Intelligent Decision Support System Based on ERP Data

Abstract

The present invention relates to the field of enterprise resource planning technology, specifically to an intelligent decision support system based on ERP data, which includes a configuration factor stability determination module, a responsibility intersection node screening module, a path load range determination module, a reverse path verification module, and a cold standby path loading module. In the present invention, by extracting the configuration call frequency and sequence and constructing a stability determination vector, multi-dimensional data association is performed in combination with job task history and material classification, and the responsibility intersection is judged by superimposing department concentration, a configuration combination with continuity and inheritance is generated, call data and module record calculation load boundaries are introduced at the path level, a multi-layer comparison mechanism of the main path and the backup path is constructed, and intelligent recommendation of path switching is realized by screening similar ratios, thereby improving the stability of the configuration combination, the rationality of path selection, and the synergy of resource calls, and enhancing the system's strategy adaptation capability and decision support efficiency in multi-role and multi-path scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise resource planning, and in particular to an intelligent decision support system based on ERP data. Background Art

[0002] The field of enterprise resource planning (ERP) involves the integrated management and coordination of various internal enterprise resources, including human, financial, material, information flows, and processes. It uses information technology to enable information exchange and resource sharing among different functional departments. Its core areas include production planning management, procurement and supply chain management, inventory control, financial accounting, human resource allocation, and customer relationship management. It serves as the foundational support system for modern enterprise informatization. By building a unified data platform and business logic system, this technology enables comprehensive monitoring and management optimization of enterprise operations, thereby improving overall operational efficiency and resource utilization effectiveness. Among them, the traditional intelligent decision support system based on ERP data refers to an auxiliary system that uses historical data and business process information accumulated in the enterprise resource planning system to provide decision-making references for the enterprise management in its business management activities. The technical matter targeted by this patent subject is how to extract data with decision-making value from the ERP system and analyze it to support strategic or tactical management decisions. The traditional decision support system based on ERP data usually adopts a fixed rule data extraction method combined with a multidimensional database structure to organize information, and displays the results through online analysis and processing and preset report templates. The data source mainly relies on accounting vouchers, inventory ledgers, order fulfillment records and personnel management data in the ERP system, and forms a static data reference view through statistical summary, ratio analysis or trend analysis.

[0003] Existing technologies rely on fixed rules and multi-dimensional structures to extract ERP data. When processing resource allocation and task scheduling, they fail to reflect the dynamic characteristics of call frequency and sequence. There is a lack of a deep cross-identification mechanism for the relationship between job history and material type. It is impossible to finely characterize cross-organizational responsibility boundaries based on actual department concentration. As a result, the path selection process lacks a systematic judgment basis for load-bearing capacity. When multiple paths exist in parallel, an effective reverse verification and backup mechanism cannot be established. As a result, in scenarios such as task switching, configuration updates or process reorganization, problems such as uneven resource allocation, increased probability of path failure and insufficient intelligent support may occur, reducing system stability and response robustness. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent decision support system based on ERP data.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: The intelligent decision support system based on ERP data includes:

[0006] The configuration factor stability determination module obtains configuration call logs, groups them by job role, extracts call frequency and sequence, takes the square mean of the difference between frequency vectors and the mean square of the position spacing between sequence vectors, generates a stability determination vector, compares it with the preset range, and selects sustainable call configuration combinations;

[0007] The responsibility intersection node screening module obtains job task history and material category based on the sustainable call configuration combination, extracts material category, department number and downstream data, calculates department concentration, screens cross-department responsibility intersections, and generates inheritable cross-department call nodes;

[0008] The path carrying range determination module obtains the main path calling data and module records based on the inheritable cross-department calling nodes, calculates the product of the total calling data and the number of modules, and screens the main path carrying range;

[0009] The reverse path verification module obtains the excluded paths and outputs based on the loadable range of the primary path, calculates a combined value, compares it with the primary path range, and generates an activatable backup path set;

[0010] The cold standby path loading module obtains the cold standby path target data based on the activatable standby path set, compares the main path target data one by one, screens similar combinations, and generates an intelligent decision support solution.

[0011] As a further solution of the present invention, the sustainable call configuration combination includes call frequency vector characteristics, call sequence vector characteristics, stable judgment vector, matching screening results, and configuration item combination set; the inheritable cross-department call node includes job task history information, material category identification code, associated department number, department concentration index, and responsibility intersection node set; the main path carrying range includes the total value of call data, module quantity index, path carrying capacity value, and reference comparison range; the activatable backup path set includes excluded path records, backup path output data, path judgment result value, and activated path set; the intelligent decision support solution includes cold standby path target parameters, main path target parameters, path similarity ratio, and recommended combination solution.

[0012] As a further solution of the present invention, the definition of the preset range is used to determine the reference interval for configuration call stability, including the expected value range of the frequency difference and the sequence spacing;

[0013] The main path's load capacity is defined as the upper limit of the call capacity that the main path can stably handle under the current module distribution and total number of calls;

[0014] The definition of the similar combination is a cold standby path combination that has a high degree of similarity with the main path target data in terms of task structure and material type.

[0015] As a further solution of the present invention, the configuration factor stability determination module includes:

[0016] The call log collection submodule obtains configuration call log data, extracts the job role information, configuration item identifiers, and corresponding timestamp sequences from the call records, classifies the call logs by job role, and generates job role call sequence groups.

[0017] The frequency sequence analysis submodule calls the configuration item sequence under the position in the position role call sequence group, calculates the call frequency for each configuration item, obtains a frequency vector, extracts the square of the average position difference of the call position of the configuration item in the sequence in turn, forms a sequence vector, and calculates the weighted combination value of the frequency deviation mean and the sequence position mean square error to obtain the call stability vector;

[0018] The stability screening submodule calls the preset configuration call stability range threshold based on the call stability vector, compares the configuration item combinations, traverses and screens the combination results under all job roles, obtains the configuration combinations within the threshold range, and establishes a sustainable call configuration combination;

[0019] The definition of the configuration combination within the threshold range is obtained, and the configuration item combination set with stable call characteristics is screened under the condition that the call stability vector meets the preset threshold.

[0020] As a further solution of the present invention, the responsibility intersection node screening module includes:

[0021] The task material extraction submodule obtains the job task history and the corresponding material category identifier based on the sustainable call configuration combination, extracts the department number corresponding to the material category and the job, and collects the downstream related job data of the job in the task history to build a job material department mapping matrix;

[0022] The department distribution calculation submodule calls the position distribution information and department number data in the position-material-department mapping matrix, performs quantitative statistics on the distribution of task histories of material categories in their respective departments, normalizes the proportion of task histories in material categories with the same department number, calculates and obtains the position distribution concentration value corresponding to the material category, calls the position coverage number under the department number and compares it with the task coverage ratio to obtain the cross-position distribution concentration trend vector;

[0023] The intersection node screening submodule detects the repeated task history paths of material categories across multiple departments based on the cross-post distribution central trend vector, screens the set of post numbers that have cross-department call trajectories at the same time, extracts the task nodes that are repeatedly called in downstream related posts, and establishes inheritable cross-department call nodes.

