WO2023015615A1 - 基于预测模型的多渠道资金调拨方法、装置、设备和介质 - Google Patents

基于预测模型的多渠道资金调拨方法、装置、设备和介质 Download PDF

Info

Publication number
WO2023015615A1
WO2023015615A1 PCT/CN2021/114678 CN2021114678W WO2023015615A1 WO 2023015615 A1 WO2023015615 A1 WO 2023015615A1 CN 2021114678 W CN2021114678 W CN 2021114678W WO 2023015615 A1 WO2023015615 A1 WO 2023015615A1
Authority
WO
WIPO (PCT)
Prior art keywords
channel
fund
transfer
balance
funds
Prior art date
Application number
PCT/CN2021/114678
Other languages
English (en)
French (fr)
Inventor
李子圣
Original Assignee
未鲲(上海)科技服务有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 未鲲(上海)科技服务有限公司 filed Critical 未鲲(上海)科技服务有限公司
Publication of WO2023015615A1 publication Critical patent/WO2023015615A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/02—Banking, e.g. interest calculation or account maintenance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present application relates to the technical field of intelligent decision-making, and in particular to a method, device, equipment and medium for multi-channel capital allocation based on a prediction model.
  • the inventor realizes that at present, fund transfer operations are mainly performed by fund operators manually generating transfer instructions.
  • fund transfer operations are mainly performed by fund operators manually generating transfer instructions.
  • the internal operating systems of a few enterprise organizations can automatically generate fund transfer instructions.
  • the actual fund situation of the fund channel is used to generate the fund transfer instruction.
  • the embodiment of the present application provides a multi-channel fund transfer method, device, equipment and medium based on a forecasting model.
  • the implementation of the embodiment of the present application enables the fund transfer operation to be intelligently carried out according to the actual fund situation of the fund channel.
  • the embodiment of the present application provides a multi-channel fund allocation method based on a forecasting model, the above-mentioned method includes:
  • the channel parameters include the fund change record and channel weight of the funding channel in the first time
  • the embodiment of the present application provides a multi-channel fund allocation device based on a forecasting model, the above-mentioned device includes:
  • An acquisition unit configured to acquire channel parameters of each of the multiple funding channels, where the channel parameters include the fund change records and channel weights of the funding channel in the first time;
  • the input unit is used to input the channel parameters of each fund channel into the fund use prediction model to obtain the first balance threshold corresponding to each fund channel, and the first balance threshold is the amount that the fund channel needs to use in the second time ;
  • the detection unit is used to detect the balance of each fund channel, and when the balance of each fund channel is less than or equal to the first balance threshold, it is determined that the fund channel is an incoming fund channel;
  • the transfer unit is configured to transfer part of the balance of the transfer-out fund channel to the transfer-in fund channel.
  • the embodiment of the present application provides an electronic device, including a processor, a memory, and computer-executed instructions stored in the memory and operable on the processor.
  • the electronic device executes based on A multi-channel funding approach to predictive modeling, the approach comprising:
  • the embodiment of the present application provides a computer-readable storage medium, in which computer instructions are stored, and when the computer instructions are run on the communication device, the communication device is made to execute multi-channel funding based on the forecasting model.
  • a transfer method the method comprising:
  • the embodiment of the present application enables the fund transfer operation to be carried out intelligently according to the actual situation of the fund channel, thereby reducing the capital operation cost of the enterprise.
  • FIG. 1A is a structural deployment diagram of a capital operation system provided by an embodiment of the present application.
  • FIG. 1B is a schematic flow diagram of a multi-channel fund allocation method based on a forecasting model provided by an embodiment of the present application;
  • Fig. 1C is a schematic structural diagram of a fund usage forecasting model provided by the embodiment of the present application.
  • Figure 1D is a structural deployment diagram of a predictive model-based capital operation system applied in the embodiment of the present application.
  • Fig. 2A is a schematic diagram of an example of a multi-channel fund allocation method based on a forecasting model provided by an embodiment of the present application;
  • Fig. 2B is a schematic diagram of an example of a multi-channel fund allocation method based on a forecasting model provided by the embodiment of the present application;
  • Fig. 2C is a schematic diagram of an example of a multi-channel fund allocation method based on a forecasting model provided by the embodiment of the present application;
  • Fig. 3 is a schematic structural diagram of a multi-channel fund allocation device based on a forecasting model provided by an embodiment of the present application;
  • FIG. 4 is a schematic diagram of a server structure of a hardware operating environment of an electronic device provided by an embodiment of the present application.
  • This application may relate to the technical field of artificial intelligence, for example, it may specifically relate to intelligent decision-making technology. For example, relevant data can be acquired and processed based on artificial intelligence technology.
  • the technical solution of the present application can be applied to various fund allocation scenarios based on prediction models, such as medical special fund allocation scenarios in digital medical care.
  • the medical system can realize the intelligent fund transfer operation based on the multi-channel fund transfer method based on the prediction model based on the actual fund situation of the fund channel.
  • the data involved in this application such as channel parameters, balance, etc., can be stored in the block chain.
  • Fig. 1A is a structural deployment diagram of a capital operation system provided by an embodiment of the present application.
  • the system includes a channel setting module, multiple transfer-in fund channels (transfer-in fund channel 1, transfer-in fund channel 2 ... transfer-in fund channel N), multiple transfer-out fund channels (transfer-out fund channel Fund channel 1, transfer-out fund channel 2... transfer-out fund channel N), fund transfer module.
  • the channel setting module is used to allow the fund operation personnel to set the transfer types and transfer targets of multiple fund channels, and obtain multiple transfer-in fund channels and multiple transfer-out fund channels.
  • a certain transfer-in fund channel among the multiple transfer-in fund channels refers to a fund channel that receives a part of the balance from at least one transfer-out fund channel through the fund transfer module.
  • a certain transfer-out fund channel among the multiple transfer-out fund channels refers to a fund channel that transfers part of the balance to at least one transfer-in fund channel through the fund transfer module.
  • the fund transfer module is used to generate a fund transfer instruction according to the requirements of the fund operation personnel, so as to transfer part of the balance of the transfer-out fund channel to the transfer-in fund channel.
  • the transfer type and transfer target of the fund channel are set by the fund operator. There is a risk that the transfer channel is still in the state of fund transfer when its own balance is lower than a certain threshold. Fund allocation is carried out according to the actual situation of the funding channel.
  • FIG. 1B is a schematic flow diagram of a multi-channel fund allocation method based on a forecast model provided by an embodiment of this application. As shown in Figure 1B, the method includes the following steps:
  • funding channels include bank accounts and third-party payment platforms.
  • the first time refers to a certain time in the past historical time. It can be past 1 month, past 1 year, etc.
  • the fund change record of the fund channel in the first time including the fund increase and decrease rate and the fund increase and decrease speed of the fund channel in the first time.
  • the first time is the past year
  • the fund change record of the fund channel in the first time includes the fund increase and decrease rate and the fund increase and decrease speed of the fund channel in different time periods every day in the past year.
  • the channel weight can be manually set according to the specific parameter information of the corresponding fund channel.
  • the specific parameter information may be set by parameter information such as the frequency of use of the funding channel, the importance of the company's product corresponding to the funding channel, and the like.
  • the channel weight is determined according to the importance of the company's product or business corresponding to the funding channel, and the importance of the company's product corresponding to the funding channel 1 is higher than that of the company's product corresponding to the funding channel 2, so the channel weight of the funding channel 1 > funds Channel weight for channel 2.
  • the second time refers to a certain time in the future. It can be 1 hour in the future, after 17:00 on the current working day until 8:00 on the next working day, etc.
  • the channel parameters of each fund channel are input into the fund use forecasting model to obtain the first balance threshold corresponding to each fund channel, that is to say, the fund use forecasting model predicts and obtains each The amount that a funding channel needs to use within a certain period of time in the future.
  • the first balance threshold is 10,000 yuan
  • the second time is 1 hour in the future, which means that the fund usage prediction model predicts that the amount that the fund channel needs to use in the next 1 hour is 10,000 yuan.
  • the second time is the time period from 17:00 on the current working day to 8:00 on the next working day
  • the capital usage forecasting model predicts that each funding channel needs The amount of money used, so as to transfer funds according to the situation of the fund channel, so as to ensure that the fund use of each fund channel during non-working hours can be in a normal state.
  • the fund usage prediction model may include an input layer, a convolution layer, a pooling layer, and a loss function layer.
  • Figure 1C is a schematic structural diagram of a fund use forecasting model provided in the embodiment of the present application.
  • the fund use forecasting model includes the following levels:
  • the input layer is used to receive the channel parameters of each funding channel as model parameters.
  • the channel parameters include the fund change record and channel weight of the funding channel in the first time, and normalize the channel parameters of each funding channel. Eliminate the dimensions of the channel parameters, and obtain the channel parameters after the initial processing of each funding channel;
  • the convolutional layer is used to extract the features of the channel parameters after the initial processing of each of the above-mentioned fund channels, and obtain the channel parameter data features of each fund channel.
  • the channel parameter data features include the daily fund increase and decrease range. Funds increase and funds decrease, the increase/decrease rate of fund increase/decrease on the day of the channel is large/small, and the deceleration is large/small;
  • Pooling layer used to compress the channel parameter data characteristics of each of the above-mentioned funding channels, and extract the main features of the channel parameters of each funding channel corresponding to the channel parameter data characteristics of each funding channel, simplifying the calculation complexity, exemplary Specifically, the main characteristics of channel parameters include the maximum increase and minimum decrease of funds in the daily increase and decrease of funds, and the maximum increase, minimum increase, maximum deceleration, and minimum decrease of the channel's fund increase and decrease on the same day;
  • Loss function layer used to obtain the predicted usage amount of each funding channel in the first historical time according to the main characteristics of the channel parameters of each funding channel above, and calculate according to the predicted usage amount and the actual usage amount in the first historical time
  • the loss value between the predicted usage amount and the actual usage amount in the first historical time during the training process of the fund usage prediction model is used to adjust the model parameters of the input fund usage prediction model.
  • the target loss function includes a Softmax cross-entropy loss function, illustratively,
  • ai is the channel parameter of the i-th funding channel
  • T is the total number of input funding channels
  • Si is the probability of the predicted usage amount of the i-th input funding channel in the first historical time
  • yi is the i-th input
  • c is the target loss value, and the smaller c is, the more accurate the prediction result is.
  • the fund use forecasting model shown in FIG. 1C is only an example of a fund use forecasting model, and in a specific application, the fund use forecasting model may also exist in the form of other layers.
  • 103 Detect the balance of each fund channel, and when the balance of the fund channel is less than or equal to the first balance threshold, determine that the fund channel is the transfer-in fund channel;
  • the transfer-in fund channel refers to the fund channel that needs to obtain the balance from the transfer-out fund channel in the process of fund transfer to supplement its own balance.
  • a fund transfer review instruction can be generated for the user terminal, and the fund transfer can be performed after the user terminal confirms the instruction.
  • the fund transfer review instruction is a mobile phone text message.
  • the fund operation system actively sends a mobile phone message to the mobile phone number of the fund operator to remind the fund operation
  • the personnel will review the fund transfer instructions, and the fund operation personnel will confirm the fund transfer instructions in the mobile phone text message before transferring the funds. While ensuring the safety of funds, the timeliness for the execution of fund transfers is also guaranteed.
  • the transfer-out fund channel refers to a fund channel that needs to transfer part of its balance to the transfer-in fund channel during the fund transfer process, so that its own balance will decrease.
  • the method in the embodiment of the present application may be applied in a user terminal
  • the user terminal may be an electronic device such as a mobile phone, a tablet computer, a personal digital assistant, or a wearable device.
  • Figure 1D is a structural deployment diagram of a forecasting model-based capital operation system applied in the embodiment of the present application.
  • the capital operation system includes a channel parameter acquisition module, a capital use prediction model, and a channel Balance detection module, fund transfer module, the first end of the fund operation system is connected to the user terminal, and the second end is connected to multiple fund channels through a firewall.
  • the function of each module may be realized by an independent server, or the functions of multiple modules may be realized by one server. Multiple servers that realize the functions of different modules communicate with each other.
