Bank note adding information determination method and device of bank self-service equipment, equipment and medium

Abstract

The invention discloses a bank note adding information determination method and device of bank self-service equipment, equipment and a medium. Comprising the following steps: acquiring a historical transaction amount of bank self-service equipment of a target equipment type in each period in a preset historical time period; establishing a plurality of transaction sample pairs according to the historical transaction amount of each period; performing sliding sampling on the transaction sample pair based on a predetermined sample window size to obtain a sample space corresponding to each iteration; performing iterative calculation on the intermediate parameters of the transaction predictor according to the sample space and the kernel function; and performing transaction amount prediction on the bank self-service equipment according to the transaction predictor after iteration is completed, and determining money adding information according to a transaction amount prediction result. Predictor parameters are adaptively updated in an iteration mode, and good prediction precision is achieved with small calculation complexity; and for equipment needing to add money, money adding information is determined according to recent transaction amount data, so that money depositing and withdrawing services are balanced to the greatest extent, and meanwhile, waste of human resources is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method, device, equipment and medium for determining cash replenishment information of bank self-service equipment. Background Art

[0002] With the development of banking business, the types of self-service equipment in bank branches are increasing, such as ATMs, super counters, intelligent low counters, etc., which greatly diverts the counter business and improves the processing efficiency of banking business and customer experience. The inventory management of self-service equipment needs to balance two key points: one is that too little or too much cash inventory will affect the normal operation of cash business. Too little cash inventory will cause customers unable to withdraw money, and too much inventory will cause the cash bin to be full and unable to deposit money, affecting customer satisfaction; the other is that frequent or excessive cash replenishment will increase the transportation and storage costs, and at the same time, it is easy to cause frequent handover of cash between the warehouse management team and the cash replenishment team during the cash replenishment process, increasing the operation pressure and reducing the economic benefits of the bank.

[0003] Driven by the digital wave, the banking industry is experiencing unprecedented transformation and upgrading challenges. At the same time, with the rapid progress of computer technology and deep learning, artificial intelligence has penetrated into all fields of traditional industries, bringing unprecedented development opportunities to the banking industry. Applying prediction algorithms to the field of bank self-service equipment, predicting the deposit and withdrawal transaction amounts of equipment in the next cycle, and adaptively generating cash replenishment plans can greatly reduce operating costs and improve service efficiency. In the prior art, machine learning algorithms are mostly used for prediction, but the algorithms based on machine learning require a large number of samples for training, and the number of bank self-service equipment is huge, and the training cost is too high. Summary of the Invention

[0004] The present invention provides a method, device, equipment and medium for determining cash replenishment information of bank self-service equipment to solve the problem of low efficiency in predicting transaction amounts of bank self-service equipment.

[0005] According to one aspect of the present invention, there is provided a method for determining cash replenishment information of bank self-service equipment, including:

[0006] Obtaining the historical transaction amounts of each cycle of bank self-service equipment of the target equipment type in a preset historical time period; wherein, the historical transaction amounts include historical withdrawal amounts and historical deposit amounts;

[0007] Establishing a plurality of transaction sample pairs according to the historical transaction amounts of each cycle; wherein, each transaction sample pair includes the corresponding relationship between the input historical transaction data of a plurality of first cycles and the output historical transaction data of a second cycle; the first cycle is a preset number of cycles before the second cycle; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs;

[0008] Perform sliding sampling on the transaction sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration;

[0009] Iteratively calculate the intermediate parameters of the transaction predictor according to the sample space and the kernel function; wherein, the transaction predictor includes a withdrawal predictor and a deposit predictor;

[0010] Predict the transaction amount of the bank self-service device according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

[0011] According to another aspect of the present invention, there is provided a device for determining cash replenishment information of a bank self-service device, including:

[0012] A historical transaction amount determination module, configured to obtain the historical transaction amounts of each cycle of the bank self-service device of the target device type within a preset historical time period; wherein, the historical transaction amounts include historical withdrawal amounts and historical deposit amounts;

[0013] A transaction sample pair establishment module, configured to establish a plurality of transaction sample pairs according to the historical transaction amounts of each cycle; wherein, each transaction sample pair includes the corresponding relationship between the input historical transaction data of a plurality of first cycles and the output historical transaction data of a second cycle; the first cycle is a preset number of cycles before the second cycle; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs;

[0014] An iterative sample space determination module, configured to perform sliding sampling on the transaction sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration;

[0015] A predictor iterative calculation module, configured to iteratively calculate the intermediate parameters of the transaction predictor according to the sample space and the kernel function; wherein, the transaction predictor includes a withdrawal predictor and a deposit predictor;

[0016] A prediction module, configured to predict the transaction amount of the bank self-service device according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

[0017] According to another aspect of the present invention, there is provided an electronic device, the electronic device includes:

[0018] At least one processor; and

[0019] A memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program executable by the at least one processor. When executed by the at least one processor, the computer program enables the at least one processor to execute the method for determining cash replenishment information of a bank self-service device according to any embodiment of the present invention.

[0021] According to another aspect of the present invention, there is provided a computer-readable storage medium storing computer instructions for causing a processor to implement the method for determining cash replenishment information of a bank self-service device according to any embodiment of the present invention when executed.

[0022] The technical solution of this embodiment avoids the matrix inversion process with high computational cost based on the kernel recursive least squares algorithm, adaptively updates the predictor parameters in an iterative manner, and achieves good prediction accuracy with relatively low computational complexity. For devices that need cash replenishment, the cash replenishment information is determined based on recent transaction amount data, maximizing the balance of deposit and withdrawal operations while reducing waste of human resources.