[0024] As a further solution of the present invention, the path carrying range determination module includes:

[0025] The main path information extraction submodule counts the total amount of call data in each path based on the main path call data and corresponding module records obtained by the inheritable cross-department call node, and extracts the module number and quantity included in each path to establish a main path call module matrix;

[0026] The load value generation submodule calls the total amount of call data and the number of modules in the main path call module matrix, performs a product operation on the data volume and the number of modules of each main path, and normalizes the call frequency offset and structural overlap between path nodes to obtain an evaluation result for each path and obtain a path load ratio vector;

[0027] The range status screening submodule compares the set load determination reference range value according to the path load ratio vector, screens the path numbers within the upper and lower limit intervals, and establishes the main path load range based on the result set composed of structural characteristics, module coverage structure and call density.

[0028] As a further solution of the present invention, the reverse path verification module includes:

[0029] The exclusion path extraction submodule obtains the path number that is not included in the path set based on the loadable range of the main path, collects the call data and output records corresponding to the path, marks the number of modules in the path, and establishes an exclusion path call record set;

[0030] The composite value combination submodule calls the total amount of call data and the number of modules in the exclusion path call record set, performs superposition processing on the output number of the corresponding path, and aggregates the product of the call data and the number of modules with the output record value according to the path number to obtain the exclusion path composite evaluation value group;

[0031] The backup path screening submodule compares the judgment conditions in the loadable range of the primary path based on the composite evaluation value group of the excluded paths, compares the evaluation values ​​corresponding to the paths to see whether they fall within the acceptable range, screens the path numbers whose evaluation values ​​meet the conditions, and establishes an activatable backup path set;

[0032] The definition of the acceptable range is the standard interval for measuring whether the composite evaluation value of the backup path has the potential to replace the main path.

[0033] As a further solution of the present invention, the cold standby path loading module includes:

[0034] The target data acquisition submodule obtains the call target data corresponding to each cold standby path based on the activatable standby path set, extracts the call number and path identification information of the target item, and simultaneously obtains the call target data of the paths within the loadable range of the primary path, and establishes a cold standby primary path target comparison set;

[0035] The path comparison and screening submodule calls the cold standby path target data and the main path target data in the cold standby main path target comparison set, compares the matching number and combination order of the same type of target items in turn, calculates the same type ratio for all comparison results, and screens the cold standby paths whose ratios meet the fitting threshold, thereby generating a same type path comparison result set;

[0036] The solution output combination submodule extracts the path number, target item structure and calling sequence from the path combinations marked as matching in the similar path comparison result set, constructs a path logical structure map that supports intelligent loading, and organizes and merges the structural information and path attributes to establish an intelligent decision support solution;

[0037] The definition of the fit threshold is used to measure whether the cold backup path and the primary path are similar enough in terms of target item type and order to match the comparison criteria;

[0038] The definition of the path logic structure map is based on the call sequence and target structure mapping constructed by matching path combinations, supporting intelligent path loading and decision making.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are:

[0040] In the present invention, by extracting the configuration call frequency and sequence and constructing a stable judgment vector, combining job task history with material classification to perform multi-dimensional data association, superimposing department concentration to judge the intersection of responsibilities, a configuration combination with continuity and inheritance is generated, and call data and module record calculation carrying boundaries are introduced at the path level. A multi-layer comparison mechanism of the main path and the backup path is constructed, and intelligent recommendation of path switching is realized through screening of similar ratios, thereby improving the stability of the configuration combination, the rationality of path selection and the coordination of resource calls, and enhancing the system's strategy adaptation capability and decision support efficiency in multi-role and multi-path scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a system flow chart of the present invention;

[0042] Figure 2 This is a flow chart of the configuration factor stability determination module of the present invention;

[0043] Figure 3 This is a flow chart of the responsibility intersection node screening module of the present invention;

[0044] Figure 4This is a flow chart of the path carrying range determination module of the present invention;

[0045] Figure 5 This is a flow chart of the reverse path verification module of the present invention;

[0046] Figure 6 This is a flow chart of the cold standby path loading module of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0049] See also Figure 1 , the intelligent decision support system based on ERP data includes:

[0050] The configuration factor stability determination module obtains configuration call log data, classifies and organizes records by job role, extracts the configuration item call frequency and sequence sequence, averages the frequency vector by the sum of squared differences in the frequency data, and calculates the mean square value of the sequence vector by the call position spacing. These vectors are combined to generate a stability determination vector, which is then compared with a preset range to select combinations and generate a sustainable call configuration combination.

[0051] The responsibility intersection node screening module obtains job task history and material category identification based on a sustainable call configuration combination, extracts material category and department number and downstream related data, processes department concentration based on the proportion of tasks with the same number, compares and screens cross-department responsibility intersections, and generates inheritable cross-department call nodes;

[0052] The path carrying range determination module obtains the main path call data and module records based on the inheritable cross-department call nodes, generates a comprehensive value by multiplying the sum of the call data and the number of modules, and compares the carrying status with the reference range to generate the main path carrying range;

[0053] The reverse path verification module obtains the excluded path data and output based on the loadable range of the primary path, combines the product of the call data and the module quantity with the sum of the output, compares it with the loadable range of the primary path to determine whether the conditions are met, and generates an activatable backup path set;

[0054] The cold standby path loading module obtains the cold standby path call target data based on the activatable standby path set, compares the cold standby path and the main path target data, compares the similar ratios in turn to screen similar combinations, and generates an intelligent decision support plan.

[0055] The sustainable call configuration combination includes call frequency vector characteristics, call sequence vector characteristics, stable judgment vector, matching screening results, and configuration item combination set. The inheritable cross-department call nodes include job task history information, material category identification code, associated department number, department concentration index, and responsibility intersection node set. The main path can carry a range that includes the total value of call data, module quantity index, path carrying capacity value, and reference comparison range. The activatable backup path set includes excluded path records, backup path output data, path judgment result value, and activated path set. The intelligent decision support plan includes cold standby path target parameters, main path target parameters, path similarity ratio, and recommended combination plan.

[0056] See also Figure 2 , the configuration factor stability determination module includes:

[0057] The call log collection submodule obtains configuration call log data, extracts the job role information, configuration item identifiers, and corresponding timestamp sequences from the call records, classifies the call logs by job role, and generates job role call sequence groups.

