshold prediction model is a time series prediction model based on the attention mechanism.

[0054] It's important to note that in deep learning, the attention mechanism can weight input data features using an importance weight vector, assigning greater weight to data features that are highly relevant to the target. This helps the model better mine key information between features and ignore irrelevant information, thereby improving the model's prediction accuracy. When time series prediction models process sequence data using the attention mechanism, they can effectively capture feature data information at any position in the sequence.

[0055] In an embodiment of the present application, optionally, feature enhancement processing is performed on the network point attribute data, smart terminal business data, network point meteorological data, location code of model input data, and timestamp code of model input data to obtain enhanced features, including: inputting the network point attribute data, smart terminal business data, and network point meteorological data into a feature selection network to add weights to each feature through the feature selection network to obtain weighted features; and determining the enhanced features based on the weighted features, location code of model input data, and timestamp code of model input data.

[0056] For example, the dynamic features composed of smart terminal business data and network meteorological data and the static features composed of network attribute data are used as the main input of the early warning threshold prediction model. They are highly correlated with the prediction target of the early warning threshold of the target business indicator. To ensure the performance of the model, the dynamic features (smart terminal business data and network meteorological data) and the static features of the network itself (network attribute data) are selected through the feature selection network. In a data-driven way, it automatically learns which features are more important for the prediction results of the early warning threshold of the target business indicator, while suppressing features that may have a negative impact on the results, thereby enhancing the input of the early warning threshold prediction model. The structural diagram of the feature selection network is shown in the figure below. Figure 4 As shown. Figure 4 It can be seen that the feature selection network is mainly composed of the gating module G, which includes: linear layer, ELU function, linear layer, gating layer, residual connection & normalization. The feature selection network obtains the selection weight w of the input feature through the gating module G and the softmax function. xt , then based on the selection weight w xt The input features after mapping transformation are weighted and assigned to obtain weighted features. The selection of input features is realized. Specifically, the network attribute data, smart terminal business data and network weather data after feature selection at time t are output for

[0057]

[0058] w xt =softmax(G(x t ));

[0059] Among them, ξ t is the feature vector of network attribute data, smart terminal business data and network weather data at time t, w xt is the selection weight, and G is the processing function of the gating module.

[0060] The input of the gate control module G is the feature vector of the network attribute data, smart terminal business data and network weather data, and its calculation is implemented as follows:

[0061] G(ξ t )=LayerNorm(ξ t +GLU(η1));

[0062] η1=W1η2+b1;

[0063] η2=ELU(W2ξ t +b2);

[0064]

[0065] Among them, GLU is the gate layer, which is a component of the gate module G; ELU is the activation function, α is a hyperparameter, which is generally set to [0, 1]; W i is the learnable weight parameter of the model; b i is the bias parameter. x can be the data after the linear layer processes the feature vectors of the network point attribute data, smart terminal business data, and network point meteorological data;

[0066] Among them, the gating layer adjusts the input of the model through the gating mechanism to increase the flexibility of the model. Its calculation method is:

[0067] GLU(x)=s(W3x+b3)☉(W4x+b4)

[0068] Where x is the input of the gating layer; σ is the Sigmoid activation function; W i is the learnable weight parameter of the model; b i is the bias parameter; ☉ is the Hadamard product.

[0069] This scheme is set up in this way, adding corresponding weights to the features of network attribute data, smart terminal business data and network meteorological data, obtaining weighted features, thereby reducing the influence of features with less impact on the warning threshold, increasing the influence of features with greater impact on the warning threshold, improving model performance, and making the predicted warning threshold finally output by the model more accurate.

[0070] In an embodiment of the present application, optionally, determining the enhanced features according to the weighted features, the position code of the model input data, and the timestamp code of the model input data includes: determining the enhanced features according to the following formula: Among them, χ t is the enhanced feature, α is the parameter factor that balances the size between the input mapping sequence and the local and global features, is the weighted feature, PE pos Position encoding for model input data, TE i Timestamp encoding for model input data.

[0071] For example, in the scenario of predicting the warning threshold of business indicators, the timestamp encoding of the model input data is an indispensable part, such as the time of day when the transaction peak occurs, the difference in the operating status of branches during holidays, etc. For the collected data, the timestamp encoding of the input data is represented by four types of time encoding: hour, day, week, and month. The specific time encoding method is as follows:

[0072]

[0073] In the formula, TE i Indicates the timestamp code of the i-th type, where i can be 1, 2, 3 or 4, representing hours, days, weeks and months respectively; t i and T i Indicates the time point and period for this type of timestamp. Taking a month as an example, t represents the month of data collection, and T is a period of 12 (there are 12 months in a year). Taking a day as an example, t represents the specific day of data collection, and T is a period of 7 (there are 7 days in a week).