  • the user terminal is the role of using the fund operation system.
  • the user terminal can manually set the channel weight of each fund channel according to the actual situation of the enterprise organization.
  • the results are entered into the treasury operating system.
  • the channel parameter acquisition module in the fund operation system is used to acquire the channel parameters of each fund channel in the multiple fund channels.
  • the channel parameters include the fund change record and channel weight of the fund channel in the first time.
  • the fund change record in the first time is provided by the fund channel, and the channel weight of the fund channel can be provided by the user terminal.
  • the fund usage prediction model in the fund operation system is used to use the channel parameters of each fund channel as model parameters, and predict the amount that the fund channel needs to use in the second time as the first balance threshold of the fund channel.
  • the channel balance detection module in the fund operation system is used to detect the balance of each fund channel, and when the balance of a certain fund channel is less than or equal to the first balance threshold, it is determined that the fund channel is a transfer-in fund channel.
  • the fund transfer module in the fund operation system is used for transferring part of the balance of the transfer-out fund channel to the transfer-in fund channel for fund transfer processing.
  • firewall 1 firewall 2 is used to prevent malicious external attacks on the fund operation system and multiple fund channels, avoid leakage of important financial data related to funds, and protect the security of the fund operation system and multiple fund channels .
  • the multiple fund channels include multiple transfer-in funds and multiple transfer-out fund channels.
  • the transfer-in fund channel refers to the fund channel that needs to obtain the balance from the transfer-out fund channel in the process of fund transfer to supplement its own balance.
  • Outgoing fund channel refers to a fund channel that needs to transfer part of its own balance to the inbound fund channel during the process of fund allocation, so that its own balance will decrease.
  • fund channel 1 and fund channel 2 wherein, fund channel 2 is a transfer-out fund channel, and the balance of fund channel 1 is 3,000 yuan, and the fund change record and channel weight of fund channel 1 in the past 1 year are obtained, Input the funds change record and channel weight of fund channel 1 in the past year into the fund use forecasting model, and predict that the amount that fund channel 1 needs to use in the next hour is 5,000 yuan, which is the first balance of fund channel 1
  • the threshold is 5,000 yuan
  • the balance of fund channel 1 is only 3,000 yuan and less than 5,000 yuan. That is to say, the balance of fund channel 1 is not enough to pay the amount it will use in the next hour.
  • the channel parameters include the fund change records and channel weights of the funding channels in the first time; Input the channel parameters into the fund usage forecasting model to obtain the first balance threshold corresponding to each fund channel.
  • the first balance threshold is the amount that the fund channel needs to use in the second time; detect the balance of each fund channel, and when the fund When the balance of the channel is less than or equal to the first balance threshold, the fund channel is determined to be the transfer-in fund channel; and part of the balance of the transfer-out fund channel is transferred to the transfer-in fund channel.
  • the channel parameters of the fund channel in the first time are input into the fund use forecasting model to obtain the amount that the fund channel needs to use in the second time, and then the fund channel with insufficient balance is used as
  • the transfer-in fund channel receives the fund supplement from the transfer-out fund channel, so that the fund transfer operation can be carried out intelligently according to the actual situation of the fund channel, reducing the capital operation cost of the enterprise.
  • the balance of the transfer-out fund channel is greater than the second balance threshold
  • transferring part of the balance of the transfer-out fund channel to the transfer-in fund channel further includes: keeping the balance of the transfer-out fund channel greater than or equal to the fund channel A corresponding second balance threshold, wherein the second balance threshold of the fund channel is greater than the first balance threshold of the fund channel.
  • the balance of the transfer-out fund channel remains greater than or equal to the second balance threshold corresponding to the fund channel.
  • the second balance threshold of the fund channel > the first balance threshold of the fund channel the purpose is to prevent the transfer-out fund channel from being changed to the transfer-in fund channel because the balance is less than or equal to the first balance threshold, by maintaining The transfer type of the transfer-out fund channel remains unchanged to ensure the stability of the fund transfer process.
  • the balance of the transferred fund channel remains greater than or equal to the second balance threshold corresponding to the fund channel and the second balance threshold of the fund channel is greater than the first balance of the fund channel Threshold, to avoid the risk of lack of balance in the transferred funds channel, while keeping the transfer type of the transferred funds channel unchanged, improving the security and stability of the funds transfer process.
  • the training process of the fund use prediction model is as follows: obtain the training data set, the training data set is the channel correlation parameter of each funding channel in the first historical time, and the channel correlation parameter is the The parameters of the initial amount allocation of the initial amount; input the training data set into the target prediction model to obtain the predicted usage amount of each funding channel in the first historical time; compare the predicted usage amount of each funding channel with the actual usage amount of the funding channel Input the target deviation function, and calculate the target deviation value according to the target deviation function; when the target deviation value is greater than the preset deviation value, detect the deviation abnormal parameter in the channel related parameters, and the deviation abnormal parameter is that the target deviation value is greater than the preset deviation value The key item of the value; use the alternative channel association parameters to replace the deviation and abnormal parameters to obtain a new training data set, and repeat the process of inputting the new training data set into the target prediction model and obtaining the predicted usage amount until the predicted usage amount and the actual usage amount are determined When the target deviation value of the amount is less than the preset difference, the channel correlation
  • the channel related parameters refer to the related parameters of the fund channel.
  • These related parameters are parameters that may affect the use of channel funds. Specifically, they may be the fund change record of the fund channel in the first historical time, channel weight, fund peak-valley value, The average value of funds, etc., among which the records of fund changes include the extent of fund increase and decrease, and the rate of fund increase and decrease.
  • the parameters that decisively affect the use of channel funds among these channel-associated parameters that is, the channel parameters.
  • the target loss function includes the Softmax cross-entropy loss function.
  • the abnormal deviation parameter refers to the parameter that does not actually have a decisive impact on the use of funds among the inferred channel-related parameters.
  • the detection of abnormal deviation parameters in the channel-related parameters may be implemented by using standardized regression coefficients to detect key items in the channel-related parameters that cause the target deviation value to be greater than the preset deviation value.
  • the standardized regression coefficient is the regression coefficient obtained by standardizing the independent variable and the dependent variable in the target deviation function at the same time, where the independent variable is the channel correlation parameter in the first historical time, and the dependent variable is each fund channel
  • the data has been standardized to eliminate the impact of differences in dimensions and orders of magnitude, making different variables comparable, so standardized regression coefficients can be used to compare the impact of different independent variables on the dependent variable , so as to determine the deviation and abnormal parameters in the channel correlation parameters.
  • the channel weight and fund peak-valley value of each fund channel in the first historical time period are obtained as the first training data set, and the first training data set is input into the target forecasting model , to obtain the first predicted use amount of each fund channel in the first historical time, input the first predicted use amount of each fund channel and the actual use amount of the fund channel into the target deviation function, and calculate according to the target deviation function to obtain the first A target deviation value, the first target deviation value is greater than the preset deviation value, and the abnormal parameter of the deviation is detected to be the peak-valley value of funds, then use the fund change record as the associated parameter of the alternative channel to replace the peak-valley value of funds to obtain a new Including the second training data set of fund change records and channel weights within the first historical time period, repeat the process of inputting the second training data set into the target prediction model and obtaining the predicted amount of use.
  • the target deviation value between the second predicted usage amount and the actual usage amount is less than the preset difference value, then determine the fund change records included in the second training data set within the first historical period, and the channel weight is the channel parameter, which is included in the first historical
  • the target prediction model with the fund change records and channel weights as model parameters within a certain period of time is the fund use forecast model. So far, the training process of the fund use forecast model has been completed.
  • the third balance threshold of the channel ⁇ the first balance threshold of the fund channel, and the third balance threshold is the amount that needs to be held by each transfer-in fund channel after fund transfer;
  • the amount of funds that can be transferred out of each transfer-out fund channel, the amount of funds that can be transferred out j the balance of the transfer-out fund channel - the second balance threshold, where j is the j-th transfer-out fund channel; calculate the funds that can be transferred in If the amount of funds that can be transferred out is greater than the preset amount, then the transfer of funds corresponding
  • the third balance threshold of the fund channel ⁇ the first balance threshold of the fund channel, the purpose is to make the amount held by each transferred fund channel after fund transfer can be maintained higher than the first balance threshold Since the first balance threshold is the amount that the fund channel needs to use in the second time, when the amount held by the transfer-in fund channel after fund transfer is higher than the first balance threshold, it can Respond more flexibly to the instability caused by possible emergencies or emergencies in the process of fund operation, and reduce the risk of fund operation.
  • the preset amount may be 0.
  • the transfer of funds corresponding to the transferable fund i will be executed; otherwise, if the amount of transferable funds is less than or equal to the preset amount, the funds will be transferred to the i-1th transfer When the balance of the transfer-in fund channel of the transfer-in fund channel reaches the third balance threshold, it will stop, and the transfer of funds to the i-th transfer-in fund channel will no longer be performed.
  • FIG. 2A is a schematic diagram of a multi-channel fund transfer method based on a forecast model provided by the embodiment of the present application.
  • the balance of the transferred-in fund channel is lower than the first balance threshold.
  • the supplementary amount h the third balance threshold is calculated.
  • the balance of the transferred-in fund channel h of the transferred-in fund channel h is updated to the third balance threshold.
  • FIG. 2B is a schematic diagram of a multi-channel fund transfer method based on a forecasting model provided by the embodiment of the present application.
  • the funds Before the transfer, the balance of its transfer-out fund channel is higher than the second balance threshold.
  • the transfer-out fund amount j transfer-out fund channel balance - the first
  • the second balance threshold value after the funds are transferred, the balance of the transfer-out fund channel j of the transfer-out fund channel j is updated to the second balance threshold value.
  • the balance of the transferred-in fund channel is supplemented to the third balance threshold after the transfer of funds, so that the amount held by the transferred-in fund channel after the transfer of funds is greater than or Equal to the first balance threshold, where, when the amount held by the transfer-in fund channel is greater than the first balance threshold, it can be more flexible to deal with the instability caused by possible emergencies or emergencies in the process of fund operation, Reduce the risk of capital operation.
  • the above method further includes: h is the transfer-in priority ranking of the transfer-in fund channel, and j is the transfer-out priority ranking of the transfer-out fund channel; wherein, transferring part of the balance of the transfer-out fund channel to The transfer-in fund channel includes: transferring part of the balance of the transfer-out fund channel to the transfer-in fund channel according to the transfer-in priority of the transfer-in fund channel and the transfer-out priority of the transfer-out fund channel.
  • the transfer-in priority of multiple transfer-in fund channels is sorted. Specifically, the higher the priority value of the transfer-in fund channel, the higher the transfer priority. The earlier the transfer process will get the balance from the transferred funds channel.
  • the transfer-out priorities of multiple transfer-out fund channels are sorted. Specifically, the higher the priority value of the transfer-out fund channel, the higher the transfer-out priority. During the transfer process, part of the balance will be transferred to the transfer-in funds channel earlier.
  • the priority value of each channel for transferring in funds is calculated according to the transfer-in difference and channel weight of each transfer-in fund channel, and the priority value of each transfer-in fund channel is obtained according to the transfer-out difference and
  • the channel weight is calculated to obtain the priority value of each transfer-out fund channel, and the transfer-in priority is sorted according to the priority value of each transfer-in fund channel, and the transfer-out priority is sorted according to the priority value of each transfer-out fund channel , so that the transfer-in fund channel whose balance is far from the first balance threshold and has a large channel weight will have a high transfer-in priority, and the transfer-out fund whose balance is far from the second balance threshold and has a large channel weight