[0023] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 is a flowchart of a method for determining cash replenishment information of a bank self-service device according to an embodiment of the present invention;

[0026] Figure 2 is a flowchart of another method for determining cash replenishment information of a bank self-service device according to an embodiment of the present invention;

[0027] Figure 3 is a schematic structural diagram of a device for determining cash replenishment information of a bank self-service device according to an embodiment of the present invention;

[0028] Figure 4 is a schematic structural diagram of an electronic device for implementing the method for determining cash replenishment information of a bank self-service device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] To enable those skilled in the art to better understand the solution 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 accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0030] It should be noted that the terms "candidate", "target", etc. in the specification and claims of the present invention and the above drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances 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 inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0031] Figure 1 This embodiment of the present invention provides a flowchart of a method for determining cash replenishment information of a bank self-service device. This embodiment is applicable to the situation of predicting the cash replenishment amount of a bank self-service device. This method can be executed by a device for determining cash replenishment information of a bank self-service device. The device for determining cash replenishment information of a bank self-service device can be implemented in the form of hardware and / or software, and the device for determining cash replenishment information of a bank self-service device can be configured in a bank self-service device. As Figure 1 shown, the method includes:

[0032] S110. Obtain the historical transaction amounts of each cycle of the bank self-service device of the target device type within a preset historical time period.

[0033] Among them, the historical transaction amount includes the historical withdrawal amount and the historical deposit amount.

[0034] Since there are many types of self-service devices in the bank that carry out cash business, and the differences in the deposit and withdrawal transaction volumes of each type of device are relatively large, the transaction predictor designed in the present invention separately processes the transaction data of different device types to obtain transaction predictors corresponding to different device types. Specifically, before initializing the predictor, determine the target device type, collect the historical transaction amounts of multiple bank self-service devices of the target device type within a preset historical time period, and statistically analyze the historical transaction amounts within the preset historical time period according to the preset cycle length to obtain the historical transaction amounts of each cycle.

[0035] Exemplarily, the preset cycle length can be one day, two days, or one week, etc., and can be determined according to the situation of the network where the bank self-service device is located. There is no limitation on the preset cycle length here. Obtain the historical withdrawal amount and historical deposit amount of the bank self-service device within the past month, and conduct statistics according to the preset cycle length. Taking one day as an example, determine the historical withdrawal amount and historical deposit amount of each day as the historical transaction amount of each cycle. For example, collect the deposit amount data and withdrawal amount data of N bank self-service devices for T days. The deposit amount A of a certain bank self-service device for T days D =[A D 1,A D 2,…A D T , and the withdrawal amount of this bank self-service device for T days is A W =[A W 1,A W 2,…A W T .

[0036] S120. Establish multiple transaction sample pairs according to the historical transaction amounts of each cycle.

[0037] Among them, each transaction sample pair includes the corresponding relationship between multiple input historical transaction data of the first cycle and output historical transaction data of the second cycle; the first cycle is a preset number of cycles before the second cycle; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs.

[0038] There is a certain correlation in the historical transaction data of the bank self-service device, that is, the transaction data within a certain period of history can reflect the characteristics of the transaction data after that period of time. However, the data before that period of history is too long from the time after that period of history, resulting in the data before that period of history being unable to accurately reflect the characteristics of the transaction data after that period of history. Therefore, in the embodiments of the present invention, the influence relationship between the historical transaction amounts of each cycle is determined according to the preset predictor order, and then the transaction sample pairs are determined according to this influence relationship. Among them, the predictor order refers to the model order of the predictor, that is, the number of parameters or the model complexity required to describe the dynamic behavior of the system. In the present invention, the predictor order refers to the number of observed values of past time points included in the prediction model, which is used to predict the value at the current time point, that is, it is considered that the deposit and withdrawal amounts in the next cycle are a linear combination of the deposit and withdrawal amounts in the previous several cycles.

[0039] Specifically, determine the predictor order m. The model order being m means that the historical transaction amount in the nth period is related to the historical transaction amounts in the n - m to n - 1 periods, and is not related to the historical transaction amounts before the n - m period, which means predicting the transaction amount in the next period based on the historical transaction amounts in the previous m periods. That is, the input historical transaction data in multiple first periods in the transaction sample pair is the historical transaction amounts in the n - m to n - 1 periods, and the corresponding output historical transaction data in the second period is the historical transaction amount in the nth period. Exemplarily, establish transaction sample pairs for the historical withdrawal amount and the historical deposit amount respectively. Based on the above example, for the historical deposit amount, a deposit sample pair can be constructed: {u(i), y(i)} M , where: u(i) = [A D i A D i+1 …A D i+m-1 , y(i) = A D i+m . For the historical withdrawal amount, a withdrawal sample pair can be constructed: {u(i), y(i)} M , where u(i) = [A W i A W i+1 …A W i+m-1 , y(i) = A W i+m . For example, if m = 3, then u(1) = [A D 1A D 2A D 3], y(1) = A D 4; u(2) = [A D 2A D 3A D4 , y(2) = A D 5; and so on, constructing transaction sample pairs for the historical transaction amounts of all periods.

[0040] S130. Perform sliding sampling on the transaction sample pairs based on a pre - determined sample window size to obtain the sample space corresponding to each iteration.

[0041] Among them, the sample window size is the maximum number of sample pairs used in one iteration calculation. When performing sliding sampling, the initial sample space is (u(1), y(1)), and the first iteration calculation is performed using the initial sample space; in the second iteration, new transaction sample pairs (u(1), y(1)), (u(2), y(2)) are added to the sample space, and it is determined whether the number of transaction sample pairs in the current sample space is greater than the sample window size. If it is less, the current sample space is determined as the second sample space, and the second iteration calculation is performed using this second sample space; if it is greater, the transaction sample pair with the longest preservation time in the current sample space is deleted to obtain the second sample space, so as to ensure that the number of transaction sample pairs in the sample space after deletion is equal to the sample window size.