[0058] The call log collection submodule obtains configuration call log data. During the implementation process, it first connects to the configuration management system or service call log recording module to extract structured call records. Each record contains fields such as "caller ID", "configuration item ID", "timestamp", "task ID", etc. The system parses the field content in each log record in turn and matches the role by "caller ID" and the position information mapping table. For example, when the caller ID is U310 and the matching result is "operation and maintenance administrator", the current record is classified into the "operation and maintenance administrator" position role category. The call log is initially classified according to this role, and then the configuration item ID field is extracted from the log. For example, the record with "configuration item ID: P022" is used as the object of this operation and continues to be archived. After completing the role mapping of all records, they are grouped and aggregated by role dimension. For each role, the aggregated data is sorted sequentially by the timestamp field. The specific date information is not considered during sorting, and the order is only sorted by time order. For example, the record order is logically numbered as t1, t2, t3, ..., and then the sorted configuration item ID field is extracted to form a sequence. For example, for the "operation and maintenance administrator" role, the configuration sequence is obtained.

[0059] ;

[0060] This sequence reflects the role's operation trajectory on the configuration item and is used as the "position role call sequence" in subsequent processing. Through this method, call sequence groups are constructed for all position roles in turn to achieve position-dimensional log collection.

[0061] The frequency sequence analysis submodule calls the configuration item sequence under the position in the position call sequence group. For each configuration item, the call frequency is calculated to obtain a frequency vector. The average position difference square of the configuration item's call position in the sequence is extracted in turn to form a sequence vector using the formula:

[0062] ;

[0063] The weighted combination value of the mean frequency deviation and the mean square error of the sequence position is obtained by operation to obtain the call stability vector;

[0064] in, Representative The calling frequency of each configuration item, is the average value of the configuration item call frequency, For the Configuration item The position index of the call, For the The total number of calls to the configuration item, is the total number of configuration items, The stability vector for the generated call;

[0065] The calculation formula for calling stability vector is:

[0066] ;

[0067] Take the configuration item sequence of the position "Database Configuration Specialist";

[0068] ;

[0069] For example, the total call sequence length is 7, and there are 4 types of configuration items involved, namely D001, D002, D003, and D004. The execution steps are as follows:

[0070] Step 1: Calculate the call frequency and its mean;

[0071] D001 appears 3 times, frequency ;

[0072] D002 appears 2 times, frequency ;

[0073] D003 appears once, frequency ;

[0074] D004 appears once, frequency ;

[0075] Calculate the average frequency ;

[0076] Step 2: Calculate the square of frequency deviation

[0077] ;

[0078] ;

[0079] Step 3: Extract the call location and calculate the mean square of the position difference

[0080] D001 appears in positions 1, 3, and 7: Difference , , the average is ;

[0081] D002 appears in positions 2 and 6: Difference , the average is ;

[0082] D003 appears once, with no difference, and is recorded as 0;

[0083] D004 appears once, with no difference, and is recorded as 0;

[0084] The mean of the square of the sequential differences is:

[0085] ;

[0086] Step 4: Synthesize the call stability vector S value

[0087] ;

[0088] The result shows that for the call sequence of the position "Database Configuration Specialist", its call stability vector is 6.514, which will be used to screen and determine whether it meets the stability combination standard.

[0089] Description of the dimensional normalization method of formula parameters

[0090] In the calculation of the call stability vector, the original participants include the configuration item call frequency , frequency mean , Configuration item call location index , and the number of configuration item calls ,These parameters originally have different dimensional units and need to be normalized to avoid affecting the validity of the results due to inconsistent units.

[0091] Configuration item call frequency : This item itself is a unitized parameter, defined as the ratio of the number of occurrences of a configuration item to the total length of the call sequence. It does not need to be normalized again and the dimension is "times / times", that is, it is a pure numerical value.

[0092] Average call frequency : It is obtained by the arithmetic average of the call frequencies of all configuration items. Its dimension is the same as Same, both are pure numbers.

[0093] Configuration item call location index : The original integer position index represents the configuration item The sequence number of the occurrence of the sequence number needs to be normalized to avoid the absolute size of the value between positions affecting the stability judgment. The normalization method is to divide all indices by the maximum position value of the sequence, that is, all Map to interval, and obtain the relative position ratio, thereby eliminating the influence of sequence length.

[0094] Squared position difference : Since the two indices have been normalized, the result is also a dimensionless value, representing a relative measure of the degree of position change between adjacent calls.

[0095] Number of calls : is the call count of a single configuration item. It is a dimensionless positive integer used as a divisor for the position difference mean calculation. It does not participate in the stability comparison and exists only as a structural denominator.

[0096] After the above normalization process, each component in the stability vector formula has been converted into a dimensionless pure numerical value, which can be directly compared between different job roles or configuration sequences.

[0097] Calling the definition of the stability vector

[0098] The call stability vector is a quantitative indicator used to measure the stability of a job role in the configuration item call behavior. Its essence is to integrate the two dimensions of "frequency stability" and "sequence regularity" of the configuration item call to form a single value, reflecting whether the configuration operation presents a highly consistent and regular usage pattern.

[0099] Among them, the frequency stability part reflects the degree of fluctuation of the calling frequency of each configuration item compared with the overall average frequency. The smaller the value, the more uniform the frequency distribution of the configuration item being called. The sequence regularity part measures whether the intervals between the positions of the same configuration item in the calling sequence are stable. If the position difference between adjacent calls is small, it means that the calling sequence of the item is relatively concentrated and regular.

[0100] Therefore, this vector is conceptually defined as the "joint variance of configuration behavior in frequency distribution and position sequence", which comprehensively expresses the uncertainty of two different dimensions in a single value, and is used for stability identification and judgment in subsequent configuration screening and combination recommendations.

[0101] The operation principle of calling stability vector

[0102] Frequency dimension calculation principle

[0103] First, the call frequency of each configuration item is calculated, that is, the ratio of the number of times a configuration item appears in the call sequence to the total length of the sequence. These frequency values ​​​​compose the frequency vector. Then, by taking the square of the difference between each element in the frequency vector and the mean, the variance is calculated to reflect the volatility of the frequency distribution.

[0104] This process corresponds to the measurement of sample uniformity or concentration in traditional statistical analysis. The smaller the value, the closer the frequency of use of each configuration item is to equilibrium. If a certain configuration is frequently called while other configurations are rarely used, the variance will increase.

[0105] Principle of sequential dimension calculation

[0106] Extract all index positions of each configuration item in the call sequence, calculate the square of the position difference between two adjacent calls, and average the square of all call differences of the configuration item to reflect whether the appearance interval of the configuration item in the sequence has a stable rhythm.

[0107] If the configuration item appears in a highly concentrated position and the difference changes little, the value of this component will be small; if the call positions are irregularly distributed in different areas of the sequence, resulting in large interval differences, the value will increase significantly.

