An intelligent financial cloud audit method and system based on big data

Abstract

The present invention discloses an intelligent financial cloud audit method and system based on big data, which relates to the technical field of financial cloud audit. The data preprocessing and partitioning module effectively processes the collected financial data, including data cleaning, deduplication and normalization. These preprocessing steps not only improve the data quality, but also generate several data blocks by time partitioning the data, laying the foundation for subsequent analysis. Such processing significantly reduces data noise and redundancy, and improves the accuracy of audit results. In the dynamic task scheduling module, a scheduling algorithm is used to decompose the data blocks into tasks, and the priority scheduling coefficient T is used to schedule the data blocks according to the priority scheduling coefficient T. i Task allocation makes task processing more orderly and efficient. By dynamically adjusting the scheduling strategy and delay processing mechanism, the system can better cope with sudden task peaks, reduce system crashes or overloads caused by uneven task processing, and improve system stability.

Description

Technical Field

[0001] The present invention relates to the technical field of financial cloud auditing, and in particular to an intelligent financial cloud auditing method and system based on big data. Background Art

[0002] In the field of big data processing and cloud computing, financial big data analysis is a specific and important application field, which involves the collection, processing and analysis of large amounts of financial data to support the financial decision-making and compliance review of enterprises. Specifically, the financial cloud audit system is a sub-field of financial big data analysis, which uses cloud computing technology and big data analysis tools to conduct comprehensive audits and risk assessments of the financial activities of enterprises to ensure the accuracy and transparency of financial data.

[0003] In the patent with application publication number CN202210641606.8, a systematic method for hospital financial auditing is disclosed, which uses a data management platform to achieve the coordination and unification of internal and external audits, greatly reducing audit costs and effectively improving audit efficiency. The outpatient settlement module, inpatient settlement module, drug management module, personnel funding module, fixed asset management module and local data center of the client management platform can greatly simplify internal audit work and shorten the audit cycle.

[0004] However, the functions implemented above are difficult to cope with the growing amount of data and complex auditing needs. Moreover, there are still deficiencies in task scheduling and load balancing. When processing high-priority tasks, unreasonable resource allocation and unbalanced load on computing nodes may occur, resulting in low overall computing efficiency and failure to meet the real-time requirements of complex auditing tasks. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides an intelligent financial cloud audit system based on big data, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent financial cloud audit system based on big data, including a data preprocessing and partitioning module, a dynamic task scheduling module, a load balancing and task distribution module, a parallel computing and execution module, a result integration and consistency verification module, and an audit report generation and feedback module;

[0007] The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to different time periods, and generate several data blocks;

[0008] The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to the computing node i for calculation, and obtains the priority scheduling coefficient T of the data block. i， At the same time, the normal scheduling coefficient P N Compare and determine priority;

[0009] The load balancing and task distribution module distributes computing tasks to different computing nodes i according to the output of the dynamic task scheduling module, and calculates the load coefficient L K , calculated through a weighted load balancing algorithm; ensuring load balancing of each node to prevent node overload and idleness;

[0010] The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the calculated results using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module;

[0011] The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution module, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm;

[0012] The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module, analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module through system logs and task scheduling feedback, and feeds back optimization suggestions to the system.

[0013] Preferably, the data preprocessing and partitioning module includes a data preprocessing unit and a time partitioning unit;

[0014] The data preprocessing unit is responsible for preprocessing the collected financial data, specifically including data cleaning, removing duplicate records and processing missing values, and then performing normalization processing to standardize data of different dimensions and ranges into the same range to form a data set;

[0015] The time partition unit divides the data set according to the time attributes in the financial data, including transaction time, billing time and financial report time, including selecting one month as the time granularity, extracts the corresponding financial records for each time period, generates independent data blocks, obtains the total number of data blocks through the statistical data blocks, and records it as the total number of tasks N;

[0016] Each data block corresponds to one month of time data;

[0017] The time partition is obtained by the following formula:

[0018] ;

[0019] In the formula, Indicates the number of the time period task unit for recording d, Indicates the timestamp of record d, Indicates the timestamp of the earliest record in the dataset. represents the time mapping function, Indicates the selected time granularity, specifically the number of seconds corresponding to a month.

[0020] Preferably, the dynamic task scheduling module includes a task priority calculation unit and a scheduling allocation unit;

[0021] The task priority calculation unit divides the data block into task units and assigns them to computing node i for calculation to obtain the priority scheduling coefficient T i :

[0022] ;

[0023] In the formula, , and represents the weight coefficient, and , , ; Indicates the data block size, represents the complexity of the task, Indicates the urgency of data processing, and N indicates the total number of tasks;

[0024] The priority scheduling coefficient T of all tasks i Arrange them in order from high to low to generate a priority scheduling list;

[0025] ;

[0026] In the formula, Represents a priority scheduling list, arranged from high to low priority. Represents the priority scheduling coefficient of all tasks;

[0027] The scheduling allocation unit calculates the priority scheduling coefficient T i Compared with the normal scheduling coefficient P set in the system N For comparison, P N Represents the system's default scheduling priority threshold, which is used to determine the urgency of the task;

[0028] ;

[0029] In the formula, Represents a set of tasks that need to be executed immediately;

[0030] If T i P N , then the task is considered high priority and is immediately assigned to the task set to process;

[0031] If T i <P N , then the task is delayed and assigned to the priority scheduling list Waiting for processing.

[0032] By calculating the priority scheduling coefficient Ti of each task, the system can generate an optimized task execution order to ensure that high-priority tasks are processed first. The priority scheduling list helps the system to reasonably allocate tasks under limited resources, avoid waste of resources and task delays. By giving priority to high-priority tasks, the system can use computing resources more effectively and improve overall computing efficiency.

[0033] Preferably, the load balancing and task distribution module includes a node load monitoring unit and a task distribution balancing unit;

[0034] The node load monitoring unit is responsible for monitoring the real-time load of each computing node i, including CPU usage, memory occupancy, disk I / O load and network bandwidth utilization, and calculates the load factor L for each computing node i. K , the load factor L K Obtained by the following formula:

[0035] ;

[0036] In the formula, , , and represents the weight coefficient, represents the CPU usage of node k, represents the memory usage of node k, represents the disk I / O load of node k, represents the network bandwidth utilization of node k;

[0037] The task distribution balancing unit is based on the load factor L provided by the computing node i. K , use the weighted load balancing algorithm to allocate tasks and obtain the allocation coefficient ; Ensure reasonable distribution of tasks and prevent node overload and idleness;

[0038] The partition coefficient By using a weighted load balancing algorithm:

[0039] ;

[0040] Where, T i represents the priority scheduling coefficient, L K represents the load factor, Represents a positive number, specifically to prevent L K Small leads to infinite distribution coefficient;

[0041] Calculate the load factor L of all computing nodes i K , and form a set of load factors :

[0042] Said Obtained by the following formula:

[0043] ;

[0044] Sort the load factor set in ascending order to get a sorted list ;

[0045] According to the sorted list subscript order, the allocation coefficients assigned to each node from 1 to n are :

[0046] ;

[0047] By calculating the allocation coefficient of node i , the task is assigned to the node with the highest allocation coefficient first, that is, the first task will be assigned to The largest node ensures load balancing.

[0048] Preferably, the parallel computing and execution module includes a parallel computing execution unit and a distributed processing unit;

[0049] The parallel computing execution unit receives computing tasks assigned by the load balancing and task distribution module, and assigns them to different computing nodes i for parallel execution. The computing tasks are processed in parallel through multi-threading and multi-process using a distributed computing framework;

[0050] After the node completes the computing task, the distributed processing unit collects the computing results of each node, processes the computing results using the cluster capability of the distributed computing framework, and transmits the processed result data back to the result integration and consistency verification module to obtain accurate and consistent computing results.

[0051] Preferably, the result integration and consistency verification module includes a result integration unit and a consistency verification unit;

[0052] The result integration unit receives the processing results of each computing node i from the parallel computing and execution module, summarizes and merges them, and obtains the integration result M:

[0053] The integration result M is obtained by the following formula:

[0054] ;

[0055] In the formula, represents the result data of computing node i, and W represents the number of nodes for parallel computing.

[0056] Preferably, the consistency check unit checks the integration result M and the result data of the computing node i. Perform hash verification and verify the consistency of data during the integration process through the distributed hash verification algorithm. The verification calculation formula is:

[0057] ;

[0058] In the formula, Represents the hash value of the integration result M, Represents the result data of computing node i By comparing the hash values ​​of the results of each node, check whether the data has been tampered with, lost or other abnormalities during transmission or processing. If the verification fails, the system will mark the abnormality and record the log for subsequent investigation. This formula is used to verify the consistency of the integrated result data M. By comparing the hash values, it can be confirmed whether the data is consistent and complete, ensuring the accuracy of the calculation results.

[0059] Preferably, the audit report generation and feedback module includes an audit report generation unit and a feedback unit;

[0060] The audit report generation unit generates a financial audit report based on the integrated data M provided by the result integration and consistency verification module. The report includes data analysis results, risk assessment, audit findings and analysis of financial status.

[0061] Preferably, the feedback unit analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module based on the financial audit report and task scheduling feedback; generates optimization suggestions based on the policy effects and feeds them back to the system to continuously improve the scheduling and load balancing strategies. The effectiveness of the existing strategies is evaluated by analyzing indicators such as task completion time, resource utilization, and balance of task allocation.

[0062] An intelligent financial cloud audit method based on big data includes the following steps:

[0063] Step 1: The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to time, and generate several data blocks;

[0064] Step 2: The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to computing nodes i for calculation, obtaining the priority scheduling coefficient T of the data blocks. i， At the same time, the normal scheduling coefficient P N Compare and determine priority;

[0065] Step 3: The load balancing and task distribution module allocates computing tasks to different computing nodes i according to the output of the dynamic task scheduling module and calculates the load factor L K , calculated by weighted load balancing algorithm;

[0066] Step 4: The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the results of the calculation using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module;

[0067] Step 5: The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution modules, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm;

[0068] Step 6: The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module. Through system logs and task scheduling feedback, it analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module, and feeds back optimization suggestions to the system.

[0069] The present invention provides an intelligent financial cloud audit method and system based on big data, which has the following beneficial effects:

[0070] (1) When the system is running, the data preprocessing and partitioning module effectively processes the collected financial data, including data cleaning, deduplication and normalization. These preprocessing steps not only improve the data quality, but also generate several data blocks by time partitioning the data, laying the foundation for subsequent analysis. Such processing significantly reduces data noise and redundancy and improves the accuracy of audit results. In the dynamic task scheduling module, a scheduling algorithm is used to decompose the data blocks and schedule them according to the priority scheduling coefficient T. i Task allocation makes task processing more orderly and efficient.

[0071] (2) By dividing the data into data blocks of the same time period, multiple data blocks can be processed in parallel, improving the overall data processing efficiency. This method avoids the computational bottleneck of full data processing and makes data processing more efficient. Clear time partitioning allows the task scheduling algorithm to more effectively allocate computing tasks and perform load balancing based on the total number of tasks N in the time period. This improvement reduces the uneven load between computing nodes i and improves the overall performance of the system. Clear data partitioning and consistent time granularity make the analysis of time series data more accurate. Data blocks that can be partitioned by month help identify cyclical trends and anomalies in financial data and improve the reliability of analysis results.

[0072] (3) Through precise priority scheduling, the system can prioritize high-priority tasks, thereby improving the overall system response speed and processing efficiency. Immediate processing of high-priority tasks can meet business needs more quickly and improve the system's service quality. The task scheduling module can reasonably allocate resources according to the priority of the task to avoid resource waste and task delays. High-priority tasks are allocated resources first to ensure the timely completion of critical tasks, while low-priority tasks are processed when resources permit, thereby optimizing resource utilization. By dynamically adjusting the scheduling strategy and delay processing mechanism, the system can better cope with sudden task peaks and reduce system crashes or overloads caused by uneven task processing. This mechanism improves the stability and reliability of the system.

[0073] (4) The system effectively eliminates redundant and inaccurate data, ensuring the quality and integrity of the data. This process reduces noise in subsequent analysis and improves the accuracy of audit results. i , the system can intelligently prioritize different tasks to ensure that urgent tasks are handled first. This scheduling mechanism improves the system's response speed and task processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a schematic diagram of a block diagram of an intelligent financial cloud audit system based on big data in the present invention;

[0075] Figure 2 This is a schematic diagram of the steps of an intelligent financial cloud auditing method based on big data in the present invention. DETAILED DESCRIPTION

[0076] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0077] Example 1

[0078] The present invention provides an intelligent financial cloud audit method and system based on big data, please refer to Figure 1 , including data preprocessing and partitioning module, dynamic task scheduling module, load balancing and task distribution module, parallel computing and execution module, result integration and consistency verification module and audit report generation and feedback module;

[0079] The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to different time periods, and generate several data blocks;

[0080] The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to the computing node i for calculation, and obtains the priority scheduling coefficient T of the data block. i， At the same time, the normal scheduling coefficient P N Compare and determine priority;

[0081] The load balancing and task distribution module distributes computing tasks to different computing nodes i according to the output of the dynamic task scheduling module, and calculates the load coefficient L K , calculated by weighted load balancing algorithm;

[0082] The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the calculated results using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module;

[0083] The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution module, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm;

[0084] The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module, analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module through system logs and task scheduling feedback, and feeds back optimization suggestions to the system.

[0085] In this embodiment: the data preprocessing and partitioning module effectively processes the collected financial data, including data cleaning, deduplication and normalization. These preprocessing steps not only improve the data quality, but also generate several data blocks by time partitioning the data, laying the foundation for subsequent analysis. Such processing significantly reduces data noise and redundancy, and improves the accuracy of audit results. In the dynamic task scheduling module, a scheduling algorithm is used to decompose the data blocks into tasks, and the priority scheduling coefficient T is used to schedule the tasks. i Task allocation makes task processing more orderly and efficient.

[0086] The load balancing and task distribution module further optimizes the task allocation and distributes the computing tasks to different computing nodes i according to the load balancing algorithm. This dynamic task scheduling and load balancing strategy ensures the efficient use of computing resources and reduces the computing bottleneck caused by resource contention, thereby improving the overall processing speed and performance of the system. The result integration and consistency verification module verifies the consistency of the data through the distributed hash verification algorithm, ensuring the accuracy and reliability of the audit results. The audit report generation and feedback module uses these accurate data to generate financial audit reports, and analyzes and optimizes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module through system logs and task scheduling feedback.

[0087] Example 2

[0088] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data preprocessing and partitioning module includes a data preprocessing unit and a time partitioning unit;

[0089] The data preprocessing unit is responsible for preprocessing the collected financial data, specifically including data cleaning, removing duplicate records and processing missing values, and then performing normalization processing to standardize data of different dimensions and ranges into the same range to form a data set;

[0090] Specifically, data cleaning ensures the quality and consistency of data, reduces noise and errors, and uses visualization tools to detect outliers in the data.

[0091] Handling missing values ​​is to ensure the integrity of the dataset and use programming tools to detect missing values.

[0092] Normalization scales the data to a uniform range, avoiding the impact of different scales of eigenvalues ​​on model training and ensuring the high quality, completeness and consistency of financial data, thereby providing a reliable data foundation for subsequent analysis and auditing.

[0093] The time partition unit divides the data set according to the time attributes in the financial data, including transaction time, billing time and financial report time, including selecting one month as the time granularity, extracts the corresponding financial records for each time period, generates independent data blocks, obtains the total number of data blocks through the statistical data blocks, and records it as the total number of tasks N;

[0094] Each data block corresponds to one month of time data;

[0095] The time partition is obtained by the following formula:

[0096] ;

[0097] In the formula, Indicates the number of the time period task unit for recording d, Indicates the timestamp of record d, Indicates the timestamp of the earliest record in the dataset. represents the time mapping function, Indicates the selected time granularity, specifically the number of seconds corresponding to a month.

[0098] In this embodiment: by dividing the data into data blocks of the same time period, multiple data blocks can be processed in parallel, improving the overall data processing efficiency. This method avoids the computational bottleneck of full data processing, making data processing more efficient. Clear time partitioning allows the task scheduling algorithm to more effectively allocate computing tasks and perform load balancing based on the total number of tasks N in the time period. This improvement reduces the uneven load between computing nodes i and improves the overall performance of the system. Clear data partitioning and consistent time granularity make the analysis of time series data more accurate. Data blocks that can be partitioned by month help identify cyclical trends and anomalies in financial data and improve the reliability of analysis results.

[0099] Example 3

[0100] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the dynamic task scheduling module includes a task priority calculation unit and a scheduling allocation unit;

[0101] The task priority calculation unit divides the data block into task units and assigns them to computing node i for calculation to obtain the priority scheduling coefficient T i :

[0102] ;

[0103] In the formula, , and represents the weight coefficient, and , , ; Indicates the data block size, represents the complexity of the task, Indicates the urgency of data processing, and N indicates the total number of tasks;

[0104] The priority scheduling coefficient T of all tasks i Arrange them in order from high to low to generate a priority scheduling list;

[0105] ;

[0106] In the formula, Represents a priority scheduling list, arranged from high to low priority. Represents the priority scheduling coefficient of all tasks;

[0107] The scheduling allocation unit calculates the priority scheduling coefficient T i Compared with the normal scheduling coefficient P set in the system N For comparison, P N Represents the system's default scheduling priority threshold, which is used to determine the urgency of the task;

[0108] ;

[0109] In the formula, Represents a set of tasks that need to be executed immediately;

[0110] If T i P N , then the task is considered high priority and is immediately assigned to the task set to process;

[0111] If T i <P N , then the task is delayed and assigned to the priority scheduling list Waiting for processing.

[0112] In this embodiment: through accurate priority scheduling, the system can give priority to high-priority tasks, thereby improving the response speed and processing efficiency of the overall system. The immediate processing of high-priority tasks can meet business needs more quickly and improve the service quality of the system. The task scheduling module can reasonably allocate resources according to the priority of the task to avoid resource waste and task delays. High-priority tasks are allocated resources first to ensure the timely completion of critical tasks, while low-priority tasks are processed when resources permit, thereby optimizing resource utilization. By dynamically adjusting the scheduling strategy and delay processing mechanism, the system can better cope with sudden task peaks and reduce system crashes or overloads caused by uneven task processing. This mechanism improves the stability and reliability of the system.

[0113] Example 4

[0114] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the load balancing and task distribution module includes a node load monitoring unit and a task distribution balancing unit;

[0115] The node load monitoring unit is responsible for monitoring the real-time load of each computing node i, including CPU usage, memory occupancy, disk I / O load and network bandwidth utilization, and calculates the load factor L for each computing node i. K , the load factor L K Obtained by the following formula:

[0116] ;

[0117] In the formula, , , and represents the weight coefficient, represents the CPU usage of node k, represents the memory usage of node k, represents the disk I / O load of node k, represents the network bandwidth utilization of node k;

[0118] The task distribution balancing unit is based on the load factor L provided by the computing node i. K , use the weighted load balancing algorithm to allocate tasks and obtain the allocation coefficient ;

[0119] The partition coefficient By using a weighted load balancing algorithm:

[0120] ;

[0121] Where, Ti represents the priority scheduling coefficient, L K represents the load factor, represents a positive number;

[0122] Calculate the load factor L of all computing nodes i K , and form a set of load factors :

[0123] Said Obtained by the following formula:

[0124] ;

[0125] Sort the load factor set in ascending order to get a sorted list ;

[0126] According to the sorted list subscript order, the allocation coefficients assigned to each node from 1 to n are :

[0127] ;

[0128] By calculating the allocation coefficient of node i , tasks are preferentially assigned to nodes with the highest distribution coefficient to ensure load balancing.

[0129] In this embodiment: the node load monitoring unit monitors multiple resource indicators (CPU usage, memory occupancy, disk I / O load and network bandwidth utilization) of the computing node in real time. This multi-dimensional load monitoring ensures a comprehensive understanding of the node load. Compared with the solution of monitoring only a single resource indicator, it can more accurately reflect the actual load status of the node. The task distribution balancing unit uses a weighted load balancing algorithm to distribute tasks. This algorithm calculates the task priority scheduling coefficient T i and node load factor L K Compared with the traditional simple polling or uniform distribution method, this weighted algorithm can more reasonably balance the node load, reduce resource competition and delay in task processing, and optimize the task allocation strategy by comprehensively monitoring and accurately calculating the load factor L. K , the system can achieve more reasonable task allocation. This optimization measure reduces node overload, improves the overall efficiency of task processing, and makes the resources of each computing node i more balanced.

[0130] Example 5

[0131] This embodiment is explained in Example 4. Please refer to Figure 1 ,Specifically: the parallel computing and execution module includes a parallel computing execution unit and a distributed processing unit;

[0132] The parallel computing execution unit receives computing tasks assigned by the load balancing and task distribution module, and assigns them to different computing nodes i for parallel execution. The computing tasks are processed in parallel through multi-threading and multi-process using a distributed computing framework;

[0133] After the node completes the computing task, the distributed processing unit collects the computing results of each node, processes the computing results using the cluster capability of the distributed computing framework, and transmits the processed result data back to the result integration and consistency verification module to obtain accurate and consistent computing results.

[0134] The result integration and consistency verification module includes a result integration unit and a consistency verification unit;

[0135] The result integration unit receives the processing results of each computing node i from the parallel computing and execution module, summarizes and merges them, and obtains the integration result M:

[0136] The integration result M is obtained by the following formula:

[0137] ;

[0138] In the formula, represents the result data of computing node i, and W represents the number of nodes for parallel computing.

[0139] The consistency check unit checks the integration result M and the result data of the computing node i. Perform hash verification and verify the consistency of data during the integration process through the distributed hash verification algorithm. The verification calculation formula is:

[0140] ;

[0141] In the formula, Represents the hash value of the integration result M, Represents the result data of computing node i The hash value of .

[0142] The audit report generation and feedback module includes an audit report generation unit and a feedback unit;

[0143] The audit report generation unit generates a financial audit report based on the integrated data M provided by the result integration and consistency verification module. The report includes data analysis results, risk assessment, audit findings and analysis of financial status.

[0144] The feedback unit analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module based on the financial audit report and task scheduling feedback; generates optimization suggestions based on the policy effects and feeds them back to the system to continuously improve the scheduling and load balancing strategies.

[0145] In this embodiment, the parallel computing execution unit uses the multi-threading and multi-process technology in the distributed computing framework to assign computing tasks to multiple computing nodes i for parallel processing. Compared with traditional serial processing, this parallel computing method significantly improves the speed and efficiency of data processing and can process large-scale financial data more quickly; through the use of parallel computing and multi-threading and multi-process, the system can significantly speed up data processing and shorten the audit cycle. This improvement enhances the system's ability to process large-scale financial data and meets the needs of real-time auditing. The result integration and consistency verification ensure the accuracy and consistency of the calculation results of each node and reduce audit errors caused by data processing errors. It improves the reliability of audit results and provides decision makers with more accurate financial data support. The generated financial audit report provides detailed data analysis and risk assessment, which helps to reveal financial problems and potential risks.

[0146] Example 6

[0147] A smart financial cloud audit method based on big data, please refer to Figure 2 , specifically: including the following steps:

[0148] Step 1: The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to time, and generate several data blocks;

[0149] Step 2: The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to computing nodes i for calculation, obtaining the priority scheduling coefficient T of the data blocks. i， At the same time, the normal scheduling coefficient P N Compare and determine priority;

[0150] Step 3: The load balancing and task distribution module allocates computing tasks to different computing nodes i according to the output of the dynamic task scheduling module and calculates the load factor L K , calculated by weighted load balancing algorithm;

[0151] Step 4: The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the results of the calculation using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module;

[0152] Step 5: The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution modules, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm;

[0153] Step 6: The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module. Through system logs and task scheduling feedback, it analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module, and feeds back optimization suggestions to the system.

[0154] In this embodiment: through steps 1 to 6, the system effectively eliminates redundant and inaccurate data, ensuring the quality and integrity of the data. This process reduces the noise in subsequent analysis and improves the accuracy of the audit results. i , the system can intelligently prioritize different tasks to ensure that urgent tasks are handled first. This scheduling mechanism improves the system's response speed and task processing efficiency.

[0155] With the normal scheduling coefficient P N By comparing the load factor L, the task scheduling strategy is optimized, so that high-priority tasks can be allocated and processed in a timely manner, thereby improving the flexibility and real-time performance of the overall task processing; through the weighted load balancing algorithm, according to the load factor L K and priority scheduling coefficient T i Calculate the allocation coefficient and allocate computing tasks reasonably. This approach optimizes resource usage, improves the efficiency of the computing process, and reduces computing bottlenecks. It uses a distributed computing framework for multi-threaded and multi-process parallel computing, which greatly speeds up data processing. Parallel computing enables the system to handle multiple computing tasks at the same time, improving processing power and response speed. Based on the integrated results, a detailed audit report is generated, including data analysis results, risk assessments, and audit findings, providing decision makers with a comprehensive analysis of the financial situation. This report helps identify potential risks and improve financial management.

[0156] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent financial cloud audit system based on big data, characterized by: It includes data preprocessing and partitioning module, dynamic task scheduling module, load balancing and task distribution module, parallel computing and execution module, result integration and consistency verification module and audit report generation and feedback module; The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to different time periods, and generate several data blocks; The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to the computing node i for calculation, and obtains the priority scheduling coefficient T of the data block. i， At the same time, the normal scheduling coefficient P N Compare and determine priorities; The dynamic task scheduling module includes a task priority calculation unit and a scheduling allocation unit; The task priority calculation unit divides the data block into task units and assigns them to computing node i for calculation to obtain the priority scheduling coefficient T i : ； In the formula, , and represents the weight coefficient, and , , ; Indicates the data block size, represents the complexity of the task, Indicates the urgency of data processing, and N indicates the total number of tasks; Set the priority scheduling coefficient T of all tasks i Arrange them in order from high to low to generate a priority scheduling list; ； In the formula, Represents a priority scheduling list, arranged from high to low priority. Represents the priority scheduling coefficient of all tasks; The scheduling allocation unit calculates the priority scheduling coefficient T i Compared with the normal scheduling coefficient P set in the system N For comparison, P N Represents the system's default scheduling priority threshold, which is used to determine the urgency of the task; ； In the formula, Represents a set of tasks that need to be executed immediately; If T i P N , then the task is considered high priority and is immediately assigned to the task set to process; If T i <P N , then the task is delayed and assigned to the priority scheduling list Waiting for processing; The load balancing and task distribution module distributes computing tasks to different computing nodes i according to the output of the dynamic task scheduling module, and calculates the load coefficient L K , calculated by weighted load balancing algorithm; The load balancing and task distribution module includes a node load monitoring unit and a task distribution balancing unit; The node load monitoring unit is responsible for monitoring the real-time load of each computing node i, including CPU usage, memory occupancy, disk I / O load and network bandwidth utilization, and calculates the load factor L for each computing node i. K , the load factor L K Obtained by the following formula: ； In the formula, , , and represents the weight coefficient, represents the CPU usage of node k, represents the memory usage of node k, represents the disk I / O load of node k, represents the network bandwidth utilization of node k; The task distribution balancing unit is based on the load factor L provided by the computing node i. K , use the weighted load balancing algorithm to allocate tasks and obtain the allocation coefficient ; The partition coefficient By using a weighted load balancing algorithm: ； Where, T i represents the priority scheduling coefficient, L K represents the load factor, represents a positive number; Calculate the load factor L of all computing nodes i K , and form a set of load factors : Said Obtained by the following formula: ； Sort the load factor set in ascending order to get a sorted list ; According to the sorted list subscript order, the allocation coefficients assigned to each node from 1 to n are : ； By calculating the allocation coefficient of node i , prioritize tasks to nodes with the highest allocation coefficient to ensure load balancing; The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the calculated results using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module; The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution module, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm; The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module, analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module through system logs and task scheduling feedback, and feeds back optimization suggestions to the system.

2. According to the big data-based intelligent financial cloud audit system of claim 1, it is characterized by: The data preprocessing and partitioning module includes a data preprocessing unit and a time partitioning unit; The data preprocessing unit is responsible for preprocessing the collected financial data, specifically including data cleaning, removing duplicate records and processing missing values, and then performing normalization processing to standardize data of different dimensions and ranges into the same range to form a data set; The time partition unit divides the data set according to the time attributes in the financial data, including transaction time, billing time and financial report time, including selecting one month as the time granularity, extracts the corresponding financial records for each time period, generates independent data blocks, obtains the total number of data blocks through the statistical data blocks, and records it as the total number of tasks N; Each data block corresponds to one month of time data; The time partition is obtained by the following formula: ； In the formula, Indicates the number of the time period task unit for recording d, Indicates the timestamp of record d, Indicates the timestamp of the earliest record in the dataset. represents the time mapping function, Indicates the selected time granularity, specifically the number of seconds corresponding to a month.

3. According to the big data-based intelligent financial cloud audit system of claim 1, it is characterized by: The parallel computing and execution module includes a parallel computing execution unit and a distributed processing unit; The parallel computing execution unit receives computing tasks assigned by the load balancing and task distribution module, and assigns them to different computing nodes i for parallel execution. The computing tasks are processed in parallel through multi-threading and multi-process using a distributed computing framework; After the node completes the computing task, the distributed processing unit collects the computing results of each node, processes the computing results using the cluster capability of the distributed computing framework, and transmits the processed result data back to the result integration and consistency verification module to obtain accurate and consistent computing results.

4. According to the big data-based intelligent financial cloud audit system of claim 1, it is characterized by: The result integration and consistency verification module includes a result integration unit and a consistency verification unit; The result integration unit receives the processing results of each computing node i from the parallel computing and execution module, summarizes and merges them, and obtains the integration result M: The integration result M is obtained by the following formula: ； In the formula, represents the result data of computing node i, and W represents the number of nodes for parallel computing.

5. According to claim 4, a smart financial cloud audit system based on big data is characterized in that: The consistency check unit checks the integration result M and the result data of the computing node i. Perform hash verification and verify the consistency of data during the integration process through the distributed hash verification algorithm. The verification calculation formula is: ； In the formula, Represents the hash value of the integration result M, Represents the result data of computing node i The hash value of .

6. The intelligent financial cloud audit system based on big data according to claim 1 is characterized by: The audit report generation and feedback module includes an audit report generation unit and a feedback unit; The audit report generation unit generates a financial audit report based on the integrated data M provided by the result integration and consistency verification module. The report includes data analysis results, risk assessment, audit findings and analysis of financial status.

7. The intelligent financial cloud audit system based on big data according to claim 6 is characterized by: The feedback unit analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module based on the financial audit report and task scheduling feedback; generates optimization suggestions based on the policy effects and feeds them back to the system to continuously improve the scheduling and load balancing strategies.

8. An intelligent financial cloud audit method based on big data, applied to an intelligent financial cloud audit system based on big data as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: Step 1: The data preprocessing and partitioning module preprocesses the collected financial data, including data cleaning, deduplication and normalization, to form a data set, partition the data set according to time, and generate several data blocks; Step 2: The dynamic task scheduling module decomposes the preprocessed data blocks through the scheduling algorithm and assigns them to computing nodes i for calculation, obtaining the priority scheduling coefficient T of the data blocks. i， At the same time, the normal scheduling coefficient P N Compare and determine priorities; Step 3: The load balancing and task distribution module allocates computing tasks to different computing nodes i according to the output of the dynamic task scheduling module and calculates the load factor L K , calculated by weighted load balancing algorithm; Step 4: The parallel computing and execution module performs parallel computing on computing node i according to the tasks assigned by the load balancing and task distribution module, processes the results of the calculation using the distributed computing framework, and transmits the result data back to the result integration and consistency verification module; Step 5: The result integration and consistency check module integrates the processing results of each computing node i according to the result data returned by the parallel computing and execution modules, obtains the integration result M, and verifies the consistency of the data through the distributed hash check algorithm; Step 6: The audit report generation and feedback module generates a financial audit report based on the result data obtained by the result integration and consistency verification module. Through system logs and task scheduling feedback, it analyzes the policy effects of the dynamic task scheduling module and the load balancing and task distribution module, and feeds back optimization suggestions to the system.

Images