[0074] The warning threshold prediction model has no recursive and convolutional structure, so it cannot obtain the position information of the sequence by iterating the state vector like a recurrent neural network. Therefore, the position encoding of the model input data needs to be added in the input link of the model to provide position information. The position encoding function PE pos The calculation formula is

[0075]

[0076] Where pos represents the position of the original data in the sequence. When it is at an even position, the sine function is used for position encoding, otherwise the cosine function is used; i represents the i-th feature dimension of the feature sequence; d represents the dimension of the multidimensional feature after being mapped by the input vector embedding layer.

[0077] Furthermore, after obtaining the position encoding of the model input data and the timestamp encoding of the model input data, the enhanced features are obtained by the following formula: This scheme is set up in such a way that the enhanced features can reflect the position coding and timestamp coding of the input data, providing more feature support for the warning threshold prediction model to predict the warning threshold, thereby improving the prediction accuracy of the warning threshold prediction model.

[0078] S230: Input the enhanced features into the encoder of the warning threshold prediction model, and then input the output of the encoder into the decoder of the warning threshold prediction model to obtain the predicted warning threshold of the target business indicator.

[0079] Exemplarily, the encoder part is composed of a plurality of stacked network layers containing attention mechanisms, and each network layer contains a multi-head attention mechanism, a feedforward network, and a residual connection. The multi-head attention mechanism divides the input feature vector into h equal parts, and then operates the divided vector through the attention function to obtain the attention matrix of each attention head. Specifically, the warning threshold prediction model maps the enhanced feature input to three subspaces: query (query, Q), key (key, K), and value (value, V). The similarity between the query and the key is measured by dot product calculation, and after softmax normalization, it is multiplied with the value vector and weighted to obtain the final attention weight matrix to realize the probability value assignment for each key. The calculation formula of the attention mechanism Attention is:

[0080]

[0081] Q=W Q x,K=W K x,V=W V x;

[0082] Where query Q, key K and value V are linear transformations of input x; W Q , W K and W V is the model learnable parameter; d k is the feature dimension of each attention head.

[0083] Then all the obtained attention matrices are concatenated to obtain the multi-head attention matrix. The implementation of MultiHead is

[0084] MultiHead(Q,K,V)=Concat(head1,...,head h );

[0085] head i =Attention(QW i Q ,KW i K ,VW i V );

[0086] Where W i Q 、W i K and W i V is a model learnable parameter; head i represents the attention matrix of the i-th attention head.

[0087] The decoder portion of the warning threshold prediction model is similar to the encoder, except that it contains two multi-head attention layers. The first attention layer performs a masking operation, which means that when predicting the threshold at time t, the data after time t is set to infinity. The key-value feature matrix of the second attention layer is derived from the encoder output. The rest of the implementation mechanism is the same as the encoder.

[0088] In addition to the attention network layer, each layer in the encoder and decoder of the warning threshold prediction model contains a feedforward network F consisting of a fully connected layer and a residual connection, which is calculated as follows:

[0089] F(x)=max(0,xW1+b1)W2+b2;

[0090] Where W1 and W2 are the learnable weight parameters of the model; b1 and b2 are the bias parameters.

[0091] Finally, the model is trained by minimizing the loss function of the prediction model to obtain the predicted value of the threshold. The loss function of the model is:

[0092]

[0093] Where y i and They respectively represent the actual value and predicted value of the threshold of the monitoring alarm indicator of the smart terminal application at the branch.

[0094] It should be noted that during the training phase of the warning threshold prediction model, the training samples are divided into a 7:3 ratio for training and testing. After the model training is completed, the predicted warning thresholds for the target business indicators for a period of time are predicted. Based on the predicted results (predicted warning thresholds), the warning thresholds are dynamically set and adjusted to ensure that the thresholds accurately reflect the actual operating status of the target business.

[0095] S240: Determine a warning result based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator.

[0096] In an embodiment of the present application, optionally, the prediction warning threshold is the warning threshold of the target business indicator at the target time; the target time is after the acquisition time of the input data of the warning threshold prediction model; accordingly, the warning result is determined based on the prediction warning threshold of the target business indicator and the actual business data of the target business indicator, including: determining the warning result based on the warning threshold of the target business indicator at the target time and the actual business data of the target business indicator at the target time.

[0097] Specifically, since the predicted warning threshold is obtained through prediction, the target time corresponding to the predicted warning threshold is after the time when the input data of the warning threshold prediction model is obtained.

[0098] Specifically, taking the current time as today and the target time as tomorrow as an example, after reaching the target time, the actual business data for tomorrow is obtained, and then the warning result is determined based on the warning threshold and the size relationship of the actual business data. This scheme is set up in this way, and the warning threshold of the target business indicator of the target time can be set as the predicted warning threshold. Since the predicted warning threshold at each time may be different, the effect of dynamically setting the warning threshold is achieved. Compared with the method of making warning judgments based on fixed warning thresholds, the warning accuracy of the embodiment of the present application is improved.