  • the fund channel will have a higher transfer-out priority, so that in the process of fund transfer, the partial balance of the transfer-out fund channel with redundant balance will be transferred to the transfer-in fund channel with a seriously scarce balance, realizing the transfer of funds
  • the above method further includes: obtaining the fund transfer amount of each time period in the N time periods within the preset number of days of the transfer-in fund channel and the transfer-out fund channel, and obtaining each of the N time periods
  • the target transfer frequency coefficient for each time period is obtained;
  • the target frequency adjustment coefficient for each time period is obtained;
  • the target refresh frequency for each time period is calculated; in each time period During the period of time, the priority value of the transfer-in priority of the transfer-in fund channel and the transfer-out priority of the transfer-out fund channel are refreshed according to the target refresh frequency.
  • the priority value refresh is used to determine each transfer-in fund channel, each transfer The updated priority value of the funding channel.
  • the transfer-in priority of the transfer-in fund channel and the transfer-out priority of the transfer-out fund channel are refreshed according to the number of priority values corresponding to the target refresh frequency.
  • the actual situation of fund transfer The priority value of the fund channel is refreshed for the corresponding number of times in line with the actual situation of fund transfer in different time periods, so as to avoid the unexpected situation that the priority value of the fund channel cannot be dealt with in the process of fund transfer due to the immobilization of the priority value of the fund channel.
  • target refresh frequency target allocation frequency coefficient * target frequency adjustment coefficient, divide a day into 12 time periods, each time period is 2 hours, that is, divide a day into 0:01-2:00, 2:01-4 :00...12:01-14:00...16:01-18:00...22:01-0:00, a total of 12 time periods, please refer to Figure 2C, Figure 2C is an example provided by the embodiment of this application A schematic diagram of an example of a multi-channel fund transfer method based on a prediction model, as shown in Figure 2C, wherein the average fund transfer amount from 2:01 to 4:00 is 500 yuan, and the number of fund transfers is 1 time, and from 16:01 to 18 The average fund transfer amount of :00 is 30,000 yuan, and the number of fund transfers is 10 times.
  • Table 1 The default mapping relationship between the average fund transfer amount and the transfer frequency coefficient is shown in Table 1:
  • the target frequency adjustment factor 1 for the 2:01-4:00 time period is 0.7
  • the target frequency adjustment factor 2 for the 16:01-18:00 time period is 1.5
  • the target for the 2:01-4:00 time period is calculated
  • 12 priority value refreshes will be performed on the transfer-in priority of the transfer-in fund channel and the transfer-out priority of the transfer-out fund channel.
  • the target refresh frequency of each time period is obtained through the average fund transfer amount and the number of fund transfer times in each time period of the transfer-in fund channel and the transfer-out fund channel in N time periods. Refresh the priority value corresponding to the target refresh frequency for the transfer-in priority of the transfer-in fund channel and the transfer-out priority of the transfer-out fund channel within each time period.
  • the priority value of the transfer-in fund channel and the transfer-out fund channel can be refreshed according to the actual situation of the average fund transfer amount and the number of fund transfer times in each time period, avoiding the priority value of the fund channel during the fund transfer process. Immobilization makes it impossible to flexibly respond to sudden changes in funds in the process of fund allocation, making the intelligent decision-making process of fund allocation more flexible.
  • the fourth balance threshold of each low-interest rate transfer-out fund channel in the fund channel the second balance threshold ⁇ (1-channel interest rate); update the second balance threshold of the transfer-out fund channel to the corresponding value of the transfer-out fund channel
  • the channel interest rate is the bank interest rate corresponding to each transfer channel, and the bank interest rate represents the ratio of interest to principal within a certain period of time, expressed as a percentage.
  • the bank rate is 0.3%.
  • the fourth balance threshold of the high-interest rate funds transfer channel the second balance threshold ⁇ (1 + channel interest rate)
  • the fourth balance threshold of the high-interest rate transfer funds channel is greater than the second balance threshold
  • the purpose is, because the bank interest rate of the high-interest transfer fund channel itself is relatively high, and the interest will be higher if the balance is large, so the high-interest rate transfer fund channel can maintain a balance slightly higher than the second balance threshold.
  • the fourth balance threshold of the low-interest rate transfer-out funds channel the second balance threshold ⁇ (1-channel interest rate)
  • the fourth balance threshold of the low-interest rate transfer-out funds channel is smaller than the second balance threshold
  • the purpose is, because the bank interest rate of the low-interest transfer fund channel itself is relatively low, even if the balance is large, the interest will be relatively small.
  • the balance in the channel should not be maintained at a relatively high level, therefore, the second balance threshold of the low-interest rate transfer-out funds channel is updated to a fourth balance threshold lower than the second balance threshold.
  • the second balance threshold of the high interest rate transfer fund channel is updated to be compared to the second balance threshold
  • update the second balance threshold of the low-interest rate transfer fund channel to the fourth balance threshold lower than the second balance threshold, so that the high-interest rate transfer fund channel can maintain
  • the low-interest transfer fund channel maintains a balance slightly lower than the second balance threshold, so that the balance of each transfer-out fund channel after fund transfer can be maintained
  • the degree of utilization of the balance of the transferred funds channel has been fully improved.
  • FIG. 3 is a schematic structural diagram of a multi-channel fund transfer device based on a forecast model provided in the embodiment of the present application, as shown in FIG. 3:
  • a multi-channel fund transfer device based on a predictive model comprising:
  • An acquisition unit configured to acquire channel parameters of each of the multiple funding channels, where the channel parameters include the fund change record and channel weight of the funding channel within the first time;
  • the input unit is used to input the channel parameters of each fund channel into the fund use prediction model to obtain the first balance threshold corresponding to each fund channel, and the first balance threshold is what the fund channel needs to use in the second time Amount;
  • the detection unit is used to detect the balance of each fund channel, and when the balance of the fund channel is less than or equal to the first balance threshold, it is determined that the fund channel is the transferred fund channel;
  • An allocation unit configured to allocate part of the balance of the transferred-out fund channel to the transferred-in fund channel.
  • the channel parameters of each fund channel in the multiple fund channels are acquired by the acquisition unit, and the channel parameters include the fund change record and channel weight of the fund channel in the first time;
  • the channel parameters of each fund channel are input into the fund use prediction model to obtain the first balance threshold corresponding to each fund channel.
  • the first balance threshold is the amount that the fund channel needs to use in the second time; through the detection unit, each The balance of each fund channel, when the balance of each fund channel is less than or equal to the first balance threshold, determine that the fund channel is the transfer-in fund channel; transfer part of the balance of the transfer-out fund channel to the transfer-in fund channel through the transfer unit .
  • the channel parameters of the fund channel in the first time are input into the fund use forecasting model to obtain the amount that the fund channel needs to use in the second time, and then the fund channel with a lack of balance is used as
  • the transfer-in fund channel receives the fund supplement from the transfer-out fund channel, so that the fund transfer operation can be carried out intelligently according to the actual situation of the fund channel, reducing the capital operation cost of the enterprise.
  • the embodiment of the present application can divide the functional units of the multi-channel fund allocation device based on the forecasting model according to the above-mentioned method examples.
  • each functional unit can be divided corresponding to each function, or two or more functions can be divided into integrated in one processing unit.
  • the above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. It should be noted that the division of units in the embodiment of the present application is schematic, and is only a logical function division, and there may be another division manner in actual implementation.
  • FIG. 4 is a schematic diagram of the server structure of the hardware operating environment of the electronic device provided by the embodiment of the present application , as shown in FIG. 4 , the electronic device includes a processor, a memory, and computer-executed instructions stored in the memory and operable on the processor. When the computer-executed instructions are executed, the electronic device executes any one of prediction model-based Multi-channel fund transfer method.
  • the processor is a CPU.
  • the memory optionally, the memory may be a high-speed RAM memory, or a stable memory, such as a disk memory.
  • the structure of the server shown in FIG. 4 is not limited thereto, and may include more or less components than those shown in the illustration, or combine some components, or arrange different components.
  • the memory may include an operating system, a network communication module, and computer-executed instructions for multi-channel fund transfer based on a forecast model.
  • the operating system is used to manage and control server hardware and software resources, and supports the operation of computer-executed instructions.
  • the network communication module is used to realize the communication between the various components inside the memory, as well as the communication with other hardware and software inside the server.
  • the communication can use any communication standard or protocol, including but not limited to GSM (Global System of Mobile communication, global Mobile Communication System), GPRS (General Packet Radio Service, General Packet Radio Service), CDMA2000 (Code Division Multiple Access 2000, Code Division Multiple Access 2000), WCDMA (Wideband Code Division Multiple Access, Wideband Code Division Multiple Access), TD-SCDMA ( Time Division-Synchronous CodeDivision Multiple Access, Time Division Synchronous Code Division Multiple Access), etc.
  • GSM Global System of Mobile communication, global Mobile Communication System
  • GPRS General Packet Radio Service, General Packet Radio Service
  • CDMA2000 Code Division Multiple Access 2000, Code Division Multiple Access 2000
  • WCDMA Wideband Code Division Multiple Access
  • TD-SCDMA Time Division-Synchronous CodeDivision Multiple Access, Time Division Synchronous Code Division Multiple Access
  • the processor is used to run the computer-executed instructions stored in the memory for personnel management, and realize the following steps: obtain the channel parameters of each of the multiple funding channels, and the channel parameters include the funding channel in Fund change records and channel weights within a period of time; input the channel parameters of each fund channel into the fund use prediction model to obtain the first balance threshold corresponding to each fund channel, and the first balance threshold is the fund channel at the second time The amount that needs to be used within; detect the balance of each fund channel, and when the balance of each fund channel is less than or equal to the first balance threshold, determine that the fund channel is the transfer-in fund channel; transfer part of the balance of the fund channel Transferred to the transferred funds channel.
  • An embodiment of the present application provides a computer-readable storage medium.
  • Computer instructions are stored in the computer-readable storage medium.
  • the communication device is made to perform the following steps: acquire each of the multiple funding channels The channel parameters of each funding channel, which include the fund change record and channel weight of the funding channel in the first time; input the channel parameters of each funding channel into the fund usage forecasting model to obtain the first balance corresponding to each funding channel Threshold, the first balance threshold is the amount that the fund channel needs to use in the second time; the balance of each fund channel is detected, and when the balance of the fund channel is less than or equal to the first balance threshold, it is determined that the fund channel is adjusted Inward fund channel; transfer part of the balance of the transferred out fund channel to the transferred in fund channel.
  • the computer described above includes electronic equipment.
  • the storage medium involved in this application such as a computer-readable storage medium, may be non-volatile or volatile.
  • electronic devices include mobile phones, tablet computers, personal digital assistants, wearable devices, etc.
  • the computer-readable storage medium may be an internal storage unit of the electronic device described in the above embodiments, such as a hard disk or a memory of the electronic device.
  • the computer-readable storage medium can also be an external storage device of the above-mentioned electronic equipment, such as a plug-in hard disk equipped on the electronic equipment, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash memory card (Flash Card) etc.
  • the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device.
  • Computer-readable storage media are used to store computer-executable instructions and other computer-executable instructions and data required by the electronic device.
  • the computer-readable storage medium can also be used to temporarily store data that has been output or will be output.
  • An embodiment of the present application provides a computer program product, wherein the computer program product includes a computer program, and the computer program is operable to make the computer as part of any predictive model-based multi-channel fund allocation method described in the above method embodiments or all steps, the computer program product may be a software installation package.
  • These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions
  • the device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.
  • the memory may include: a flash disk, a read-only memory (English: Read-Only Memory, ROM for short), a random access device (English: Random Access Memory, RAM for short), a magnetic disk or an optical disk, and the like.
  • the memory may include: flash disk, read-only memory (English: Read-Only Memory, abbreviated: ROM), random access device (English: Random Access Memory, abbreviated: RAM), magnetic disk or optical disc, etc. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Technology Law (AREA)
  • General Business, Economics & Management (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

