A parameter-driven banking system front-end data system and method

Abstract

The present invention discloses a bank system front-end data system and method based on parameter drive, which relates to the field of parameter control technology, including: obtaining bank system front-end parameter data at historical moments to establish a parameter state data set; obtaining bank system front-end operation and maintenance logs at corresponding historical moments to establish an operation and maintenance behavior data set; obtaining bank system front-end parameter prediction values ​​according to a bank system front-end parameter prediction model, generating a predicted parameter state set, inputting it into a bank system front-end operation and maintenance decision model, obtaining a bank system front-end operation and maintenance plan, and completing bank system front-end adaptive operation and maintenance management; performing dynamic operation and maintenance behavior based on a current moment parameter deviation value X, a first threshold value, and a second threshold value, obtaining a current moment bank system front-end operation and maintenance optimization plan, and completing bank system front-end dynamic operation and maintenance optimization management. The method in the present application can complete adaptive operation and maintenance management in a bank system front-end system.

Description

Technical Field

[0001] The present invention relates to the field of parameter control technology, and in particular to a parameter-driven banking system front-end data system and method. Background Art

[0002] The bank system front-end (system) is a middleware platform for data exchange between the bank's internal system and internal and external systems. Its main functions include receiving transaction requests and query requests from external channels such as ATM, POS, online banking, mobile banking, etc., performing preliminary verification and processing operations such as identity verification and account balance check on transaction requests, forwarding the processed transaction requests to the back-end core banking system for further processing, and returning the core system's response to external channels. Therefore, the bank system front-end (system) should have stability, efficiency, scalability and manageability in processing data.

[0003] The Chinese patent application with the publication number CN116051257A discloses an intelligent operation and maintenance anomaly detection method, device and server, which obtains the real-time operation data of the bank network system and the bank network system detection time series data, pre-processes the bank network system detection time series data to obtain a multi-dimensional training data set, and conducts online training in the long short-term memory network model through the multi-dimensional feature data to obtain a long short-term memory network prediction model. According to the obtained long short-term memory network prediction model, the difference between the long short-term memory network prediction data and the real-time acquired bank network data is compared. When the difference is greater than the preset value, it is determined that the bank network system has an anomaly. Although the application can use the long short-term memory network prediction data and compare it with the real-time acquired bank network data to determine whether there is an anomaly, the long short-term memory network prediction model does not consider the complex non-stationary time series characteristics of the bank network system parameters, nor does it consider how to perform automated and efficient operation and maintenance when the bank network system has an anomaly.

[0004] The Chinese patent application with the publication number CN117370013A discloses a method and thread pool for dynamically maintaining a thread pool based on a neural network, including: when there are no idle threads in the thread pool, using negative value distribution to predict the probability of idle threads appearing in the thread pool; determining the current load of the thread pool, increasing the number of core threads of the thread pool or expanding the thread pool according to the current load of the thread pool; and predicting the future load of the thread pool based on the neural network, increasing the number of core threads of the thread pool or expanding the thread pool according to the future load of the thread pool. Although the application can perform dynamic operation and maintenance according to the thread pool parameters predicted by the neural network and improve the resource allocation utilization rate, it fails to realize the adaptive operation and maintenance adjustment of the system at the same time as the dynamic operation and maintenance.

[0005] To this end, the present invention proposes a parameter-driven banking system front-end data system and method. Summary of the invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a parameter-driven bank system front-end data system and method, which can accurately predict the bank system front-end operation parameters and realize the adaptive operation and maintenance management and dynamic operation and maintenance optimization management of the bank system front-end.

[0007] The present invention aims at solving the technical problems existing in the prior art and provides a parameter-driven banking system front-end data system and method, the specific steps of which are as follows:

[0008] S1. Obtain the parameter data of the bank system front-end at the historical moment to establish a parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish an operation and maintenance behavior data set;

[0009] S2. According to the parameter state data set, a bank system front-end parameter prediction model is obtained through time series prediction model training; the time series prediction model is an end-to-end network model composed of Transformer and probability time series feature extraction modules; the probability time series feature extraction module is composed of a mixed density network and a fully connected layer; according to the parameter state data set and the operation and maintenance behavior data set, a bank system front-end operation and maintenance decision model is obtained through deep reinforcement learning training;

[0010] S3. According to the bank system front-end machine parameter prediction model, obtain the bank system front-end machine parameter prediction value, generate the prediction parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management;

[0011] S4, obtaining the actual parameter value of the bank system front-end at the current moment and performing parameter deviation detection on the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment;

[0012] S5. Set the first threshold value Y according to the bank system front-end parameter prediction value 1 and the second threshold Y 2 , based on the current parameter deviation value X and the first threshold value Y 1 and the second threshold Y 2 Perform dynamic operation and maintenance behaviors, obtain the operation and maintenance optimization plan of the bank system front-end machine at the current moment, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

[0013] In particular, the parameter data of the bank system front-end processor at the historical moment is obtained to establish a parameter status data set; the operation and maintenance log of the bank system front-end processor at the corresponding historical moment is obtained and feature processed to establish an operation and maintenance behavior data set, wherein:

[0014] The bank system front-end machine parameter data at the historical moment includes system operation parameters and external environment parameters;

[0015] The system operation parameters include system resource allocation parameters, thread pool parameters, connection pool parameters and concurrent processing parameters; the external environment parameters include banking business transaction volume, user access volume, network environment and external system call frequency;

[0016] The bank system front-end operation and maintenance log corresponding to the historical moment includes the system operation parameter operation and maintenance log and the external environment parameter operation and maintenance log;

[0017] The system operation parameter operation and maintenance log includes resource allocation operation records, thread pool management operation records, connection pool management operation records and concurrent processing optimization operation records; the external environment parameter operation and maintenance log includes external environment adaptation operation records;

[0018] The feature processing includes operation type recognition, time feature extraction, operation amplitude quantification and operation effect quantification;

[0019] The operation and maintenance behavior data set includes operation type characteristics, time characteristics, operation amplitude characteristics and operation effect characteristics;

[0020] The quantification of the operational effect is based on the maximization of resource utilization of the bank system front-end machine as a quantitative indicator.