[0042] Exemplarily, the sample space with new transaction sample pairs added during the iteration process has reached the sample window size. In order to add new samples to the sample space while keeping the sample space size unchanged, it is necessary to discard the outdated samples, that is, delete the transaction sample pair with the longest preservation time in the sample space. Assume that the sample with the longest preservation time in the sample space is u(l), and the kernel dictionary is updated as D(i) = D(i - 1) \ u(l), where "\ " represents deletion.

[0043] S140. Iteratively calculate the intermediate parameters of the transaction predictor according to the sample space and the kernel function; among them, the transaction predictor includes a withdrawal predictor and a deposit predictor.

[0044] Extract the input features of the input historical transaction data in the sample space as the first intermediate parameter through the kernel function, and determine the input-output correlation feature as the second intermediate parameter according to the first intermediate parameter and the output historical transaction data in the sample space. Determine the third intermediate parameter according to the input historical transaction data in the sample space and the input historical transaction data of the next transaction sample pair in the sample space; as the sample space is continuously updated, the first intermediate parameter, the second intermediate parameter, and the third intermediate parameter are continuously updated until the sliding sampling ends, and the transaction predictor is determined according to the first intermediate parameter, the second intermediate parameter, and the third intermediate parameter corresponding to the last iteration.

[0045] Specifically, perform sliding sampling on the withdrawal sample pairs based on the pre-determined sample window size to obtain the withdrawal sample space corresponding to each iteration; iteratively calculate the intermediate parameters of the withdrawal predictor according to the withdrawal sample space and the kernel function to obtain the final withdrawal predictor; perform sliding sampling on the deposit sample pairs based on the pre-determined sample window size to obtain the deposit sample space corresponding to each iteration; iteratively calculate the intermediate parameters of the deposit predictor according to the deposit sample space and the kernel function to obtain the final deposit predictor.

[0046] When using the kernel recursive least squares algorithm to iteratively calculate the intermediate parameters of the trading predictor based on the sample space and the kernel function, in view of the problem that the network structure of the traditional kernel recursive least squares algorithm grows with the increase of samples, resulting in an increasing amount of computation when processing continuously arriving signals, this embodiment proposes to window the sample space to control the number of samples used in each iterative prediction, limit the size of the network structure, and improve the iterative calculation efficiency of the trading predictor.

[0047] The kernel recursive least squares algorithm is an online kernel regression algorithm. Its main feature is that it can process one sample at a time and construct a training dictionary to approximately approximate the function. This algorithm utilizes the "kernel trick" and does not need to explicitly know the mapping of the training samples in the feature space. Instead, it directly calculates the inner product in the feature space through the kernel function, thus simplifying the entire calculation process. By using the matrix inversion lemma to simplify the process of inverting the kernel matrix, it effectively reduces the algorithm complexity and prediction efficiency, enabling it to process the sample sequence online in real time. The kernel recursive least squares algorithm, based on the idea of SVM, maps the sample data to a high-dimensional space using Mercer's theorem and performs linear regression in this high-dimensional feature space. The linear regression process uses the idea of RLS for iteration, thus reducing the amount of computation and ensuring the performance of nonlinear prediction. It is applicable to predicting the deposit and withdrawal transaction amounts of self-service devices.

[0048] In a feasible embodiment, S140 includes:

[0049] Determine the initial sample space corresponding to the first iteration, and determine the initial intermediate matrix according to the calculation result of the kernel function for the input historical transaction data in the transaction sample pairs within the initial sample space; determine the initial coefficient according to the initial intermediate matrix and the output historical transaction data in the transaction sample pairs within the initial sample space; determine the initial kernel vector according to the calculation result of the kernel function for the input historical transaction data in the transaction sample pairs within the initial sample space and the input historical transaction data in the next transaction sample pair corresponding to the initial sample space.

[0050] Determine the predicted value corresponding to the first iteration according to the initial kernel vector and the initial coefficient, and determine the prediction error corresponding to the first iteration according to the predicted value and the output historical transaction data in the next transaction sample pair corresponding to the initial sample space.

[0051] Determine the updated sample space corresponding to the subsequent iterations, and iteratively update the initial intermediate matrix, the initial coefficient, and the initial kernel vector according to the updated sample space and the prediction error to obtain the updated intermediate matrix, the updated coefficient, and the updated kernel vector.

[0052] At the initial iteration, a kernel function is selected. Let κ(·) denote the kernel function. Commonly used kernel functions include the Gaussian radial basis kernel function, the linear kernel function, and the polynomial basis. Let i = 1. Determine the initial intermediate parameters of the transaction predictor according to the initial sample space, including the initial intermediate matrix Q(1), the initial coefficient α(1), the kernel vector k(2), and the sample space D(1).

[0053] Specifically, add [u(1)] to the sample space, i.e., D(1) = (u(1), y(1)), and Q(1) = [κ(u(1), u(1))] -1 , α(1) = Q(1)y(1), the kernel vector k(2) = [κ(u(2), u(1))], and output the predicted value of the first iteration The prediction error corresponding to the first iteration Exemplarily, take the historical transaction amounts corresponding to the period A1 - A3 as u(1), and the historical transaction amount corresponding to A4 as y(1); take the historical transaction amounts corresponding to A2 - A4 as u(2), and the calculated corresponding to the predicted value of the transaction amount corresponding to A5. Take the difference between the predicted value and the historical transaction amount corresponding to the real A5 as the prediction error calculated in the first iteration. That is, the intermediate parameters corresponding to the i-th iteration are the intermediate matrix Q(i), the coefficient α(i), the kernel vector k(i + 1), the sample space D(i), the prediction error e(i + 1), and the predicted value

[0054] Add the new transaction sample pair (u(i), y(i)) to the kernel dictionary, i.e., D(i) = [D(i - 1)(u(i), y(i))], where D(i - 1) is the sample space corresponding to the previous iteration. Perform iterative calculations according to the above process to continuously update the intermediate parameters.