[0099] In an embodiment of the present application, optionally, the warning result is determined based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator, including: if it is determined that the predicted warning threshold and the actual business data have met the warning triggering condition, then a warning prompt information is sent to the warning receiving entity, so that the warning processing entity processes this warning according to the warning prompt information.

[0100] The warning triggering conditions can be set according to actual conditions, and the embodiments of the present application do not limit this. For example, if it is determined that the predicted warning threshold and the actual business data have met the warning triggering conditions, a warning prompt message is sent to the warning receiving entity, including: if it is determined that the predicted warning threshold is greater than the actual business data, a warning prompt message is sent to the warning receiving entity; or, if it is determined that the predicted warning threshold is less than or equal to the actual business data, a warning prompt message is sent to the warning receiving entity; or, if it is determined that the actual business data is within the data interval reflected by the predicted warning threshold, a warning prompt message is sent to the warning receiving entity.

[0101] The technical solution of the embodiment of the present application dynamically predicts the warning threshold from multiple feature dimensions by using branch attribute data, smart terminal business data, branch weather data, location encoding of model input data, and timestamp encoding of model input data. This multi-dimensional prediction method can more comprehensively evaluate the operating status of the system, effectively reduce false positives and missed negatives, and improve the accuracy and reliability of warnings. Moreover, by comprehensively considering the combined effects of multiple feature factors, it can more comprehensively reflect the actual operating status of the bank branch application system, thereby obtaining more accurate warning threshold prediction results, which can effectively improve the accuracy of warnings. In addition, the prediction model is data-driven and learns the feature correlation between data. It has high generalization ability and is suitable for dynamic setting of warning thresholds in different scenarios. It does not require operation and maintenance personnel to excessively design and adjust alarm rules, reducing operation and maintenance pressure. This technical solution introduces timestamp features, which enables the model to learn the detailed differences in monitoring warning indicators in the time dimension, thereby refining warning thresholds at different time points, such as transaction peak thresholds, holiday thresholds, etc., making the prediction results of warning thresholds more reasonable and effectively reducing false positives and missed negatives in branch warnings.

[0102] Example 3

[0103] Figure 5 This is a schematic diagram of the structure of a warning device for business indicators of a network intelligent terminal provided in the third embodiment of this application. The device can execute the warning method for business indicators of a network intelligent terminal provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. Figure 5 As shown, the device includes:

[0104] Data acquisition module 310, used to obtain network attribute data, smart terminal business data and network weather data;

[0105] The prediction and warning threshold determination module 320 is used to input the network point attribute data, smart terminal business data and network point weather data into the warning threshold prediction model to obtain the prediction and warning threshold of the target business indicator;

[0106] The warning result determination module 330 is configured to determine a warning result based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator.

[0107] The technical solution of the embodiment of the present application includes: a data acquisition module 310 for acquiring network point attribute data, smart terminal service data, and network point meteorological data; a prediction and warning threshold determination module 320 for inputting the network point attribute data, smart terminal service data, and network point meteorological data into a warning threshold prediction model to obtain a prediction and warning threshold for a target business indicator; and a warning result determination module 330 for determining a warning result based on the prediction and warning threshold for the target business indicator and the actual business data of the target business indicator. This technical solution processes the network point attribute data, smart terminal service data, and network point meteorological data through the warning threshold prediction model to predict a dynamic warning threshold, thereby greatly improving the accuracy of warnings when issuing warnings based on the warning threshold.

[0108] Optionally, the prediction warning threshold is a warning threshold of a target business indicator at a target time; the target time is after the time when input data of the warning threshold prediction model is acquired;

[0109] Accordingly, the warning result determination module 330 includes:

[0110] The warning result determination unit is used to determine the warning result according to the warning threshold of the target business indicator at the target time and the actual business data of the target business indicator at the target time.

[0111] Optionally, the warning result determination module 330 includes:

[0112] The warning prompt information sending unit is used to send warning prompt information to the warning receiving entity if it is determined that the predicted warning threshold and the actual business data have met the warning triggering condition, so that the warning processing entity can process this warning according to the warning prompt information.

[0113] Optionally, the prediction and warning threshold determination module 320 includes:

[0114] A feature enhancement unit is used to perform feature enhancement processing on the network point attribute data, smart terminal service data, network point meteorological data, location code of model input data, and timestamp code of model input data to obtain enhanced features;

[0115] The prediction and warning threshold determination unit is used to input the enhanced features into the encoder of the warning threshold prediction model, and then input the output of the encoder into the decoder of the warning threshold prediction model to obtain the prediction and warning threshold of the target business indicator.