[0108] Joint calculation method

[0109] Finally, the mean square error of the frequency dimension and the mean square error of the sequence dimension are averaged over the number of configuration items, and then weighted or directly summed to form a single stability vector value, which serves as an indicator to determine whether the current configuration item sequence has stable usage characteristics.

[0110] Numerical interpretation and application

[0111] The smaller the vector value, the more stable the configuration item call behavior. Conversely, a larger vector value indicates abnormal behavior such as excessive frequency deviation or confusion in call locations. This indicator can be used to screen configuration combinations and identify anomalies. This indicator has a clear numerical comparison meaning and is suitable for automated stability assessment and screening operations under unified rules.

[0112] The stability screening submodule calls the preset configuration call stability range threshold based on the call stability vector, compares the configuration item combinations, and traverses and screens the combination results under all job roles to obtain the configuration combinations within the threshold range and establish a sustainable call configuration combination;

[0113] The stability screening submodule compares the calculated call stability vector S value with the preset stability threshold to screen out stable configuration combinations. The threshold setting process can refer to the S value distribution of the historical sequence and set the interval by adding or subtracting the standard deviation. For example, if the mean S value in the historical sequence is 6.2 and the standard deviation is 0.9, the reasonable screening interval can be set as

[0114] ;

[0115] Right now ;

[0116] The judgment criteria are: if the calculated S value falls within the interval, the current configuration combination is considered to meet the stable call characteristics, otherwise it is marked as an unstable combination and excluded. The S value obtained in the current paragraph 2 is 6.514, which is within the interval Among them, it is determined to be a stable combination and enters the subsequent configuration template screening result set. In the batch screening process, the system will perform a fixed-step sliding window grouping process on the call sequence of each position role. For example, a combination with a step length of 3 is grouped in the sequence.

[0117] ;

[0118] Generate a combined sequence such as:

[0119] ;

[0120] ;

[0121] ;

[0122] Execute the frequency-sequence joint calculation process described in paragraph 2 again for these combinations respectively, obtain the S value corresponding to each group, compare it with the set interval, and determine one by one which combinations meet the conditions and which should be discarded. After all combinations are analyzed, output a list of mapping relationships between job roles and their stable configuration combinations for subsequent calling modules to match and recommend configuration sets.

[0123] See also Figure 3 , the responsibility intersection node screening module includes:

[0124] The task material extraction submodule obtains the job task history and the corresponding material category identifier based on a sustainable call configuration combination, extracts the department number corresponding to the material category and job, and collects the downstream related job data of the job in the task history to build a job-material-department mapping matrix;

[0125] The task material extraction submodule obtains the job task history and the corresponding material category identification based on the sustainable call configuration combination. In the execution process, the system first loads the screened stable configuration combination index, and combines it with the job role mapping table to extract the task number sequence bound to each job. For example, job P001 is bound to 5 tasks T001 to T005. The system extracts the "material category identification" field from the task record one by one. Assuming that they are M001, M001, M002, M001, and M002 respectively, the material category mapping result corresponding to job P001 is M001×3 and M002×2. Continuing to process job P002, the corresponding materials in its task history T006 to T010 are M001, M003, M001, M001, and M003. Then, the corresponding material category mapping result for P002 is M001×3 and M003×2. After accumulating multiple jobs, the initial job-material category binary mapping matrix is ​​formed. Then the system queries the job information table and finds that the department to which P001 belongs is D01. P002 is D02, P003 is D01, and so on. This constructs a position-department mapping vector. These two dimensions are then merged into a "position-material-department" ternary combination, such as P001-M001-D01 and P002-M001-D02. After obtaining this basic mapping structure, the system loads the position task flow table and extracts the "downstream associated position field" for each position. This field may have a multi-value structure. For example, if P001's downstream positions are P004 and P005, these fields are added to the position successor table. The downstream positions are then traversed and extracted by position, ultimately forming a "position-material-department-downstream position" quaternary combination structure. Multiple records are automatically categorized, deduplicated, and counted to form a position-material-department mapping matrix. Its cell values ​​can represent, for example, that position P001 processes material M001, which belongs to department D01 and has two downstream positions, P004 and P005. After this process is completed, the system possesses complete basic data for dynamic mappings between material and position organizational structures.

[0126] The department distribution calculation submodule calls the position distribution information and department number data in the position-material-department mapping matrix, counts the distribution of task histories of material categories in their respective departments, and normalizes the proportion of task histories with the same department number in the material category using the formula:

[0127] ;

[0128] Calculate and obtain the position distribution concentration value corresponding to the material category, call the position coverage number under the department number and compare it with the task coverage ratio to obtain the cross-position distribution concentration trend vector;

[0129] in, Indicates the The job distribution concentration value of this type of material, For the Classification of materials The number of task records of each department, For the The total number of task histories for this type of material, For the The total number of tasks for each department, is the total number of tasks for all departments, For the Classification of materials The number of positions in each department, For the The total number of departments associated with this type of material;

[0130] The department distribution calculation submodule counts the distribution of task histories of each type of material in each department based on the mapping matrix obtained in the previous section, and calculates the job distribution concentration value according to the following formula:

[0131] ;

[0132] Now let's take material category M002 as an example to perform the complete calculation process:

[0133] Assume that the material involves two departments: D01 and D03, that is ;

[0134] The number of task records of M002 in D01 is 30, recorded as , which is 70 in D03, recorded as , the sum of the two is ;

[0135] The total number of department tasks for D01 is 200, and for D03 is 400. , the corresponding department tasks account for , ;

[0136] The number of positions for processing M002 in D01 is 12, and in D03 is 6, which are recorded as , ;

[0137] Substitute in the terms and perform the operation:

[0138] The first correction term is calculated as follows:

[0139] ;

[0140] Absolute value of difference: ;

[0141] Weight modifier: ;

[0142] Item value: ;

[0143] The second correction term is calculated as follows:

[0144] ;

[0145] Absolute value of difference: ;

[0146] Weight modifier: ;

[0147] Item value: ;

[0148] Substituting the two terms into the overall structure of the formula, we get:

[0149] ;

[0150] The final job distribution concentration value of material M002 is 0.0976 (rounded to four decimal places). This result shows that there is a certain deviation between the proportion of M002's task history in the two departments and the proportion of departmental tasks, and the degree of deviation amplification is determined by the number of positions in the department that handle the material. If a department has more positions, the impact of the difference will be magnified, reflecting the relative concentration or dispersion of resources allocated in the department. The innovation of the formula lies in introducing the root weight correction of the number of positions into the proportional error value, which can avoid abnormal judgments caused by a single proportion offset and improve the rationality of judgment.