[0055] S150. Predict the transaction amount of the bank self-service equipment according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

[0056] Determine the intermediate parameters corresponding to the last iteration according to the transaction predictor after the iteration is completed, including the target intermediate matrix, the target coefficient, and the target sample space. Determine the corresponding prediction input value according to the current prediction period. Determine the prediction kernel vector according to the calculation result of the input historical transaction data and the prediction input value in the transaction sample pair in the last sample space by the kernel function; determine the transaction amount prediction result according to the prediction kernel vector and the target coefficient, and determine the cash replenishment information according to the transaction amount prediction result. For example, if the predicted result of the withdrawal amount in the transaction amount prediction result is greater than the predicted result of the deposit amount, then perform cash replenishment, and the cash replenishment amount is the difference between the predicted result of the withdrawal amount and the predicted result of the deposit amount.

[0057] Further, determine the actual transaction amount corresponding to the current prediction period, determine the current prediction error according to the prediction input value and the actual transaction amount, further update the target sample space according to the corresponding prediction input value and actual transaction amount of the current prediction period, and then continue to update the target intermediate matrix, target coefficient, and target kernel vector according to the updated target sample space and the current prediction error, so as to predict the transaction amount for the next prediction period according to the updated target intermediate matrix, target coefficient, and target kernel vector.

[0058] By continuously updating the predicted value and the actual value, the transaction predictor is continuously updated. And since the transaction sample pairs in the iteration process are determined based on the mapping relationship and the size of the sample space in the iteration process is fixed, the timeliness and accuracy of the prediction are guaranteed.

[0059] In a feasible embodiment, S150 includes:

[0060] Use the historical transaction data of multiple periods before the prediction period as the input value of the transaction predictor to obtain the transaction amount prediction result corresponding to the prediction period; wherein, the transaction amount prediction result includes the deposit amount prediction result and the withdrawal amount prediction result;

[0061] Determine whether the bank self-service device is a cash replenishment device according to the deposit amount prediction result and the withdrawal amount prediction result;

[0062] If so, determine the cash replenishment amount according to the total capacity of the cash cassettes of the bank self-service device, the deposit amount prediction result, and the withdrawal amount prediction result;

[0063] Among them, the cash replenishment amount is determined according to the following formula:

[0064]

[0065] Wherein, A add represents the cash replenishment amount, C represents the total capacity of the cash cassettes of the bank self-service device, a represents the preset cash replenishment coefficient, A W represents the withdrawal amount prediction result corresponding to the preset period, and A D represents the deposit amount prediction result corresponding to the preset period.

[0066] Specifically, the transaction predictor is iteratively calculated based on the historical transaction amounts from the first period to the T-th period, with a prediction period of T + 1. The input value of the transaction predictor is determined according to the number of the first period in the transaction sample pair, that is, the historical transaction amounts from the (T - m + 1)-th period to the T-th period are used as the input value of the transaction predictor. Referring to the calculation method of the predicted value corresponding to each iteration in the above iterative process, the transaction amount prediction result is calculated. Among them, the deposit amount prediction result is calculated according to the deposit input value and the deposit predictor, and the withdrawal amount prediction result is calculated according to the withdrawal input value and the withdrawal predictor.

[0067] Determine that the deposit amount prediction result for the future prediction period is The withdrawal amount prediction result is Set the cash replenishment threshold X, where X represents the threshold of the cash box amount for maintaining basic cash operations. The current balance of the cash box of the device is A, and the predicted value of the device balance for the prediction period is calculated. If Then it is considered that the bank self-service device will be out of cash in the prediction period, and the device is determined as a cash replenishment device, and a cash replenishment plan is generated for the device. Specifically, the cash replenishment amount can be calculated in the following way: Assume that the total capacity of the cash box of the device is C, then the cash replenishment plan amount corresponding to this prediction period is:

[0068] Based on the predicted deposit and withdrawal amounts, screen the devices that need cash replenishment, and formulate a differentiated cash replenishment plan according to the recent deposit and withdrawal transaction amounts of the devices to balance the deposit and withdrawal operations as much as possible.

[0069] The technical solution of this embodiment avoids the matrix inversion process with high computational cost through the kernel recursive least squares algorithm, adaptively updates the predictor parameters in an iterative manner, and achieves good prediction accuracy with relatively low computational complexity; and for the devices that need cash replenishment, the cash replenishment information is determined according to the recent transaction amount data, balancing the deposit and withdrawal operations to the greatest extent and reducing the waste of human resources at the same time.

[0070] Figure 2 It is a flowchart of another method for determining the cash replenishment information of a bank self-service device provided by an embodiment of the present invention. In this embodiment, the iterative update process of the predictor in the above embodiment is further refined. As Figure 2 shown, the method includes:

[0071] S201. Obtain the historical transaction amounts of each period of the bank self-service device of the target device type in a preset historical time period.

[0072] S202. Establish multiple transaction sample pairs according to the historical transaction amounts of each period.

[0073] S203. Perform sliding sampling on the trading sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration.

[0074] S204. Determine the initial sample space corresponding to the first iteration. Determine the initial intermediate matrix based on the calculation results of the input historical trading data in the trading sample pairs within the initial sample space according to the kernel function; determine the initial coefficient based on the initial intermediate matrix and the output historical trading data in the trading sample pairs within the initial sample space; determine the initial kernel vector based on the calculation results of the input historical trading data in the trading sample pairs within the initial sample space and the input historical trading data in the next trading sample pair corresponding to the initial sample space according to the kernel function.

[0075] S205. Determine the predicted value corresponding to the first iteration according to the initial kernel vector and the initial coefficient, and determine the prediction error corresponding to the first iteration based on the predicted value and the output historical trading data in the next trading sample pair corresponding to the initial sample space.