本申请实施例公开了一种基于预测模型的多渠道资金调拨方法、装置、设备和介质，其中方法的实现包括：获取多个资金渠道中的每个资金渠道的渠道参数；将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；将调出资金渠道的部分余额调拨给调入资金渠道。采用本申请实施例的方法，通过资金使用预测模型预测资金渠道在第二时间内需要使用的金额额度，使得资金调拨操作得以智能化地根据资金渠道的实际情况来进行。

Description

基于预测模型的多渠道资金调拨方法、装置、设备和介质
本申请要求于2021年8月11日提交中国专利局、申请号为202110918567.7，发明名称为“基于预测模型的多渠道资金调拨方法、装置、设备和介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及智能决策技术领域，特别是涉及一种基于预测模型的多渠道资金调拨方法、装置、设备和介质。
背景技术
为了充分、快速、准确地进行资金运营，降低资金流动性风险，资金运营人员需要登录不同银行机构的网银进行资金调拨操作，保障企业机构业务能够正常运营。
发明人意识到，目前，资金调拨操作主要由资金运营人员手动地生成调拨指令来进行。随着金融科技和企业信息化的发展，少数企业机构的内部运营系统可以自动地生成资金调拨指令，但是，调入、调出资金渠道往往是固定的，缺少灵活性，无法智能化地根据各个资金渠道的实际资金情况来生成资金调拨指令。
发明内容
本申请实施例提供了一种基于预测模型的多渠道资金调拨方法、装置、设备和介质，实施本申请实施例，使得资金调拨操作得以智能化地根据资金渠道的实际资金情况来进行。
第一方面，本申请实施例提供了一种基于预测模型的多渠道资金调拨方法，上述方法包括：
获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；
检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；
将调出资金渠道的部分余额调拨给调入资金渠道。
第二方面，本申请实施例提供了一种基于预测模型的多渠道资金调拨装置，上述装置包括：
获取单元，用于获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
输入单元，用于将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；
检测单元，用于检测每个资金渠道的余额，当每个资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；
调拨单元，用于将调出资金渠道的部分余额调拨给调入资金渠道。
第三方面，本申请实施例提供了一种电子设备，包括处理器、存储器以及存储在存储器上并可在处理器上运行的计算机执行指令，当计算机执行指令被运行时，使得电子设备执行基于预测模型的多渠道资金调拨方法，所述方法包括：
获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
将调出资金渠道的部分余额调拨给所述调入资金渠道。
第四方面，本申请实施例提供了一种计算机可读存储介质，计算机可读存储介质中存储有计算机指令，当计算机指令在通信装置上运行时，使得通信装置执行基于预测模型的多渠道资金调拨方法，所述方法包括：
获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
将调出资金渠道的部分余额调拨给所述调入资金渠道。
本申请实施例使得资金调拨操作得以智能化地根据资金渠道的实际情况来进行，降低企业的资金运营成本。
附图说明
图1A是本申请实施例提供的一种资金运营系统的结构部署图；
图1B是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的流程示意图；
图1C是本申请实施例提供的一种资金使用预测模型的结构示意图；
图1D是本申请实施例应用的一种基于预测模型的资金运营系统的结构部署图；
图2A是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图；
图2B是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图；
图2C是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图；
图3是本申请实施例提供的一种基于预测模型的多渠道资金调拨装置的结构示意图；
图4是本申请的实施例提供的一种电子设备的硬件运行环境的服务器结构示意图。
具体实施方式
本申请可涉及人工智能技术领域，如可具体涉及智能决策技术。比如可以基于人工智能技术对相关的数据进行获取和处理。
可选的，本申请的技术方案可应用于各种基于预测模型的资金调拨场景，如数字医疗中的医疗专项资金调拨场景。例如，医疗系统可基于预测模型的多渠道资金调拨方法来实现根据资金渠道的实际资金情况智能化进行资金调拨操作。进一步可选的，本申请涉及的数据如渠道参数、余额等可存储于区块链中。
下面结合附图对本申请实施例中所涉及的设备进行介绍。
图1A是本申请实施例提供的一种资金运营系统的结构部署图。如图1A所示，该系统包括渠道设定模块、多个调入资金渠道(调入资金渠道1、调入资金渠道2……调入资金渠道N)、多个调出资金渠道(调出资金渠道1、调出资金渠道2……调出资金渠道N)、资金调拨模块。
其中，渠道设定模块，用于让资金运营人员对多个资金渠道的调拨类型、调拨目标进行设定，得到多个调入资金渠道和多个调出资金渠道。
其中，多个调入资金渠道中的某一调入资金渠道，指通过资金调拨模块接收来自至少一个调出资金渠道的部分余额的资金渠道。
其中，多个调出资金渠道中的某一调出资金渠道，指通过资金调拨模块将部分余额调拨给至少一个调入资金渠道的资金渠道。
其中，资金调拨模块，用于根据资金运营人员的要求生成资金调拨指令，从而将调出资金渠道的部分余额调拨给调入资金渠道。
在上述系统进行资金调拨的过程中，资金渠道的调拨类型、调拨目标是资金运营人员 进行设定的，存在调出渠道在自身余额已低于一定阈值却仍然处于资金调出状态的风险，无法根据资金渠道的实际情况来进行资金调拨。
基于此，本申请实施例提供了一种基于预测模型的多渠道资金调拨方法，请参阅图1B，图1B是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的流程示意图，如图1B所示，该方法包括以下步骤：
101：获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
其中，资金渠道，包括银行账户、第三方支付平台。
其中，第一时间，指过往历史时间中的某一时间。可以是过往1个月、过往1年等。
其中，资金渠道在第一时间内的资金变化记录，包括资金渠道在第一时间内的资金增减幅度情况、资金增减速度情况。
示例性地，第一时间为过往1年，则资金渠道在第一时间内的资金变化记录，包括资金渠道在过往1年内每日不同时间段的资金增减幅度情况、资金增减速度情况。
其中，渠道权重，可以由人工根据对应资金渠道的具体参数信息来设定。示例性地，具体参数信息可以是资金渠道的使用频率高低、资金渠道所对应的公司产品的重要性等参数信息来设定。
示例性地，渠道权重根据资金渠道对应的公司产品或业务的重要性来确定，资金渠道1对应的公司产品的重要性高于资金渠道2对应的公司产品，则资金渠道1的渠道权重＞资金渠道2的渠道权重。
102：将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；
其中，第二时间，指未来的某一时间。可以是未来1小时、当前工作日17:00之后直到下一个工作日的8：00前等。
其中，将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，即是说，资金使用预测模型根据每个资金渠道当前的渠道参数，预测得到每个资金渠道在未来的某一时间内需要使用的金额额度。
示例性地，第一余额阈值为10000元，第二时间为未来的1小时，则表示资金使用预测模型预测得出资金渠道在未来的1小时内需要使用的金额额度为10000元。
又一示例性地，第二时间为当前工作日17:00之后直到下一个工作日的8：00前的这一时间段，资金使用预测模型预测得到每个资金渠道在这一时间段内需要使用的金额额度，从而根据资金渠道的情况进行资金调拨，进而保障非工作时段中各个资金渠道的资金使用情况能够处于正常状态。
其中，资金使用预测模型可以包括输入层、卷积层、池化层、损失函数层。
示例性地，请参阅图1C，图1C是本申请实施例提供的一种资金使用预测模型的结构示意图，如图1C所示，该资金使用预测模型包括以下层次：
输入层，用于接收每个资金渠道的渠道参数作为模型参数，渠道参数包括资金渠道在第一时间内的资金变化记录、渠道权重，并对每个资金渠道的渠道参数进行归一化处理，消除渠道参数的量纲，得到每个资金渠道初步处理后的渠道参数；
卷积层，用于对上述每个资金渠道初步处理后的渠道参数进行特征提取，得到每个资金渠道的渠道参数数据特征，示例性地，渠道参数数据特征包括每日资金增减幅度中的资金增大、资金减小，渠道当日的资金增减速度的增速大/小、减速大/小；
池化层：用于对上述每个资金渠道的渠道参数数据特征进行压缩，提取出每个资金渠道的渠道参数数据特征对应的每个资金渠道的渠道参数主要特征，简化计算复杂度，示例性地，渠道参数主要特征包括每日资金增减幅度中的资金增大最大幅度、资金减小最小幅 度，渠道当日的资金增减速度的最大增速、最小增速、最大减速、最小减速；
损失函数层：用于根据上述每个资金渠道的渠道参数主要特征得到每个资金渠道在第一历史时间内的预测使用金额，并根据第一历史时间内的预测使用金额和实际使用金额，计算资金使用预测模型的训练过程的在第一历史时间内的预测使用金额和实际使用金额之间的亏损值，从而对输入资金使用预测模型的模型参数进行调整。目标损失函数包括Softmax交叉熵损失函数，示例性地，
Figure PCTCN2021114678-appb-000001
其中，ai为第i个资金渠道的渠道参数，T为输入的资金渠道总数，Si为第i个输入的资金渠道在第一历史时间内的预测使用金额的概率，yi为第i个输入的资金渠道在第一历史时间内的实际使用金额的概率分布，c为目标损失值，其中c越小则表示预测结果越准确。
需要说明的是，如图1C所示的资金使用预测模型仅作为一种资金使用预测模型的示例，在具体的应用中，资金使用预测模型还可以以其他层次组成的形式存在。
103：检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；
其中，调入资金渠道，指在资金调拨过程中需要得到来自调出资金渠道的余额从而对自身余额进行补充的资金渠道。
104：将调出资金渠道的部分余额调拨给调入资金渠道。
其中，在将调出资金渠道的部分余额调拨给调入资金渠道之前，可以生成资金调拨审核指令给用户终端，得到用户终端的确认指令后，再进行资金调拨。
示例性地，本申请实施例的方法应用在资金运营系统中，资金调拨审核指令是手机短信，在生成资金调拨指令后，资金运营系统主动发手机短信给资金运营人员的手机号，提醒资金运营人员对资金调拨指令进行审核，在资金运营人员对手机短信中的资金调拨指令进行了确认之后再进行资金调拨，在确保资金安全的情况下同时保证了资金调拨的执行时效。
其中，调出资金渠道，指在资金调拨过程中需要将自身部分余额调拨给调入资金渠道从而自身余额会减少的资金渠道。
其中，本申请实施例的方法可以应用在用户终端中，用户终端可以是手机、平板电脑、个人数字助理、穿戴式设备等电子设备。
下面结合附图对本申请实施例中所涉及的设备进行介绍。
请参阅图1D，图1D是本申请实施例应用的一种基于预测模型的资金运营系统的结构部署图，如图1D所示，该资金运营系统包括渠道参数获取模块、资金使用预测模型、渠道余额检测模块、资金调拨模块，资金运营系统的第一端与用户终端连接、第二端通过防火墙与多个资金渠道连接。其中，每个模块的功能可以由单独的服务器来实现，也可以是多个模块的功能由一个服务器实现。实现不同模块功能的多个服务器互相通信连接。
其中，用户终端是使用资金运营系统的角色，用户终端可以根据企业机构的实际情况人工地设定每个资金渠道的渠道权重，同时可以确定出资金渠道中的调出资金渠道，并将设定结果输入到资金运营系统中。
其中，资金运营系统中的渠道参数获取模块，用于获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重，资金渠道在第一时间内的资金变化记录由资金渠道提供，资金渠道的渠道权重可以由用户终端提 供。
其中，资金运营系统中的资金使用预测模型，用于将每个资金渠道的渠道参数作为模型参数，预测出资金渠道在第二时间内需要使用的金额额度作为资金渠道的第一余额阈值。
其中，资金运营系统中的渠道余额检测模块，用于检测每个资金渠道的余额，并且在某个资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道。
其中，资金运营系统中的资金调拨模块，用于进行将调出资金渠道的部分余额调拨给调入资金渠道的资金调拨处理。
其中，防火墙(防火墙1、防火墙2)用于防止资金运营系统和多个资金渠道遭到恶意外来攻击，避免与资金相关的重要金融数据遭到泄露，保护资金运营系统和多个资金渠道的安全。
其中，多个资金渠道包括多个调入资金和多个调出资金渠道，调入资金渠道指在资金调拨过程中需要得到来自调出资金渠道的余额从而对自身余额进行补充的资金渠道，调出资金渠道指在资金调拨过程中需要将自身部分余额调拨给调入资金渠道从而自身余额会减少的资金渠道。