[0021] In particular, the steps of obtaining the bank system front-end parameter prediction model through time series prediction model training based on the parameter status data set are:

[0022] Step S11: Preprocessing the parameter state data set to obtain a parameter prediction data set;

[0023] Step S12: dividing the parameter prediction data set into a training set and a validation set, inputting the data set into a time series prediction model for training, and obtaining a bank system front-end parameter prediction model;

[0024] The preprocessing operation includes dividing the time step, denoising, standardization and counting the probability of parameters in different value intervals within the unit time step;

[0025] The standardization process adopts the Z-Score standardization process method;

[0026] In particular, the bank system front-end parameter prediction model is obtained by training the time series prediction model through the parameter prediction data set; the time series prediction model is an end-to-end network model composed of Transformer and probabilistic time series feature extraction modules, including Embedding, position encoding, probabilistic time series feature extraction module, encoder, decoder and logistic regression layer submodules;

[0027] Furthermore, the probability time series feature extraction module is composed of a mixed density network and a fully connected layer; the mixed density network is composed of an input layer, a hidden layer, a preprocessing layer and an output layer;

[0028] The time series prediction model training phase adopts a training strategy combining a pre-trained probability time series feature extraction module and a joint training time series prediction model;

[0029] The prediction result of the bank system front-end parameter prediction model is the probability of the parameter appearing in different value ranges within a unit time step at a future moment.

[0030] In particular, the steps of obtaining the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set are:

[0031] Step S21: Based on the parameter status data set and the operation and maintenance behavior data set, construct an expert experience sequence A(s t , a t , r, s t+1 ); the t is the parameter state set, a t is the set of operation and maintenance actions, r is the set of operation effect reward values, s t+1 is the parameter state set for the next moment;

[0032] Step S22: constructing an expert experience pool based on the expert experience sequence;

[0033] Step S23: Using the expert experience pool, train the deep reinforcement learning model to obtain the banking system front-end operation and maintenance decision model;

[0034] Furthermore, the steps of using the expert experience pool to train the deep reinforcement learning model to obtain the bank system front-end operation and maintenance decision model are as follows:

[0035] Step S31: constructing a prediction network and a target network using a BP neural network;

[0036] Step S32: constructing a deep reinforcement learning model based on the prediction network, the target network and the experience pool;

[0037] Step S33: Initialize the experience pool to an expert experience pool; the experience pool capacity is N;

[0038] Step S34: Initialize the prediction network using random prediction network weights θ and use random target network weights θ * =θ initialize the target network and initialize the parameter state set s t , batch n and step size m;

[0039] Step S35: Randomly obtain expert experience sequence A(s) in batches n from the experience pool t , a t , r, s t+1 ) for training, where t is a variable, n <N;

[0040] Prediction network input parameter state set s t and operation and maintenance action set a t , output prediction value function Q(s t , a t |θ) and the next moment operation and maintenance prediction action set a t+1 * ;

[0041] The target network inputs the parameter state set s at the next moment t+1 , output the target value function F, the calculation formula of F is:

[0042]

[0043] In the formula, For the parameter state set s t Time operation and maintenance action set a t The operation effect reward value, γ is the discount factor and γ∈(0,1), max Q(s t+1 , a t+1 * ,θ * ) is the parameter state set s at the next moment t+1 When the next moment operation and maintenance prediction action set a is selected t+1 * The maximum predicted value function obtained;

[0044] Step S36: Every m steps, the weight parameter θ of the prediction network is copied to the weight parameter θ of the target network. * ;

[0045] Step S37: Use the gradient descent method to minimize the MSE loss function and update the network model parameters θ and θ * Until convergence, the front-end operation and maintenance decision model of the banking system is obtained;

[0046] Furthermore, the BP neural network consists of an input layer, a hidden layer and an output layer;

[0047] The number of neurons in the input layer is equal to the number of parameter states in the parameter state set;

[0048] The number of neurons in the output layer is equal to the number of operation and maintenance actions in the operation and maintenance action set.

[0049] In particular, the actual parameter value of the bank system front-end at the current moment is obtained and the parameter deviation detection is performed with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment, where:

[0050] The parameter deviation detection only detects the deviation of system operation parameters;

[0051] The parameter deviation detection adopts a difference calculation method;

[0052] The calculation formula for obtaining the parameter deviation value X at the current moment is:

[0053] X=a 1 -a 2 ;

[0054] In the formula, a 1 is the actual parameter value of the bank system front-end at the current moment, a 2 is the parameter prediction value corresponding to the current moment.

[0055] In particular, the first threshold value Y is set according to the bank system front-end parameter prediction value 1 and the second threshold Y 2 ,in:

[0056] The first threshold value Y 1 and the second threshold Y 2 The calculation formula is:

[0057] Y 1 =b+b×α%;

[0058] Y 2 =bb×β%;

[0059] Where b is the predicted value of the bank system front-end parameter, and α is the first threshold Y 1 The threshold amplitude factor, β is the second threshold Y 2 The threshold amplitude factor.

[0060] In particular, the current moment parameter deviation value X, the first threshold value Y 1 and the second threshold Y 2 Perform dynamic operation and maintenance to obtain the current bank system front-end machine operation and maintenance optimization plan. The steps to complete the dynamic operation and maintenance optimization management of the bank system front-end machine are as follows:

[0061] Step S41: Obtain the actual parameter value of the bank system front-end at the current moment, the parameter deviation value X at the current moment, and the first threshold value Y 1 and the second threshold Y 2 ;

[0062] Step S42: performing abnormality determination on the actual parameter values ​​of the bank system front-end processor at the current moment;

[0063] Step S43: updating the current banking system front-end operation and maintenance optimization plan according to the abnormality judgment result;

[0064] Step S44: completing dynamic operation and maintenance optimization management of the bank system front-end processor according to the current bank system front-end processor operation and maintenance optimization plan;

[0065] Furthermore, the step of determining the abnormality of the actual parameter value of the bank system front-end processor at the current moment is as follows:

[0066] If the actual parameter value of the bank system front-end at the current moment is between the first threshold value Y 1 and the second threshold Y 2 In the meantime, the bank system front-end machine refuses to perform dynamic operation and maintenance optimization management of the bank system front-end machine;

[0067] If the actual parameter value of the bank system front-end at the current moment is greater than or equal to the first threshold value Y 1 , then the current moment parameter deviation value X is recorded in the current moment parameter deviation value table, and the current moment banking system front-end operation and maintenance optimization plan is updated according to the current moment parameter deviation value table;

[0068] If the actual parameter value of the bank system front-end at the current moment is less than or equal to the second threshold value Y 2 The bank system front-end will issue parameter warnings and provide feedback to manual operation and maintenance personnel.