[0076] S206. Determine the updated sample space corresponding to subsequent iterations, and determine whether there are discarded trading sample pairs in the updated sample space.

[0077] If the sample space with newly added trading sample pairs discards outdated samples because it reaches the sample window size, it is determined that there are discarded trading sample pairs in this updated sample space. Specifically, refer to the description in step 140 of the above-mentioned embodiment.

[0078] S207. If there are no discarded trading sample pairs, determine the forgetting matrix corresponding to the current iteration according to the forgetting factor, and determine the updated intermediate matrix and updated coefficient corresponding to the current iteration based on the forgetting matrix, the updated sample space, and the intermediate matrix, coefficient, kernel vector, and prediction error corresponding to the previous iteration.

[0079] The forgetting factor is a parameter introduced to eliminate data saturation. It is considered that the samples closer to the prediction time have greater value. To strengthen the influence of the most recent data and reduce the influence of historical data, introducing this parameter can accelerate the convergence of the model and improve the accuracy of the algorithm.

[0080] If there are no discarded trading sample pairs, directly perform the calculation of the current iteration based on the intermediate matrix, coefficient, kernel vector corresponding to the previous iteration, and the updated sample space. Since the updated sample space includes at least two trading sample pairs, determine the forgetting matrix brought by the time gap of at least two trading sample pairs according to the forgetting factor, and in the iterative calculation, determine the updated intermediate matrix and updated coefficient corresponding to the current iteration in combination with this forgetting matrix.

[0081] In a feasible embodiment, determining the updated intermediate matrix and updated coefficient corresponding to the current iteration according to the forgetting matrix, updated sample space, and the intermediate matrix, coefficient, kernel vector, and prediction error corresponding to the previous iteration includes:

[0082] Determining the updated intermediate matrix and updated coefficient corresponding to the current iteration according to the following formula:

[0083]

[0084] where β represents the forgetting factor, Q(i - 1) represents the intermediate matrix corresponding to the previous iteration, Q(i) represents the updated intermediate matrix corresponding to the current iteration, B(i - 1) represents the forgetting matrix corresponding to the previous iteration, k(i) represents the kernel vector corresponding to the previous iteration, κ(·) represents the kernel function, u(i) represents the input historical transaction data of the first period of the newly added transaction sample pair in the updated sample space in the current iteration, α(i) represents the updated coefficient corresponding to the current iteration, α(i - 1) represents the coefficient corresponding to the previous iteration, and e(i) represents the prediction error corresponding to the previous iteration.

[0085] The newly added transaction sample pair in the updated sample space corresponding to the current iteration is {u(i), y(i)}, the new input vector is u(i + 1), and it is judged whether the current updated sample space is less than the sample sliding window size M. If so, the new transaction sample pair is added to the sample space, that is, D(i) = [D(i - 1)(u(i), y(i))], and the prediction error e(i) = y(i) - k(i) corresponding to the previous iteration is calculated according to y(i) T α(i - 1), and since the number of elements in the sample space increases, the forgetting matrix is updated B(1) = [1], and the reference matrix is calculated: The intermediate matrix is updated according to the reference matrix The coefficient is updated: And the predicted value corresponding to the current iteration is output

[0086] S208. Determine the updated kernel vector corresponding to the current iteration according to the calculation results of the input historical transaction data in the transaction sample pairs in the updated sample space and the input historical transaction data in the next transaction sample pair corresponding to the updated sample space by the kernel function.

[0087] The input historical transaction data in the transaction sample pairs in the current updated sample space is u(l), u(l + 1), …, u(L), and L - l + 1 is the number of transaction sample pairs in the current updated sample space. Then the updated kernel vector k(i + 1) corresponding to the current iteration = [κ(u(l), u(i + 1)) κ(u(l + 1), u(i + 1)) … κ(u(L), u(i + 1))]T 。

[0088] S209. If there are discarded transaction sample pairs, modify the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous iteration.

[0089] Since discarded sample pairs are generated in the updated sample space due to the newly added transaction sample pairs in the current iteration, when determining the result of the current iteration based on the result of the previous iteration, in order to improve the accuracy of the iteration, it is necessary to delete and modify the results caused by the discarded transaction samples in the result of the previous iteration. Specifically, reduce the weight of the factors caused by the discarded transaction samples in the intermediate matrix corresponding to the previous iteration to obtain the modified intermediate matrix corresponding to the previous iteration, or directly delete the factors caused by the discarded transaction samples in the intermediate matrix corresponding to the previous iteration to obtain the modified intermediate matrix corresponding to the previous iteration.

[0090] In a feasible embodiment, modifying the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous iteration includes:

[0091] Delete the calculation results of the kernel functions related to the discarded transaction sample pairs in the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous iteration.

[0092] Since the number of transaction sample pairs in the updated sample space does not increase, the forgetting matrix remains unchanged, that is, B(i) = B(i - 1). Although the size of the updated sample space remains unchanged, the space elements are updated. Q(i - 1) can be regarded as the inverse matrix of B(i - 1)K(i - 1), and it is necessary to correspondingly delete the kernel function information of u(l) in the deleted transaction sample pairs.

[0093] Specifically, since u(l) has the longest preservation time in the updated sample space, that is, the time distance between the corresponding period and the current new period is the largest, the kernel function information related to u(l) in B(i - 1)K(i - 1) is located in the first row and the first column. Let B' = diag(β M-2 , β M-3 , …, 1), then Let K'(i - 1) denote the matrix obtained by removing the first row and the first column from K(i - 1), and let k'(i - 1) denote the vector obtained by removing the first element from k(i - 1), then there is:

[0094] Then, removing the first row and the first column of B(i - 1)K(i - 1) can obtain the matrix B'(i - 1)K'(i - 1). Using the inverse of the block matrix [B'(i - 1)K'(i - 1)] -1It can be calculated by the following formula: The intermediate matrix corresponding to the previous iteration is block-represented as Then the modified intermediate matrix [B'(i - 1)K'(i - 1)] corresponding to the previous iteration -1 = G - ff T / e.