[0151] Description of the dimension normalization method of each parameter in the formula

[0152] In the process of calculating the concentration value of job distribution, parameters with different units and data types must be dimensionalized or normalized to ensure logical rationality and comparability when participating in weighting, product, difference and other calculations. The specific processing methods are as follows:

[0153] parameter :Indicates the Classification of materials The number of task records of each department, the original unit is "item", which is a pure count value. Therefore, the ratio normalization is adopted, which is , after normalization, the range is limited to Convert to dimensionless ratio value.

[0154] parameter :Indicates the Total task history of this type of material, unit and The same is a "bar", and its role is only to serve as the denominator, to achieve normalization in the ratio, and it no longer participates in independent calculations.

[0155] parameter :Indicates the The total number of all tasks of a department, the unit is also "item", which is the total number of tasks of the department. It also adopts the normalization processing method and is compared with the total number of tasks of all departments. The ratio is , and its normalized result is limited to interval, eliminating the impact of differences in department size.

[0156] parameter : Represents the total number of task histories of all departments. As a normalized denominator, it does not participate in absolute calculations and is only used to unify the relative proportions of task volumes of each department.

[0157] parameter :Indicates the Classification of materials The number of processing positions in a department, the original unit is "piece", this item participates in the square root operation in the formula , essentially constitutes a weight adjustment term to amplify or reduce the impact of the proportional difference, but does not participate in the ratio calculation, so no normalization is required.

[0158] parameter :For the The number of departments associated with the material type is a positive integer. It is used as the divisor of the final total average and directly participates in the arithmetic average operation without involving physical quantity dimensions.

[0159] In summary, all ratio items are normalized by comparing them with the total number. The number of positions is used as an adjustment factor to retain its original magnitude and participate in weight adjustment, ensuring that the final calculated value has consistent dimensional logic. The result is a purely dimensionless concentration value.

[0160] Job distribution concentration value Definition of

[0161] Job distribution concentration value This comprehensive metric measures the balance of a particular material's tasks across departments. Essentially, it weights the difference between the proportion of a material's task history in each department and the department's total task load. By comparing the history ratio and task load across departments, it examines the consistency of resource allocation and task execution across organizational units.

[0162] More specifically, if a material is widely distributed across multiple departments, but its task history is highly concentrated, or its task proportion is significantly mismatched with the department's overall task load, then its position distribution concentration value will be high. Conversely, if its distribution is more balanced and the proportion of history is consistent with departmental capacity, then the value will be low. This indicator reflects whether there is a risk of "resource accumulation" or "position mismatch" in material tasks and is an important quantitative indicator of the coordination of departmental and position resource structures.

[0163] The calculation principle of job distribution concentration value:

[0164] The calculation of job distribution concentration value follows the following logic:

[0165] Local ratio comparison logic:

[0166] First, for each involved department, calculate two ratios: one is the department's share of the target material's history, i.e., the ratio of the number of tasks for that material within the department to the total number of tasks for that material; the other is the department's share of the total number of tasks in the entire task system, i.e., the ratio of the department's overall tasks to the total number of tasks in the entire system. The smaller the difference between these two ratios, the more consistent the department's allocation of tasks for that material is with its task capacity.

[0167] Absolute difference determination:

[0168] The absolute value of the difference between the above two ratios is taken to eliminate the positive and negative biases, and only the degree of deviation is retained to reflect the degree of deviation of the material history delivery.

[0169] Correction of weight of square root of number of posts:

[0170] By taking the square root of the number of positions in the department that handle the material task and multiplying it into the difference term as a weight factor, if a department has more processing positions, its amplification weight for the deviation value will be greater, that is, the more resources it inputs, the greater the error it bears. This mechanism strengthens the deviation impact of large-scale departments and weakens the disturbance of small-scale departments on the results.

[0171] Average integration across sectors:

[0172] Sum the corrected differences across all departments and divide by the number of departments involved to arrive at a centralized offset value that is homogenized at the department level. This provides a unified measure of the overall offset across departments, eliminating the issue of offset dilution caused by the number of related departments.

[0173] The final result This value is a dimensionless number representing the degree of centralization and deviation in the distribution of material tasks across departments. Smaller values ​​indicate a more balanced and coordinated distribution, while larger values ​​indicate a tendency toward centralization or inconsistency. This metric can be used as a standard for identifying cross-job tasks, configuring early warning mechanisms, and implementing material load balancing strategies in a system.

[0174] The intersection node screening submodule detects repeated task history paths of material categories across multiple departments based on the cross-post distribution central tendency vector, screens the set of post numbers that simultaneously have cross-department call trajectories, extracts task nodes that are repeatedly called in downstream related posts, and establishes inheritable cross-department call nodes.

[0175] After completing the calculation of the distribution central tendency vector, the intersection node screening submodule compares all materials with the set distribution reference value according to their corresponding Zq value. For example, if the current screening threshold is set to 0.08, the corresponding value of M002 calculated in paragraph 2 is 0.0976, which is determined to be "cross-post distribution material". The system then enters the intersection path determination process. In this process, the system first locks the set of all task numbers associated with M002, assuming that it includes tasks such as T105, T107, T108, and T110, and then tracks them according to the task-post execution mapping. For example, T105 is executed at post P011 of department D01, and T110 is executed at post P018 of D03. Since these two posts belong to different departments and perform the same task M002, the task is determined to be a "cross-department task history path". The system then screens these tasks to see if there are repeated trigger records in multiple positions. For example, after T105 is called for the first time by P011, it is processed twice by P012 and P017 and recorded in the resume table. In this case, task T105 is judged as a "repeated task node". The system extracts all such task node sets and includes all the position numbers connected to them as a "repeated call position number set". For example, P011, P012, and P017 constitute a set. All positions in this set are recorded as position numbers that can carry cross-departmental task nodes. The mapping between these task nodes and positions is then used to construct an "cross-departmental task inheritance node" index set, which is subsequently used in position configuration optimization and task path inheritance analysis scenarios, ultimately forming a three-layer structure index of "material category-cross-departmental task node-repeated position number".

[0176] See also Figure 4 , the path carrying range determination module includes:

[0177] The main path information extraction submodule counts the total amount of call data in each path based on the main path call data and corresponding module records obtained from the inheritable cross-department call node, and extracts the module number and quantity included in each path to establish the main path call module matrix;

[0178] The main path information extraction submodule, based on the acquired inheritable cross-department call nodes, first identifies all data chain structures that can be used to construct the main path and extracts the call data involved in these paths from the task flow. Each path is treated as a continuous task sequence. During the path resolution process, the system extracts the call logs corresponding to each node in the path order and counts the total number of calls. For example, path P01 consists of nodes N1 to N5, and the call counts for each node are 180, 220, 210, 190, and 200, respectively, resulting in a total of 1000 calls. Path P02 has a total of 800 calls, and path P03 has a total of 1500. Simultaneously, the system associates the call log records with the module mapping table, extracts the module numbers involved in each node call, and performs duplicate counting. For example, path P01 calls modules M02 to M05, totaling five modules; path P02 calls modules M02 to M05, totaling four modules; and path P03 involves modules M02 to M06, totaling six modules. In this way, the system creates a record for each path, including the path number, the total amount of call data, the module numbers involved, and the number of modules involved. This record is then constructed into a primary path call module matrix, where each path is represented as a row, including the primary key path number and its structural statistics fields. This matrix is ​​used for subsequent module evaluation and path load analysis.