[0069] Another aspect of the present application provides a parameter-driven banking system front-end data system, the system comprising:

[0070] Data processing module: used to obtain the parameter data of the bank system front-end at the historical moment to establish the parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish the operation and maintenance behavior data set;

[0071] Parameter prediction model training module: used to obtain the bank system front-end parameter prediction model through time series prediction model training based on the parameter status data set;

[0072] Operation and maintenance decision model training module: used to obtain the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set;

[0073] Adaptive operation and maintenance management module: used to obtain the predicted values ​​of the bank system front-end machine parameters according to the bank system front-end machine parameter prediction model, generate a predicted parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management;

[0074] Parameter deviation detection module: used to obtain the actual parameter value of the bank system front-end at the current moment and perform parameter deviation detection with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment;

[0075] Dynamic operation and maintenance optimization management module: used to set the first threshold value Y according to the predicted value of the bank system front-end machine parameter 1 and the second threshold Y 2 , based on the current parameter deviation value X and the first threshold value Y 1 and the second threshold Y 2 Perform dynamic operation and maintenance behaviors, obtain the operation and maintenance optimization plan of the bank system front-end machine at the current moment, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

[0076] Another aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the above-mentioned parameter-driven bank system front-end data system and method are implemented.

[0077] Another aspect of the present application provides a readable storage medium storing a computer program, wherein the computer program is suitable for loading by a processor to implement the steps of the above-mentioned parameter-driven banking system front-end data system and method.

[0078] Beneficial technical effects of the present invention:

[0079] The present invention uses the historical parameter data and historical operation and maintenance logs of the bank system front-end machine to train the bank system front-end machine parameter prediction model and the bank system front-end machine operation and maintenance decision model. Based on the prediction model and the operation and maintenance decision model, the bank system front-end machine parameters in the future time period can be predicted, and the bank system front-end machine operation and maintenance plan can be set and updated in advance. The invention can perform adaptive operation and maintenance management on the bank system front-end machine and realize the automated operation and maintenance of the bank system front-end machine. The key to the bank system front-end machine to realize adaptive operation and maintenance management and dynamic operation and maintenance optimization management is to accurately predict the complex non-stationary time series characteristics during its operation. The time series prediction model based on the probability time series feature extraction module proposed by the present invention can achieve the effect of accurately predicting the complex non-stationary time series characteristics during the operation of the bank system front-end machine.

[0080] The adaptive operation and maintenance management of the bank system front-end is an operation and maintenance behavior made according to the predicted parameters. It is difficult to cover all operation and maintenance situations in view of the complex and changeable banking business. In order to solve the above problems, the present invention makes an abnormal judgment on the actual parameter value of the bank system front-end at the current moment, and updates the operation and maintenance optimization plan of the bank system front-end at the current moment according to the abnormal judgment result, so as to complete the dynamic operation and maintenance optimization management of the bank system front-end. This method not only supplements the adaptive operation and maintenance management of the bank system front-end, but also records the operation and maintenance optimization plan of the bank system front-end under abnormal circumstances in the operation and maintenance log, continuously improving the accuracy of the operation and maintenance decision model of the bank system front-end, so as to further improve the accuracy of the automated operation and maintenance of the bank system front-end. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 It is a flow chart of a parameter-driven banking system front-end data system and method according to Embodiment 1 of the present invention.

[0082] Figure 2 It is a structural diagram of a time series prediction model of Example 1 of the present invention.

[0083] Figure 3 It is a structural diagram of a probabilistic time series feature extraction module of Example 1 of the present invention.

[0084] Figure 4 This is a step of training a bank system front-end operation and maintenance decision model in Embodiment 1 of the present invention.

[0085] Figure 5 This is a step of training a deep reinforcement learning model in Example 1 of the present invention.

[0086] Figure 6 This is a flow chart of adaptive operation and maintenance and optimized management of a banking system front-end processor according to Embodiment 1 of the present invention.

[0087] Figure 7 A parameter-driven banking system front-end data system is provided in Example 2 of the present invention.

[0088] Figure 8 This is a schematic diagram of the structure of an electronic device provided in Example 3 of the present invention.

[0089] Fig. 9 It is a schematic diagram of the structure of a computer-readable storage medium provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0090] In order to better understand the present application, a more detailed description will be made of various aspects of the present application with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application, and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0091] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0092] It should also be understood that expressions such as "include", "including", "have", "contain" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0093] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0094] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0095] Example 1

[0096] like Figure 1 to Figure 6 As shown, this embodiment provides a parameter-driven banking system front-end data system and method;

[0097] like Figure 1 As shown, the steps of the parameter-driven banking system front-end data system and method include:

[0098] S1. Obtain the parameter data of the bank system front-end at the historical moment to establish a parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish an operation and maintenance behavior data set;

[0099] In particular, the bank system front-end machine parameter data at the historical moment includes system operation parameters and external environment parameters;

[0100] The system operation parameters include system resource allocation parameters, thread pool parameters, connection pool parameters and concurrent processing parameters; the external environment parameters include banking business transaction volume, user access volume, network environment and external system call frequency;

[0101] In some preferred embodiments, the system resource allocation parameters may include but are not limited to CPU resource allocation, memory resource allocation, storage resource allocation and network resource allocation, etc.; the thread pool parameters may include but are not limited to thread pool size, number of active threads, number of queued tasks, thread pool utilization, thread creation or destruction frequency, etc.; the connection pool parameters may include but are not limited to connection pool capacity, number of active connections, number of idle connections, connection pool utilization, connection creation or release frequency, etc.; the concurrent processing parameters may include but are not limited to the number of concurrent requests, number of queued requests, request processing delay and execution of concurrency control strategy, etc.

[0102] The bank system front-end operation and maintenance log corresponding to the historical moment includes the system operation parameter operation and maintenance log and the external environment parameter operation and maintenance log;

[0103] The system operation parameter operation and maintenance log includes resource allocation operation records, thread pool management operation records, connection pool management operation records and concurrent processing optimization operation records; the external environment parameter operation and maintenance log includes external environment adaptation operation records;

[0104] The feature processing includes operation type recognition, time feature extraction, operation amplitude quantification and operation effect quantification;

[0105] The operation and maintenance behavior data set includes operation type characteristics, time characteristics, operation amplitude characteristics and operation effect characteristics;

[0106] The quantification of the operational effect is based on the maximization of resource utilization of the bank system front-end machine as a quantitative indicator.