[0095] S210. Determine the updated intermediate matrix, updated coefficient, and updated kernel vector corresponding to the current iteration according to the modified intermediate matrix and the updated sample space.

[0096] Calculate where u(l + 1), u(l + 2), …, u(L) are elements in the current kernel dictionary D(i), then The updated intermediate matrix is represented as where, A -1 = β -1 [B′(i - 1)K′(i - 1)] -1 , g = (d - b T A -1 b) -1 , the updated coefficient α(i) = Q(i)y(i), y(i) is composed of the outputs corresponding to the input samples in the kernel dictionary, y(i) = [y(l), y(l + 1), … y(L)] T , and output the predicted value of the next deposit amount

[0097] S211. Predict the transaction amount of the bank self-service equipment according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

[0098] The technical solution of this embodiment mines the correlation between the deposit and withdrawal amount data and the historical data through the improved kernel recursive least squares algorithm that introduces the forgetting factor and the sample sliding window, predicts the future deposit and withdrawal transaction amount in an adaptive iterative manner, improves the prediction accuracy while reducing the algorithm complexity; and determines the cash replenishment amount according to the prediction result to complete the formulation of the cash replenishment plan, thereby reducing the high cost of traditional manual management, effectively preventing situations such as service suspension caused by out-of-cash and deposit rejection caused by over-full cash, and greatly improving the operation efficiency and response speed. And in this embodiment, on the basis of windowing, a forgetting factor is introduced to strengthen the weight coefficient of the recent observation value and weaken the weight coefficient of the historical moment, thereby improving the model convergence speed and optimizing the system prediction performance.

[0099] Figure 3 is a schematic structural diagram of a device for determining cash replenishment information of a bank self-service equipment provided by an embodiment of the present invention. As Figure 3 shown, this device includes:

[0100] A historical transaction amount determination module 310 is configured to obtain the historical transaction amounts of each period of a bank self-service device of a target device type within a preset historical time period; wherein, the historical transaction amounts include historical withdrawal amounts and historical deposit amounts;

[0101] A transaction sample pair establishment module 320 is configured to establish a plurality of transaction sample pairs according to the historical transaction amounts of each period; wherein, each transaction sample pair includes the corresponding relationship between the input historical transaction data of a plurality of first periods and the output historical transaction data of a second period; the first period is a preset number of periods before the second period; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs;

[0102] An iterative sample space determination module 330 is configured to perform sliding sampling on the transaction sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration;

[0103] A predictor iterative calculation module 340 is configured to perform iterative calculation on the intermediate parameters of a transaction predictor according to the sample space and a kernel function; wherein, the transaction predictor includes a withdrawal predictor and a deposit predictor;

[0104] A prediction module 350 is configured to perform transaction amount prediction on the bank self-service device according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

[0105] The technical solution of this embodiment avoids the matrix inversion process with high computational cost based on the kernel recursive least squares algorithm, adaptively updates the predictor parameters in an iterative manner, and achieves good prediction accuracy with relatively low computational complexity; and for the devices that need cash replenishment, determines the cash replenishment information according to the recent transaction amount data, maximally balances the deposit and withdrawal operations, and at the same time reduces the waste of human resources.

[0106] Optionally, the predictor iterative calculation module includes:

[0107] An initial parameter determination unit is configured to determine an initial sample space corresponding to the first iteration, determine an initial intermediate matrix according to the calculation result of the input historical transaction data in the transaction sample pairs within the initial sample space according to the kernel function; determine an initial coefficient according to the initial intermediate matrix and the output historical transaction data in the transaction sample pairs within the initial sample space; and determine an initial kernel vector according to the calculation results of the input historical transaction data in the transaction sample pairs within the initial sample space and the input historical transaction data in the next transaction sample pair corresponding to the initial sample space according to the kernel function;

[0108] A prediction error determination unit, configured to determine a prediction value corresponding to the first iteration according to the initial kernel vector and the initial coefficient, and determine a prediction error corresponding to the first iteration according to the prediction value and the output historical transaction data in the next transaction sample pair corresponding to the initial sample space;

[0109] A parameter iterative update unit, configured to determine an updated sample space corresponding to subsequent iterations, and iteratively update the initial intermediate matrix, the initial coefficient, and the initial kernel vector according to the updated sample space and the prediction error to obtain an updated intermediate matrix, an updated coefficient, and an updated kernel vector.

[0110] Optionally, the predictor iterative calculation module includes an updated sample space judgment sub-module. Before iteratively updating the initial intermediate matrix, the initial coefficient, and the initial kernel vector according to the updated sample space and the prediction error, it includes:

[0111] A judgment unit, configured to determine whether there are discarded transaction sample pairs in the updated sample space;

[0112] A first update unit, configured to, if not, determine a forgetting matrix corresponding to the current iteration according to a forgetting factor, and determine an updated intermediate matrix and an updated coefficient corresponding to the current iteration according to the forgetting matrix, the updated sample space, and the intermediate matrix, coefficient, kernel vector, and prediction error corresponding to the previous iteration; determine an updated kernel vector corresponding to the current iteration according to the calculation result of the input historical transaction data in the transaction sample pair in the updated sample space and the input historical transaction data in the next transaction sample pair corresponding to the updated sample space by using the kernel function.