示例性地，有资金渠道1和资金渠道2，其中，资金渠道2是调出资金渠道，资金渠道1的余额为3000元，获取到资金渠道1在过往1年内的资金变化记录和渠道权重，并将资金渠道1在过往1年内的资金变化记录和渠道权重输入资金使用预测模型，预测出资金渠道1在未来1小时内需要使用的金额额度为5000元，即资金渠道1的第一余额阈值为5000元，而资金渠道1的余额只有3000元小于5000元，即是说，此时资金渠道1的余额不足以支付其在未来1小时内使用的金额额度，为了保障资金渠道1的资金正常运作，资金渠道1还需要得到5000-3000＝2000元的资金补充，因此，确定资金渠道1为调入资金渠道，并且将作为资金渠道2中的2000元调拨给资金渠道1，使得资金渠道1的余额达到第一余额阈值即5000元，从而完成资金调拨。
可以看出，本申请实施例中，通过获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；将调出资金渠道的部分余额调拨给调入资金渠道。通过基于预测模型的多渠道资金调拨方法，将资金渠道在第一时间内的渠道参数输入资金使用预测模型，得到资金渠道在第二时间内需要使用的金额额度，进而将余额匮乏的资金渠道作为调入资金渠道以接收来自调出资金渠道的资金补充，使得资金调拨操作得以智能化地根据资金渠道的实际情况来进行，降低企业的资金运营成本。
在一个可能的示例中，调出资金渠道的余额大于第二余额阈值，将调出资金渠道的部分余额调拨给调入资金渠道还包括：调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值，其中该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值。
其中，调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值，目的是防止调出资金渠道在余额匮乏时却依然处于资金调出状态、导致资金调拨的过程中存在资金周转风险的问题。
其中，该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值，目的是防止调出资金渠道由于余额小于或等于第一余额阈值而被更改为调入资金渠道，通过保持调出资金渠道的调拨类型不变来确保资金调拨过程的稳定性。
可以看出，本申请实施例中，通过调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值且该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值，避免调出资金渠道出现余额匮乏的风险、同时保持了调出资金渠道的调拨类型不变，提高资金调拨 过程的安全性和稳定性。
在一个可能的示例中，资金使用预测模型的训练过程如下：获取训练数据集，训练数据集为每个资金渠道在第一历史时间内的渠道关联参数，渠道关联参数为可能影响每个资金渠道的初始金额调拨情况的参数；将训练数据集输入目标预测模型，获得每个资金渠道在第一历史时间内的预测使用金额；将每个资金渠道的预测使用金额与该资金渠道的实际使用金额输入目标偏差函数，根据目标偏差函数计算得到目标偏差值；在目标偏差值大于预设偏差值的情况下，检测渠道关联参数中的偏差异常参数，偏差异常参数为导致目标偏差值大于预设偏差值的关键项；使用备选渠道关联参数替换偏差异常参数，得到新的训练数据集，重复将新的训练数据集输入目标预测模型、获得预测使用金额的过程，直到确定预测使用金额与实际使用金额的目标偏差值小于预设差值时，确定新的训练数据集中包括的渠道关联参数为渠道参数，包括渠道参数作为模型参数的目标预测模型为资金使用预测模型。
其中，渠道关联参数指资金渠道的关联参数，这些关联参数为可能影响渠道资金使用情况的参数，具体可能为资金渠道在第一历史时间内的资金变化记录、渠道权重、资金峰-谷值、资金平均值等，其中资金变化记录包括资金增减幅度情况、资金增减速度情况。在模型训练过程中，需要确定这些渠道关联参数中决定性影响渠道资金使用情况的参数，即渠道参数。
其中，第一历史时间内的预测使用金额，在具体实现中，可以是根据每个资金渠道在第一历史时间内的初始金额调拨情况和最终金额调拨情况来进行计算得到，预测使用金额＝最终金额调拨情况-初始金额调拨情况。
其中，目标损失函数包括Softmax交叉熵损失函数。
其中，偏差异常参数是指推测出来的渠道关联参数中，实际并不对资金使用情况产生决定性影响的参数。
其中，检测渠道关联参数中的偏差异常参数，在具体实现中，可以是利用标准化回归系数来来检测渠道关联参数中导致目标偏差值大于预设偏差值的关键项。标准化回归系数，是在对目标偏差函数中的自变量和因变量同时进行标准化处理后所得到的回归系数，其中，自变量为第一历史时间内的渠道关联参数，因变量为每个资金渠道在第一历史时间内的预测使用金额，数据经过标准化处理后消除了量纲、数量级等差异的影响，使得不同变量之间具有可比性，因此可以用标准化回归系数来比较不同自变量对因变量的作用大小，从而确定出渠道关联参数中的偏差异常参数。
示例性地，训练资金使用预测模型的过程中，获取每个资金渠道在第一历史时间内的渠道权重和资金峰-谷值作为第一训练数据集，将第一训练数据集输入目标预测模型，获得每个资金渠道在第一历史时间内的第一预测使用金额，将每个资金渠道的第一预测使用金额与该资金渠道的实际使用金额输入目标偏差函数，根据目标偏差函数计算得到第一目标偏差值，第一目标偏差值大于预设偏差值，检测到偏差异常参数是资金峰-谷值，则使用作为备选渠道关联参数的资金变化记录替换资金峰-谷值，得到新的包括在第一历史时间内的资金变化记录、渠道权重的第二训练数据集，重复将第二训练数据集输入目标预测模型、获得预测使用金额的过程，此时确定到在第一历史时间内的第二预测使用金额与实际使用金额的目标偏差值小于预设差值，则确定第二训练数据集中包括的第一历史时间内的资金变化记录、渠道权重为渠道参数，包括在第一历史时间内的资金变化记录、渠道权重作为模型参数的目标预测模型为资金使用预测模型，至此，完成资金使用预测模型的训练过程。
可以看出，本申请实施例中提供的资金使用预测模型训练过程，只有确定了真正影响资金使用情况的渠道参数，并将这些参数应用到资金使用预测模型中，才能够保证模型预测的准确性。进一步地，使用通过本申请实施例提供的训练过程训练得到的资金使用预测 模型进行资金调拨，能够在解放人力的同时保障资金调拨过程中智能决策的准确性，降低资金调拨过程的风险性，减少企业机构的人力成本和资金运营成本。
在一个可能的示例中，将调出资金渠道的部分余额调拨给调入资金渠道，具体为：计算多个调入资金渠道中的每个调入资金渠道的调入资金渠道余额距离该调入资金渠道对应的第三余额阈值的差值，得到补充额度h＝第三余额阈值-调入资金渠道余额，其中h为第h个调入资金渠道且1≤h≤i，其中该资金渠道的第三余额阈值≥该资金渠道的第一余额阈值，第三余额阈值为每个调入资金渠道经过资金调拨后需要持有的金额额度；计算多个调出资金渠道中的每个调出资金渠道的可调出资金量，可调出资金量j＝调出资金渠道余额-第二余额阈值，其中j为第j个调出资金渠道；计算可调入资金
Figure PCTCN2021114678-appb-000002
若可调出资金量大于预设额度，则执行可调入资金i对应的资金调拨，可调入资金i为第i个调入资金渠道的调入资金量。
其中，该资金渠道的第三余额阈值≥该资金渠道的第一余额阈值，目的是使得每个调入资金渠道经过资金调拨后持有的金额额度可以维持在高于第一余额阈值的较高水平，由于第一余额阈值为资金渠道在第二时间内需要使用的金额额度，因此，调入资金渠道在经过资金调拨后持有的金额额度高于第一余额阈值时可以更灵活地应对资金运营过程中可能的突发情况或紧急情况造成的不稳定性，降低资金运营的风险。
其中，预设额度可以是0。
其中，若可调出资金量大于预设额度，则执行可调入资金i对应的资金调拨，反之，若可调出资金量小于或等于预设额度，则资金调拨至第i-1个调入资金渠道的调入资金渠道余额为第三余额阈值时便停止，不再对第i个调入资金渠道进行资金调拨。
示例性地，请参阅图2A，图2A是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图，如图2A所示，对于调入资金渠道h而言，在资金调拨之前，其调入资金渠道余额低于第一余额阈值，为了把调入资金渠道h的调入资金渠道余额补充到第三余额阈值，计算得出补充额度h＝第三余额阈值-调入资金渠道余额，在经过资金调拨之后，调入资金渠道h的调入资金渠道余额更新为第三余额阈值。
示例性地，请参阅图2B，图2B是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图，如图2B所示，对于调出资金渠道j而言，在资金调拨之前，其调出资金渠道余额高于第二余额阈值，为了把可调出资金量j调拨给其他调入资金渠道，计算得出可调出资金量j＝调出资金渠道余额-第二余额阈值，在经过资金调拨之后，调出资金渠道j的调出资金渠道余额更新为第二余额阈值。
可以看出，本申请实施例中，通过资金调拨后将调入资金渠道的调入资金渠道余额补充到第三余额阈值，使得调入资金渠道在经过资金调拨后持有的金额额度大于或等于第一余额阈值，其中，在调入资金渠道持有的金额额度大于第一余额阈值时，可以更灵活地应对资金运营过程中可能的突发情况或紧急情况造成的不稳定性，降低资金运营的风险。
在一个可能的示例中，上述方法还包括：h为调入资金渠道的调入优先级排序，j为调出资金渠道的调出优先级排序；其中，将调出资金渠道的部分余额调拨给调入资金渠道，包括：按照调入资金渠道的调入优先级排序和调出资金渠道的调出优先级排序，将调出资金渠道的部分余额调拨给调入资金渠道。
可以看出，本申请实施例中，通过按照调入资金渠道的调入优先级排序和调出资金渠道的调出优先级排序，将调出资金渠道的部分余额调拨给调入资金渠道，使得在资金调拨的过程中，优先级排序高的调出资金渠道将优先把部分余额调拨给优先级排序高的调入资金渠道，实现了资金调拨的过程中对于调入渠道、调出渠道的智能决策。
在一个可能的示例中，在将调出资金渠道的部分余额调拨给调入资金渠道之前，上述 方法还包括：获取多个调入资金渠道中的每个调入资金渠道、多个调出资金渠道中的每个调出资金渠道的渠道权重；根据多个调入资金渠道中的每个调入资金渠道的余额和对应的第一余额阈值，计算出每个调入资金渠道的调入差值＝(第一余额阈值-余额)/第一余额阈值，且根据多个调出资金渠道中的每个调出资金渠道的余额和对应的第二余额阈值，计算出每个调出资金渠道的调出差值＝(余额-第二余额阈值)/第二余额阈值；根据每个调入资金渠道的渠道权重和调入差值，计算出每个调入资金渠道的优先值＝渠道权重*调入差值，且根据每个调出资金渠道的渠道权重和调出差值，计算出每个调出资金渠道的优先值＝渠道权重*调出差值；根据每个调入资金渠道的优先值对多个调入资金渠道的调入优先级进行排序，根据每个调出资金渠道的优先值对多个调出资金渠道的调出优先级进行排序。
其中，根据每个调入资金渠道的优先值对多个调入资金渠道的调入优先级进行排序，具体为，调入资金渠道的优先值越高则调入优先级排序越高、在资金调拨的过程中将越早得到来自调出资金渠道的余额。
其中，根据每个调出资金渠道的优先值对多个调出资金渠道的调出优先级进行排序，具体为，调出资金渠道的优先值越高则调出优先级排序越高、在资金调拨的过程中将越早将部分余额调拨给调入资金渠道。
可以看出，本申请实施例中，根据每个调入资金渠道的调入差值和渠道权重计算得到每个调入资金渠道的优先值、根据每个调出资金渠道的调出差值和渠道权重计算得到每个调出资金渠道的优先值，并根据每个调入资金渠道的优先值对调入优先级进行排序、根据每个调出资金渠道的优先值对调出优先级进行排序，从而余额距离第一余额阈值相差较大、且渠道权重大的调入资金渠道将具有较高的调入优先级，余额距离第二余额阈值超出较大、且渠道权重大的调出资金渠道将具有较高的调出优先级，进而使得资金调拨的过程中优先将具有冗余余额的调出资金渠道的部分余额调拨给余额处于严重匮乏状态的调入资金渠道，实现了资金调拨的过程中对于调入渠道、调出渠道的资金调拨过程的智能决策，使得资金调拨的过程更符合资金渠道的实际资金情况。
在一个可能的示例中，上述方法还包括：获取调入资金渠道和调出资金渠道在预设天数内的N个时间段中每个时间段的资金调拨量，得到N个时间段中每个时间段的平均资金调拨量，按照预设的平均资金调拨量和调拨频次系数之间的映射关系，得到每个时间段的目标调拨频次系数；获取每个时间段的资金调拨次数，按照预设的资金调拨次数和频次调节系数之间的映射关系，得到每个时间段的目标频次调节系数；根据目标调拨频次系数和目标频次调节系数，计算出每个时间段的目标刷新频次；在每个时间段内对调入资金渠道的调入优先级和调出资金渠道的调出优先级进行目标刷新频次对应次数的优先值刷新，优先值刷新用于确定每个调入资金渠道、每个调出资金渠道的更新后的优先值。
其中，每个时间段的平均资金调拨量＝每个时间段内调出资金渠道调拨给调入资金渠道的资金量＝每个时间段内调入资金渠道接收的来自调出资金渠道的资金量。
其中，每个时间段的资金调拨次数＝每个时间段内调出资金渠道调拨给调入资金渠道的次数＝每个时间段内调入渠道接收来自调出资金渠道的资金的次数。
其中，根据目标调拨频次系数和目标频次调节系数，计算出每个时间段的目标刷新频次，可以是：目标刷新频次＝目标调拨频次系数*目标频次调节系数。
其中，在每个时间段内对调入资金渠道的调入优先级和调出资金渠道的调出优先级进行目标刷新频次对应次数的优先值刷新，目的是，根据资金渠道在不同时间段的资金调拨实际情况对资金渠道的优先值进行符合不同时间段的资金调拨实际情况的对应次数的刷新，避免资金渠道优先值的固定化导致资金调拨过程中无法应对资金变化的突发情况。