[0116] Optionally, the feature enhancement unit includes:

[0117] A weighted processing subunit, configured to input the network point attribute data, smart terminal service data, and network point meteorological data into a feature selection network, so as to add weights to each feature through the feature selection network to obtain weighted features;

[0118] The feature enhancement subunit is used to determine the enhanced features according to the weighted features, the position encoding of the model input data, and the timestamp encoding of the model input data.

[0119] Optionally, the feature enhancer unit is specifically used to:

[0120] The enhanced features are determined according to the following formula:

[0121]

[0122] Among them, χ t is the enhanced feature, α is the parameter factor that balances the size between the input mapping sequence and the local and global features, is the weighted feature, PE pos Position encoding for model input data, TE i Timestamp encoding for model input data.

[0123] Optionally, the warning threshold prediction model is a time series prediction model based on an attention mechanism.

[0124] An early warning device for business indicators of a network intelligent terminal provided in an embodiment of the present application can execute an early warning method for business indicators of a network intelligent terminal provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0125] Example 4

[0126] Figure 6 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0127] like Figure 6As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0128] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0129] Processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors that run machine learning model algorithms, digital signal processors (DSPs), and any other suitable processor, controller, or microcontroller. Processor 11 executes the various methods and processes described above, such as the early warning method for service indicators of smart terminals at branch locations.

[0130] In some embodiments, the early warning method for service indicators of a network intelligent terminal can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the early warning method for service indicators of a network intelligent terminal described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to execute the early warning method for service indicators of a network intelligent terminal in any other appropriate manner (e.g., via firmware).

[0131] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0133] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0135] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0136] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0137] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0138] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for early warning of business indicators of a network intelligent terminal, characterized in that: include: Obtain network attribute data, smart terminal business data, and network weather data; Inputting the network point attribute data, smart terminal business data and network point weather data into the warning threshold prediction model to obtain the prediction warning threshold of the target business indicator; The warning result is determined based on the predicted warning threshold of the target business indicator and the actual business data of the target business indicator.

2. The method according to claim 1, characterized in that The prediction warning threshold is the warning threshold of the target business indicator at the target time; the target time is after the acquisition time of the input data of the warning threshold prediction model; Accordingly, the warning results are determined based on the predicted warning thresholds of the target business indicators and the actual business data of the target business indicators, including: The warning result is determined based on the warning threshold of the target business indicator at the target time and the actual business data of the target business indicator at the target time.

3. The method according to claim 1, characterized in that The warning results are determined based on the predicted warning thresholds of the target business indicators and the actual business data of the target business indicators, including: If it is determined that the predicted warning threshold and the actual business data have satisfied the warning triggering condition, a warning prompt message is sent to the warning receiving entity, so that the warning processing entity processes the warning according to the warning prompt message.

4. The method according to claim 1, wherein Inputting the network point attribute data, smart terminal business data, and network point weather data into the warning threshold prediction model to obtain the prediction warning threshold of the target business indicator, including: Performing feature enhancement processing on the network point attribute data, smart terminal service data, network point meteorological data, location code of model input data, and timestamp code of model input data to obtain enhanced features; The enhanced features are input into the encoder of the warning threshold prediction model, and the output of the encoder is input into the decoder of the warning threshold prediction model to obtain the predicted warning threshold of the target business indicator.

5. The method according to claim 4, characterized in that Performing feature enhancement processing on the network point attribute data, smart terminal service data, network point meteorological data, location code of model input data, and timestamp code of model input data to obtain enhanced features, including: Inputting the network point attribute data, smart terminal service data, and network point weather data into a feature selection network, so as to add weights to each feature through the feature selection network to obtain weighted features; The enhanced features are determined according to the weighted features, the position encoding of the model input data, and the timestamp encoding of the model input data.

6. The method according to claim 5, characterized in that The enhanced features are determined based on the weighted features, the position encoding of the model input data, and the timestamp encoding of the model input data, including: The enhanced features are determined according to the following formula: Among them, χ t is the enhanced feature, α is the parameter factor that balances the size between the input mapping sequence and the local and global features, is the weighted feature, PE pos Position encoding for model input data, TE i Timestamp encoding for model input data.

7. The method according to claim 1, characterized in that The warning threshold prediction model is a time series prediction model based on the attention mechanism.

8. An early warning device for business indicators of a network intelligent terminal, characterized in that: include: Data acquisition module, used to obtain network attribute data, smart terminal business data and network weather data; A prediction and warning threshold determination module is used to input the network point attribute data, smart terminal business data and network point meteorological data into the warning threshold prediction model to obtain the prediction and warning threshold of the target business indicator; The early warning result determination module is used to determine the early warning result based on the predicted early warning threshold of the target business indicator and the actual business data of the target business indicator.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the early warning method for business indicators of the network intelligent terminal according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the early warning method for business indicators of a network intelligent terminal according to any one of claims 1 to 7 when executed.

Images