[0107] By collecting the system operating parameters and external environment parameters of the bank system front-end machine, a parameter status dataset with complex internal and external states of the system was established. At the same time, feature extraction and processing were performed on the bank system front-end machine operation and maintenance logs to obtain a rich operation and maintenance behavior dataset, which laid a solid data foundation for the subsequent training of the bank system front-end machine parameter prediction model and the bank system front-end machine operation and maintenance decision model.

[0108] S2. According to the parameter state data set, a bank system front-end parameter prediction model is obtained through time series prediction model training; the time series prediction model is an end-to-end network model composed of Transformer and probability time series feature extraction modules; the probability time series feature extraction module is composed of a mixed density network and a fully connected layer; according to the parameter state data set and the operation and maintenance behavior data set, a bank system front-end operation and maintenance decision model is obtained through deep reinforcement learning training;

[0109] In particular, the steps of obtaining the bank system front-end parameter prediction model through time series prediction model training based on the parameter status data set are:

[0110] Step S11: Preprocessing the parameter state data set to obtain a parameter prediction data set;

[0111] Step S12: dividing the parameter prediction data set into a training set and a validation set, inputting the data set into a time series prediction model for training, and obtaining a bank system front-end parameter prediction model;

[0112] The preprocessing operation includes dividing the time step, denoising, standardization and counting the probability of parameters in different value intervals within the unit time step;

[0113] The standardization process adopts the Z-Score standardization process method;

[0114] See also Figure 2 In particular, the bank system front-end parameter prediction model is obtained by training the time series prediction model through the parameter prediction data set; the time series prediction model is an end-to-end network model composed of Transformer and probability time series feature extraction modules, including Embedding, position encoding, probability time series feature extraction module, encoder, decoder and logistic regression layer sub-modules;

[0115] See also Figure 3 ,Furthermore, the probabilistic time series feature extraction module is composed of a mixed density network and a fully connected layer; the mixed density network is composed of an input layer, a hidden layer, a preprocessing layer and an output layer;

[0116] In some preferred embodiments, the specific steps of training the time series prediction model are as follows: the parameter prediction data set is mapped into a feature vector matrix h by Embedding. i , after position encoding, the output is the feature vector matrix h containing position information j , after the probability time series feature extraction module, the output generates a feature vector sequence h containing probability time series features k , as the input data of the encoder-decoder submodule in Transformer, the decoder output is passed through the logistic regression layer to output the probability of the parameter appearing in different value intervals within the unit time step at the future moment.

[0117] In some preferred embodiments,

[0118] The mixture density network receives the feature vector matrix h containing position information j , the eigenvector matrix h j It can be expressed as h j =[x 1 , x 2 , …, x T ], where each x is a feature vector containing time series attributes and location information; the dimension of the matrix x is (T, D), where T is the time step and D is the dimension of the feature vector.

[0119] The output of the mixture density network includes the mixing coefficient ω, mean μ and variance σ parameters, which reflect the probability distribution characteristics of the time series data at different positions; K represents the complexity of the Gaussian mixture model used by the mixture density network to model the time series data, which means that the distribution of the time series data can be modeled as a mixture of K Gaussian distributions. The value of K is adjusted and optimized through the cross-validation method.

[0120] For the tth time step:

[0121] ω t =[ω 1 ,ω 2 ,…,ω k ],

[0122] Among them, t∈T, ω t Represents the mixing coefficient of K Gaussian components, the mixing coefficient ω t is a K-dimensional vector, representing the mixing coefficients of K Gaussian components at this time step;

[0123] μ t =[μ k1 , μ k2 ,…,μ kD ],

[0124] Where t∈T, μ tRepresents the mean vector of K Gaussian components; mean μ t It is a K×D matrix, each row corresponds to the D-dimensional mean vector of a Gaussian component;

[0125] σ t =[σ k1 , σ k2 , …, σ kD ],

[0126] Among them, t∈T,σ t represents the variance vector of K Gaussian components; variance σ t It is a K×D matrix, each row corresponds to the D-dimensional variance vector of a Gaussian component;

[0127] The mixed density network learns to extract the statistical distribution characteristics of the time series in the feature vector matrix x, and the mixing coefficient ω corresponding to each time step t t , mean μ t and variance σ t are all based on the feature vector x at that time step t The extracted time series information encoded in the feature vector matrix x is converted into corresponding probability distribution parameters by the mixture density network.

[0128] The fully connected layer converts the probability distribution parameters into a fixed-dimensional feature vector sequence h k , as the input of the Transformer encoder and decoder; the fully connected layer connects the hybrid density network and the Transformer modules, and inputs the probability distribution features extracted by the hybrid density network into the Transformer to achieve cross-module feature transfer, so that the entire end-to-end network can fully combine and utilize the advantages of the two modules, thereby improving the prediction accuracy of the parameter prediction model of the banking system front-end machine.

[0129] The probabilistic time series feature extraction module utilizes the combined structure of a mixed density network and a fully connected layer, which can effectively capture and extract the probability distribution features in the input time series data. Compared with the traditional deterministic time series feature extraction method, the feature representation based on probability distribution can more accurately describe the complex non-stationary time series features existing in the operation of the bank system front-end.

[0130] The probabilistic time series feature extraction module can effectively separate and extract the stationary time series features and non-stationary time series features in the operating parameters of the bank system front-end. Even if the proportion of non-stationary time series is small, the module can accurately capture this part of key non-stationary features and provide key support for subsequent parameter prediction.

[0131] The time series prediction model training phase adopts a training strategy combining a pre-trained probability time series feature extraction module and a joint training time series prediction model;

[0132] The prediction result of the bank system front-end parameter prediction model is the probability of the parameter appearing in different value ranges within a unit time step at a future moment.

[0133] The key to the bank system front-end machine to achieve adaptive operation and maintenance management and dynamic operation and maintenance optimization management is to accurately predict the complex non-stationary time series characteristics during its operation. The time series prediction model based on the probability time series feature extraction module proposed in the present invention can achieve the effect of accurately predicting the complex non-stationary time series characteristics during the operation of the bank system front-end machine. Based on the probability time series characteristics, the end-to-end time series prediction model can better model the complex non-stationarity during the dynamic operation of the bank system front-end machine, thereby significantly improving the prediction accuracy, and providing strong support for the subsequent bank system front-end machine to achieve adaptive operation and maintenance management and dynamic operation and maintenance optimization management.

[0134] In particular, see Figure 4 The steps of obtaining the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set are:

[0135] Step S21: Based on the parameter status data set and the operation and maintenance behavior data set, construct an expert experience sequence A(s t , a t , r, s t+1 ); the t is the parameter state set, a t is the set of operation and maintenance actions, r is the set of operation effect reward values, s t+1 is the parameter state set for the next moment;

[0136] Step S22: constructing an expert experience pool based on the expert experience sequence;

[0137] Step S23: Using the expert experience pool, train the deep reinforcement learning model to obtain the banking system front-end operation and maintenance decision model;

[0138] Further, see Figure 5 The steps of using the expert experience pool to train the deep reinforcement learning model and obtain the bank system front-end operation and maintenance decision model are as follows:

[0139] Step S31: constructing a prediction network and a target network using a BP neural network;

[0140] Step S32: constructing a deep reinforcement learning model based on the prediction network, the target network and the experience pool;

[0141] Step S33: Initialize the experience pool to an expert experience pool; the experience pool capacity is N;

[0142] Step S34: Initialize the prediction network using random prediction network weights θ and use random target network weights θ * =θ initialize the target network and initialize the parameter state set s t , batch n and step size m;

[0143] Step S35: Randomly obtain expert experience sequence A(s) in batches n from the experience pool t , a t , r, s t+1 ) for training, where t is a variable, n <N;

[0144] Prediction network input parameter state set s t and operation and maintenance action set a t , output prediction value function Q(s t , a t |θ) and the next moment operation and maintenance prediction action set a t+1 * ;

[0145] The target network inputs the parameter state set s at the next moment t+1 , output the target value function F, the calculation formula of F is:

[0146]

[0147] In the formula, For the parameter state set s t Time operation and maintenance action set a t The operation effect reward value, γ is the discount factor and γ∈(0,1), max Q(s t+1 , a t+1 * ,θ * ) is the parameter state set s at the next moment t+1 When the next moment operation and maintenance prediction action set a is selected t+1 * The maximum predicted value function obtained;

[0148] Step S36: Every m steps, the weight parameter θ of the prediction network is copied to the weight parameter θ of the target network. * ;

[0149] Step S37: Use the gradient descent method to minimize the MSE loss function and update the network model parameters θ and θ * Until convergence, the front-end operation and maintenance decision model of the banking system is obtained;

[0150] Furthermore, the BP neural network consists of an input layer, a hidden layer and an output layer;

[0151] The number of neurons in the input layer is equal to the number of parameter states in the parameter state set;

[0152] The number of neurons in the output layer is equal to the number of operation and maintenance actions in the operation and maintenance action set.

[0153] In particular, the actual parameter value of the bank system front-end at the current moment is obtained and the parameter deviation detection is performed with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment, where:

[0154] The parameter deviation detection only detects the deviation of system operation parameters;

[0155] The parameter deviation detection adopts a difference calculation method;

[0156] The calculation formula for obtaining the parameter deviation value X at the current moment is:

[0157] X=a 1 -a 2 ;

[0158] In the formula, a 1 is the actual parameter value of the bank system front-end at the current moment, a 2 is the parameter prediction value corresponding to the current moment.

[0159] S3. See Figure 6 , according to the bank system front-end machine parameter prediction model, obtain the bank system front-end machine parameter prediction value, generate the prediction parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management;

[0160] In the process of adaptive operation and maintenance management of the front-end processor of the banking system, the time series prediction model and deep reinforcement learning model are used to realize the intelligentization of parameter prediction and operation and maintenance decision-making. Through time series analysis, the parameter prediction model of the front-end processor of the banking system can accurately predict the change trend of various key parameters of the front-end processor in the future time period. Based on the parameter status data set and the operation and maintenance behavior data set, the operation and maintenance decision model of the front-end processor of the banking system can learn the optimal resource allocation and operation and maintenance strategy of the front-end processor.

[0161] By combining the parameter prediction model with the operation and maintenance decision model, and iteratively updating the system operation data and external environment data of the bank system front-end machine, the parameter prediction model can continuously give predicted values ​​of future states. These predicted parameter states are input into the operation and maintenance decision model. By leveraging the advantages of deep reinforcement learning, the operation and maintenance strategy is continuously optimized to complete the adaptive operation and maintenance management of the bank system front-end machine, achieving the effect of automated operation and maintenance of the system.

[0162] S4, obtaining the actual parameter value of the bank system front-end at the current moment and performing parameter deviation detection on the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment;

[0163] In particular, the actual parameter value of the bank system front-end at the current moment is obtained and the parameter deviation detection is performed with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment, where:

[0164] The parameter deviation detection only detects the deviation of system operation parameters;

[0165] The parameter deviation detection adopts a difference calculation method;

[0166] The calculation formula for obtaining the parameter deviation value X at the current moment is:

[0167] X=a 1 -a 2 ;

[0168] In the formula, a 1 is the actual parameter value of the bank system front-end at the current moment, a 2 is the parameter prediction value corresponding to the current moment.

[0169] S5. See Figure 6 , set the first threshold Y according to the bank system front-end parameter prediction value 1 and the second threshold Y 2 , based on the current parameter deviation value X and the first threshold value Y 1 and the second threshold Y 2 Perform dynamic operation and maintenance behaviors, obtain the operation and maintenance optimization plan of the bank system front-end machine at the current moment, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

[0170] In particular, the first threshold value Y is set according to the bank system front-end parameter prediction value 1 and the second threshold Y 2 ,in:

[0171] The first threshold value Y 1 and the second threshold Y 2 The calculation formula is:

[0172] Y 1=b+b×α%;

[0173] Y 2 =bb×β%;

[0174] Where b is the predicted value of the bank system front-end parameter, and α is the first threshold Y 1 The threshold amplitude factor, β is the second threshold Y 2 The threshold amplitude factor.

[0175] In particular, the current moment parameter deviation value X, the first threshold value Y 1 and the second threshold Y 2 Perform dynamic operation and maintenance to obtain the current bank system front-end machine operation and maintenance optimization plan. The steps to complete the dynamic operation and maintenance optimization management of the bank system front-end machine are as follows:

[0176] Step S41: Obtain the actual parameter value of the bank system front-end at the current moment, the parameter deviation value X at the current moment, and the first threshold value Y 1 and the second threshold Y 2 ;

[0177] Step S42: performing abnormality determination on the actual parameter values ​​of the bank system front-end processor at the current moment;

[0178] Step S43: updating the current banking system front-end operation and maintenance optimization plan according to the abnormality judgment result;

[0179] Step S44: completing dynamic operation and maintenance optimization management of the bank system front-end processor according to the current bank system front-end processor operation and maintenance optimization plan;

[0180] Furthermore, the step of determining the abnormality of the actual parameter value of the bank system front-end processor at the current moment is as follows:

[0181] If the actual parameter value of the bank system front-end at the current moment is between the first threshold value Y 1 and the second threshold Y 2 In the meantime, the bank system front-end machine refuses to perform dynamic operation and maintenance optimization management of the bank system front-end machine;

[0182] If the actual parameter value of the bank system front-end at the current moment is greater than or equal to the first threshold value Y 1 , then the current moment parameter deviation value X is recorded in the current moment parameter deviation value table, and the current moment banking system front-end operation and maintenance optimization plan is updated according to the current moment parameter deviation value table;

[0183] If the actual parameter value of the bank system front-end at the current moment is less than or equal to the second threshold value Y 2 The bank system front-end will issue parameter warnings and provide feedback to manual operation and maintenance personnel.

[0184] In some preferred embodiments, if there is a sudden major anomaly in the external environment of the bank system front-end machine, such as banking business transaction volume, user access volume, network environment, and external system call frequency, the bank system front-end machine adaptive operation and maintenance management will experience downtime, so it is necessary to optimize the operation and maintenance of the bank system front-end machine based on the abnormal situation.

[0185] In some preferred embodiments, at the current moment, the actual parameter value of the bank system front-end processor is greater than or equal to the first threshold value Y 1 In this case, the current parameter deviation value X is recorded in the current parameter deviation value table, and the current banking system front-end machine operation and maintenance optimization plan is updated according to the current parameter deviation value table. This scenario supplements the abnormal situation of the banking system front-end machine adaptive operation and maintenance management, and can eliminate the influence of the prediction error of the banking system front-end machine parameter prediction model on the accuracy of the overall banking system front-end machine adaptive operation and maintenance management, thereby improving the accuracy of the banking system front-end machine system operation and maintenance.

[0186] In some preferred embodiments, at the current moment, the actual parameter value of the bank system front-end processor is less than or equal to the second threshold value Y 2 , the bank system front-end machine issues parameter warnings and feeds back to the manual operation and maintenance personnel. At this moment, the actual parameter value of the bank system front-end machine is less than or equal to the second threshold value Y 2 There may be a situation where external environmental parameters such as banking business transaction volume, user access volume, network environment and external system call frequency are greatly abnormal, or even human interference occurs. In this case, manual operation and maintenance is required to determine whether the actual parameter value of the bank system front-end at the current moment is less than or equal to the second threshold value Y 2 reasons and make operational decisions.

[0187] This application uses the historical parameter data and historical operation and maintenance logs of the bank system front-end machine to train the bank system front-end machine parameter prediction model and the bank system front-end machine operation and maintenance decision model. Based on the prediction model and the operation and maintenance decision model, the bank system front-end machine parameters in the future time period can be predicted, and the bank system front-end machine operation and maintenance plan can be set and updated in advance. This application can perform adaptive operation and maintenance management on the bank system front-end machine and realize the automated operation and maintenance of the bank system front-end machine.

[0188] The adaptive operation and maintenance management of the bank system front-end is an operation and maintenance behavior based on the predicted parameters. It is difficult to cover all operation and maintenance situations in view of the complex and changeable banking business. In order to solve the above problems, this application makes an abnormal judgment on the actual parameter value of the bank system front-end at the current moment, and updates the operation and maintenance optimization plan of the bank system front-end at the current moment according to the abnormal judgment result, and completes the dynamic operation and maintenance optimization management of the bank system front-end. This method not only supplements the adaptive operation and maintenance management of the bank system front-end, but also records the operation and maintenance optimization plan of the bank system front-end under abnormal circumstances in the operation and maintenance log, continuously improving the accuracy of the operation and maintenance decision model of the bank system front-end, so as to further improve the accuracy of the automated operation and maintenance of the bank system front-end.

[0189] Example 2

[0190] Figure 7 This is a parameter-driven banking system front-end data system provided by an embodiment of the present application. The system as a whole can be divided into:

[0191] Data processing module: used to obtain the parameter data of the bank system front-end at the historical moment to establish the parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish the operation and maintenance behavior data set;

[0192] Parameter prediction model training module: used to obtain the bank system front-end parameter prediction model through time series prediction model training based on the parameter status data set;

[0193] Operation and maintenance decision model training module: used to obtain the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set;

[0194] Adaptive operation and maintenance management module: used to obtain the predicted values ​​of the bank system front-end machine parameters according to the bank system front-end machine parameter prediction model, generate a predicted parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management;

[0195] Parameter deviation detection module: used to obtain the actual parameter value of the bank system front-end at the current moment and perform parameter deviation detection with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment;

[0196] Dynamic operation and maintenance optimization management module: used to set the first threshold value Y according to the predicted value of the bank system front-end machine parameter 1 and the second threshold Y 2 , based on the current parameter deviation value X and the first threshold value Y 1 and the second threshold Y 2Perform dynamic operation and maintenance behaviors, obtain the operation and maintenance optimization plan of the bank system front-end machine at the current moment, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

[0197] Example 3

[0198] Figure 8 Schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 8 As shown, according to another aspect of the present application, an electronic device 100 is also provided. The electronic device 100 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the parameter-driven banking system front-end data system and method as described above may be executed.

[0199] The method or device according to the embodiment of the present application can also be used by Figure 8 The electronic device architecture shown in FIG. Figure 8 As shown, the electronic device 100 may include a bus 101, one or more CPUs 102, a read-only memory (ROM) 103, a random access memory (RAM) 104, a communication port 105 connected to a network, an input / output component 106, a hard disk 107, etc. The storage device in the electronic device 100, such as the ROM 103 or the hard disk 107, may store a parameter-driven bank system front-end data system and method provided in the present application. A parameter-driven bank system front-end data system and method may, for example, include the following steps: obtaining the bank system front-end parameter data at historical moments to establish a parameter status data set; obtaining the bank system front-end operation and maintenance logs at corresponding historical moments and performing feature processing to establish an operation and maintenance behavior data set; according to the parameter status data set, a bank system front-end parameter prediction model is obtained by training a time series prediction model; the time series prediction model is an end-to-end network model composed of a Transformer and a probability time series feature extraction module; the probability time series feature extraction module is composed of a mixed density network and a fully connected layer; according to The parameter state data set and the operation and maintenance behavior data set are used to obtain the bank system front-end operation and maintenance decision model through deep reinforcement learning training; according to the bank system front-end parameter prediction model, the bank system front-end parameter prediction value is obtained, and the prediction parameter state set is generated and input into the bank system front-end operation and maintenance decision model to obtain the bank system front-end operation and maintenance plan, and complete the bank system front-end adaptive operation and maintenance management; the actual parameter value of the bank system front-end at the current moment is obtained and the parameter deviation detection is performed with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment; the first threshold value Y is set according to the bank system front-end parameter prediction value 1 and the second threshold Y 2 , based on the current parameter deviation value X and the first threshold value Y1 and the second threshold Y 2 Perform dynamic operation and maintenance behaviors, obtain the operation and maintenance optimization plan of the bank system front-end machine at the current moment, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

[0200] Furthermore, the electronic device 100 may further include a user interface 108. Figure 8 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 8 One or more components of an electronic device are shown.

[0201] Example 4

[0202] Fig. 9 Schematic diagram of the computer-readable storage medium structure provided by an embodiment of the present application. Fig. 9 As shown, a computer-readable storage medium 200 according to an embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 200. When the computer-readable instructions are executed by the processor, a parameter-driven bank system front-end data system and method according to an embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium 200 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0203] In addition, according to the implementation of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be run by a processor to execute instructions corresponding to the method steps provided by the present application, and when the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0204] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0205] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0206] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0207] The above preset parameters or preset thresholds are all set by technicians in this field according to actual conditions or obtained through large-scale data simulation.

[0208] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A parameter-driven banking system front-end data method, characterized in that: The following steps are involved: S1. Obtain the parameter data of the bank system front-end at the historical moment to establish a parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish an operation and maintenance behavior data set; S2. According to the parameter state data set, a bank system front-end parameter prediction model is obtained through time series prediction model training; the time series prediction model is an end-to-end network model composed of Transformer and probability time series feature extraction modules; the probability time series feature extraction module is composed of a mixed density network and a fully connected layer; according to the parameter state data set and the operation and maintenance behavior data set, a bank system front-end operation and maintenance decision model is obtained through deep reinforcement learning training; S3. According to the bank system front-end machine parameter prediction model, obtain the bank system front-end machine parameter prediction value, generate the prediction parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management; S4, obtaining the actual parameter value of the bank system front-end at the current moment and performing parameter deviation detection on the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment; S5. Set the first threshold value Y1 and the second threshold value Y2 according to the predicted value of the bank system front-end machine parameter, perform dynamic operation and maintenance based on the current parameter deviation value X, the first threshold value Y1 and the second threshold value Y2, obtain the current bank system front-end machine operation and maintenance optimization plan, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

2. A parameter-driven banking system front-end data method according to claim 1, characterized in that: The parameter state data set is established by obtaining the parameter data of the bank system front-end machine at the historical moment; the operation and maintenance log of the bank system front-end machine at the corresponding historical moment is obtained and feature processed to establish the operation and maintenance behavior data set, wherein: The bank system front-end machine parameter data at the historical moment includes system operation parameters and external environment parameters; The system operation parameters include system resource allocation parameters, thread pool parameters, connection pool parameters and concurrent processing parameters; the external environment parameters include banking business transaction volume, user access volume, network environment and external system call frequency; The bank system front-end operation and maintenance log corresponding to the historical moment includes the system operation parameter operation and maintenance log and the external environment parameter operation and maintenance log; The system operation parameter operation and maintenance log includes resource allocation operation records, thread pool management operation records, connection pool management operation records and concurrent processing optimization operation records; the external environment parameter operation and maintenance log includes external environment adaptation operation records; The feature processing includes operation type recognition, time feature extraction, operation amplitude quantification and operation effect quantification; The operation and maintenance behavior data set includes operation type characteristics, time characteristics, operation amplitude characteristics and operation effect characteristics; The quantification of the operational effect is based on the maximization of resource utilization of the bank system front-end machine as a quantitative indicator.

3. A parameter-driven banking system front-end data method according to claim 2, characterized in that: The steps of obtaining the bank system front-end parameter prediction model through time series prediction model training according to the parameter status data set are: Step S11: Preprocessing the parameter state data set to obtain a parameter prediction data set; Step S12: dividing the parameter prediction data set into a training set and a validation set, inputting the data set into a time series prediction model for training, and obtaining a bank system front-end parameter prediction model; The preprocessing operation includes dividing the time step, denoising, standardization and counting the probability of parameters in different value intervals within the unit time step; The standardization process adopts the Z-Score standardization process method; The bank system front-end parameter prediction model is obtained by training the time series prediction model through the parameter prediction data set; the time series prediction model is an end-to-end network model composed of Transformer and probabilistic time series feature extraction modules, including Embedding, position encoding, probabilistic time series feature extraction module, encoder, decoder and logistic regression layer submodules; The probability time series feature extraction module is composed of a mixed density network and a fully connected layer; the mixed density network is composed of an input layer, a hidden layer, a preprocessing layer and an output layer; The time series prediction model training phase adopts a training strategy combining a pre-trained probability time series feature extraction module and a joint training time series prediction model; The prediction result of the bank system front-end parameter prediction model is the probability of the parameter appearing in different value ranges within the future unit time step.

4. A parameter-driven banking system front-end data method according to claim 3, characterized in that: The steps of obtaining the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set are: Step S21: Based on the parameter status data set and the operation and maintenance behavior data set, construct an expert experience sequence A(s t , a t , r, s t+1 ); the t is the parameter state set, a t is the set of operation and maintenance actions, r is the set of operation effect reward values, s t+1 is the parameter state set for the next moment; Step S22: constructing an expert experience pool based on the expert experience sequence; Step S23: Using the expert experience pool, train the deep reinforcement learning model to obtain the banking system front-end operation and maintenance decision model; The steps of using the expert experience pool to train the deep reinforcement learning model and obtain the bank system front-end operation and maintenance decision model are as follows: Step S31: constructing a prediction network and a target network using a BP neural network; Step S32: constructing a deep reinforcement learning model based on the prediction network, the target network and the experience pool; Step S33: Initialize the experience pool to an expert experience pool; the experience pool capacity is N; Step S34: Initialize the prediction network using random prediction network weights θ and use random target network weights θ * =θ initialize the target network and initialize the parameter state set s t , batch n and step size m; Step S35: Randomly obtain expert experience sequence A(s) in batches n from the experience pool t , a t , r, s t+1 ) for training, where t is a variable, n <N; Prediction network input parameter state set s t and operation and maintenance action set a t , output prediction value function Q(s t , a t |θ) and the next moment operation and maintenance prediction action set a t+1 * ; The target network input parameter state set s at the next moment t+1 , output the target value function F, the calculation formula of F is: In the formula, For the parameter state set s t Time operation and maintenance action set a t The operation effect reward value, γ is the discount factor and γ∈(0,1), maxQ(s t+1 , a t+1 * ,θ * ) is the parameter state set s at the next moment t+1 When the next moment operation and maintenance prediction action set a is selected t+1 * The maximum predicted value function obtained; Step S36: Every m steps, the weight parameter θ of the prediction network is copied to the weight parameter θ of the target network. * ; Step S37: Use the gradient descent method to minimize the MSE loss function and update the network model parameters θ and θ * Until convergence, the front-end operation and maintenance decision model of the banking system is obtained; The BP neural network consists of an input layer, a hidden layer and an output layer; The number of neurons in the input layer is equal to the number of parameter states in the parameter state set; The number of neurons in the output layer is equal to the number of operation and maintenance actions in the operation and maintenance action set.

5. A parameter-driven banking system front-end data method according to claim 4, characterized in that: The actual parameter value of the bank system front-end at the current moment is obtained and the parameter deviation detection is performed with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment, where: The parameter deviation detection only detects the deviation of system operation parameters; The parameter deviation detection adopts a difference calculation method; The calculation formula for obtaining the parameter deviation value X at the current moment is: X = a1-a2; Wherein, a1 is the actual parameter value of the bank system front-end at the current moment, and a2 is the predicted parameter value corresponding to the current moment.

6. A parameter-driven banking system front-end data method according to claim 5, characterized in that: The first threshold value Y1 and the second threshold value Y2 are set according to the predicted value of the bank system front-end parameter, wherein: The calculation formulas for the first threshold value Y1 and the second threshold value Y2 are: Y1=b+b×α%; Y2=b―b×β%; Wherein, b is the predicted value of the bank system front-end parameter, α is the threshold amplitude factor of the first threshold value Y1, and β is the threshold amplitude factor of the second threshold value Y2.

7. A parameter-driven banking system front-end data method according to claim 6, characterized in that: The steps of performing dynamic operation and maintenance behavior based on the current parameter deviation value X, the first threshold value Y1 and the second threshold value Y2 to obtain the current banking system front-end machine operation and maintenance optimization plan and completing the dynamic operation and maintenance optimization management of the banking system front-end machine are as follows: Step S41: obtaining the actual parameter value of the bank system front-end processor at the current moment, the parameter deviation value X at the current moment, the first threshold value Y1 and the second threshold value Y2; Step S42: performing abnormality determination on the actual parameter values ​​of the bank system front-end processor at the current moment; Step S43: updating the current banking system front-end operation and maintenance optimization plan according to the abnormality judgment result; Step S44: completing dynamic operation and maintenance optimization management of the bank system front-end processor according to the current bank system front-end processor operation and maintenance optimization plan; The steps of judging abnormality of the actual parameter value of the bank system front-end processor at the current moment are as follows: If the actual parameter value of the bank system front-end processor at the current moment is between the first threshold value Y1 and the second threshold value Y2, the bank system front-end processor refuses to perform dynamic operation and maintenance optimization management of the bank system front-end processor; If the actual parameter value of the bank system front-end at the current moment is greater than or equal to the first threshold value Y1, the parameter deviation value X at the current moment is recorded in the parameter deviation value table at the current moment, and the operation and maintenance optimization plan of the bank system front-end at the current moment is updated according to the parameter deviation value table at the current moment; If the actual parameter value of the bank system front-end processor at the current moment is less than or equal to the second threshold value Y2, the bank system front-end processor issues a parameter warning and feeds back to the manual operation and maintenance personnel.

8. A parameter-driven bank system front-end data system, which is implemented based on a parameter-driven bank system front-end data method according to any one of claims 1 to 7, characterized in that: The system comprises: Data processing module: used to obtain the parameter data of the bank system front-end at the historical moment to establish the parameter status data set; obtain the operation and maintenance log of the bank system front-end at the corresponding historical moment and perform feature processing to establish the operation and maintenance behavior data set; Parameter prediction model training module: used to obtain the bank system front-end parameter prediction model through time series prediction model training based on the parameter status data set; Operation and maintenance decision model training module: used to obtain the bank system front-end operation and maintenance decision model through deep reinforcement learning training based on the parameter status data set and the operation and maintenance behavior data set; Adaptive operation and maintenance management module: used to obtain the predicted values ​​of the bank system front-end machine parameters according to the bank system front-end machine parameter prediction model, generate a predicted parameter state set, input it into the bank system front-end machine operation and maintenance decision model, obtain the bank system front-end machine operation and maintenance plan, and complete the bank system front-end machine adaptive operation and maintenance management; Parameter deviation detection module: used to obtain the actual parameter value of the bank system front-end at the current moment and perform parameter deviation detection with the parameter prediction value corresponding to the current moment to obtain the parameter deviation value X at the current moment; Dynamic operation and maintenance optimization management module: used to set the first threshold value Y1 and the second threshold value Y2 according to the predicted value of the bank system front-end machine parameters, perform dynamic operation and maintenance based on the current parameter deviation value X, the first threshold value Y1 and the second threshold value Y2, obtain the current bank system front-end machine operation and maintenance optimization plan, and complete the dynamic operation and maintenance optimization management of the bank system front-end machine.

9. An electronic device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of a parameter-driven banking system front-end data method as described in any one of claims 1-7.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in a parameter-driven banking system front-end data method as described in any one of claims 1-7.

Images