示例性地，目标刷新频次＝目标调拨频次系数*目标频次调节系数，将一天分成12个时段，每个时段的时长为2小时，即将一天分成0:01-2:00、2:01-4:00……12:01-14:00…… 16:01-18:00……22:01-0:00共12个时间段，请参阅图2C，图2C是本申请实施例提供的一种基于预测模型的多渠道资金调拨方法的举例示意图，如图2C所示，其中，2:01-4:00的平均资金调拨量为500元、资金调拨次数为1次，16:01-18:00的平均资金调拨量为30000元、资金调拨次数为10次，预设的平均资金调拨量和调拨频次系数之间的映射关系如表1所示：
平均资金调拨量 调拨频次系数
1≤平均资金调拨量＜5000 3
5000≤平均资金调拨量＜15000 5
15000≤平均资金调拨量＜50000 8
表1
则2:01-4:00时间段的目标调拨频次系数1为3，16:01-18:00时间段的目标调拨频次系数2为8，预设的资金调拨次数和频次调节系数之间的映射关系如表2所示：
资金调拨次数 频次调节系数
1≤资金调拨次数＜3 0.7
3≤资金调拨次数＜5 1
5≤资金调拨次数＜15 1.5
表2
则2:01-4:00时间段的目标频次调节系数1为0.7，16:01-18:00时间段的目标频次调节系数2为1.5，计算出2:01-4:00时间段的目标刷新频次1＝目标调拨频次系数1*目标频次调节系数1＝3*0.7＝2.1≈2，计算出16:01-18:00时间段的目标刷新频次2＝目标调拨频次系数2*目标频次调节系数2＝8*1.5＝12，因此，在2:01-4:00时间段内对调入资金渠道的调入优先级和调出资金渠道的调出优先级进行2次优先值刷新，在16:01-18:00时间段内对调入资金渠道的调入优先级和调出资金渠道的调出优先级进行12次优先值刷新。
可以看出，本申请实施例中，通过调入资金渠道和调出资金渠道在N个时间段中每个时间段的平均资金调拨量和资金调拨次数，得到每个时间段的目标刷新频次，在每个时间段内对调入资金渠道的调入优先级和调出资金渠道的调出优先级进行目标刷新频次对应次数的优先值刷新。使得调入资金渠道和调出资金渠道的优先值能够根据每个时间段的平均资金调拨量和资金调拨次数的实际情况来进行对应次数的刷新，避免了资金调拨过程中由于资金渠道优先值的固定化导致无法灵活地应对资金调拨过程中资金变化的突发情况，使得资金调拨的智能决策过程更具备灵活性。
在一个可能的示例中，上述方法还包括：获取多个调出资金渠道中每个调出资金渠道的渠道利率；将渠道利率大于预设利率的多个调出资金渠道确定为高利率调出资金渠道，将渠道利率小于或等于预设利率的多个调出资金渠道确定为低利率调出资金渠道；根据多个调出资金渠道中每个调出资金渠道的渠道利率和第二余额阈值，计算出多个高利率调出资金渠道中每个高利率调出资金渠道的第四余额阈值＝第二余额阈值^(1+渠道利率)，且计算出多个低利率调出资金渠道中每个低利率调出资金渠道的第四余额阈值＝第二余额阈值^(1-渠道利率)；将调出资金渠道的第二余额阈值更新为该调出资金渠道对应的第四余额阈值。
其中，渠道利率，即每个调出资金渠道对应的银行利率，银行利率表示一定时期内利息与本金的比率，用百分比表示。示例性地，银行利率是0.3％。
其中，高利率调出资金渠道的第四余额阈值＝第二余额阈值^(1+渠道利率)，可以看出，高利率调出资金渠道的第四余额阈值大于第二余额阈值，目的是，由于高利率调出资金渠道本身的银行利率较高，余额较多的话利息也相应较多，因此高利率调出资金渠道可以保持有相比第二余额阈值略高的余额。
其中，低利率调出资金渠道的第四余额阈值＝第二余额阈值^(1-渠道利率)，可以看出，低利率调出资金渠道的第四余额阈值小于第二余额阈值，目的是，由于低利率调出资金渠道本身的银行利率较低，即便余额较多的话利息也相应较少，为了能够从资金盈利等角度充分运用好资金渠道中的余额，低利率调出资金渠道中的余额不应当维持在较高的水平，因此，将低利率调出资金渠道的第二余额阈值更新为相比第二余额阈值较低的第四余额阈值。
可以看出，本申请实施例中，通过获取多个调出资金渠道中每个调出资金渠道的渠道利率，将高利率调出资金渠道的第二余额阈值更新为相比第二余额阈值较高的第四余额阈值，将低利率调出资金渠道的第二余额阈值更新为相比第二余额阈值较低的第四余额阈值，从而高利率调出资金渠道可以保持有相比第二余额阈值略高的余额，同时低利率调出资金渠道则保持相比第二余额阈值略低的余额，通过使得每个调出资金渠道经过资金调拨后的余额能够保持在符合各自实际情况的合理水平，充分提高了调出资金渠道的余额的利用程度。
与上述图1B所示的实施例一致的，请参阅图3，图3是本申请实施例提供的一种基于预测模型的多渠道资金调拨装置的结构示意图，如图3所示：
一种基于预测模型的多渠道资金调拨装置，上述装置包括：
301：获取单元，用于获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
302：输入单元，用于将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；
303：检测单元，用于检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；
304：调拨单元，用于将调出资金渠道的部分余额调拨给调入资金渠道。
可以看出，本申请实施例中，通过获取单元获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；通过输入单元将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；通过检测单元检测每个资金渠道的余额，当每个资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；通过调拨单元将调出资金渠道的部分余额调拨给调入资金渠道。通过基于预测模型的多渠道资金调拨装置，将资金渠道在第一时间内的渠道参数输入资金使用预测模型，得到资金渠道在第二时间内需要使用的金额额度，进而将余额匮乏的资金渠道作为调入资金渠道以接收来自调出资金渠道的资金补充，使得资金调拨操作得以智能化地根据资金渠道的实际情况来进行，降低企业的资金运营成本。
具体地，本申请实施例可以根据上述方法示例对基于预测模型的多渠道资金调拨装置进行功能单元的划分，例如，可以对应各个功能划分各个功能单元，也可以将两个或两个以上的功能集成在一个处理单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。需要说明的是，本申请实施例中对单元的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。
与上述图1B所示的实施例一致的，本申请实施例提供了一种电子设备，请参阅图4，图4是本申请的实施例提供的一种电子设备的硬件运行环境的服务器结构示意图，如图4所示，电子设备包括处理器、存储器以及存储在存储器上并可在处理器上运行的计算机执行指令，当计算机执行指令被运行时，使得电子设备执行任一项基于预测模型的多渠道资金调拨方法。
其中，处理器为CPU。
其中，存储器，可选的，存储器可以为高速RAM存储器，也可以是稳定的存储器，例如磁盘存储器。
本领域技术人员可以理解，图4中示出的服务器的结构并不构成对其的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。
如图4所示，存储器中可以包括操作系统、网络通信模块以及基于预测模型的多渠道资金调拨的计算机执行指令。操作系统用于管理和控制服务器硬件和软件资源，支持计算机执行指令的运行。网络通信模块用于实现存储器内部各组件之间的通信，以及与服务器内部其他硬件和软件之间通信，通信可以使用任一通信标准或协议，包括但不限于GSM(Global System of Mobile communication，全球移动通讯系统)、GPRS(General Packet Radio Service，通用分组无线服务)、CDMA2000(CodeDivision Multiple Access 2000，码分多址2000)、WCDMA(Wideband Code DivisionMultiple Access,宽带码分多址)、TD-SCDMA(Time Division-Synchronous CodeDivision Multiple Access，时分同步码分多址)等。
在图4所示的服务器中，处理器用于运行存储器中存储的人员管理的计算机执行指令，实现以下步骤：获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；检测每个资金渠道的余额，当每个资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；将调出资金渠道的部分余额调拨给调入资金渠道。
本申请涉及的服务器的具体实施可参见上述基于预测模型的多渠道资金调拨方法的各实施例，在此不做赘述。
本申请实施例提供了一种计算机可读存储介质，计算机可读存储介质中存储有计算机指令，当计算机指令在通信装置上运行时，使得通信装置执行以下步骤：获取多个资金渠道中的每个资金渠道的渠道参数，渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；将每个资金渠道的渠道参数输入资金使用预测模型，得到每个资金渠道对应的第一余额阈值，第一余额阈值为资金渠道在第二时间内需要使用的金额额度；检测每个资金渠道的余额，当资金渠道的余额小于或等于第一余额阈值时，确定该资金渠道为调入资金渠道；将调出资金渠道的部分余额调拨给调入资金渠道。上述计算机包括电子设备。
可选的，本申请涉及的存储介质如计算机可读存储介质可以是非易失性的，也可以是易失性的。
其中，电子设备包括手机、平板电脑、个人数字助理、穿戴式设备等。
其中，计算机可读存储介质可以是上述实施例所述的电子设备的内部存储单元，例如电子设备的硬盘或内存。计算机可读存储介质也可以是上述电子设备的外部存储设备，例如电子设备上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。进一步地，计算机可读存储介质还可以既包括电子设备的内部存储单元也包括外部存储设备。计算机可读存储介质用于存储计算机执行指令以及电子设备所需的其他计算机执行指令和数据。计算机可读存储介质还可以用于暂时地存储已经输出或者将要输出的数据。
本申请涉及的计算机可读存储介质的具体实施可参见上述基于预测模型的多渠道资金调拨方法的各实施例，在此不做赘述。
本申请实施例提供了一种计算机程序产品，其中，计算机程序产品包括计算机程序，计算机程序可操作来使计算机如上述方法实施例中记载的任何一种基于预测模型的多渠道资金调拨方法的部分或全部步骤，该计算机程序产品可以是一个软件安装包。
需要说明的是，对于前述的任一种基于预测模型的多渠道资金调拨方法的实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本申请并不受所描述的动作顺序的限制，因为依据本申请，某些步骤可以采用其他顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作并不一定是本申请所必须的。
以上对本申请实施例进行了详细介绍，本文中应用了具体个例对本申请一种基于预测模型的多渠道资金调拨方法、装置、设备和介质的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本申请的方法及其核心思想；同时，对于本领域的一般技术人员，依据本申请一种基于预测模型的多渠道资金调拨方法、装置、设备和介质的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本申请的限制。
本申请是参照本申请实施例的方法、硬件产品和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。存储器可以包括：闪存盘、只读存储器(英文：Read-Only Memory，简称：ROM)、随机存取器(英文：Random Access Memory，简称：RAM)、磁盘或光盘等。
尽管在此结合各实施例对本申请进行了描述，然而，在实施所要求保护的本申请过程中，本领域技术人员通过查看附图、公开内容、以及所附权利要求书，可理解并实现所公开实施例的其他变化。在权利要求中，“包括”(comprising)一词不排除其他组成部分或步骤，“一”或“一个”不排除多个的情况。相互不同的从属权利要求中记载了某些措施，但这并不表示这些措施不能组合起来产生良好的效果。
本领域普通技术人员可以理解上述任一种基于预测模型的多渠道资金调拨方法的方法实施例的各种方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，该程序可以存储于计算机可读存储器中，存储器可以包括：闪存盘、只读存储器(英文：Read-Only Memory，简称：ROM)、随机存取器(英文：Random Access Memory，简称：RAM)、磁盘或光盘等。
显然，本领域的技术人员可以对本申请提供的一种基于预测模型的多渠道资金调拨方法、装置、设备和介质进行各种改动和变型而不脱离本申请的精神和范围。这样，倘若本申请的这些修改和变型属于本申请权利要求及其等同技术的范围之内，则本申请也意图包含这些改动和变型在内。