[0113] Optionally, the first update unit includes a first update sub-unit, configured to determine an updated intermediate matrix and an updated coefficient corresponding to the current iteration according to the forgetting matrix, the updated sample space, and the intermediate matrix, coefficient, kernel vector, and prediction error corresponding to the previous iteration. Specifically, it is configured to:

[0114] Determine the updated intermediate matrix and the updated coefficient corresponding to the current iteration according to the following formula:

[0115]

[0116] Among them, β represents the forgetting factor, Q(i - 1) represents the intermediate matrix corresponding to the previous round of iteration, Q(i) represents the updated intermediate matrix corresponding to the current iteration, B(i - 1) represents the forgetting matrix corresponding to the previous round of iteration, k(i) represents the kernel vector corresponding to the previous round of iteration, κ(·) represents the kernel function, u(i) represents the input historical transaction data of the first cycle of the newly added transaction samples in the updated sample space in the current iteration, α(i) represents the update coefficient corresponding to the current iteration, α(i - 1) represents the coefficient corresponding to the previous round of iteration, and e(i) represents the prediction error corresponding to the previous round of iteration.

[0117] Optionally, the updated sample space judgment sub-module further includes a second update unit, which is specifically used for: after determining whether there are discarded transaction sample pairs in the updated sample space.

[0118] If so, modify the intermediate matrix corresponding to the previous round of iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous round of iteration.

[0119] Determine the updated intermediate matrix, update coefficient, and updated kernel vector corresponding to the current iteration according to the modified intermediate matrix and the updated sample space.

[0120] Optionally, the second update unit includes an intermediate matrix modification sub-unit, which is used to modify the intermediate matrix corresponding to the previous round of iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous round of iteration, and is specifically used for:

[0121] Delete the calculation results of the kernel functions related to the discarded transaction sample pairs in the intermediate matrix corresponding to the previous round of iteration based on the discarded transaction sample pairs to obtain the modified intermediate matrix corresponding to the previous round of iteration.

[0122] Optionally, the prediction module is specifically used for:

[0123] Use the historical transaction data of multiple cycles before the prediction cycle as the input value of the transaction predictor to obtain the transaction amount prediction result corresponding to the prediction cycle; among them, the transaction amount prediction result includes the deposit amount prediction result and the withdrawal amount prediction result.

[0124] Determine whether the bank self-service equipment is a cash replenishment equipment according to the deposit amount prediction result and the withdrawal amount prediction result.

[0125] If so, determine the cash replenishment amount according to the total capacity of the cash cassette of the bank self-service equipment, the deposit amount prediction result, and the withdrawal amount prediction result.

[0126] Among them, the cash replenishment amount is determined according to the following formula:

[0127]

[0128] Among them, A add represents the cash replenishment amount, C represents the total capacity of the cash cassette of the bank self-service equipment, a represents the preset cash replenishment coefficient, and A W represents the predicted result of the withdrawal amount corresponding to the preset period, and A D represents the predicted result of the deposit amount corresponding to the preset period.

[0129] The cash replenishment information determination device of the bank self-service equipment provided by the embodiment of the present invention can execute the cash replenishment information determination method of the bank self-service equipment provided by any embodiment of the present invention, and has corresponding function modules and beneficial effects for executing the method.

[0130] In the technical solution of this application, the acquisition, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations and do not violate public order and good customs.

[0131] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0132] Figure 4 FIG. shows a schematic structural diagram of an electronic device 10 that can be used to implement the embodiments of the present invention. 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 processors, 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 only examples and are not intended to limit the implementation of the present invention described herein and / or claimed.

[0133] As Figure 4 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. Among them, the memory stores a computer program executable by at least one processor. The processor 11 can execute 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. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0134] 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 disk, an optical disc, 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.

[0135] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as determining the cash replenishment information of an ATM.

[0136] In some embodiments, determining the cash replenishment information of an ATM can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of determining the cash replenishment information of an ATM described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the determination of the cash replenishment information of an ATM by any other suitable means (e.g., by means of firmware).

[0137] The various embodiments of the systems and technologies described above herein 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 can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0138] A computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can 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 programs are executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer programs can be executed entirely on the machine, partially on the machine, executed partially on the machine and partially on a remote machine as an independent software package, or executed entirely on a remote machine or server.

[0139] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0140] 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 a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds 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, speech input, or tactile input).

[0141] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes switching components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, switching components, or frontend components. The components of the system can be interconnected with each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0142] The computing system can include a client and a server. The client and the server are generally far from each other and typically interact through a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0143] In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product that includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes program code for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network through the communication unit 19, or installed from the storage unit 18, or installed from the ROM 12. When the computer program is executed by the processor 11, the above-mentioned functions defined in the method of the embodiment of the present invention are performed.

[0144] It should be understood that various forms of the flows shown above can be used, steps can be reordered, added, or deleted. For example, the steps recited in the present invention can be executed 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, and no limitation is made herein.

[0145] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for determining cash replenishment information of a bank self-service device, characterized in that, The method includes: Obtaining the historical transaction amounts of each cycle of bank self-service devices of the target device type within a preset historical time period; wherein, the historical transaction amounts include historical withdrawal amounts and historical deposit amounts; Establishing a plurality of transaction sample pairs according to the historical transaction amounts of each cycle; wherein, each transaction sample pair includes the corresponding relationship between the input historical transaction data of a plurality of first cycles and the output historical transaction data of a second cycle; the first cycle is a preset number of cycles before the second cycle; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs; Performing sliding sampling on the transaction sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration; Performing iterative calculation on the intermediate parameters of the transaction predictor according to the sample space and the kernel function; wherein, the transaction predictor includes a withdrawal predictor and a deposit predictor; Predicting the transaction amount of the bank self-service device according to the transaction predictor after the iteration is completed, and determining the cash replenishment information according to the transaction amount prediction result.