[0179] The load value generation submodule calls the total amount of call data and the number of modules in the main path call module matrix, performs a product operation on the data volume and the number of modules of each main path, and normalizes the call frequency offset and structural overlap between path nodes to obtain the evaluation result of each path and obtain the path load ratio vector;

[0180] The load-bearing value generation submodule receives the main path's module call matrix as input and calculates a basic load-bearing value for each path based on its data volume and number of modules. Specifically, this calculation multiplies the path's total call data by the number of modules involved to determine the path's total load-bearing value under static structural conditions. For example, path P01 has a total call volume of 1000 and a module count of 5, resulting in a load-bearing value of 5000; path P02 has a total call volume of 800 and a module count of 4, resulting in a load-bearing value of 3200; and path P03 has a load-bearing value of 1500 multiplied by 6, which equals 9000. Subsequently, the system introduces two path characteristic factors to dynamically correct the load value. The first is the call frequency offset factor, which calculates the relative deviation between the maximum and minimum call times in the path and adjusts the load value based on the offset value. For example, if the maximum call of the node on path P01 is 220 times, the minimum call is 180 times, and the average is 200 times, the offset ratio is 20%, which is mapped to a frequency correction factor of 1.2 in the system. The second is the structural overlap factor, which is used to correct the degree of duplication of each module in the path in other paths. For example, two of the five modules in path P01 also appear in other paths, with an overlap ratio of 40%. The system maps this to a structural correction coefficient of 0.8. The two correction factors are multiplied together to normalize the load value. The final load result is 5000 multiplied by 0.96, which is 4800. The final load value for P02 is 3168, and for P03 is 17550, which constitutes the final path load ratio vector. This processing flow introduces real path characteristic parameters to avoid the deviation of ignoring structural complexity and call volatility by simply multiplying the call volume by the number of modules.

[0181] The range status screening submodule compares the path load ratio vector with the set load determination reference range value, filters the path numbers within the upper and lower limits, and establishes the main path load range based on the result set composed of structural characteristics, module coverage structure and call density;

[0182] The range status screening submodule screens each path in the path load ratio vector. The system first loads a set load ratio reference range. This range is set based on statistical analysis of historical path operation data. The lower and upper limits are typically calculated from the mean and standard deviation of the historical path load ratios. The lower limit is equal to the mean minus one to two standard deviations, which is used to exclude underloaded paths; the upper limit is equal to the mean plus one to two standard deviations, which is used to identify overloaded paths. For example, if the historical path load ratio is 8000 and the standard deviation is 4100, the screening range is set to 3080 to 12920. The system compares each path's load ratio against this range. For example, the load ratio of path P01 is 4800, and that of path P02 is 3168, both within the range. Path P03 has a value of 17550, which exceeds the upper limit of the range and is marked as non-compliant. The system then calls the structure verification module to analyze path structure integrity, module coverage, and call density. Path P01 involves five modules and includes two core modules defined by the system. Each module has an average call count of 200, indicating reasonable structural coverage and balanced call distribution. Path P02 involves four modules but includes key module numbers, with an average call count of 200, also considered balanced. Path P03, while covering more modules, was excluded due to excessive load capacity. The system ultimately marks paths P01 and P02 as loadable primary paths and records their path numbers in the loadable path set for subsequent task scheduling and execution path generation.

[0183] See also Figure 5 , the reverse path verification module includes:

[0184] The exclusion path extraction submodule obtains the path numbers that are not included in the path set based on the loadable range of the main path, collects the call data and output records corresponding to the path, marks the number of modules in the path, and establishes an exclusion path call record set;

[0185] The exclusion path extraction submodule obtains the path numbers that are not included in the path set based on the main path's carrying range. The system first obtains the full set of all path numbers from the path master table, and calls the main path carrying list to exclude all selected path numbers. The remaining path numbers are the candidate exclusion path number set. For example, if the full set of paths includes numbers P01 to P08, and the main path number set is P01, P02, and P03, then the exclusion path set is P04, P05, P06, P07, and P08. The system processes these exclusion path numbers one by one and performs a call data extraction operation on each path. The specific process is to call the call log table and filter all associated call records by path number. The number of calls is summed up. For example, the call data volume of P04 is 600, that of P05 is 750, and that of P06 is 500. The system then parses the module structure information of each path from the path structure table, de-duplicates the module number set covered by it, and counts the number of modules. The number of modules covered by P04 is 3, P05 is 5, and P06 is 4. The number of output records corresponding to each path is then extracted through the path execution log. For example, there are 20 output records for P04, 18 for P05, and 12 for P06. Finally, the system constructs an exclusion path call record set with the path number as the index, and records the total call data, the number of modules, and the number of outputs as fields. This record set is then used in the subsequent composite value evaluation operation stage.

[0186] The composite value combination submodule calls the total amount of call data and the number of modules in the exclusion path call record set, performs superposition processing on the output number of the corresponding path, and aggregates the product of the call data and the number of modules with the output record value according to the path number to obtain the exclusion path composite evaluation value group;

[0187] The composite value combines the three basic fields in the path call record set to exclude sub-module calls: the total amount of call data, the number of modules, and the number of output records. The system first calculates the product of the call data and the number of modules for each path, which is used as the basic load value evaluation of the path. For example, the call volume of path P04 is 600, the number of modules is 3, and the basic load value is 1800. The path P05 is 750 multiplied by 5, which equals 3750, and the path P06 is 500 multiplied by 4, which equals 2000. The system then introduces the number of path output records as an extended load coefficient, which is directly added to the basic load value to represent the comprehensive operation intensity of the path. For example, the composite value of P04 is 1800 plus 20, which equals 1820, P05 is 3750 plus 18, which equals 3768, and P06 is 2000 plus 12, which equals 2012. This process is a standard multiplication and addition operation. Path dimensions do not cross or merge. The calculation logic of each path is completely independent. Ultimately, all calculation results are aggregated by path number to form a composite value evaluation record set. This set is used for subsequent comparison and screening with the load-bearing conditions of the main path. The evaluation value reflects the comprehensive resource consumption level of the path in terms of data processing, structural load-bearing, and output efficiency. The higher the path composite value, the greater the execution intensity corresponding to the path. The closer it is to or exceeds the main path mean threshold, the more likely the path is to replace the main path.

[0188] The backup path screening submodule compares the judgment conditions in the loadable range of the primary path based on the composite evaluation value group of the excluded paths, compares the evaluation values ​​corresponding to the paths to see if they fall within the acceptable range, screens the path numbers whose evaluation values ​​meet the conditions, and establishes an activatable backup path set;

[0189] After the backup path screening submodule obtains the composite evaluation value of the exclusion path, the system needs to set the screening interval and compare it according to the judgment conditions in the loadable range of the main path. The judgment condition comes from the composite value interval statistical parameters in the main path set. The system counts the composite value set of the main path at the initial operation and determines the upper and lower limits of the screening by calculating the median, standard deviation, maximum and minimum values ​​and other indicators. For example, the composite value samples of the main path are 4200, 4600, 5100, 5700, and 6100. The system calculates the median to be 5100 and the standard deviation to be about 700. The screening interval is set to the median plus or minus the standard deviation, that is, 4400 to 5800. The interval boundary is used as the acceptable value range. The system compares and judges the composite values ​​of the exclusion paths one by one. For example, the composite value of P04 is 1820. It is lower than the lower limit and is judged as not meeting the standard; P05 is 3768, which is still lower than the lower limit of the interval and is judged as underloaded; P06 is 2012 and is also excluded. If there is another path P07 with a composite value of 5400, it is between 4400 and 5800, and the system marks it as a compliant backup path. Then, the system confirms its module number, key module coverage, and path node connection integrity based on the P07 path structure information. For example, if its number of modules is 5, including the system key modules M03 and M04, and there are stable call chains between all nodes, with an average call frequency of more than 150 times per module, the system determines that its structure is complete and the density is balanced, and adds the P07 path number to the set of activatable backup paths to form a backup path index table for scheduling modules to call when the primary path is unavailable.

[0190] See also Figure 6 , the cold standby path loading module includes:

[0191] The target data acquisition submodule obtains the call target data corresponding to each cold standby path based on the set of activatable backup paths, extracts the call number and path identification information of the target item, and simultaneously obtains the call target data of the paths within the loadable range of the primary path, and establishes a cold standby primary path target comparison set;

[0192] The target data acquisition submodule obtains the call target data corresponding to each cold standby path based on the activatable backup path set. The system first filters out the path number list marked as "backup" or "cold standby" from the path structure table, and reads each path number in the list one by one. The system calls the call log database and extracts all target item call records corresponding to the path number. These target item records usually contain fields such as call number, target item identification and call timestamp. The system retains the call number and target item identification fields as the core content, and at the same time associates the structure table of the path to obtain the path identification information, forming a multi-field mapping record of "path number-target item number-call number". The system then switches to the main path loadable set and executes exactly the same extraction process for each main path to obtain the target item call data and path identification information of the main path, and also retains the target item number and call number fields. Finally, the system matches the cold backup path target data set with the main path target data set according to the path number dimension, unifies the standardized field names and field formats, establishes a unified field naming rule for all target item records, and constructs a target comparison set with the structure of "path number, path type, target item number, call number". The path type field is used to distinguish whether the record source is cold backup or main path. The system stores the target comparison set in the internal comparison cache area for the call structure comparison task in the subsequent path fit determination stage.

[0193] The path comparison and screening submodule calls the cold standby path target data and the main path target data in the cold standby main path target comparison set, compares the matching number and combination order of the same type of target items in turn, calculates the same type ratio for all comparison results, and screens the cold standby paths whose ratios meet the fit threshold to generate a same type path comparison result set;

[0194] The path comparison and screening submodule calls the cold standby path target data and the main path target data in the cold standby main path target comparison set to match all path combinations. During the comparison process, the system uses the target item number as the core comparison element and the path number as the matching unit. It performs a one-to-one comparison operation of "cold standby path to main path". In each round of comparison, the system first counts the number of identical target items between the cold standby path and the target main path. For example, the target items of the cold standby path B01 are T101, T102, and T105. The main path M01 consists of T101, T102, and T104. Their common targets are T101 and T102, resulting in two matches. The cold backup path has three targets, so the match ratio is 2 divided by 3. The resulting match ratio represents the degree of overlap in the target item content of the paths. The system then determines whether the call order of the target items in the two paths is consistent, comparing whether the order in which the target items are executed in the paths is exactly the same. If they are completely consistent, the sequence fit is marked as 1; if they are completely misaligned, it is marked as 0; and if they are only partially misaligned, it is set to 0.5. The system sets a 40% weight for sequence fit and a 60% weight for target item overlap. This weighted calculation generates the overall fit ratio of the cold backup path to the main path. After the fitting ratios of all paths are completed, the system generates a set of fitting value vectors according to the path combination, and then sets the fitting degree screening threshold. The threshold is derived from the historical path comparison data. The calculation method is to take the average of all fitting values ​​in the past actual path replacement comparison, and then float a stability factor upward in combination with the deviation range as the judgment upper limit. For example, if the historical average ratio is 0.68, the system sets the stable floating factor to 0.05, and the screening threshold is set to 0.73. The system retains the path combination with a fitting value greater than or equal to 0.73 as a qualified matching combination, and adds the cold standby path number to the fitting result set.

[0195] The solution output combination submodule extracts the path number, target item structure, and call sequence from the path combinations marked as matching in the result set of similar path comparison, constructs a path logical structure map that supports intelligent loading, and organizes and merges the structural information and path attributes to establish an intelligent decision support solution;

[0196] The solution output combination submodule reads the information of each combination item in sequence based on the path combinations marked as matching in the result set of similar path comparison. The system first extracts the cold standby path number and the corresponding target item structure in the combination, that is, the list of target item numbers and their arrangement order. For example, the cold standby path B02 and the main path M02 are judged to be a qualified combination. Their target items are both T103 and T108, and the order is exactly the same. The system records their path number as the combination key value and generates the path edge structure with T103→T108 as the sequence field. Then the system loads the structure template and maps the logical relationship of the path into the node information of the structure graph. Each target item number is set as a node, and directed edges are formed between adjacent target items. The path number is used as the upper-level node to anchor the path attributes, such as path type, source status, and whether to replace the main path. After uniformly organizing the logical structure diagram for each combined path, the system adds attribute fields to all path nodes, such as metadata fields such as target item call frequency, module coverage number, and number of path modules. All structure diagrams are formatted and encapsulated, ultimately forming a graph record set consisting of "path number - node structure - call sequence - path attributes." All qualified path structure diagrams are uniformly output and stored in the intelligent path support system, enabling path call decision combinations based on the identification of similar backup and primary path structures.

[0197] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. Intelligent decision support system based on ERP data, characterized by: The system comprises: The configuration factor stability determination module obtains configuration call logs, groups them by job role, extracts call frequency and sequence, takes the square mean of the difference between frequency vectors and the mean square of the position spacing between sequence vectors, generates a stability determination vector, compares it with the preset range, and selects sustainable call configuration combinations; The responsibility intersection node screening module obtains job task history and material category based on the sustainable call configuration combination, extracts material category, department number and downstream data, calculates department concentration, screens cross-department responsibility intersections, and generates inheritable cross-department call nodes; The path carrying range determination module obtains the main path calling data and module records based on the inheritable cross-department calling nodes, calculates the product of the total calling data and the number of modules, and screens the main path carrying range; The reverse path verification module obtains the excluded paths and outputs based on the loadable range of the primary path, calculates a combined value, compares it with the primary path range, and generates an activatable backup path set; The cold standby path loading module obtains the cold standby path target data based on the activatable standby path set, compares the target data of the main path one by one, screens similar combinations, and generates an intelligent decision support solution; The sustainable call configuration combination includes call frequency vector characteristics, call sequence vector characteristics, stable judgment vector, matching screening results, and configuration item combination set; the inheritable cross-department call node includes job task history information, material category identification code, associated department number, department concentration index, and responsibility intersection node set; the main path carrying range includes the total value of call data, module quantity index, path carrying capacity value, and reference comparison range; the activatable backup path set includes excluded path records, backup path output data, path judgment result value, and activated path set; the intelligent decision support solution includes cold standby path target parameters, main path target parameters, path similarity ratio, and recommended combination solution.

2. The intelligent decision support system based on ERP data according to claim 1, characterized in that: The definition of the preset range is used to determine the reference interval for configuration call stability, including the expected value range of the frequency difference and the sequence spacing; The main path's load capacity is defined as the upper limit of the call capacity that the main path can stably handle under the current module distribution and total number of calls; The definition of the same type combination is a cold standby path combination that has a high degree of fit with the main path target data in terms of task structure and material type.

3. The intelligent decision support system based on ERP data according to claim 1 is characterized in that: The configuration factor stability determination module includes: The call log collection submodule obtains configuration call log data, extracts the job role information, configuration item identifiers, and corresponding timestamp sequences from the call records, classifies the call logs by job role, and generates job role call sequence groups. The frequency sequence analysis submodule calls the configuration item sequence under the position in the position role call sequence group, calculates the call frequency for each configuration item, obtains a frequency vector, extracts the square of the average position difference of the call position of the configuration item in the sequence in turn, forms a sequence vector, and calculates the weighted combination value of the frequency deviation mean and the sequence position mean square error to obtain the call stability vector; The stability screening submodule calls the preset configuration call stability range threshold based on the call stability vector, compares the configuration item combinations, traverses and screens the combination results under all job roles, obtains the configuration combinations within the threshold range, and establishes a sustainable call configuration combination; The definition of the configuration combination within the threshold range is obtained, and the configuration item combination set with stable call characteristics is screened under the condition that the call stability vector meets the preset threshold.

4. The intelligent decision support system based on ERP data according to claim 3 is characterized in that: The responsibility intersection node screening module includes: The task material extraction submodule obtains the job task history and the corresponding material category identifier based on the sustainable call configuration combination, extracts the department number corresponding to the material category and the job, and collects the downstream related job data of the job in the task history to build a job material department mapping matrix; The department distribution calculation submodule calls the position distribution information and department number data in the position-material-department mapping matrix, performs quantitative statistics on the distribution of task histories of material categories in their respective departments, normalizes the proportion of task histories in material categories with the same department number, calculates and obtains the position distribution concentration value corresponding to the material category, calls the position coverage number under the department number and compares it with the task coverage ratio to obtain the cross-position distribution concentration trend vector; The intersection node screening submodule detects the repeated task history paths of material categories across multiple departments based on the cross-post distribution central trend vector, screens the set of post numbers that have cross-department call trajectories at the same time, extracts the task nodes that are repeatedly called in downstream related posts, and establishes inheritable cross-department call nodes.

5. The intelligent decision support system based on ERP data according to claim 4 is characterized in that: The path carrying range determination module includes: The main path information extraction submodule counts the total amount of call data in each path based on the main path call data and corresponding module records obtained by the inheritable cross-department call node, and extracts the module number and quantity included in each path to establish a main path call module matrix; The load value generation submodule calls the total amount of call data and the number of modules in the main path call module matrix, performs a product operation on the data volume and the number of modules of each main path, and normalizes the call frequency offset and structural overlap between path nodes to obtain an evaluation result for each path and obtain a path load ratio vector; The range status screening submodule compares the set load determination reference range value according to the path load ratio vector, screens the path numbers within the upper and lower limit intervals, and establishes the main path load range based on the result set composed of structural characteristics, module coverage structure and call density.

6. The intelligent decision support system based on ERP data according to claim 5, characterized in that: The reverse path verification module includes: The exclusion path extraction submodule obtains the path number that is not included in the path set based on the loadable range of the main path, collects the call data and output records corresponding to the path, marks the number of modules in the path, and establishes an exclusion path call record set; The composite value combination submodule calls the total amount of call data and the number of modules in the exclusion path call record set, performs superposition processing on the output number of the corresponding path, and aggregates the product of the call data and the number of modules with the output record value according to the path number to obtain the exclusion path composite evaluation value group; The backup path screening submodule compares the judgment conditions in the loadable range of the primary path based on the composite evaluation value group of the excluded paths, compares the evaluation values ​​corresponding to the paths to see whether they fall within the acceptable range, screens the path numbers whose evaluation values ​​meet the conditions, and establishes an activatable backup path set; The definition of the acceptable range is the standard interval for measuring whether the composite evaluation value of the backup path has the potential to replace the main path.

7. The intelligent decision support system based on ERP data according to claim 6, characterized in that: The cold standby path loading module includes: The target data acquisition submodule obtains the call target data corresponding to each cold standby path based on the activatable standby path set, extracts the call number and path identification information of the target item, and simultaneously obtains the call target data of the paths within the loadable range of the primary path, and establishes a cold standby primary path target comparison set; The path comparison and screening submodule calls the cold standby path target data and the main path target data in the cold standby main path target comparison set, compares the matching number and combination order of the same type of target items in turn, calculates the same type ratio for all comparison results, and screens the cold standby paths whose ratios meet the fitting threshold, thereby generating a same type path comparison result set; The solution output combination submodule extracts the path number, target item structure and calling sequence from the path combinations marked as matching in the similar path comparison result set, constructs a path logical structure map that supports intelligent loading, and organizes and merges the structural information and path attributes to establish an intelligent decision support solution; The definition of the fitting threshold is used to measure whether the cold backup path and the primary path are similar enough in terms of target item type and sequence to match the comparison criteria; The definition of the path logic structure map is based on the call sequence and target structure mapping constructed by matching path combinations, supporting intelligent path loading and decision making.

Images