Claims (20)

  1. 一种基于预测模型的多渠道资金调拨方法，所述方法包括：
    获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
    将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
    检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
    将调出资金渠道的部分余额调拨给所述调入资金渠道。
  2. 根据权利要求1所述的方法，其中，所述调出资金渠道的余额大于第二余额阈值，所述将调出资金渠道的部分余额调拨给所述调入资金渠道还包括：所述调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值，其中该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值。
  3. 根据权利要求1或2所述的方法，其中，所述资金使用预测模型的训练过程如下：
    获取训练数据集，所述训练数据集为所述每个资金渠道在第一历史时间内的渠道关联参数，所述渠道关联参数为可能影响所述每个资金渠道的初始金额调拨情况的参数；
    将所述训练数据集输入目标预测模型，获得所述每个资金渠道在所述第一历史时间内的预测使用金额；
    将所述每个资金渠道的预测使用金额与该资金渠道的实际使用金额输入目标偏差函数，根据所述目标偏差函数计算得到目标偏差值；
    在所述目标偏差值大于预设偏差值的情况下，检测所述渠道关联参数中的偏差异常参数，所述偏差异常参数为导致所述目标偏差值大于预设偏差值的关键项；
    使用备选渠道关联参数替换所述偏差异常参数，得到新的训练数据集，重复将所述新的训练数据集输入所述目标预测模型、获得预测使用金额的过程，直到确定所述预测使用金额与所述实际使用金额的目标偏差值小于预设差值时，确定所述新的训练数据集中包括的渠道关联参数为所述渠道参数，包括所述渠道参数作为模型参数的目标预测模型为所述资金使用预测模型。
  4. 根据权利要求2所述的方法，其中，所述将调出资金渠道的部分余额调拨给所述调入资金渠道，具体为：
    计算所述多个调入资金渠道中的每个调入资金渠道的调入资金渠道余额距离该调入资金渠道对应的第三余额阈值的差值，得到补充额度h＝第三余额阈值-调入资金渠道余额，其中h为第h个调入资金渠道且1≤h≤i，其中该资金渠道的第三余额阈值≥该资金渠道的第一余额阈值，所述第三余额阈值为所述每个调入资金渠道经过资金调拨后需要持有的金额额度；
    计算所述多个调出资金渠道中的每个调出资金渠道的可调出资金量，所述可调出资金量j＝调出资金渠道余额-第二余额阈值，其中j为第j个调出资金渠道；
    计算可调入资金
    Figure PCTCN2021114678-appb-100001
    若所述可调出资金量大于预设额度，则执行所述可调入资金i对应的资金调拨，所述可调入资金i为第i个调入资金渠道的调入资金量。
  5. 根据权利要求4所述的方法，其中，所述方法还包括：
    所述h为所述调入资金渠道的调入优先级排序，所述j为所述调出资金渠道的调出优先级排序；
    其中，所述将调出资金渠道的部分余额调拨给所述调入资金渠道，包括：
    按照所述调入资金渠道的调入优先级排序和所述调出资金渠道的调出优先级排序，将所述调出资金渠道的部分余额调拨给所述调入资金渠道。
  6. 根据权利要求5所述的方法，其中，在所述将调出资金渠道的部分余额调拨给所述调入资金渠道之前，所述方法还包括：
    获取所述多个调入资金渠道中的每个调入资金渠道、所述多个调出资金渠道中的每个调出资金渠道的渠道权重；
    根据所述多个调入资金渠道中的每个调入资金渠道的余额和对应的第一余额阈值，计算出所述每个调入资金渠道的调入差值＝(第一余额阈值-余额)/第一余额阈值，且根据所述多个调出资金渠道中的每个调出资金渠道的余额和对应的第二余额阈值，计算出所述每个调出资金渠道的调出差值＝(余额-第二余额阈值)/第二余额阈值；
    根据所述每个调入资金渠道的渠道权重和调入差值，计算出每个调入资金渠道的优先值＝渠道权重*调入差值，且根据所述每个调出资金渠道的渠道权重和调出差值，计算出每个调出资金渠道的优先值＝渠道权重*调出差值；
    根据所述每个调入资金渠道的优先值对所述多个调入资金渠道的调入优先级进行排序，根据所述每个调出资金渠道的优先值对所述多个调出资金渠道的调出优先级进行排序。
  7. 根据权利要求6所述的方法，其中，所述方法还包括：
    获取所述调入资金渠道和所述调出资金渠道在预设天数内的N个时间段中每个时间段的资金调拨量，得到所述N个时间段中每个时间段的平均资金调拨量，按照预设的平均资金调拨量和调拨频次系数之间的映射关系，得到每个时间段的目标调拨频次系数；
    获取所述每个时间段的资金调拨次数，按照预设的资金调拨次数和频次调节系数之间的映射关系，得到每个时间段的目标频次调节系数；
    根据所述目标调拨频次系数和所述目标频次调节系数，计算出每个时间段的目标刷新频次；
    在所述每个时间段内对所述调入资金渠道的调入优先级和所述调出资金渠道的调出优先级进行目标刷新频次对应次数的优先值刷新，所述优先值刷新用于确定所述每个调入资金渠道、所述每个调出资金渠道的更新后的优先值。
  8. 一种基于预测模型的多渠道资金调拨装置，所述装置包括：
    获取单元，用于获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
    输入单元，用于将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
    检测单元，用于检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
    调拨单元，用于将调出资金渠道的部分余额调拨给所述调入资金渠道。
  9. 一种电子设备，包括处理器、存储器以及存储在所述存储器上并可在所述处理器上运行的计算机执行指令，当所述计算机执行指令被运行时，使得所述电子设备执行基于预测模型的多渠道资金调拨方法，所述方法包括：
    获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
    将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
    检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
    将调出资金渠道的部分余额调拨给所述调入资金渠道。
  10. 根据权利要求9所述的电子设备，其中，所述调出资金渠道的余额大于第二余额阈值，所述将调出资金渠道的部分余额调拨给所述调入资金渠道还包括：所述调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值，其中该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值。
  11. 根据权利要求9或10所述的电子设备，其中，所述资金使用预测模型的训练过程如下：
    获取训练数据集，所述训练数据集为所述每个资金渠道在第一历史时间内的渠道关联参数，所述渠道关联参数为可能影响所述每个资金渠道的初始金额调拨情况的参数；
    将所述训练数据集输入目标预测模型，获得所述每个资金渠道在所述第一历史时间内的预测使用金额；
    将所述每个资金渠道的预测使用金额与该资金渠道的实际使用金额输入目标偏差函数，根据所述目标偏差函数计算得到目标偏差值；
    在所述目标偏差值大于预设偏差值的情况下，检测所述渠道关联参数中的偏差异常参数，所述偏差异常参数为导致所述目标偏差值大于预设偏差值的关键项；
    使用备选渠道关联参数替换所述偏差异常参数，得到新的训练数据集，重复将所述新的训练数据集输入所述目标预测模型、获得预测使用金额的过程，直到确定所述预测使用金额与所述实际使用金额的目标偏差值小于预设差值时，确定所述新的训练数据集中包括的渠道关联参数为所述渠道参数，包括所述渠道参数作为模型参数的目标预测模型为所述资金使用预测模型。
  12. 根据权利要求10所述的电子设备，其中，执行所述将调出资金渠道的部分余额调拨给所述调入资金渠道，包括：
    计算所述多个调入资金渠道中的每个调入资金渠道的调入资金渠道余额距离该调入资金渠道对应的第三余额阈值的差值，得到补充额度h＝第三余额阈值-调入资金渠道余额，其中h为第h个调入资金渠道且1≤h≤i，其中该资金渠道的第三余额阈值≥该资金渠道的第一余额阈值，所述第三余额阈值为所述每个调入资金渠道经过资金调拨后需要持有的金额额度；
    计算所述多个调出资金渠道中的每个调出资金渠道的可调出资金量，所述可调出资金量j＝调出资金渠道余额-第二余额阈值，其中j为第j个调出资金渠道；
    计算可调入资金
    Figure PCTCN2021114678-appb-100002
    若所述可调出资金量大于预设额度，则执行所述可调入资金i对应的资金调拨，所述可调入资金i为第i个调入资金渠道的调入资金量。
  13. 根据权利要求12所述的电子设备，其中，还包括：
    所述h为所述调入资金渠道的调入优先级排序，所述j为所述调出资金渠道的调出优先级排序；
    其中，所述将调出资金渠道的部分余额调拨给所述调入资金渠道，包括：
    按照所述调入资金渠道的调入优先级排序和所述调出资金渠道的调出优先级排序，将所述调出资金渠道的部分余额调拨给所述调入资金渠道。
  14. 根据权利要求13所述的电子设备，其中，在所述将调出资金渠道的部分余额调拨给所述调入资金渠道之前，还包括：
    获取所述多个调入资金渠道中的每个调入资金渠道、所述多个调出资金渠道中的每个调出资金渠道的渠道权重；
    根据所述多个调入资金渠道中的每个调入资金渠道的余额和对应的第一余额阈值，计 算出所述每个调入资金渠道的调入差值＝(第一余额阈值-余额)/第一余额阈值，且根据所述多个调出资金渠道中的每个调出资金渠道的余额和对应的第二余额阈值，计算出所述每个调出资金渠道的调出差值＝(余额-第二余额阈值)/第二余额阈值；
    根据所述每个调入资金渠道的渠道权重和调入差值，计算出每个调入资金渠道的优先值＝渠道权重*调入差值，且根据所述每个调出资金渠道的渠道权重和调出差值，计算出每个调出资金渠道的优先值＝渠道权重*调出差值；
    根据所述每个调入资金渠道的优先值对所述多个调入资金渠道的调入优先级进行排序，根据所述每个调出资金渠道的优先值对所述多个调出资金渠道的调出优先级进行排序。
  15. 一种计算机可读存储介质，其中，所述计算机可读存储介质中存储有计算机指令，当所述计算机指令在通信装置上运行时，使得所述通信装置执行基于预测模型的多渠道资金调拨方法，所述方法包括：
    获取多个资金渠道中的每个资金渠道的渠道参数，所述渠道参数包括资金渠道在第一时间内的资金变化记录和渠道权重；
    将所述每个资金渠道的渠道参数输入资金使用预测模型，得到所述每个资金渠道对应的第一余额阈值，所述第一余额阈值为所述资金渠道在第二时间内需要使用的金额额度；
    检测所述每个资金渠道的余额，当所述资金渠道的余额小于或等于所述第一余额阈值时，确定该资金渠道为调入资金渠道；
    将调出资金渠道的部分余额调拨给所述调入资金渠道。
  16. 根据权利要求15所述的计算机可读存储介质，其中，所述调出资金渠道的余额大于第二余额阈值，所述将调出资金渠道的部分余额调拨给所述调入资金渠道还包括：所述调出资金渠道的余额保持大于或等于该资金渠道对应的第二余额阈值，其中该资金渠道的第二余额阈值＞该资金渠道的第一余额阈值。
  17. 根据权利要求15或16所述的计算机可读存储介质，其中，所述资金使用预测模型的训练过程如下：
    获取训练数据集，所述训练数据集为所述每个资金渠道在第一历史时间内的渠道关联参数，所述渠道关联参数为可能影响所述每个资金渠道的初始金额调拨情况的参数；
    将所述训练数据集输入目标预测模型，获得所述每个资金渠道在所述第一历史时间内的预测使用金额；
    将所述每个资金渠道的预测使用金额与该资金渠道的实际使用金额输入目标偏差函数，根据所述目标偏差函数计算得到目标偏差值；
    在所述目标偏差值大于预设偏差值的情况下，检测所述渠道关联参数中的偏差异常参数，所述偏差异常参数为导致所述目标偏差值大于预设偏差值的关键项；
    使用备选渠道关联参数替换所述偏差异常参数，得到新的训练数据集，重复将所述新的训练数据集输入所述目标预测模型、获得预测使用金额的过程，直到确定所述预测使用金额与所述实际使用金额的目标偏差值小于预设差值时，确定所述新的训练数据集中包括的渠道关联参数为所述渠道参数，包括所述渠道参数作为模型参数的目标预测模型为所述资金使用预测模型。
  18. 根据权利要求16所述的计算机可读存储介质，其中，执行所述将调出资金渠道的部分余额调拨给所述调入资金渠道，包括：
    计算所述多个调入资金渠道中的每个调入资金渠道的调入资金渠道余额距离该调入资金渠道对应的第三余额阈值的差值，得到补充额度h＝第三余额阈值-调入资金渠道余额，其中h为第h个调入资金渠道且1≤h≤i，其中该资金渠道的第三余额阈值≥该资金渠道的第一余额阈值，所述第三余额阈值为所述每个调入资金渠道经过资金调拨后需要持有的金额额度；
    计算所述多个调出资金渠道中的每个调出资金渠道的可调出资金量，所述可调出资金量j＝调出资金渠道余额-第二余额阈值，其中j为第j个调出资金渠道；
    计算可调入资金
    Figure PCTCN2021114678-appb-100003
    若所述可调出资金量大于预设额度，则执行所述可调入资金i对应的资金调拨，所述可调入资金i为第i个调入资金渠道的调入资金量。
  19. 根据权利要求18所述的计算机可读存储介质，其中，还包括：
    所述h为所述调入资金渠道的调入优先级排序，所述j为所述调出资金渠道的调出优先级排序；
    其中，所述将调出资金渠道的部分余额调拨给所述调入资金渠道，包括：
    按照所述调入资金渠道的调入优先级排序和所述调出资金渠道的调出优先级排序，将所述调出资金渠道的部分余额调拨给所述调入资金渠道。
  20. 根据权利要求19所述的计算机可读存储介质，其中，在所述将调出资金渠道的部分余额调拨给所述调入资金渠道之前，还包括：
    获取所述多个调入资金渠道中的每个调入资金渠道、所述多个调出资金渠道中的每个调出资金渠道的渠道权重；
    根据所述多个调入资金渠道中的每个调入资金渠道的余额和对应的第一余额阈值，计算出所述每个调入资金渠道的调入差值＝(第一余额阈值-余额)/第一余额阈值，且根据所述多个调出资金渠道中的每个调出资金渠道的余额和对应的第二余额阈值，计算出所述每个调出资金渠道的调出差值＝(余额-第二余额阈值)/第二余额阈值；
    根据所述每个调入资金渠道的渠道权重和调入差值，计算出每个调入资金渠道的优先值＝渠道权重*调入差值，且根据所述每个调出资金渠道的渠道权重和调出差值，计算出每个调出资金渠道的优先值＝渠道权重*调出差值；
    根据所述每个调入资金渠道的优先值对所述多个调入资金渠道的调入优先级进行排序，根据所述每个调出资金渠道的优先值对所述多个调出资金渠道的调出优先级进行排序。
PCT/CN2021/114678 2021-08-11 2021-08-26 基于预测模型的多渠道资金调拨方法、装置、设备和介质 WO2023015615A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110918567.7 2021-08-11
CN202110918567.7A CN113610634A (zh) 2021-08-11 2021-08-11 基于预测模型的多渠道资金调拨方法、装置、设备和介质