2. The method according to claim 1, wherein Performing iterative calculation on the intermediate parameters of the transaction predictor according to the sample space and the kernel function, including: Determining the initial sample space corresponding to the first iteration, determining the initial intermediate matrix according to the calculation result of the input historical transaction data in the transaction sample pairs in the initial sample space by the kernel function; determining the initial coefficient according to the initial intermediate matrix and the output historical transaction data in the transaction sample pairs in the initial sample space; determining the initial kernel vector according to the calculation result of the input historical transaction data in the transaction sample pairs in the initial sample space and the input historical transaction data in the next transaction sample pair corresponding to the initial sample space; Determining the prediction value corresponding to the first iteration according to the initial kernel vector and the initial coefficient, and determining the prediction error corresponding to the first iteration according to the prediction value and the output historical transaction data in the next transaction sample pair corresponding to the initial sample space; Determining the updated sample space corresponding to the subsequent iteration, and performing iterative update on the initial intermediate matrix, the initial coefficient, and the initial kernel vector according to the updated sample space and the prediction error to obtain an updated intermediate matrix, an updated coefficient, and an updated kernel vector.

3. The method according to claim 2, wherein Before performing iterative update on the initial intermediate matrix, the initial coefficient, and the initial kernel vector according to the updated sample space and the prediction error, the method includes: Determining whether there are discarded transaction sample pairs in the updated sample space; If not, determining the forgetting matrix corresponding to the current iteration according to the forgetting factor, and determining the updated intermediate matrix and the updated coefficient corresponding to the current iteration according to the forgetting matrix, the updated sample space, and the intermediate matrix, the coefficient, the kernel vector, and the prediction error corresponding to the previous iteration; Determining the updated kernel vector corresponding to the current iteration according to the calculation result of the input historical transaction data in the transaction sample pairs in the updated sample space and the input historical transaction data in the next transaction sample pair corresponding to the updated sample space.

4. The method according to claim 3, wherein Determining an updated intermediate matrix and updated coefficients corresponding to the current iteration based on the forgetting matrix, the updated sample space, and the intermediate matrix, coefficients, kernel vectors, and prediction errors corresponding to the previous iteration, includes: Determining an updated intermediate matrix and updated coefficients corresponding to the current iteration according to the following formula: Among them, β represents the forgetting factor, Q(i - 1) represents the intermediate matrix corresponding to the previous round of iteration, Q(i) represents the updated intermediate matrix corresponding to the current iteration, B(i - 1) represents the forgetting matrix corresponding to the previous round of iteration, k(i) represents the kernel vector corresponding to the previous round of iteration, κ(·) represents the kernel function, u(i) represents the input historical transaction data of the first cycle of the newly added transaction samples in the updated sample space in the current iteration, α(i) represents the update coefficient corresponding to the current iteration, α(i - 1) represents the coefficient corresponding to the previous round of iteration, and e(i) represents the prediction error corresponding to the previous round of iteration.

5. The method according to claim 3, characterized in that, After determining whether there are discarded transaction sample pairs in the updated sample space, the method further includes: If so, modifying the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain a modified intermediate matrix corresponding to the previous iteration; Determining an updated intermediate matrix, updated coefficients, and updated kernel vectors corresponding to the current iteration according to the modified intermediate matrix and the updated sample space.

6. The method according to claim 5, characterized in that, Modifying the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain a modified intermediate matrix corresponding to the previous iteration, includes: Deleting the calculation results of the kernel functions related to the discarded transaction sample pairs in the intermediate matrix corresponding to the previous iteration based on the discarded transaction sample pairs to obtain a modified intermediate matrix corresponding to the previous iteration.

7. The method according to claim 1, characterized in that Predicting the transaction amount of the bank self-service device according to the transaction predictor after the iteration is completed, and determining the cash replenishment information according to the transaction amount prediction result, includes: Using the historical transaction data of multiple periods before the prediction period as the input values of the transaction predictor to obtain the transaction amount prediction result corresponding to the prediction period; wherein, the transaction amount prediction result includes a deposit amount prediction result and a withdrawal amount prediction result; Determining whether the bank self-service device is a cash replenishment device according to the deposit amount prediction result and the withdrawal amount prediction result; If so, determining the cash replenishment amount according to the total capacity of the cash cassettes of the bank self-service device, the deposit amount prediction result, and the withdrawal amount prediction result; Wherein, the cash replenishment amount is determined according to the following formula: Among them, A add represents the cash replenishment amount, C represents the total capacity of the cash cassettes of the bank self-service equipment, a represents the preset cash replenishment coefficient, and A W represents the predicted result of the withdrawal amount corresponding to the preset period, and A D represents the predicted result of the deposit amount corresponding to the preset period.

8. A device for determining cash replenishment information of a bank self-service device, characterized in that, The device includes: A historical transaction amount determination module, configured to obtain the historical transaction amounts of each period of the bank self-service device of the target device type within a preset historical time period; wherein, the historical transaction amounts include historical withdrawal amounts and historical deposit amounts; A transaction sample pair establishment module, configured to establish a plurality of transaction sample pairs according to the historical transaction amounts of each period; wherein, each transaction sample pair includes the corresponding relationship between the input historical transaction data of a plurality of first periods and the output historical transaction data of a second period; the first period is a preset number of periods before the second period; the transaction sample pairs include withdrawal sample pairs and deposit sample pairs; An iterative sample space determination module, configured to perform sliding sampling on the transaction sample pairs based on a pre-determined sample window size to obtain a sample space corresponding to each iteration; A predictor iterative calculation module, configured to perform iterative calculation on the intermediate parameters of the transaction predictor according to the sample space and the kernel function; wherein, the transaction predictor includes a withdrawal predictor and a deposit predictor; A prediction module, configured to predict the transaction amount of the bank self-service device according to the transaction predictor after the iteration is completed, and determine the cash replenishment information according to the transaction amount prediction result.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor is enabled to execute the method for determining the cash replenishment information of the bank self-service device 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 for implementing the method for determining the cash replenishment information of the bank self-service device according to any one of claims 1-7 when the computer instructions are executed by a processor.

Images