Publications (1)

Publication Number Publication Date
WO2023015615A1 true WO2023015615A1 (zh) 2023-02-16

Family

ID=78340254

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/114678 WO2023015615A1 (zh) 2021-08-11 2021-08-26 基于预测模型的多渠道资金调拨方法、装置、设备和介质

Country Status (2)

Country Link
CN (1) CN113610634A (zh)
WO (1) WO2023015615A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114693436A (zh) * 2022-05-30 2022-07-01 云账户技术(天津)有限公司 头寸资金调拨管理系统、方法和网络设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108764863A (zh) * 2018-05-24 2018-11-06 腾讯科技（深圳）有限公司 一种虚拟资源转移方法、装置、服务器和存储介质
CN111583046A (zh) * 2020-05-20 2020-08-25 北京九恒星科技股份有限公司 金融资源的处理方法、装置、设备以及系统
CN111815316A (zh) * 2020-07-08 2020-10-23 银联商务股份有限公司 自有资金额度预测方法、装置、电子设备及存储介质
US20200402156A1 (en) * 2019-06-24 2020-12-24 Accenture Global Solutions Limited Liquidity management system
US10990980B1 (en) * 2019-06-24 2021-04-27 Square, Inc. Predicting capital needs

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108764863A (zh) * 2018-05-24 2018-11-06 腾讯科技（深圳）有限公司 一种虚拟资源转移方法、装置、服务器和存储介质
US20200402156A1 (en) * 2019-06-24 2020-12-24 Accenture Global Solutions Limited Liquidity management system
US10990980B1 (en) * 2019-06-24 2021-04-27 Square, Inc. Predicting capital needs
CN111583046A (zh) * 2020-05-20 2020-08-25 北京九恒星科技股份有限公司 金融资源的处理方法、装置、设备以及系统
CN111815316A (zh) * 2020-07-08 2020-10-23 银联商务股份有限公司 自有资金额度预测方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN113610634A (zh) 2021-11-05

Similar Documents

Publication Publication Date Title
US9721316B2 (en) Change convergence risk mapping
US9177138B2 (en) Change convergence risk planning and avoidance
US11367075B2 (en) Method, apparatus and electronic device for identifying risks pertaining to transactions to be processed
WO2020248883A1 (zh) 一种催收方法、系统及装置
WO2020258994A1 (zh) 一种节点数据预测方法及装置
CN109784703A (zh) 业务数据处理方法、装置、计算机设备和存储介质
US20210264337A1 (en) Risk decision-making method and apparatus
US11157963B2 (en) Methods and systems for offering financial products
US20190251492A1 (en) Cognitive optimization of work permit application and risk assessment
US20230087106A1 (en) Tokenization request handling at a throttled rate in a payment network
WO2023015615A1 (zh) 基于预测模型的多渠道资金调拨方法、装置、设备和介质
US20200357059A1 (en) Multi-layer machine learning validation of income values
CN110197426B (zh) 一种信用评分模型的建立方法、装置及可读存储介质
KR102042442B1 (ko) 디지털 컴플라이언스 및 위험 관리를 위한 레그테크 플랫폼 장치, 금융거래 위험 관리 방법 및 이를 위한 컴퓨터 프로그램
CN112162762B (zh) 灰度发布方法、灰度发布装置和电子设备
CN112416558A (zh) 基于区块链的业务数据处理方法、装置及存储介质
US20240005276A1 (en) Methods and systems for generating a checklist in smart city based on an internet of things
US20230091063A1 (en) Systems and methods for real-time processing of resource requests
CN115641166A (zh) 虚拟资源分配策略创建方法和系统
US20230153326A9 (en) Space partitioning method for database table, device and storage medium
CN112907362A (zh) 贷款业务的处理方法、装置、电子设备和存储介质
US20230106705A1 (en) System and method for real-time processing of resource transfers
CN111402018B (zh) 一种上报资源预算的方法和系统
US11521215B1 (en) Computerized content dissemination
US20240095832A1 (en) System and method for incorporating a mortality risk in a financial instrument

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21953226

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE